volume2 issue2 december 2009 final

Contributing AuthorsOliver Tiemann, Ludwig-Maximilians-Universität München, GermanyJonas Schreyögg, Ludwig-Maximilians-Universität München, Germany Alexandra Maßbaum, University of Paderborn, Germany Caren Sureth, University of Paderborn, GermanyLutz Kruschwitz, Freie Universität Berlin, GermanyAndreas Löffler, University of Paderborn, GermanyUlf Schrader, Technical University of Berlin, GermanyThorsten Hennig-Thurau, Bauhaus-University of Weimar, GermanyAlf Kimms, University of Duisburg-Essen, GermanyJulia Drechsel, University of Duisburg-Essen, GermanyChristiane Prange, Lyon Business School, FranceBodo B. Schlegelmilch, WU Wien, Austria, and Leeds University Business School, UK

Vo l u m e 2 | I s su e 2 | D e ce m b e r 20 0 9 | w w w. b u s i n e s s - r e s e a r c h . o r g | I SSN 18 6 6 - 8 658

Accounting | Finance | Management | Marketing | Operations and Information Systems

Jan Bouwens Tilburg University, The Netherlands

Jörg Budde Rheinische Friedrich-Wilhelms-Universität Bonn, Germany

Willem Buijink Tilburg University, The Netherlands

Joachim Gassen Humboldt-Universität zu Berlin, Germany

Martin Glaum Justus-Liebig-Universität Gießen, Germany

Christian Hofmann Universität Mannheim, Germany

Jochen Hundsdoerfer Freie Universität Berlin, Germany

Vesa Kanniainen University of Helsinki, Finland

Laurence van Lent Tilburg University, The Netherlands

Frank Moers Maastricht University, The Netherlands

Lasse Niemi Helsinki School of Economics, Finland

Paolo Panteghini Università degli Studi di Brescia, Italy

Thomas Pfeiffer University of Vienna, Austria

Stefan Reichelstein Stanford University, USA

Christian Riegler WU Vienna, Austria

Ulf Schiller Universität Bern, Switzerland

Dirk Simons Universität Mannheim, Germany

Peter Birch Sørensen University of Copenhagen, Denmark

Ulrike Stefani Universität Konstanz, Germany

Hervé Stolowy HEC Paris, France

Caren Sureth Universität Paderborn, Germany

Stefan Wielenberg Leibniz Universität Hannover, Germany

Hannes Winner Universität Salzburg, Austria

Accounting

Rainer Niemann (Graz)[email protected]

Finance

Christian Schlag (Frankfurt am Main)[email protected]

Management

Peter Walgenbach (Jena)[email protected]

Sönke Albers (Kiel)[email protected]

Franklin Allen Wharton School, Philadelphia, USA

Laura Ballotta Cass Business School, City University, London, UK

Nicole Branger Westfälische Wilhelms-Universität Münster, Germany

Engelbert Dockner WU Vienna, Austria

Ralf Elsas LMU Munich, Germany

Michael F. Gallmeyer Texas A&M University, USA

Robert Gillenkirch Universität Osnabrück, Germany

Joachim Grammig Eberhard-Karls-Universität Tübingen, Germany

Bruce Grundy University of Melbourne, Australia

Dirk Hackbarth University of Illinois at Urbana-Champaign, USA

Holger Kraft Goethe-Universität Frankfurt am Main, Germany

Andreas Löffler Universität Paderborn, Germany

Claus Munk University of Southern Denmark, Odense, Denmark

Alessandro Sbuelz Università degli Studi di Verona, Italy

Carsten Sørensen Copenhagen Business School, Denmark

Hans Stoll Vanderbilt University, Nashville, USA

Marketing

Adamantios Diamantopoulos (Vienna) [email protected]

Operations and Information Systems

Karl Inderfurth (Magdeburg)[email protected]

Rainer Niemann (Graz)[email protected]

Christian Schlag (Frankfurt am Main)[email protected]

Editor-in-Chief

Finance

Department Editors

Accounting

107

George Balabanis Cass Business School, City University, London, UK

Hans Baumgartner Pennsylvania State University, USA

Suzanne Beckmann Copenhagen Business School, Denmark

Daniel Bello Georgia State University, USA

Albert Bemmaor ESSEC Business School, France

Rod Brodie University of Auckland, New Zealand

John Cadogan Loughborogh University, UK

Tamer Cavusgil Georgia State University, USA

René Darmon ESSEC Business School, France

Timothy Devinney Australian Graduate School of Management, Australia

Susan Douglas New York University, USA

Lutz Hildebrandt Humboldt Universität zu Berlin, Germany

Roy Howell Texas Tech University, USA

Harald Hruschka Universität Regensburg, Germany

Tomas Hult Michigan State University, USA

Constantine Katsikeas University of Leeds, UK

Daniel Klapper Goethe-Universität Frankfurt am Main, Germany

Harley Krohmer Universität Bern, Switzerland

Peter Leeflang University of Groningen, The Netherlands

David Midgley INSEAD, France

Vince Mitchell Cass Business School, City University, London, UK

Neil Morgan Indiana University, USA

Saeed Samiee University of Tulsa, USA

Henrik Sattler Universität Hamburg, Germany

Bodo Schlegelmilch WU Vienna, Austria

Judy Siguaw Nanyang Technological University, Singapore

Jagdip Singh Case Western Reserve University, USA

Bernd Skiera Goethe-Universität Frankfurt am Main, Germany

Rajan Varadarajan Texas A&M University, USA

Udo Wagner University of Vienna, Austria

Berend Wierenga Erasmus University, The Netherlands

Russell Winer New York University, USA

Heidi Winklhofer University of Nottingham, UK

Adamantios Diamantopoulos (Vienna)[email protected]

Marketing

Andreas Al-Laham TU Kaiserslautern, Germany

Giuseppe Delmestri University of Bergamo, Italy

Martha S. Feldman UC Irvine, USA

Peer Fiss University of Southern California, USA

Mike Geppert University of Surrey, UK

Paul Gooderham The Norwegian School of Economics and

Business Administration, Norway

Christian Grund Julius-Maximilians-Universität Würzburg, Germany

Axel Haunschild University of London, UK

Eric von Hippel MIT Sloan School of Management, USA

Bernd Irlenbusch London School of Economics, UK

Dirk Matten York University, Schulich School of Business, Canada

Michael Mayer University of Bath, UK

Renate E. Meyer WU Vienna, Austria

Guido Möllering MPI Cologne, Germany

Philippe Monin EM Lyon, France

Denise Rousseau Carnegie Mellon University, USA

Kuno Schedler University of St. Gallen, Switzerland

Andreas Scherer University of Zurich, Switzerland

David Seidl LMU Munich, Germany

Udo Staber University of Canterbury, New Zealand

Roy Suddaby University of Alberta, Canada

Jörg Sydow Freie Universität Berlin, Germany

Janne Tienari Helsinki School of Economics, Finland

Anja Tuschke LMU Munich, Germany

Filippo Carlo Wezel University of Lugano, Switzerland

Michael Woywode RWTH Aachen, Germany

Peter Walgenbach (Jena)[email protected]

Management

Freimut Bodendorf Universität Erlangen-Nürnberg, Germany

Peter Chamoni Universität Duisburg-Essen, Germany

Charles Corbett University of California, USA

Werner Delfmann Universität zu Köln, Germany

Harald Dyckhoff RWTH Aachen, Germany

Bernhard Fleischmann Universität Augsburg, Germany

Martin Grunow Technical University of Denmark, Denmark

Daniel R. Guide, Jr. Pennsylvania State University, USA

Richard Hartl University of Vienna, Austria

Werner Jammernegg WU Wien, Austria

Peter Kelle Louisiana State University, USA

Alf Kimms Universität Duisburg-Essen, Germany

Herbert Kotzab Copenhagen Business School, Denmark

Heinrich Kuhn Universität Eichstätt-Ingolstadt, Germany

Rainer Lasch TU Dresden, Germany

Ronald Maier Universität Innsbruck, Austria

Dirk C. Mattfeld TU Braunschweig, Germany

Kaj Rosling Växjö University, Sweden

Detlef Schoder Universität zu Köln, Germany Thomas Spengler TU Braunschweig, Germany

Leena Suhl Universität Paderborn, Germany

Gerhard Wäscher Universität Magdeburg, Germany

Matthieu van der Heijden University of Twente, The Netherlands

Karl Inderfurth (Magdeburg)[email protected]

Operations and Information Systems

108

Effects of Ownership on Hospital Efficiency in Germany

Oliver Tiemann, Jonas Schreyögg ..................................................................................................................................................... 115

Thin Capitalization Rules and Entrepreneurial Capital Structure Decisions

Alexandra Maßbaum, Caren Sureth ..................................................................................................................................................147

Do Taxes Matter in the CAPM?

Lutz Kruschwitz, Andreas Löffler .......................................................................................................................................................171

VHB-JOURQUAL2: Method, Results, and Implications of the German Academic Association for

Business Research‘s Journal Ranking

Ulf Schrader, Thorsten Hennig-Thurau ............................................................................................................................................ 180

Cost Sharing under Uncertainty: An Algorithmic Approach to Cooperative Interval-Valued Games

Alf Kimms, Julia Drechsel ..................................................................................................................................................................206

The Role of Ambidexterity in Marketing Strategy Implementation: Resolving the Exploration-Exploitation Dilemma

Christiane Prange, Bodo B. Schlegelmilch ........................................................................................................................................215


109


Oliver TiemannMunich School of Management, Ludwig-Maximilians-Universität München

Jonas SchreyöggMunich School of Management, Ludwig-Maximilians-Universität München

Several studies have compared the efficiency of public, private

for-profit, and private non-profit hospitals. However, most of

these studies have data and methodological limitations. In ad-

dition, none of the studies has considered parameters for the

quality of care together with the number of cases as tradition-

al output parameter. In the German hospital sector different

ownership types have co-existed for decades, making this an

interesting field for investigating the effects of ownership on

efficiency. The objective of the present study was to evaluate

the relative efficiency of public, private for-profit, and private

non-profit hospitals in Germany. First, bootstrapped data en-

velopment analysis (DEA) was used to evaluate the efficiency

of a panel (n = 1,046) of public, private for-profit, and pri-

vate non-profit hospitals between 2002 and 2006. This was

followed by a second-step truncated linear regression model

with bootstrapped DEA efficiency scores as dependent vari-

able. In this model we explored the impact of hospital organi-

zational and environmental characteristics while controlling

for patient heterogeneity. The results of our analysis show

that public hospitals performed significantly better than their

private for-profit and non-profit counterparts. We also found

a significant positive association between hospital size and

efficiency, and that competitive pressure had a significant

negative impact on hospital efficiency. In addition, we found

a number of important interaction effects, between compe-

tition and ownership as well as hospital size and ownership.

In particular, we found that private for-profit hospitals are

especially inefficient in regions with strong competition and

private for-profit hospitals are more efficient among the very

large hospitals with more than 1,000 beds. Furthermore,

when quality of care was considered as an additional output

in our efficiency models, our findings indicate an inevitable

trade-off between efficiency and quality of care.

p 115

110

Do Taxes Matter in the CAPM?Lutz KruschwitzDepartment of Banking and Finance, Freie Universität Berlin

Andreas LöfflerDepartment of Finance and Investment, University of Paderborn

Authors contributing to the tax CAPM discussion usually

neglect the fact that the Treasury typically redistributes the

tax revenues that were collected from investors. Whereas this

may be acceptable for a partial model, it is certainly unaccept-

able for an equilibrium model. Surprisingly, this aspect has

been overlooked by a large number of authors. Based on a total

model that contains the redistribution we can show that equi-

librium security prices are not affected by taxation if either

the riskless rate vanishes or investors show constant absolute

risk aversion. Both results can be explained by economic rea-

soning.

However, neither vanishing interest rates nor CARA utility

functions are realistic cases. Since there seems to be no way

to extend our results beyond those two special cases, we con-

clude that taxes do have an impact on security prices in the

CAPM as a rule.

p 171

Thin Capitalization Rules and Entrepreneurial Capital Structure DecisionsAlexandra MaßbaumFaculty of Business Administration and Economics,University of Paderborn

Caren SurethFaculty of Business Administration and Economics,University of Paderborn

From a tax planner’s point of view, it is often attractive to

choose debt over equity financing. As this has led to an in-

crease in debt financing of corporations, many countries have

introduced thin capitalization rules to secure their tax rev-

enues. Thin capitalization rules are regulations that limit the

corporate tax deductibility of interest paid to shareholders. We

analyze the influence of these regulations on investors’ divi-

dend and financing decisions. We consider the Italian, Ger-

man and Belgian thin capitalization rules in 2007 as examples

of regulations with a given permitted debt-equity ratio. In

2008 both Italy and Germany replaced their thin capitaliza-

tion rules with an interest barrier. We find that full retention

is always the optimal dividend policy. Further, often no gen-

eral statement regarding the profitability of either financing

option can be made. The effects depend on tax and non-tax

parameters. In Belgium, equity capital is always beneficial

under the given tax rates and debt-equity ratio. For an Italian-

type tax system, the financing decision mainly depends on the

ratio of the internal after-tax yield and the aftertax market

rate of return. Assuming that these rates are almost equal,

debt financing will be beneficial for most investors. With a lo-

cal business tax and thin capitalization rules that distinguish

between long- and short-term debt (German type), the results

depend significantly on the debt’s time structure. If the frac-

tion of long-term debt is large, equity capital is likely to be fa-

vorable. An advantage of debt capital arises if the fraction of

long-term debt is small.

The analysis helps investors not only to make their dividend

and financing decisions under the respective tax systems but

also to adjust their decisions when tax laws change. Our analy-

sis is representative of all thin capitalization rules that refer to

a permitted debtequity ratio. As such regulations exist in sev-

eral countries, our results can be relatively easily transferred

to dividend and financing decisions in various countries to

support investor’s decisions and help treasuries to decide

what thin capitalization type to implement under their given

tax regime.

p 147Volume 2 | Issue 2 | December 2009 | www.business-research.org | ISSN 1866-8658

111

Cost Sharing under Uncertainty: An Algorithmic Approach to Cooperative Interval-Valued GamesAlf KimmsMercator School of Management,University of Duisburg-Essen

Julia DrechselMercator School of Management,University of Duisburg-Essen

In cooperative business settings like supply networks where

actors can form coalitions, the partners within a coalition

have to negotiate the distribution of the outcome shares in

advance. But this is a problem if the outcome is uncertain.

It seems that the best one can do is to negotiate rules with

which the outcome is shared. As an illustrative example we

use purchasing alliances where partners cooperate to reduce

procurement costs. Each partner has to fulfil his demand by

placing orders alone or several partners may cooperate to

place orders together, e.g. in order to save fixed costs. The

demand of each partner may be uncertain and so the total

demand of a coalition may be uncertain as well. Two simple

rules may be used for specifying a fair way of sharing ex post

procurement costs. The first rule states that the total procure-

ment costs are distributed among the partners who cooperate.

And the second rule makes sure that no (sub)group of partners

can perform better by leaving the grand coalition where all

actors are involved. Any allocation of costs that follows these

rules is named an element of the so-called core of this type

of cooperative game. If the uncertain demand of each partner

can be described by an interval of possible demand values, we

face a so-called interval valued game. In this case we can do a

bit more than just stipulating the two rules. We can compute

an interval of cost share values for each partner in advance

such that also in case of uncertain outcomes a cooperation

between the partners can be viewed as stable. The respec-

tive computation of these intervals is everything but trivial,

and the paper describes the technical details of doing this.

p 206

VHB-JOURQUAL2: Method, Results, and Implications of theGerman Academic Association for Business Research‘s JournalRankingUlf SchraderInstitute of Vocational Education and Work Studies,Technical University of Berlin

Thorsten Hennig-ThurauDepartment of Marketing and Media ResearchBauhaus-University of Weimar

VHB-JOURQUAL represents the official journal ranking of

the German Academic Association for Business Research.

Since its introduction in 2003, the ranking has become the

most influential journal evaluation approach in German-

speaking countries, impacting several key managerial de-

cisions of German, Austrian, and Swiss business schools.

This article describes the concept of VHB-JOURQUAL, the

method it uses to generate a quality index for an extensive

list of academic journals, and the results of the most recent

edition of VHB-JOUQUAL. VHB-JOURQUAL is based on an

individualized online-survey of members (i.e. professors and

researchers with Ph.D.) of the German Academic Associa-

tion for Business Research (Verband der Hochschullehrer für

Betriebswirtschaftslehre – VHB). The journal ranking mea-

sures the scientific quality of academic journals according to

two dimensions: (1) scientific quality of the published articles

(outcome quality) and (2) quality of the peer review process

(process quality). The approach weights the individual assess-

ments according to the expertise of the respondents.

For VHB-JOURQUAL2, 1,011 VHB-members evaluated the

scientific quality of more than 1,633 international and Ger-

man-language academic journals. The final ranking com-

prises 666 business research journals which are all assessed

by at least 10 respondents. Within the top ten journals are

four marketing and three finance journals, with the Journal

of Finance being listed as the number one journal. The best

German-language journal is Wirtschaftsinformatik, ranked

169, followed by Schmalenbachs Zeitschrift für betrieb-

swirtschaftliche Forschung (zfbf), ranked 177.

The validity of ranking is supported by strong and significant

correlations with the first edition of VHB-JOURQUAL and

with established rankings from the U.S., U.K., France, and the

Netherlands.

The authors acknowledge the threat that the importance of

VHB-JOURQUAL might be carried to excess. They see it as a

valid and useful tool to measure the scientific quality of jour-

nal publications, but argue that it must not be confused with a

measure of academics’ overall performance. At the same time,

further development of the VHB-JOURQUAL approach offers

promising research perspectives.

p 180Volume 2 | Issue 2 | December 2009 | www.business-research.org | ISSN 1866-8658

112

Volume 2 | Issue 2 | December 2009 | www.business-research.org | ISSN 1866-8658

p 215The Role of Ambidexterity in Marketing Strategy Implemen- tation: Resolving the Exploration-Exploitation DilemmaChristiane PrangeLyon Business School

Bodo B. SchlegelmilchWU Wien and Leeds University Business School

Marketing managers today face many contradictory challeng-

es when formulating and implementing strategies. Most of

these strategies are not intuitively compatible, such as, the re-

quirements of business development versus R&D, incremental

versus radical innovation or cost-driven versus high-quality

strategies. A recent stream of research reinterprets these chal-

lenges as exploration versus exploitation activities and offers

some solutions by suggesting the concept of ambidexterity.

Ambidextrous organization designs are suggested to overcome

intrinsically conflicting demands into organizational proce-

dures and toolsets that help to solve marketing dilemmas. Sev-

eral variants of ambidexterity may be used, such as separating

tasks between organizational units (structural ambidexterity)

or pursuing one task after the other (sequential ambidexter-

ity). Also, individuals may exhibit ambidextrous behaviors

in that they allocate their time to different tasks (contextual

ambidexterity) or develop a preference for one strategy so that

eventually managers must be exchanged (peripatric ambidex-

terity). In this article, we suggest a multi-level and temporal

framework that synthesizes previous research and helps man-

agers to decide upon the most suitable type of ambidexterity.

Examples from four firms further demonstrate that knowl-

edge of ambidextrous designs, associated tools and proce-

dures is a prominent requirement when companies want to

solve their marketing dilemmas. Each of the ambidextrous de-

signs facilitates the solution of a different marketing dilemma

and requires a specific, partly overlapping, set of implementa-

tion factors. These implementation factors relate to both the

individual and organizational level and require that managers

take a holistic view on elements, such as, goals, structures,

cultures, communication, incentives, recruiting leadership is-

sues, behavioral complexity, and organizational context. Each

type of ambidexterity requires a specific combination of these

elements. Further, as firms may not only use single variants

of ambidexterity but focus on their dynamic interplay, the ad-

justment of implementation factors is even more pertinent.

113


Effects of Ownership on Hospital Efficiency in Germany Oliver Tiemann, Jonas Schreyögg

V o l u m e 2 | I s s u e 2 | D e c e m b e r 2 0 0 9 | w w w . b u s i n e s s - r e s e a r c h . o r g | I S S N 1 8 6 6 - 8 6 5 8

BuR - Business Research Official Open Access Journal of VHB Verband der Hochschullehrer für Betriebswirtschaft e.V. Volume 2 | Issue 2 | December 2009 | 115-145

115

1 Introduction A considerable number of empirical studies have investigated whether public, private for-profit, and private non-profit hospitals differ in terms of effi-ciency. However, most of these studies have data and methodological problems limiting the gener-alizability of the findings. Furthermore, none of the studies has considered parameters for the quality of care together with the number of cases as traditional output parameter. The German hospital sector is large, and several different ownership types have co-existed for decades, making this a fruitful field for studying the effects of ownership on efficiency. Be-cause of increasing cost pressure, the hospital sector in Germany has been subject over the past two dec-ades to a variety of healthcare reforms aiming to stabilize expenditures at sustainable levels. The most significant reform in recent years was the in-troduction of a new system of reimbursement based on diagnosis-related groups (DRGs). The chief mo-tivation behind this fundamental overhaul of the old

reimbursement system, which was based on per-diem charges, was to set financial incentives that would increase the efficiency of German hospitals (Schreyögg, Tiemann, and Busse 2006). Due to substantial overcapacities and the rapid changes currently taking place in the regulatory and competitive environment, the German hospital sector is now facing an extensive process of consoli-dation and reorganization. In this context, hospitals are considering mergers, acquisitions, and coopera-tive agreements as ways to improve competitive-ness. Over the past decade, the total number of hos-pitals in Germany has decreased and a growing number of hospitals have been privatized. Although there are several possible reasons for this develop-ment, the main driver has been the need to increase hospital efficiency (Megginson, Nash, and van Ran-denborgh 1994). The objective of the present study was to evaluate the relative efficiency of public, private for-profit, and private non-profit hospitals in Germany. To do


Oliver Tiemann, Munich School of Management, Ludwig-Maximilians-Universität München and Helmholtz Zentrum München, Germany

E-Mail: [email protected]

Jonas Schreyögg, Munich School of Management, Ludwig-Maximilians-Universität München and Helmholtz Zentrum München, Germany

E-Mail: [email protected]

Abstract The objective of our study was to evaluate the efficiency of public, private for-profit, and private non-profit hospitals in Germany. First, bootstrapped data envelopment analysis (DEA) was used to evaluate the effi-ciency of a panel (n = 1,046) of public, private for-profit, and private non-profit hospitals between 2002 and 2006. This was followed by a second-step truncated linear regression model with bootstrapped DEA efficiency scores as dependent variable. The results show that public hospitals performed significantly better than their private for-profit and non-profit counterparts. In addition, we found a significant positive association between hospital size and efficiency, and that competitive pressure had a significant negative impact on hospital efficiency. Keywords: public hospitals, ownership, data envelopment analysis, performance measurement, non-parametric technique, truncated regression, Germany Manuscript received September 23, 2008, accepted by Peter Walgenbach (Management) September 29, 2009.


116

so, we used a bootstrapped DEA approach followed by a second-step truncated linear regression model while controlling for patient heterogeneity and ex-ploring the impact of hospital organizational and environmental characteristics. Failing to take prop-er account of these characteristics can lead to seri-ously flawed conclusions (Fried, Knox Lovell, Schmidt, and Yaisawarng 2002). The paper is structured as follows. The next (i.e. second) section reviews the relevant theoretical and empirical literature on the effects of ownership on efficiency and quality of care. The third section pre-sents the setting, data, and methodology used in this paper to explore this relationship. The fourth sec-tion describes and discusses the estimated results, and the final section draws conclusions and makes suggestions for future research.

2 Theoretical and empirical background

The hospital industry in Germany is a fruitful field for studying the effects of ownership on hospital performance. It is one of the few sectors where dif-ferent types of ownership have co-existed for dec-ades, and it is large, ensuring an appropriate sample size. When selecting a central performance criterion for our analysis, we followed the example of other theoretical and empirical studies that have dealt with the comparison of ownership types. Standard performance measures such as return on invest-ment and profitability were found to be inappropri-ate for public and private non-profit entities (Leibenstein 1966; Feldstein 1967; Rees 1988; Pes-tieau and Tulkens 1990). Consequently, we focus our analysis on technical efficiency, which is a key concept in measuring performance as it refers to the optimal use of resources in the production process. In particular, technical efficiency (i.e. productive efficiency) is a measure of how well an organization produces output from a given amount of input, or alternatively produces a given amount of output with minimum quantities of inputs. In order to address a key limitation of existing studies, our study examines both the productive efficiency im-plications and patient health outcomes (i.e. quality of care) of hospital ownership. There have been many theoretical contributions on the effects of hospital ownership on efficiency and quality of care. The following discussion highlights the main argu-ments.

2.1 Theoretical Perspectives and Development of Hypotheses

By ownership, hospitals can be generally divided into private hospitals (i.e. private for-profit and private non-profit), invested and owned by private entities (i.e. individual or several private owners) and public hospitals, invested and owned mainly by public entities such as governments. Agency theory and property-rights theory, as well as public choice theory provide different explanations for a common outcome. Private for-profit ownership, they hy-pothesize, is superior to public and private non-profit ownership because private for-profit owner-ship is associated with a higher productive effi-ciency. These theories emphasize that differences in efficiency are due to substantial dissimilarities in objectives, incentives, and control mechanisms between ownership types. The following discussion highlights the main arguments. The agency theory assumes that managers (or agents) seek to maximize their own utility rather than that of the organization or its owners (or prin-cipals). As a result, in all three types of hospital ownership owners face a principal-agent problem with those whom they hire to do the managing. Although both public and private hospitals face this problem, it is assumed that private for-profit hospi-tals have better means to solve the principal-agent dilemma and therefore their performance in terms of productive efficiency is expected to differ signifi-cantly. The owners of a profit-seeking hospital have profits as their measure of the manager’s success. The owner can limit divergences from his interest, by making the manager's compensation a positive function of the profits (i.e. a correlation between profits and managerial salaries and promotions). In addition, the income of executive physicians in pri-vate for-profit hospitals might also be tied to hospi-tal's financial performance. Within the public and private non-profit hospitals, individual decision makers rarely have their income tied to the hospi-tal's performance (e.g., pay scales designed for civil servants); therefore no individual has a strong in-centive to enforce efficient behavior. Accordingly, it is expected that private for-profit hospitals realize a higher level of efficiency than their public and pri-vate non-profit counterparts (Reder 1965; Rice 1966; Newhouse 1970; Lee 1971; Pauly and Redisch 1973; Foster 1974; Sloan 1976; Fama and Jensen 1983; Jacobs and Rapoport 2003).


117

According to the property-rights theory, ownership of a firm involves two essential rights: (1) the right to control the firm and (2) the right to appropriate the firm's profits (i.e. financial surplus). Accord-ingly, the defining difference is that public and non-profit hospitals are precluded from distributing, in financial form, its surplus to those in control of the organization (i.e. non-distribution constraint). Within for-profit hospitals, assigning some of the financial surplus to the individual who manage the hospital thus provides a way to monitor his activi-ties. In this case, it is expected that the monitoring is automatic and self-imposed by the manager and that managers will have strong incentives to behave in the interests of the owners (Jacobs 1974; Clark-son 1972; Hansmann 1988). In addition, potential divergences of interests between owners and man-agers in private for-profit organizations are further reduced by external mechanisms, including (a) a market for ownership rights that enables the owners to sell their shares if they are not satisfied with ma-nagerial performance; (b) the threat of takeover; (c) the threat of bankruptcy; and (d) an extensive ma-nagerial labor market (Villalonga 2000). Thus, property-rights theory assumes that private for-profit ownership is associated with a higher effi-ciency compared to the other types of ownership. As part of public choice theories Buchanan and Tollison (1972), Niskanen (1975), and Shleifer and Vishny (1994) argue that politicians impose their objectives on public organizations in order to gain votes, which may conflict with profit maximization and, therefore, with productive efficiency. In the case of private non-profit providers Newhouse (1970) and Weisbrod (1988) argue that, because they lack the incentive to maximize profit, they should be expected to diverge from strict cost- or inefficiency-minimizing behavior and instead max-imize quality, quantity, and/or prestige. Sloan (2000) supports this view and adds that productive efficiency will decay if objectives are vague and con-tradictory, which is typically the case in public and private non-profit hospitals. According to the theories mentioned above, private for-profit hospitals are expected to maximize profits on the basis of a high level of productive efficiency. It is emphasized that dissimilarities in efficiency and quality of care are due to substantial differences in objectives, incentives, and control mechanisms. Owners may differ as some are more willing to sup-ply more than the profit-maximizing services than

are others. Such hospitals produce services that are not likely to be produced by a for-profit institution, unless, of course, dedicated subsidies exist for these services (e.g., supplementary for highly specialized services or centers in Germany). A profit-maximizing hospital will produce services to the point where the marginal costs equal the marginal revenue. Public and private non-profit hospitals will expand their output at least to the point where the hospital just breaks even – that is, where the total cost equals total revenue. Much of the rationale for public and private non-profit hospital ownership is based on the welfare implications of these owner-ship types (Hall 1995; Steinberg 2006; Rathgeb Smith and Gronbjerg 2006).

Hypothesis 1: Private for-profit hospitals are more efficient than private non-profit and public hospitals.

The character of the services that are provided by hospitals implies that quality of care is one major objective for hospitals in addition to efficiency. For the hospital sector, a common theoretical assump-tion is that better quality of care requires more re-sources and therefore reduces efficiency. It is fur-ther assumed that due to information asymmetries between hospitals and other actors (e.g., patients and purchasers) hospitals may be able to vary their quality of care (Newhouse 1970; Weisbrod 1988). However, differences in the trade-off between pro-ductive efficiency and quality of care among public and private non-profit providers have attracted little attention from a theoretical point of view. Given the strong efficiency focus of private for-profit providers it is argued that due to the trade-off those hospitals offer a lower quality of care than hospitals of other ownership types. In contrast, private non-profit and public hospitals allow more room for objectives other than efficiency and are therefore able to pro-vide a higher quality of care than private for-profit hospitals. Another group of theorists argues that particularly physicians represent a group that is typically acting driven by high intrinsic motivation (Arrow 1963; Zismer 1999; Feess and Ossig 2007). Thus, physi-cians are intrinsically motivated to deliver high quality of care. The enhanced behavioral discretion conditioned by the non-distribution constraint of public and private non-profit hospitals might pro-vide a fruitful field for intrinsic motivation of physi-cians to act in the patient’s best interest. In contrast,


118

the extrinsic motivation conditioned by financial incentive schemes usually provided in private for-profit hospitals can be expected to result in a crowd-ing out of intrinsic motivation (Osterloh and Frey 2000; Frey and Jegen 2001). This effect lends sup-port to the notion that private for-profit hospitals provide a lower quality of care. However, as physi-cians have to act in accordance with regulations and ethical rules (e.g., clinical guidelines, Hippocratic Oath) a certain minimum standard of quality of care will be ensured.

Hypothesis 2: The efficiency differences between ownership types will decrease if quality of care is taken into account to determine efficiency (i.e. quality-adjusted efficiency) due to an inevitable trade-off between efficiency and quality of care.

2.2 Empirical Evidence There is a rich body of literature on hospital per-formance in the United States and Europe. Two major reviews have been published in recent years. Hollingsworth (2008) reviewed the literature on frontier efficiency measurement techniques in healthcare. From the early 1980s up to mid-2006 a total of 317 studies had been published, almost 80 % of which made use of non-parametric DEA while most of the other studies used parametric stochastic frontier analysis (SFA). Reviewing the studies on efficiency differences of hospital ownership Hol-lingsworth found 39 studies and concluded that public hospitals in Europe and the United States appear to be more efficient than their private for-profit and non-profit counterparts. Shen, Eggleston, Lau, and Schmid (2007) applied meta-analytic me-thods to conduct a quantitative review of the em-pirical literature on hospital ownership published between January 1990 and July 2004 (i.e. 16 em-pirical studies). The authors concluded that owner-ship had an impact on efficiency, but the conven-tional assumption that private for-profit hospitals operate more efficiently was not supported in the review by Shen, Eggleston, Lau, and Schmid (2007), who indeed rather observed the opposite. Taken together Hollingsworth (2008) and Shen, Eggleston, Lau, and Schmid (2007) reviewed eleven international studies that have compared all three different types of ownership in terms of efficiency. In four studies public hospitals were found to be less efficient than their counterparts (Zuckerman, Had-ley, and Iezzoni 1994; Ferrier and Valdmanis 1996;

Brown 2003; Chang, Chang, Das, and Li 2004), while six studies showed that publicly owned hospi-tals are more efficient than private for-profit and non-profit hospitals (Ozcan, Luke, and Haksever 1992; Burgess and Wilson 1996; Koop, Osiewalski, and Steel 1997; Chirikos and Sear 2000; McKay, Deily, and Dorner 2003; Sari 2003). Burgess and Wilson (1998) found no significant efficiency differ-ences associated with ownership. Out of these ele-ven studies, seven studies have a large nation-wide sample and are based on comparable large data samples (i.e. sample sizes between 382 and 4,075 hospitals). Among all eleven studies, none of the studies have used patient-level information to con-trol for patient heterogeneity and to ensure the comparability of the observed hospitals which may be due to the lack of availability of adequate meas-ures for patient heterogeneity between ownership types. Only two studies have used a two-stage analy-sis (i.e. DEA followed by some form of regression analysis), despite the fact that in recent years it has become the state-of-the-art approach (Burgess and Wilson 1998; Ferrier and Valdmanis 1996). In addi-tion, the study periods of all eleven studies are dated back to the late 1980s or 1990s, thus limiting the generalizability of the results. Empirical data on the efficiency of the German hos-pital sector, and especially on the effect of owner-ship status on hospital efficiency, are scarce. The efficiency of German hospitals has been investigated primarily using DEA. Staat and Hammerschmidt (2000) were the first to employ DEA to German hospitals, based on data of 160 hospitals in 1994. The authors found that DEA efficiency scores dif-fered significantly between ownership types and that private non-profit hospitals were, on the aver-age, less efficient than their public and private for-profit counterparts. Using aggregate state-level data from 1991 to 1996, Helmig and Lapsley (2001) showed that public and non-profit hospitals appear to use relatively fewer resources than private for-profit hospitals (n= 288). Staat (2006) applied a refined DEA approach to the same sample of 160 hospitals for the year 1994 and found no significant efficiency differences associated with ownership. Herr (2008) employed SFA to investigate the effi-ciency of about 1,500 German hospitals between 2001 and 2003. Herr’s empirical results for the years 2000 through 2003 indicate that private for-profit and non-profit hospitals were less efficient than publicly owned hospitals.


119

Of the few studies that have investigated the effi-ciency of the German hospital sector to date, all have important drawbacks. One of these is the lack of detailed data, which means that the quality and the quantity of the information used to assess effi-ciency is often limited (e.g., aggregate state-level data, small sample size), thus limiting the gener-alizability of results. Another drawback of these studies is their use of DEA. From a methodological point of view, some authors argue that results of DEA analysis are much more robust when study samples are very large (Shen, Eggleston, Lau, and Schmid 2007; Jacobs, Smith, and Street 2006). All of the DEA studies that have investigated the effi-ciency of the German hospital sector thus far have used DEA alone without using a two-stage ap-proach. In addition, none of the German studies used patient-level information on diagnoses, proce-dures, and age to control for patient heterogeneity and to ensure the comparability of the observed hospitals. Our review of empirical studies suggests that, in contrast to the assumed behavior outlined in theory, there is no evidence that private ownership is asso-ciated with higher efficiency compared to other ownership types. In fact, our review indicates that public hospitals use relatively fewer resources than private for-profit and non-profit hospitals. In addi-tion, none of the mentioned studies has considered parameters for the quality of care together with the number of cases as traditional output parameter. The output of the hospitals’ production process has been measured in terms of physical units (e.g., number of patients treated or patient days). Al-though an increasing number of studies have made adjustments for case mix, the intermediate hospital outputs (i.e. number of cases) are not adjusted for final health outcomes/quality of care (e.g., mortality rates or readmission rates). One reason for this might be the paucity of validated measures of qual-ity. The absence of quality measures requires the implicit assumption that there are no systematic variations in quality, or that variations in quality do not systematically affect efficiency. However, studies examining the trade-off between efficiency and quality of care have provided evidence of an inevita-ble trade-off between these two measures. Morey, Fine, Loree, Retzlaff-Roberts, and Tsubakitani (1992), Picone, Chou, and Sloan (2002) and Deily and McKay (2006) found that hospital efficiency

was positively associated with the observed in-hospital mortality rate. While testing the above mentioned hypotheses we also aimed to address methodological limitations of previous studies. First, we used a two-step method-ology integrating covariates representing organiza-tional and environmental characteristics in the sec-ond stage regression that may have impact on hos-pital efficiency. Second, we controlled for patient heterogeneity among hospitals by including vari-ables representing case-mix complexity. Third, we relied on a large data sample to generate robust results. Fourth, we performed a number of robust-ness checks to investigate the validity of our results.

3 Research design and methods

3.1 Setting and Data In Germany, approximately 1,800 hospitals provide inpatient care and receive DRG payments from social health insurance funds and private health insurance companies. The data for our study were derived from the annual hospital reports collected and administered by the Research Data Centre of the Statistical Offices of the Länder (For-schungsdatenzentrum der Statistischen Landesäm-ter 2008). This rich dataset covers all public, private for-profit, and private non-profit hospitals in Ger-many and contains hospital-level information on costs and hospital infrastructure, as well as patient-level information on age, diagnoses, and certain procedures performed per case. Our study is based on data from the fiscal years 2002-2006, and the unit of analysis was the hospital. Because of data privacy issues, we were able to obtain randomly selected data from only two-thirds of German acute care hospitals (n = 1318). To ensure the comparabil-ity of the hospitals in the sample, hospitals provid-ing only psychiatric care, day clinics, and hospitals with fewer than 50 beds were excluded from further analysis. In addition, content-based plausibility checks were conducted to reveal measurement er-rors. Finally, a total of 1,046 hospitals remained in the sample and due to missing values in some years our sample for the years 2002 through 2006 is based on 4,902 observations. German hospitals can have public (PUBLIC), pri-vate for-profit (PRIVATE-FP), or private non-profit (PRIVATE-NP) ownership status. Between 1993 and 2006, the share of private for-profit hospitals


120

rose from 16 % to 28 %, whereas the share of public hospitals decreased from 43 % to 34 %. During the same period, the share of private non-profit hospi-tals remained relatively constant (Forschungsdaten-zentrum der Statistischen Landesämter 2008).

3.2 First stage: Data Envelopment Analysis In the hospital sector, data envelopment analysis (DEA) has been the most frequently used approach for measuring efficiency (Hollingsworth 2008). DEA is a linear programming technique for evaluat-ing the relative efficiency of individual organizations based on observed data. The relative efficiency of an organization is defined as the ratio of the weighted sum of its outputs to the weighted sum of its inputs. The weights are not pre-assigned, but rather deter-mined by the model, thus avoiding any bias result-ing from subjectively assigned weights. DEA as-sesses the efficiency of organizations in two stages. First, the location and the shape of the efficiency frontier are determined based either on organiza-tions that use the lowest input mix to produce their outputs or on organizations that achieve the highest output mix given their inputs. The efficiency frontier is constructed by joining these observations and all linear observations in the input-output space. In our study, we used an input-oriented DEA approach to address the following question: "To what extent can the input factors, defined as supplies and labor, be reduced proportionally without changing the output quantities of German hospitals?" Second, DEA measures inefficiency as the radial distance from the inefficient unit to the frontier and produces an effi-ciency score that reflects the relative efficiency of each unit (Cooper, Seiford, and Zhu 2004). DEA allows multiple inputs and outputs to be con-sidered simultaneously, which seems particularly well-suited for measuring the efficiency of hospitals. In contrast to parametric methods, where a specific pre-defined functional form is assumed to apply to each observation, DEA has the advantage of requir-ing no assumptions about the functional form of the production or cost frontier. Although this reduces the need for a theoretical exposition of the model specification, there are other important considera-tions. The DEA results are sensitive to the number of variables included in the model. In general, the number of inputs and outputs is limited by the sam-ple size, which should not exceed one-third of the number of observed units (Banker, Charnes, Coo-per, Swarts, and Thomas 1989). It is important to

recognize that the inputs and outputs are collected routinely by hospital accounting departments, mak-ing measurement errors less relevant. Based on our understanding of the market constraints within the German hospital sector, we assumed variable re-turns to scale, which may be appropriate when it is impossible to assume that all observed units are operating at an optimal scale (Banker, Charnes, and Cooper 1984). In the healthcare sector, imperfect competition and budgetary constraints, as well as regulatory constraints on entry, mergers, and exits, may often result in organizations operating at an inefficient scale size (Jacobs, Smith, and Street 2006). When selecting inputs and outputs, we followed the example of other studies that developed DEA frameworks for measuring hospital efficiency (Pi-lyavsky, Aaronson, Bernet, Rosko, Valdmanis, and Golubchikov 2006; Jacobs, Smith, and Street 2006; Burgess and Wilson 1998). For the purposes of our study, six inputs and two outputs were considered. The first input variable (SUPPLIES) is the amount spent on supplies per year, including operational expenses, but excluding payroll, capital, and depre-ciation expenses. Taking into account the relative importance of resource use in terms of labor in the hospital production process, additional input vari-ables were the number of full-time equivalents (FTE) for the following personnel categories: clinical staff (CLIN), nursing staff (NURS), medical-technical staff (MEDTECH), administrative staff (ADMIN), and other staff members (OTHER). The first output variable (INPATIENT) reflects the number of treated inpatient cases per year in each hospital. To adjust for variations in the quality of care between hospitals, we used the average mortal-ity rates per year in each hospital. Therefore, the second output represents 1 minus the average hos-pital mortality rate per year (1 - MORTALITY). This could lead to the concern that case mix might vary systematically across the hospitals in our sample, which would be problematic because hospitals with a more complex case mix should receive lower effi-ciency scores in the first stage of our analysis. To help address this potential issue, we included case-mix measures in our regression analysis as control variables (see further details in section 3.2). A de-scriptive overview of the inputs and outputs used for our DEA model is given in Table 1. A correlation analysis (Appendix 1) shows that our multiple inputs are positively correlated with our


121

output set. This is an important prerequisite for applying DEA. In addition, subsets of inputs and outputs are often correlated. In our study especially the input variables are highly correlated. This might suggest that a limited number of inputs might ade-quately represent the selected input set in our effi-ciency assessment. However, several authors (e.g., Dyson, Allen, Camanho, Podinovski, Sarrico, and Shale 2001; Jacobs, Smith, and Street 2006) argue that omission of a highly correlated variable can lead to significant changes in the efficiency esti-mates. They emphasize that correlation is an aggre-gate measure of the closeness of two sets of ob-served data. Therefore, variations of the input levels of individual hospitals may have little effect on the correlation, but significant effect on the measured efficiency. It may also give evidence for a possible production technology that is common among all decision-making units. In addition Dyson, Allen, Camanho, Podinovski, Sarrico, and Shale (2001) argue that the omission of variables in order to in-crease discrimination is less relevant for large data samples. Thus, we used all input variables for our DEA model.

Table 1: Variable definitions and descriptive statistics for first-stage analysis*

Variable name Definition Mean SD

CLIN Number of clinical staff in FTE

78.17 106.54

PUBLIC 112.7 182.8

PRIVATE-NP 52.4 42.1

PRIVATE-FP 47.9 63.1

NURS Number of nursing staff in FTE

225.8 225.3

PUBLIC 296.6 353.4



MEDTECH Number of medical technicians in FTE

163.3 231.3

PUBLIC 240.9 415.9



ADMIN Number of adminis-trative staff in FTE

43.1 56.0

PUBLIC 60.0 100.8



OTHER Number of other staff in FTE

87.4 106.7

PUBLIC 127.7 184.9



SUPPLIES Amount of supplies in million euros

16.1 23.0

PUBLIC 22.8 38.8

PRIVATE-NP 10.6 9.5


INPATIENT Number of treated cases

10,852 9,912

PUBLIC 13,941 14,731

PRIVATE-NP 8,809 5,839

PRIVATE-FP 7,340 7,180

1-MORTALITY 1 minus the average mortality rates per year

0.974 0.015

PUBLIC 0.973 0.011



DEA I** Average DEA effi-ciency scores

0.634 0.138

PUBLIC 0.649 0.130



DEA II** Average DEA effi-ciency scores 0.636 0.137

PUBLIC 0.654 0.128



DEA III** Average DEA effi-ciency scores 0.662 0.133

PUBLIC 0.674 0.125



*Pooled sample including university hospitals and hospitals with beds ≥50. **DEA models are specified in table 2.

Three model specifications served as a sensitivity analysis to test whether the efficiency scores and ranks remained stable when a specific variable (quality of care as second output) or specific provid-ers (university hospitals) were removed or added. Efficiency models I and II both used INPATIENT as the only output variable, whereas university hospi-tals were removed in the second analysis. The third


122

DEA model reflects quality of care using 1-MOR-TALITY as a second output; here, too, university hospitals were excluded. The specifications are summarized in Table 2.

Table 2: Specification of DEA models*

Models Trimming Output

DEA I With university hospi-tals

INPATIENT

DEA II Without university hospitals

INPATIENT

DEA III Without university hospitals

INPATIENT &

1-MORTALITY

*Models are estimated per year (2002-2006)

3.3 Second Stage: Truncated Linear Regression Model

An important assumption of our study was that environmental and organizational factors may in-fluence the relative efficiency of hospitals in addi-tion to ownership. By considering the impact of co-variates reflecting environmental and organizational factors on hospital efficiency we believe to provide a better explanation of the variation of efficiency and more robust results about the effects of ownership than previous studies that did not control for these effects. We therefore used the DEA efficiency scores obtained in the first stage of our analysis as depend-ent variable in a truncated linear regression model, which became the favored approach owing to the censored distribution of the DEA-based relative efficiency estimates (Simar and Wilson 2007). This model is appropriate for these data, as these are bounded at both ends of the 0-1 distribution (Ja-cobs, Smith, and Street 2006). However, this ap-proach has been found to result in inconsistent es-timates unless the DEA efficiency scores are cor-rected by a bootstrapping procedure. The procedure applied in the present study follows the bootstrap approach developed by Simar and Wilson (1998, 2000). For our study, the bias-corrected scores were derived from 250 bootstrap iterations, which al-lowed us to estimate a robust regression model as the second-stage analysis (Simar and Wilson 2007). Based on our theoretical framework, the following empirical model was used in the analysis. In the input-oriented case, the variable returns to scale

variant of the BCC (Banker, Charnes, and Cooper) model can be formulated as a linear programming problem as shown below (Banker, Charnes, and

Cooper 1984). Let iθ , i = 1,…,n , be the hospital’s

efficiency where n represents the number of obser-vations (i.e. the number of hospitals). Matrix

k x nX refers to k observed inputs of n com-pared hospitals and matrix r x nY refers to r observed outputs of the compared hospitals. Vectors

ki x and k

i y present the inputs and

outputs of unit i , i.e. the i th columns of matrix X and Y respectively. Furthermore, 1 refers to a column vector of ones with a suitable dimension. The DEA efficiency score, which is the reciprocal of

the inefficiency, iθ can be obtained by solving the following BCC linear programming model:

(1) Maxi i ,

(2) 0s.t. 0, Y y

(3) 0 0,i xX

(4) 0 0,i xX

(5) T 1, 0.1

In the second step, we used the following model specifications for our regression analysis. Let iZ be each corresponding vector of covariates (i.e. types of ownership, hospital characteristics, environmental characteristics, and patient heterogeneity). Accord-ing to Simar and Wilson (2007), we applied a trun-cated linear regression to model hospital’s efficien-cy. Therefore, we assumed iθ as being distributed based upon a set of m n normally distributed

random variables, j , j 1 , , m with

(6) j j j , Z

where jZ refers to a vector of covariates, repre-

sents a parameter vector, and j is a normally dis-

tributed error term (i.e. 2N 0j , ). On this

basis 1 n, , is defined as the truncated set of

1 m, , with 1j , and i jZ Z for

i j . We applied linear regression analysis to assess whether different types of ownership (PUBLIC, PRIVATE-FP, and PRIVATE-NP) led to differences in efficiency; we also considered a number of control variables. The use of control variables is of particu-lar importance in the healthcare context because there are usually certain structural or regulatory


123

determinants of hospital efficiency that a hospital cannot influence. Heterogeneity in hospital characteristics was cov-ered by the following variables. The first of these was the number of licensed and staffed beds (BEDS), an approach taken in previous studies to control for hospital size (Carey and Burgess 1999; Dudley, Johansen, Brand, Rennie, and Milstein 2000; Harrison, Coppola, and Wakefield 2004). In the context of strict hospital planning in Germany, the number of beds per hospital can be seen, at least in the medium term, as an exogenous factor outside of hospital management's control (Busse and Ries-berg 2004). To account for higher resource con-sumption due to differences in teaching activities, we included a variable (TEACH) for the training of non-medical staff. These activities are represented by the ratio of trainee positions to the sum of all non-medical personnel. Another important point is that hospitals may hire out beds to self-employed ambulatory physicians (e.g., for ambulatory sur-gery). The estimated DEA efficiency scores in the first stage of our analysis were higher for these hos-pitals because the referring cases were counted as hospital output, whereas the corresponding re-source use in terms of physicians was not consid-ered on the input side. To control for this fact, we considered the proportion of all hospital beds that had been hired (HIRED BEDS) as variable in the regression models. The dummy variable "ambula-tory care" (AMBULATORY) is related to missing data from the first stage of our analysis, which only considers the inputs for the production of ambula-tory care, whereas the outputs of the ambulatory activities (e.g., ambulatory visits) were not captured in our data set. Because of the ongoing trend to-wards privatization in Germany, one could argue that a large number of the existing private for-profit hospitals were recently converted from public and non-profit ownership status and that most of these hospitals represent inefficient units. Therefore, we included a dummy variable (CONVERSION) to account for public and non-profit hospitals that were privatized during the study period (n = 68). The set of explanatory variables representing the different environmental characteristics were as follows. The most important regressor is the Hir-schman-Herfindahl index (HHI), which measures competitive pressure in a hospital's market, a stan-dard economic measure of industry concentration. The market area was defined as the county in which

a hospital was located, which is a frequently used definition in hospital studies (Chang, Chang, Das, and Li 2004; Rosko 1999, 2001, 2004; Rosko and Chilingerian 1999; Chirikos and Sear 1994). Al-though there has been some controversy about the appropriate definition of a hospital's market area, Garnick, Luft, Robinson, and Tetreault (1987) re-ported that, for the purpose of measuring competi-tive activity, it made little difference whether a hos-pital's market was defined as a county or as a radius. The HHI is obtained by squaring the regional mar-ket share of a hospital (reflected by the distribution of treated cases), and then summing the market shares of admissions for all of the hospitals in the county. The higher the HHI, the more concentrated the regional market. We used HHI to measure the effects through the changes over time in hospitals' competitive environment. This specification allowed us to differentiate between the effects of ownership and the effects of changes in market structure re-sulting from recent healthcare reforms. Further-more, variations may also result from the fact that hospitals located in eastern or western Germany are different in terms of their infrastructure. After Ger-man reunification in 1990, hospitals in eastern Germany received comparably higher subsidies from the federal government to upgrade their infra-structure (Busse and Riesberg 2004). Thus, we defined being located in eastern Germany (includ-ing all of Berlin) as a dummy variable (EAST). Because resource consumption can vary substan-tially between patients, we also included 26 vari-ables to control for variations associated with case-mix complexity. We used patient-level information on diagnoses, procedures, and age to control for patient heterogeneity. To control for case-mix com-plexity, we compiled a comprehensive list of co-morbidities that have been found in other studies to affect mortality and resource use. In doing so, we relied on the Charlson Comorbidity Index (Sunda-rarajan, Henderson, Perry, Muggivan, Quan, and Ghali 2004). Thus, the full set of case-mix measures are included in all of the regression models de-ployed in our second-stage analysis. Another com-mon approach would be to use the case-mix index whose weight reflects the relative costliness of DRGs. Due to data privacy issues, we were not per-mitted to match the referring data. However, Carey (2000, 2002) reported that individual-level meas-ures represent a vast improvement over aggregate case-mix measures and that the DRG case-mix in-


124

dex is therefore a relatively weak measure of sick-ness to control for patient heterogeneity. To check the robustness of our estimates, we per-formed three regression models using the DEA effi-ciency scores from three different DEA specifica-

tions as dependent variables (see Table 1). Table 3 provides the definitions of variables for our second-stage analysis and the overall descriptive statistics of our sample with respect to the dependent and inde-pendent measures.

Table 3: Variable definitions and descriptive statistics for second-stage analysis*

Variable name Definition Mean SD

HHI Index for concentration of hospital cases per county 0.345 0.219

PUBLIC 0.402 0.247



BEDS Number of beds per hospital 330.3 279.3

PUBLIC 417.7 418.6



EAST Dummy variable for hospitals located in the eastern part of Germany 0.183 0.379

PUBLIC 0.178 0.382



AMBULATORY Dummy variable for hospitals that provide ambulatory care 0.783 0.395

PUBLIC 0.856 0.351



HIRED BEDS Ratio of hired out beds to all beds 0.068 0.141

PUBLIC 0.065 0.115



TEACH Ratio of apprenticeship training positions 0.269 0.426

PUBLIC 0.302 0.518



CONVERSION Dummy variable for hospitals that were privatized during study period

68

PUBLIC 50

PRIVATE-NP 18

*Pooled sample including university hospitals and hospitals with beds ≥50.


125

4 Findings and discussion The regression results for the three regression mod-els are summarized in Table 4. A correlation analy-sis of our explanatory variables suggested that mul-ti-collinearity was not an issue in our study (see variance inflation factors in Appendix 3). The coeffi-cients can be interpreted as marginal effects, and private for-profit hospitals served as a reference category. Throughout the models, ownership was coded as a dummy variable. The regression results for all three models showed that public hospitals operate at a significantly higher level of efficiency than their counterparts. According to this result our first hypothesis has to be rejected. However, the impact of public ownership was lower for the first model, which included university hospitals (effect of 1.9 %; P≤0.01), than it was for the second model (effect of 2.3 %; P≤0.001). University hospitals were clearly associated with lower efficiency in our first model. This can be explained by the fact that univer-sity hospitals produce multiple outputs (i.e. a com-bination of patient care, education, and research) (Schreyögg and Reitzenstein 2008); as a result, their production process is not adequately captured by our DEA models. In addition to the different model variations per-formed for sensitivity purposes we checked the ro-bustness of our findings in several ways. To begin with, we re-estimated our first- and second-stage models with different windows for hospital sizes at the lower end of the sample. We performed four different models including all hospitals above 30 beds, above 50 beds, above 70 beds, and above 100 beds. Efficiency differences between private hospi-tals and the two other hospital types tended to be slightly larger when 30 beds were used as the lowest hospital size included in our study while the results hardly changed when sizes of 70 beds or 100 beds were used as the minimum number of beds per hospital instead of 50 beds. Second, we run the second-stage models modifying the number of peri-ods used in our regression (2 years, 3 years, 4 years, and 5 years). The modifications had very little effect on the coefficients representing different ownership types (see Appendix 4 and 5 for regression results). Looking at the development of efficiency scores over time it turns out that efficiency of all ownership types slightly increases over the years to a similar extent. Finally, in our third DEA model we used 1

minus the average hospital mortality rate per year, which represents an index variable. Syrjänen (2004) Table 4: Regression results for each model

Independent variables

DEA I DEA II DEA III

Coefficients

PUBLIC 0.019** 0.023*** 0.018**

PRIVATE-NP 0.009 0.010 0.006

PRIVATE-FP Served as reference category

HHI 0.086*** 0.085*** 0.069***

BEDS (in 1,000) 0.061*** 0.085*** 0.064***

EAST 0.012* 0.012* 0.024***

AMBULATORY -0.018*** -0.019*** -0.028***

HIRED BEDS 0.049* 0.054** 0.038*

TEACH 0.004 0.003 0.001

CONVERSION -0.002 0.002 0.009

26 Case-mix va-riables

Included Included Included

* P≤0.05; ** P≤0.01; *** P≤0.001

found that mixing index and volume measures in DEA may lead to biased results for the most com-monly used constant returns to scale variant of the BM model (Banker and Moorey model) and the CCR model (Charnes, Cooper, and Rhodes model). However, this problem does not apply in our con-text as we used a variable returns to scale variant of the BCC model. Thus, in our case the use of an index variable leads to robust results (Hollingsworth and Smith 2003; Syrjänen 2004). According to our second hypothesis a trade-off be-tween efficiency and quality of care would lead to smaller efficiency differences between ownership types. Compared to DEA models I and II the relative efficiency of PUBLIC hospitals decreased in the DEA III model when quality was considered as an additional output. Thus, we indeed observed a trade-off between efficiency and quality of care sup-porting hypothesis 2. However, the regression anal-ysis that used quality-adjusted DEA efficiency scores as the dependent variable (i.e. DEA III) revealed significantly higher efficiency for public ownership compared to other ownership types. Our general finding that public hospitals perform better than hospitals with other forms of ownership is consistent with the results of several earlier stu-


126

dies which, however, as mentioned earlier, have certain methodological limitations. Among the con-trol variables, market concentration (HHI) and hospital size (BEDS) were important exogenous market effects, and the regression results revealed a significant positive association with efficiency in all three models (P≤0.001). In empirical studies on hospital efficiency, the variable "beds per hospital" is often used as a proxy for hospital size. However, as Breyer (1987) indicated, the variable also con-tains information on capital inputs. If beds are in-terpreted as a measure of size, we can conclude that larger hospitals perform significantly better. How-ever, studies on the relationship between hospital costs and hospital size measured in terms of beds have generally identified a U-curve, because mar-ginal costs tend to decline with increased size and then rise again, as indicated by a negative sign for the variable "squared beds per hospital" in the re-gression (Carr and Feldstein 1967). In our regres-sion, the variable "squared beds per hospital" and also the variable "cubed beds", which would imply a cubic relation, were not significant. Thus, our study shows that, in contrast to the evident relationship between costs and hospital size, the relationship between efficiency and hospital size is linear. In this context, it is important to emphasize that this find-ing is not biased by our scale assumptions. In our DEA model specification we used variable returns to scale because constant returns to scale are assumed to represent a hospital planning view or a govern-mental view (Steinmann, Dietrich, Karmann, and Zweifel 2004). In their review of the literature, Scherer and Ross (1990) and Button and Weyman-Jones (1992) found that the degree of competitiveness in a firm's market was a potential source of lower efficiency. Classical economic theory predicts that organiza-tions in monopolistic or oligopolistic markets re-strict output and have higher average costs than competitive organizations. In our study, market competition was measured using the Hirschman-Herfindahl index, which is defined over a range between zero and one such that increases in HHI correspond to decreases in hospital competition. Therefore, the positive coefficient lends support to the notion that a hospital’s behavior is highly af-fected by competition in the hospital sector, though in a direction that is inverse to that seen in markets for most other goods and services. In the German context, this finding is also related to the strict regu-

lations in place regarding hospital planning. Com-petition between hospitals in Germany does not occur primarily in terms of individual patients, but with regard to the optimal fit of demand and supply in terms of hospital infrastructure (e.g., specialties, departments, number of beds). Our findings indi-cate that hospitals operating in monopolistic or oligopolistic markets are more likely to agree with the hospital planning authorities on a hospital infra-structure that enables a convergence of demand and supply in the referring county. Overcapacities in urban areas result in greater competitive pressure (i.e. cutthroat competition), whereas counties with lower competitive pressure are likely to be more rural. As a regulatory instrument, hospital planning is supposed to lead to optimal hospital infrastruc-ture and efficiency. Our results indicate, however, that this is a questionable assumption that needs to be addressed by further research. However, Rosko (1999, 2001, 2004) and Rosko and Chilingerian (1999) found efficiency in the US hospital sector to be negatively related to market competition, a find-ing similar to that in our own study. Figure 1 shows the DEA efficiency scores and qual-ity-adjusted DEA efficiency scores as predicted val-ues (from the DEA II and DEA III models) versus hospital size and market concentration for the dif-ferent types of ownership. Generally, there was a linear relationship between hospital size and DEA efficiency scores and quality-adjusted DEA effi-ciency scores. Taking a closer look at the three different curves, it becomes clear that public hospitals outperformed their private for-profit and non-profit counterparts up to a size of approximately 1,000 beds. From 1,000 beds onwards, the private for-profit hospitals operated with greater efficiency, which also holds after taking quality into account. However, most private for-profit providers in Germany operate within a size range of 50 to 800 beds. Indeed, of the hospitals in our sample that had more than 1,000 beds, only 15 were private for-profit hospitals, whe-reas 119 were public hospitals. Generally, the relationship between market concen-tration and efficiency / quality-adjusted efficiency was also linear, although the public hospitals in our sample operated at a substantially higher level of efficiency in all of the various settings of competitive pressure (Figure 1). Private for-profit hospitals show a comparably low level of performance in very competitive markets.


127

Here, it is important to recognize that private for-profit and non-profit hospitals operate primarily in urban and other more competitive areas, whereas public hospitals operate both in urban and non-competitive regions (Figure 2). Furthermore, it is striking that in Figure 1, particu-lar in relation to market concentration, the effi-ciency curves and quality-adjusted efficiency curves of private for-profit hospitals are descending first and then raise linear. This descending curve can be explained by an interaction effect between private ownership, hospital size and market concentration. Those private for-profit hospitals operating in more competitive regions have comparably smaller enti-ties in these regions than hospitals of other owner-ship types. Therefore, the hospitals represented by the descending curve combine three characteristics that are associated with lower efficiency. If private for-profit hospitals operate in more concentrated regions the size of the entities approaches that of

other ownership types and thus differences between efficiency scores of different ownership types be-come smaller. As expected, other control variables also had a sig-nificant impact on efficiency (Table 4); the case-mix variables, in particular, had a major impact in this regard (see Appendix 2 for the full model specifica-tion). Our set of case-mix variables accounted for approximately two-thirds of the explained variance in the three full models (based on ordinary least square estimates), which indicates the importance of adjusting for patient heterogeneity. As expected, the operational efficiency of hospitals located in eastern Germany (EAST) was significantly greater than that of their western counterparts. This can be explained by the large investments made to mod-ernize hospital infrastructure in eastern Germany after the German reunification in 1990. In addition, hiring out a larger share of beds to ambulatory phy-sicians was associated with higher efficiency

Figure 1: DEA efficiency scores and quality-adjusted DEA efficiency scores as predicted values versus hospital size and market concentration

.55

.6.6

5.7

.75

Effi

cien

cy

0 .2 .4 .6 .8 1

Market concentration (HHI).5

5.6

.65

.7.7

5

0 500 1000 1500 2000

Hospital size (BEDS)

.62

.66

.7.7

4Q

ualit

y-ad

just

ed e

ffici

ency

0 .2 .4 .6 .8 1

Market concentration (HHI)

.62

.66

.7.7

4

0 500 1000 1500 2000


Private-FP Public Private-NP

.55

.6.6

5.7

.75

Effi

cien

cy

0 .2 .4 .6 .8 1

Market concentration (HHI).5

5.6

.65

.7.7

5

0 500 1000 1500 2000


.62

.66

.7.7

4Q

ualit

y-ad

just

ed e

ffici

ency

0 .2 .4 .6 .8 1


.62

.66

.7.7

4

0 500 1000 1500 2000


Private-FP Public Private-NP


128

(HIRED BEDS), and the existence of ambulatory care activities was associated with lower efficiency (AMBULATORY). Several of our findings are likely to be thought-provoking, because they are counter-intuitive and are not in line with the arguments put forward by authors in the field of agency theory and property-rights theory, as well as public choice theory. This may be explained by a number of specific character-istics of the hospital market in Germany. Our results show that for-profit status was associated with lower efficiency. However, for-profit hospitals may have found a different way to maximize their profits (i.e. financial surplus) than hospitals with other forms of ownership. Indeed, they may seek to maximize their profits by maximizing revenues instead of minimiz-ing inputs at a given output, which was defined as a measure of efficiency in our study. Wörz (2008) supports this view, having found that private for-profit hospitals (and especially hospital chains) were able to generate significantly higher revenues per

case on the average than hospitals with other forms of ownership. This study of a large German hospital sample (n = 1,614) with data from 2004 reflects the pre-DRG era in Germany, during which prices could be negotiated. However, even after the introduction of DRGs, today there are still a substantial number of additional reimbursement components being paid on the top of DRGs that can be negotiated at the hospital level (e.g., certain expensive drugs). Indeed, these additional components account for approximately 20 % of total reimbursements for non-psychiatric inpatient care (Schreyögg, Tie-mann, and Busse 2006). Shen, Eggleston, Lau, and Schmid (2007) found comparable results for the US hospital sector, concluding that the mission of pri-vate for-profit hospitals puts greater emphasis on earning profits (i.e. higher revenues per case due to higher prices) compared to public hospitals, which focus primarily on efficiency. Private for-profit providers are more efficient among the very large hospitals with more than

Figure 2: Histogram with hospital density versus market concentration

01

23

Den

sity

0 .2 .4 .6 .8 1

Private-NP

01

23

Den

sity

0 .2 .4 .6 .8 1

Public

01

23

Den

sity

0 .2 .4 .6 .8 1

Private-FP


01

23

Den

sity

0 .2 .4 .6 .8 1

Private-NP

01

23

Den

sity

0 .2 .4 .6 .8 1

Public

01

23

Den

sity

0 .2 .4 .6 .8 1

Private-FP



129

1,000 beds. This finding is in line with the theories mentioned above. Private for-profit providers have stronger incentive schemes and control mecha-nisms, which are of crucial importance for the man-agement of larger organizations (Bishop and Thompson 1992). Our results also indicate that private for-profit hos-pitals provide a higher quality of care compared to other types of ownership. As mentioned above, theoretical approaches assume that information asymmetries exist in the hospital market and thus particularly for-profit hospitals have the incentive (i.e. profit-seeking) to increase productive efficiency at the expense of quality of care. However, in the German hospital sector information asymmetry has decreased over the last decade due to a variety of healthcare reforms aiming at quality assurance (e.g., mandatory publication of quality reports). Further-more, the mentioned theoretical approaches do not account for the strategic importance of quality of care in markets with substantial overcapacities (i.e. cutthroat competition). There is evidence that pri-vate for-profit hospitals (and especially private for-profit hospital chains) operating in more competi-tive regions have improved their quality manage-ment and hospital outcomes in order to attract pa-tients (Busse, Nimptsch, and Mansky 2009). Our study has a number of strengths plus adds value compared to previous approaches. First, it applies a more refined approach to investigate the effects of ownership on hospital efficiency in Germany. To our knowledge, it is the first study to examine the rela-tionship between ownership status and efficiency using a panel data approach based on bootstrapped DEA efficiency scores. Second, our panel (n = 1,046) is large and covers the majority of all German acute-care hospitals over five years, allowing us to control for serial correlation and providing greater statisti-cal power than in previous studies leading to more robust estimates. Third, the sample is rich, contain-ing information at the patient and hospital levels. Within our two step approach, this allowed us to control appropriately for case-mix and other envi-ronmental and organizational characteristics, and is likely to have yielded more consistent results. Fourth, this is the first ownership study in the hos-pital sector to incorporate quality measures as out-puts in DEA models. Our study also has several important limitations. First, additional inputs and outputs (e.g., ambula-tory cases as an additional output, or capital as an

additional input) would have helped us capture more of the resources required in and all of the out-put produced from the hospital production process. Considering the number of outpatient cases in addi-tion to inpatient cases is generally recommended in order to measure patient care output (Jacobs, Smith, and Street 2006). We intended to include a proxy for hospital outpatient activities (e.g., outpa-tient surgery). However, data inconsistencies and measurement errors did not allow us to use this information for further analysis. Including other explanatory factors in addition to environmental and organizational characteristics might have provided a better explanation of varia-tion in our estimates, thus potentially affecting our interpretation of the relationship between owner-ship and efficiency. Another limitation may be that this study uses mortality as the only indicator for quality of care. Moreover, we were not able to take account of hospitals that were privatized during the 3 years prior to our study period. Nevertheless, our study results gave no indication that the hospitals privatized during our study period were primarily inefficient units. Finally, our study employed only DEA, although it would have been possible to use SFA in addition to DEA. SFA was not included in the analysis, however, because it relies on assump-tions about the functional form of the production or cost frontier that we aimed to avoid. Further, Linna (1998) found that both methods yielded comparable results for individual efficiency.

5 Summary and outlook In this paper, we investigated the effects of owner-ship on hospital efficiency in Germany. Our findings show that public ownership was associated with significantly higher efficiency than other forms of ownership; private for-profit ownership, in particu-lar, was associated with lower efficiency. These key findings remained unchanged after conducting a number of sensitivity checks. Our results suggest that private for-profit hospitals place greater em-phasis on earning profits (i.e. higher revenues per case due to higher prices), whereas public hospitals, because of resource constraints, focus primarily on input efficiency. We also found a significant positive association between hospital size and efficiency, and that competitive pressure had a significant negative impact on hospital efficiency. From a strategic man-agement point of view, private for-profit hospitals


130

may be well advised to change their acquisition strategy in terms of choice of hospital size and loca-tion. In addition, the ongoing trend towards privati-zation in Germany may not be an appropriate way to ensure the best use of the scarce resources in the hospital sector, because public hospitals use rela-tively fewer resources than private for-profit hospi-tals. Additional longitudinal studies are thus needed to measure and compare the efficiency of privatized hospitals. This could be a fruitful way to gain a bet-ter understanding of the consequences of ongoing privatization in the hospital sector.

Acknowledgements The authors are grateful for the insightful comments by two anonymous BuR reviewers on previous ver-sions of the paper. Furthermore, we would like to thank the Research Data Centre of the Statistical Offices of the Länder for providing the data for our analysis (No. 842-2008). Our research was sup-ported within the Munich Center of Health Sciences (MC-Health) as part of LMUinnovativ.

Appendix 1: Correlation matrix for the input and output variables

Variables 1 2 3 4 5 6 7 8

1 - CLIN 1

2 - NURS 0.936 1

3 - MEDTECH 0.971 0.915 1

4 - ADMIN 0.950 0.893 0.940 1

5 - OTHER 0.870 0.890 0.872 0.874 1

6 - SUPPLIES 0.960 0.909 0.951 0.907 0.835 1

7 - INPATIENT 0.911 0.948 0.856 0.841 0.821 0.862 1

8 - 1-MORTALITY 0.084 0.067 0.088 0.073 0.074 0.093 0.075 1

Appendix 2: Correlation matrix for the organizational and environmental variables

Variables 1 2 3 4 5 6 7

1 - HHI 1

2 - BEDS 0.106 1

3 - EAST 0.359 0.101 1

4 - AMBULATORY 0.041 0.162 -0.072 1

5 - HIRED BEDS -0.041 -0.197 -0.168 -0.081 1

6 - TEACH 0.001 0.134 -0.060 0.018 -0.077 1

7 - CONVERSION 0.062 0.003 0.040 -0.035 -0.042 0.043 1


131

Appendix 3: Regression results for each model - full model specification

Independent vari-ables


Coeffi-cients

Standard error

z-value p-value Coeffi-cients

Standard error

z-value p-value Variance inflation factor

Coeffi-cients

Standard error

z-value p-value

Ownership

PUBLIC 0.019 0.007 2.850 0.004 0.023 0.007 3.380 0.001 0.018 0.007 2.720 0.007

PRIVATE-NP 0.009 0.007 1.340 0.180 0.010 0.007 1.450 0.147 0.006 0.006 0.890 0.376

PRIVATE-FP served as reference category

Organizational and environmental vari-ables

HHI 0.086 0.010 8.970 0.000 0.085 0.010 8.860 0.000 1.29 0.069 0.009 7.420 0.000

BEDS (in thousands) 0.061 0.010 6.380 0.000 0.085 0.011 8.040 0.000 1.62 0.064 0.010 6.500 0.000

EAST 0.012 0.006 2.020 0.043 0.012 0.006 2.060 0.040 1.34 0.024 0.006 4.220 0.000

AMBULATORY -0.018 0.005 -3.480 0.001 -0.019 0.005 -3.750 0.000 1.15 -0.028 0.005 -5.690 0.000

HIRED BEDS 0.049 0.017 2.950 0.003 0.054 0.016 3.250 0.001 1.12 0.038 0.016 2.380 0.017

TEACH 0.004 0.005 0.940 0.347 0.003 0.005 0.750 0.454 1.15 0.001 0.005 0.190 0.853

CONVERSION -0.002 0.009 -0.180 0.857 0.002 0.009 0.180 0.859 1.12 0.009 0.009 0.950 0.344


132

Appendix 3 continued: Regression results for each model - full model specification



Coeffi-cients

Standard error


Standard error


Coeffi-cients

Standard error

z-value p-value

Case-mix variables

AVAILAB. PET (yes/no)

-0.046 0.012 -4.020 0.000 0.008 0.014 0.530 0.599 1.08 0.010 0.014 0.730 0.466

AVAILAB. CARD. CATH. LAB. (yes/no)

0.027 0.006 4.370 0.000 0.024 0.006 3.820 0.000 1.68 0.020 0.006 3.320 0.001

AGE ≥ 65a -0.099 0.025 -3.890 0.000 -0.110 0.025 -4.370 0.000 2.04 -0.185 0.024 -7.610 0.000

ARTIFICIAL RES-PIRATIONa

-0.345 0.048 -7.180 0.000 -0.355 0.048 -7.330 0.000 1.19 -0.377 0.046 -8.230 0.000

SPECIAL FACILITY HIVa

-12.52 2.386 -5.250 0.000 -22.39 4.256 -5.260 0.000 1.74 -22.74 4.067 -5.590 0.000

SPECIAL FACILITY CYSTIC FYBROSISa

-3.519 2.802 -1.260 0.209 -3.343 2.885 -1.160 0.247 1.04 -4.633 2.790 -1.660 0.097

SPECIAL FACILITY BURN INJURYa

-10.15 2.287 -4.440 0.000 -11.15 2.280 -4.890 0.000 1.05 -12.11 2.195 -5.520 0.000

SPECIAL FACILITY BRAIN INJURYa

-0.128 0.074 -1.730 0.084 -0.122 0.073 -1.660 0.098 1.11 -0.143 0.071 -2.020 0.043

TRANSPLANTA-TIONa

-2.973 1.085 -2.740 0.006 -4.001 1.151 -3.480 0.001 1.06 -4.719 1.107 -4.260 0.000

ACUTE MYOCAR-DIAL INFARCTIONa

-2.410 0.689 -3.500 0.000 -2.235 0.686 -3.260 0.001 1.48 -3.125 0.663 -4.710 0.000

CONGESTIVE HEART FAILUREa

0.393 0.212 1.860 0.064 0.463 0.210 2.200 0.028 1.50 0.368 0.202 1.820 0.068


133




Coeffi-cients

Standard error


Standard error


Coeffi-cients

Standard error

z-value p-value

Case-mix variables

PERIPHERAL VAS-CULAR DISEASEa

-2.218 0.791 2.800 0.005 -1.766 0.790 -2.240 0.025 1.02 -1.996 0.765 -2.610 0.009

CEREBRAL VAS-CULAR ACCIDENTa

0.407 0.212 1.920 0.055 0.398 0.210 1.900 0.058 1.40 0.078 0.203 -0.390 0.699

DEMENTIAa -0.792 0.427 -1.860 0.063 -0.760 0.422 -1.800 0.072 1.20 0.556 0.417 1.330 0.183

PULMONARY DISEASEa

1.086 0.339 3.200 0.001 1.158 0.336 3.450 0.001 1.10 0.618 0.320 1.930 0.053

CONNECTIVE TIS-SUE DISORDERa

-0.469 0.094 -5.010 0.000 -0.455 0.093 -4.910 0.000 1.05 0.026 0.092 0.280 0.779

PEPTIC ULCERa 3.898 1.384 2.820 0.005 3.063 1.371 2.230 0.025 1.46 -1.795 1.323 -1.360 0.175

LIVER DISEASE a 0.795 0.494 1.610 0.108 1.028 0.500 2.050 0.040 1.33 0.721 0.465 1.550 0.121

DIABETESa 0.777 0.249 3.120 0.002 0.748 0.244 3.060 0.002 1.06 0.325 0.231 1.410 0.160

DIABETES COM-PLICATIONSa

-0.920 0.946 -0.970 0.331 -0.391 0.944 -0.410 0.679 1.11 -0.014 0.922 -0.020 0.988

PARAPLEGIAa -8.831 1.393 -6.340 0.000 -8.758 1.376 -6.360 0.000 1.08 -7.220 1.207 -5.980 0.000

RENAL DISEASEa -0.172 0.199 -0.860 0.387 -0.153 0.197 -0.780 0.438 1.17 0.327 0.211 1.550 0.121

CANCERa -0.715 0.419 -1.710 0.088 -0.299 0.427 -0.700 0.484 1.19 -1.032 0.413 -2.500 0.013

METASTATIC CAN-CERa

-0.373 0.352 -1.060 0.290 -0.490 0.350 -1.400 0.161 1.18 -1.072 0.339 -3.160 0.002

SEVERE LIVER DISEASEa

6.619 5.888 1.120 0.261 7.752 5.923 1.310 0.191 1.40 -0.591 5.731 -0.100 0.918


134




Coeffi-cients

Standard error


Standard error


Coeffi-cients

Standard error

z-value p-value

Case-mix variables

HIVa -166.8 128.2 -1.300 0.193 -138.6 130.5 -1.060 0.288 1.74 -144.0 126.4 -1.140 0.255

Years

YEAR 2003 0.017 0.006 2.720 0.006 0.017 0.006 2.750 0.006 1.61 0.014 0.006 2.300 0.021

YEAR 2004 0.022 0.006 3.480 0.000 0.027 0.006 4.280 0.000 1.68 0.026 0.006 4.230 0.000

YEAR 2005 0.024 0.006 3.750 0.000 0.030 0.006 4.710 0.000 1.83 0.036 0.006 5.710 0.000

YEAR 2006 0.052 0.007 8.010 0.000 0.054 0.007 8.280 0.000 2.04 0.040 0.006 6.290 0.000

INTERCEPT 0.576 0.010 58.03 0.000 0.567 0.010 57.06 0.000 0.662 0.010 68.53 0.000

PSEUDO-R² 0.146 0.161 0.157

a Share of all cases per hospital * p≤0.05; ** p≤0.01; *** p≤0.001


135

Appendix 4: Regression results – modification of the number of periods useda


YEARS 2002-2006 YEARS 2003-2006 YEARS 2004-2006 YEARS 2005-2006

Coeffi-cients

Standard error


Standard error


Standard error


Standard error

z-value p-value

Ownership

PUBLIC 0.023 0.007 3.380 0.001 0.021 0.007 2.880 0.004 0.019 0.009 2.150 0.031 0.020 0.011 1.870 0.062

PRIVATE-NP 0.010 0.007 1.450 0.147 0.008 0.007 1.080 0.279 0.007 0.009 0.760 0.446 0.008 0.011 0.710 0.477



HHI 0.085 0.010 8.860 0.000 0.084 0.011 7.910 0.000 0.085 0.012 6.990 0.000 0.091 0.015 6.130 0.000

BEDS (in thousands) 0.085 0.011 8.040 0.000 0.076 0.012 6.530 0.000 0.087 0.013 6.510 0.000 0.088 0.016 5.340 0.000

EAST 0.012 0.006 2.060 0.040 0.012 0.006 1.940 0.052 0.011 0.007 1.470 0.141 0.005 0.009 0.570 0.570

AMBULATORY -0.019 0.005 -3.750 0.000 -0.020 0.006 -3.480 0.001 -0.018 0.007 -2.660 0.008 -0.016 0.008 -1.960 0.050

HIRED BEDS 0.054 0.016 3.250 0.001 0.038 0.018 2.090 0.037 0.026 0.021 1.210 0.226 0.025 0.026 0.970 0.330

TEACH 0.003 0.005 0.750 0.454 0.006 0.005 1.210 0.227 0.007 0.005 1.340 0.179 0.008 0.005 1.440 0.150

CONVERSION 0.002 0.009 0.180 0.859 0.002 0.011 0.200 0.839 0.001 0.013 0.070 0.948 -0.003 0.016 -0.160 0.875


136

Appendix 4 continued: Regression results – modification of the number of periods useda



Coeffi-cients

Standard error


Standard error


Standard error


Standard error

z-value p-value

Case-mix variables


0.008 0.014 0.530 0.599 0.013 0.016 3.220 0.403 0.002 0.018 0.120 0.907 0.002 0.021 0.100 0.924


0.024 0.006 3.820 0.000 0.022 0.007 0.840 0.001 0.016 0.008 1.950 0.051 0.005 0.010 0.540 0.592

AGE ≥ 65b -0.110 0.025 -4.370 0.000 -0.109 0.028 -3.880 0.000 -0.151 0.033 -4.580 0.000 -0.167 0.039 -4.230 0.000

ARTIFICIAL RES-PIRATIONb

-0.355 0.048 -7.330 0.000 -0.331 0.049 -6.710 0.000 -0.338 0.059 -5.700 0.000 -0.365 0.087 -4.200 0.000

SPECIAL FACILITY HIVb

-22.39 4.256 -5.260 0.000 -9.821 2.965 -3.310 0.001 -28.92 6.291 -4.600 0.000 -33.98 8.217 -4.140 0.000

SPECIAL FACILITY CYSTIC FYBROSISb

-3.343 2.885 -1.160 0.247 -2.067 3.203 -0.650 0.519 -2.015 3.923 -0.510 0.608 0.062 5.208 0.010 0.990

SPECIAL FACILITY BURN INJURYb

-11.15 2.280 -4.890 0.000 -13.02 2.624 -4.960 0.000 -13.49 3.104 -4.350 0.000 -10.84 4.183 -2.590 0.010

SPECIAL FACILITY BRAIN INJURYb

-0.122 0.073 -1.660 0.098 -0.119 0.074 -1.600 0.109 -0.102 0.075 -1.360 0.173 -0.093 0.077 -1.210 0.228

TRANSPLANTA-TIONb

-4.001 1.151 -3.480 0.001 -3.813 1.144 -3.330 0.001 -3.709 1.178 -3.150 0.002 -2.655 1.725 -1.540 0.124

ACUTE MYOCAR-DIAL INFARCTIONb

-2.235 0.686 -3.260 0.001 -2.384 0.748 -3.190 0.001 -2.188 0.906 -2.410 0.016 -1.684 1.303 -1.290 0.196

CONGESTIVE HEART FAILUREb

0.463 0.210 2.200 0.028 0.568 0.283 2.010 0.045 1.622 0.458 3.540 0.000 2.112 0.599 3.530 0.000


137




Coeffi-cients

Standard error


Standard error


Standard error


Standard error

z-value p-value

Case-mix variables

PERIPHERAL VAS-CULAR DISEASEb

-1.766 0.790 -2.240 0.025 -2.307 0.974 -2.370 0.018 -2.328 1.323 -1.760 0.079 -5.261 2.437 -2.160 0.031

CEREBRAL VAS-CULAR ACCIDENTb

0.398 0.210 1.900 0.058 0.319 0.216 1.480 0.139 0.204 0.247 0.820 0.410 0.138 0.290 0.480 0.633

DEMENTIAb -0.760 0.422 -1.800 0.072 -0.887 0.457 -1.940 0.052 -0.723 0.500 -1.450 0.148 -0.823 0.573 -1.440 0.151

PULMONARY DISEASEb

1.158 0.336 3.450 0.001 1.140 0.385 2.960 0.003 1.659 0.495 3.350 0.001 1.199 0.554 2.160 0.031

CONNECTIVE TIS-SUE DISORDERb

-0.455 0.093 -4.910 0.000 -0.416 0.101 -4.110 0.000 -0.400 0.113 -3.520 0.000 -0.402 0.141 -2.860 0.004

PEPTIC ULCERb 3.063 1.371 2.230 0.025 2.783 1.496 1.860 0.063 3.217 1.722 1.870 0.062 4.739 2.141 2.210 0.027

LIVER DISEASEb 1.028 0.500 2.050 0.040 1.172 0.588 2.000 0.046 0.710 0.708 1.000 0.316 0.430 0.951 0.450 0.651

DIABETESb 0.748 0.244 3.060 0.002 0.656 0.265 2.480 0.013 0.580 0.309 1.880 0.060 0.750 0.490 1.530 0.126

DIABETES COM-PLICATIONSb

-0.391 0.944 -0.410 0.679 -0.131 1.286 -0.100 0.919 0.780 1.497 0.520 0.602 1.215 1.767 0.690 0.492

PARAPLEGIAb -8.758 1.376 -6.360 0.000 -7.137 1.253 -5.700 0.000 -6.023 1.312 -4.590 0.000 -4.660 1.413 -3.300 0.001

RENAL DISEASEb -0.153 0.197 -0.780 0.438 -0.171 0.284 -0.600 0.548 -1.305 1.445 -0.900 0.366 -4.263 2.582 -1.650 0.099

CANCERb -0.299 0.427 -0.700 0.484 -0.738 0.387 -1.910 0.057 -0.492 0.464 -1.060 0.289 -0.528 0.652 -0.810 0.418

METASTATIC CAN-CERb

-0.490 0.350 -1.400 0.161 -0.370 0.389 -0.950 0.342 -0.264 0.467 -0.570 0.571 -0.690 0.741 -0.930 0.352

SEVERE LIVER DISEASEb

7.752 5.923 1.310 0.191 7.262 6.534 1.110 0.266 5.102 7.798 0.650 0.513 6.268 12.99 0.480 0.629


138




Coeffi-cients

Standard error


Standard error


Standard error


Standard error

z-value p-value

Case-mix variables

HIVb -138.6 130.5 -1.060 0.288 -15.76 71.84 -0.220 0.826 -65.83 79.80 -0.830 0.409 -19.65 108.9 -0.180 0.857

Years

YEAR 2003 0.017 0.006 2.750 0.006

YEAR 2004 0.027 0.006 4.280 0.000 0.010 0.006 1.630 0.103

YEAR 2005 0.030 0.006 4.710 0.000 0.012 0.006 1.960 0.050 0.003 0.006 0.430 0.666

YEAR 2006 0.054 0.007 8.280 0.000 0.036 0.006 5.580 0.000 0.027 0.006 4.110 0.000 0.024 0.006 3.950 0.000

INTERCEPT 0.567 0.010 57.06 0.000 0.591 0.011 54.25 0.000 0.603 0.013 47.94 0.000 0.606 0.015 40.33 0.000

PSEUDO-R² 0.161 0.136 0.142 0.142

a DEA II - Efficiency model including hospitals with beds ≥50, without university hospitals. b Share of all cases per hospital.


139

Appendix 5: Regression results - modification in terms of hospital sizea


BEDS>30 BEDS>50 BEDS>70 BEDS>100

Coeffi-cients

Standard error


Standard error


Standard error


Standard error

z-value p-value

Ownership

PUBLIC 0.028 0.006 4.340 0.000 0.023 0.007 3.380 0.001 0.022 0.006 3.510 0.000 0.027 0.007 3.780 0.000

PRIVATE-NP 0.012 0.006 1.840 0.065 0.010 0.007 1.450 0.147 0.005 0.006 0.750 0.456 0.016 0.007 2.190 0.029



HHI 0.076 0.009 8.180 0.000 0.085 0.010 8.860 0.000 0.070 0.009 7.490 0.000 0.046 0.010 4.790 0.000

BEDS (in thousands) 0.109 0.011 10.16 0.000 0.085 0.011 8.040 0.000 0.024 0.021 1.160 0.247 0.058 0.021 2.700 0.007

EAST 0.022 0.006 3.910 0.000 0.012 0.006 2.060 0.040 0.013 0.006 2.270 0.023 0.026 0.006 4.300 0.000

AMBULATORY -0.010 0.005 -1.980 0.047 -0.019 0.005 -3.750 0.000 -0.019 0.005 -3.970 0.000 -0.016 0.005 -3.060 0.002

HIRED BEDS 0.072 0.017 4.340 0.000 0.054 0.016 3.250 0.001 0.060 0.016 3.730 0.000 0.171 0.022 7.650 0.000

TEACH 0.010 0.005 2.160 0.031 0.003 0.005 0.750 0.454 0.010 0.005 2.150 0.031 0.021 0.005 4.700 0.000

CONVERSION 0.001 0.009 0.070 0.942 0.002 0.009 0.180 0.859 0.001 0.009 0.130 0.897 0.014 0.009 1.490 0.136


140

Appendix 5 continued: Regression results - modification in terms of hospital sizea



Coeffi-cients

Standard error


Standard error


Standard error


Standard error

z-value p-value

Case-mix variables


0.009 0.014 0.630 0.527 0.008 0.014 0.530 0.599 0.026 0.014 1.810 0.070 0.020 0.014 1.470 0.142


0.026 0.006 4.270 0.000 0.024 0.006 3.820 0.000 0.013 0.006 2.210 0.027 0.012 0.006 1.980 0.048

AGE ≥ 65b -0.196 0.026 -7.640 0.000 -0.110 0.025 -4.370 0.000 -0.124 0.020 -6.040 0.000 -0.387 0.033 -11.70 0.000

ARTIFICIAL RES-PIRATIONb

-0.264 0.043 -6.070 0.000 -0.355 0.048 -7.330 0.000 -0.463 0.049 -9.380 0.000 -0.201 0.082 -2.460 0.014

SPECIAL FACILITY HIVb

-4.945 1.890 -2.620 0.009 -22.39 4.256 -5.260 0.000 -23.91 4.180 -5.720 0.000 -26.11 3.942 -6.620 0.000

SPECIAL FACILITY CYSTIC FYBROSISb

-2.711 2.806 -0.970 0.334 -3.343 2.885 -1.160 0.247 -4.083 2.818 -1.450 0.147 -4.324 2.645 -1.630 0.102

SPECIAL FACILITY BURN INJURYb

-11.05 2.274 -4.860 0.000 -11.15 2.280 -4.890 0.000 -8.842 2.246 -3.940 0.000 -10.76 2.129 -5.050 0.000

SPECIAL FACILITY BRAIN INJURYb

-0.191 0.073 -2.610 0.009 -0.122 0.073 -1.660 0.098 -0.104 0.068 -1.530 0.127 -0.869 0.286 -3.040 0.002

TRANSPLANTA-TIONb

-4.183 1.144 -3.660 0.000 -4.001 1.151 -3.480 0.001 -4.297 1.081 -3.970 0.000 -4.570 1.013 -4.510 0.000

ACUTE MYOCAR-DIAL INFARCTIONb

-2.091 0.660 -3.170 0.002 -2.235 0.686 -3.260 0.001 0.000 0.000 -0.560 0.000 0.000 0.000 -1.360 0.173

CONGESTIVE HEART FAILUREb

0.246 0.208 1.180 0.237 0.463 0.210 2.200 0.028 0.000 0.000 2.660 0.008 0.000 0.000 5.590 0.000


141




Coeffi-cients

Standard error


Standard error


Standard error


Standard error

z-value p-value

Case-mix variables -1.791 0.782 -2.290 0.022 -1.766 0.790 -2.240 0.025 0.000 0.000 -0.460 0.649 0.000 0.000 0.120 0.906

PERIPHERAL VAS-CULAR DISEASEb

0.236 0.202 1.170 0.243 0.398 0.210 1.900 0.058 0.000 0.000 1.380 0.168 0.000 0.000 2.570 0.010

CEREBRAL VAS-CULAR ACCIDENTb

-1.509 0.420 -3.590 0.000 -0.760 0.422 -1.800 0.072 0.000 0.000 -3.870 0.000 -0.001 0.000 -6.210 0.000

DEMENTIAb 0.457 0.326 1.400 0.161 1.158 0.336 3.450 0.001 0.000 0.000 4.120 0.000 0.000 0.000 3.940 0.000

PULMONARY DISEASEb

-0.695 0.100 -6.970 0.000 -0.455 0.093 -4.910 0.000 0.000 0.000 -5.630 0.000 0.000 0.000 -4.940 0.000

CONNECTIVE TIS-SUE DISORDERb

7.734 1.345 5.750 0.000 3.063 1.371 2.230 0.025 0.000 0.000 2.420 0.015 0.001 0.000 5.160 0.000

PEPTIC ULCERb 0.516 0.465 1.110 0.267 1.028 0.500 2.050 0.040 0.000 0.000 2.940 0.003 0.000 0.000 1.600 0.111

LIVER DISEASEb -0.217 0.360 -0.600 0.547 0.748 0.244 3.060 0.002 0.000 0.000 1.970 0.048 0.000 0.000 1.560 0.118

DIABETESb -0.624 0.931 -0.670 0.503 -0.391 0.944 -0.410 0.679 0.000 0.000 0.480 0.632 0.000 0.000 0.430 0.671

DIABETES COM-PLICATIONSb

-12.21 1.452 -8.410 0.000 -8.758 1.376 -6.360 0.000 -0.002 0.000 -9.180 0.000 -0.003 0.000 -10.15 0.000

PARAPLEGIAb -0.133 0.196 -0.680 0.498 -0.153 0.197 -0.780 0.438 0.000 0.000 -2.020 0.043 0.000 0.000 -2.050 0.041

RENAL DISEASEb -0.693 0.324 -2.140 0.033 -0.299 0.427 -0.700 0.484 0.000 0.000 1.120 0.265 0.000 0.000 0.060 0.955

CANCERb -1.299 0.335 -3.880 0.000 -0.490 0.350 -1.400 0.161 0.000 0.000 -2.230 0.026 0.000 0.000 -3.420 0.001

METASTATIC CAN-CERb

2.741 5.592 0.490 0.624 7.752 5.923 1.310 0.191 0.001 0.001 1.890 0.058 0.002 0.001 2.840 0.005

SEVERE LIVER DISEASEb

-1.791 0.782 -2.290 0.022 -1.766 0.790 -2.240 0.025 0.000 0.000 -0.460 0.649 0.000 0.000 0.120 0.906


142




Coeffi-cients

Standard error


Standard error


Standard error


Standard error

z-value p-value

Case-mix variables

HIVb -34.45 66.75 -0.520 0.606 -138.6 130.5 -1.060 0.288 -0.026 0.010 -2.680 0.007 -0.034 0.010 -3.530 0.000

Years

YEAR 2003 -0.035 0.006 -5.920 0.000 0.017 0.006 2.750 0.006 0.015 0.006 2.540 0.011 0.013 0.006 2.160 0.030

YEAR 2004 0.027 0.006 4.450 0.000 0.027 0.006 4.280 0.000 0.026 0.006 4.190 0.000 0.032 0.006 5.040 0.000

YEAR 2005 0.016 0.006 2.610 0.009 0.030 0.006 4.710 0.000 0.030 0.006 4.790 0.000 0.040 0.007 6.110 0.000

YEAR 2006 0.066 0.006 10.39 0.000 0.054 0.007 8.280 0.000 0.054 0.006 8.520 0.000 0.069 0.007 10.33 0.000

INTERCEPT 0.560 0.010 58.21 0.000 0.567 0.010 57.060 0.000 0.580 0.009 61.16 0.000 0.581 0.012 49.21 0.000

PSEUDO-R² 0.188 0.161 0.158 0.206

a DEA II - Efficiency model including hospitals with beds ≥50, without university hospitals. b Share of all cases per hospital.


143

References

Arrow, Kenneth Joseph (1963): Uncertainty and the welfare economics of medical care, American Economic Review, 53 (5): 941-973.

Banker, Rajiv D., Abraham Charnes, William W. Cooper, John Swarts, and Dave Thomas (1989): An introduction to data envel-opment analysis with some of its models and their uses, Research in Government and Nonprofit Accounting, 5: 125-163.

Banker, Rajiv D., Abraham Charnes, and William W. Cooper (1984): Some models for estimating technical and scale inefficien-cies in data envelopment analysis, Management Science, 30 (9): 1078-1092.

Bishop, Matthew and David Thompson (1992): Regulatory reform and productivity growth in the UK's public utilities, Applied Economics, 24 (11): 1181-1190.

Breyer, Friedrich (1987): The specification of a hospital cost function. A comment on the recent literature, Journal of Health Economics, 6 (2): 147-157.

Brown, H. Shelton (2003): Managed care and technical efficiency, Health Economics, 12 (2): 149-158.

Buchanan, James M. and Robert D. Tollison (1972): Theory of public choice, University of Michigan Press: Ann Arbor.

Burgess, James F., Jr. and Paul W. Wilson (1996): Hospital own-ership and technical inefficiency, Management Science, 42 (1): 110-123.

Burgess, James F., Jr. and Paul W. Wilson (1998): Variation in inefficiency among US hospitals, Canadian Journal of Opera-tional Research and Information Processing, 36 (3): 84–102.

Busse, Reinhard and Annette Riesberg (2004): Health care sys-tems in transition: Germany, European observatory on health care systems: Copenhagen.

Busse, Reinhard, Ulrike Nimptsch, and Thomas Mansky (2009): Measuring, Monitoring, and Managing Quality in Germany’s Hospitals. Health Affairs, 28 (2): 294-304.

Button, Kenneth J. and Thomas G. Weyman-Jones (1992): Own-ership structure, institutional organization and measured X-efficiency, American Economic Review, 82 (2): 439-445.

Carey, Kathleen (2000): A multilevel modelling approach to analysis of patient costs under managed care, Health Economics, 9 (5): 435-446.

Carey, Kathleen (2002): Hospital Length of Stay and Cost: A Multilevel Modeling Analysis, Health Services and Outcomes Research Methodology, 3 (1): 41-56.

Carey, Kathleen and James F. Burgess, Jr. (1999): On measuring the hospital cost/quality trade-off, Health Economics, 8 (6): 509-520.

Carr, John W. and Paul J. Feldstein (1967): The relationship of cost to hospital size, Inquiry, 4 (2): 45-65.

Chang, Hsihui, Wen- Jing Chang, Somnath Das, and Shu- Hsing Li (2004): Health care regulation and the operating efficiency of hospitals: Evidence from Taiwan, Journal of Accounting and Public Policy, 23 (6): 483-510.

Chirikos, Thomas N. and Alan M. Sear (1994): Technical effi-ciency and the competitive behaviour of hospitals, Socio-Economic Planning Sciences, 28 (4): 219-227.

Chirikos, Thomas N. and Alan M. Sear (2000): Measuring hospi-tal efficiency: a comparison of two approaches, Health Services Research, 34 (6): 1389-1408.

Clarkson, Kenneth (1972): Some Implications of Property Rights in Hospital Management, Journal of Law and Economics, 15 (2): 363-384.

Cooper, William W., Lawrence M. Seiford, and Joe Zhu (2004): Sensitivity analysis in DEA, in: William W. Cooper, Lawrence M. Seiford, and Joe Zhu (eds.): Handbook on data envelopment analysis, Kluwer Academic: Boston, 75-97.

Deily, Mary E. and Niccie L. McKay (2006): Cost inefficiency and mortality rates in Florida hospitals, Health Economics, 15 (4): 419-431.

Dudley, R. Adams, Kirsten L. Johansen, Richard Brand, Deborah J. Rennie, and Arnold Milstein (2000): Selective referral to high-volume hospitals: estimating potentially avoidable deaths, The Journal of the American Medical Association, 283 (9): 1159-1166.

Dyson, R.G., R. Allen, A.S. Camanho, V.V. Podinovski, C.S. Sar-rico, and E.A. Shale (2001): Pitfalls and protocols in DEA, Euro-pean Journal of Operational Research, 132 (2): 245-259.

Fama, Eugene F. and Michael C. Jensen (1983): Separation of ownership and control, The Journal of Law and Economics, 26 (2): 301-325.

Feess, Eberhard and Sonja Ossig (2007): Reimbursement schemes for hospitals, malpractice liability, and intrinsic motiva-tion, International Review of Law and Economics, 27 (4): 423-441.

Feldstein, Martin Stuart (1967): Economic Analysis for Health Service Efficiency, North-Holland: Amsterdam.

Ferrier, Gary D. and Vivian G. Valdmanis (1996): Rural hospital performance and its correlates, The Journal of Productivity Analysis, 7 (1): 63-80.

Forschungsdatenzentrum der Statistischen Landesämter (2008): Krankenhausstatistik (Teil I: Grunddaten, Teil II: Diagnosen und Teil III: Kostennachweis) der Jahre 2002–2006, Antrag 842-2008.

Foster, Richard W. (1974): Economic Models of the Hospital, Hospital Administration, 19 (4): 87-93.

Frey, Bruno S. and Reto Jegen (2001): Motivation crowding theory, Journal of Economic Surveys, 15 (5): 589-611.

Fried, Harold O., C.A. Knox Lovell, Shelton S. Schmidt, and Sut-hathip Yaisawarng (2002): Accounting for environmental effects and statistical noise in data envelopment analysis, Journal of Productivity Analysis, 17 (1-2): 157-174.

Garnick, Deborah W., Harold S. Luft, James C. Robinson, and Janice Tetreault (1987): Appropriate measures of hospital market areas, Health Services Research, 22 (1): 69-89.

Hall, Peter D. (1995): Theories and Institutions, Nonprofit & Voluntary Sector Quarterly, 24 (1), 5-13.

Hansmann, Henry (1988): Ownership of the Firm, Journal of Law, Economics, & Organization, 4 (2): 267-304.

Harrison, Jeffrey P., M. Nicholas Coppola, and Mark Wakefield (2004): Efficiency of federal hospitals in the United States, Jour-nal of Medical Systems, 28 (5): 411-422.


144

Helmig, Bernd and Irvine Lapsley (2001): On the efficiency of public, welfare and private hospitals in Germany over time: a sectoral data envelopment analysis study, Health Services Man-agement Research, 14 (4): 263-274.

Herr, Annika (2008): Cost and technical efficiency of German hospitals: does ownership matter?, Health Economics, 17 (9): 1057-1071.

Hollingsworth, Bruce (2003): Non-parametric and parametric applications measuring efficiency in health care, Health Care Management Science, 6 (4): 203-218.

Hollingsworth, Bruce (2008): The measurement of efficiency and productivity of health care delivery, Health Economics, 17 (10): 1107-1128.

Jacobs, Philip (1974): A survey of economic models of hospitals, Inquiry, 11 (2): 83-97.

Jacobs, Philip and John Rapoport (2003): The economics of health and medical care, Jones and Bartlett: Sudbury, MA.

Jacobs, Rowena, Peter C. Smith, and Andrew Street (2006): Measuring efficiency in health care: analytic techniques and health policy, Cambridge University Press: Cambridge.

Koop, Gary, Jacek Osiewalski, and Mark F.J. Steel (1997): Bayes-ian efficiency analysis through individual effects: hospital cost frontiers, Journal of Econometrics, 76 (1-2): 77-105.

Lee, Maw Lin (1971): A conspicuous production theory of hospital behavior, Southern Economic Journal, 38 (1): 48-58.

Leibenstein, Harvey (1966): Allocative efficiency vs. X-efficiency, American Economic Review, 56 (3): 392-415.

Linna, Miika (1998): Measuring hospital cost efficiency with panel data models, Health Economics, 7 (5): 415-427.

McKay, Niccie, Mary E. Deily, and Fred H. Dorner (2003): Own-ership and changes in hospital inefficiency, 1866-1991, Inquiry, 39 (4): 388-399.

Megginson, William L., Robert C. Nash, and Matthias van Ran-denborgh (1994): The financial and operating performance of newly privatized firms: an international empirical analysis, Jour-nal of Finance, 49 (2): 403-452.

Morey, Richard C., David J. Fine, Stephen W. Loree, Donna L. Retzlaff-Roberts, and Shigeru Tsubakitani (1992): The trade-off between hospital cost and quality of care. An exploratory empiri-cal analysis, Medical Care, 30 (8): 677-698.

Newhouse, Joseph P. (1970): Toward a theory of nonprofit Insti-tutions: an economic model of a hospital, American Economic Review, 60 (1): 64-74.

Niskanen, William A. (1975): Bureaucrats and politicians, Journal of Law and Economics, 18 (3): 617-643.

Osterloh, Margit and Bruno S. Frey (2000): Motivation, knowl-edge transfer, and organizational forms, Organization Science, 11 (5): 538-550.

Ozcan, Yasar A., Roice D. Luke, and Cengiz Haksever (1992): Ownership and organizational performance: a comparison of technical efficiency across hospital types, Medical Care, 30 (9): 781-794.

Pauly, Mark V. and Michael S. Redisch (1973): The Not-for-Profit Hospital as a Physician's Cooperative, The American Economic Review, 63 (1): 87-99.

Pestieau, Pierre and Henry Tulkens (1990): Assessing the per-formance of public sector activities: some recent evidence from the productive efficiency viewpoint, Discussion Paper, Center for Operations Research and Econometrics, Université catholique de Louvain.

Picone, Gabriel, Shin-Yi Chou and Frank Sloan (2002): Are for-profit hospital conversions harmful to patients and to medicare?, RAND Journal of Economics, 33 (3): 507-523.

Pilyavsky, Anatoly I., William E. Aaronson, Patrick M. Bernet, Michael D. Rosko, Vivian G. Valdmanis and Mikhail V. Golubchi-kov (2006): East-west: does it make a difference to hospital effi-ciencies in Ukraine?, Health Economics, 15 (11): 1173-1186.

Rathgeb Smith, Steven and Kirsten A. Gronbjerg (2006): Scope and theory of government-nonprofit relations, in: Walter W. Powell and Richard Steinberg (eds.): The nonprofit sector: a research handbook, Yale University Press: New Haven, 221-242.

Rees, Ray (1988): Inefficiency, public enterprise and privatisation, European Economic Review, 32 (2-3): 422-431.

Reder, Melvin W. (1965): Some problems in the economics of hospitals, The American Economic Review, 55 (1-2): 472-480.

Rice, Rip G. (1966): Analysis of the hospital as an economic or-ganism, Modern Hospital, 106 (4): 87-91.

Rosko, Michael D. (1999): Impact of internal and external envi-ronmental pressures on hospital inefficiency, Health Care Man-agement Science, 2 (2): 63-74.

Rosko, Michael D. (2001): Cost efficiency of US hospitals: a sto-chastic frontier approach, Health Economics, 10 (6): 539-551.

Rosko, Michael D. (2004): Performance of U.S. teaching hospi-tals: a panel analysis of cost inefficiency, Health Care Manage-ment Science, 7 (1): 7-16.

Rosko, Michael D. and Jon A. Chilingerian (1999): Estimating hospital inefficiency: does case mix matter?, Journal of Medical Systems, 23 (1): 57-71.

Sari, Nazmi (2003): Efficiency outcomes of market concentration and managed care, International Journal of Industrial Organiza-tion, 21 (10): 1571–1589.

Scherer, Frederic M. and David Ross (1990): Industrial market structure and economic performance, Houghton Mifflin: Boston.

Schreyögg, Jonas, Oliver Tiemann, and Reinhard Busse (2006): Cost accounting to determine prices: how well do prices reflect costs in the German DRG-system?, Health Care Management Science, 9 (3): 269-279.

Schreyögg, Jonas and Constantin von Reitzenstein (2008): Stra-tegic groups and performance differences among academic medi-cal centers, Health Care Management Review, 33 (3): 225-233.

Shen, Yu-Chu, Karen Eggleston, Joseph Lau, and Christopher Schmid (2007): Hospital ownership and financial performance: what explains the different findings in the empirical literature?, Inquiry, 44 (1): 41-68.

Shleifer, Andrei and Robert W. Vishny (1994): Politicians and Firms, The Quarterly Journal of Economics, 109 (4): 995-1025.

Simar, Léopold and Paul W. Wilson (1998): Sensitivity analysis of efficiency scores: How to bootstrap in nonparametric frontier models, Management Science, 44 (1): 49-61.


145

Simar, Léopold and Paul W. Wilson (2000): A general methodol-ogy for bootstrapping in non-parametric frontier models, Journal of Applied Statistics, 27 (6): 779-802.

Simar, Léopold and Paul W. Wilson (2007): Estimation and inference in two-stage, semi-parametric models of production processes, Journal of Econometrics, 136 (1): 31-64.

Sloan, Frank A. (1976): Physician fee inflation: evidence from the late 1960's, in: Richard N. Rosett (ed.): The role of health insur-ance in the health services sector, National Bureau of Economic Research: New York, 321-362

Sloan, Frank A. (2000): Not-for-profit ownership and hospital behaviour, in: Anthony J. Culyer and Joseph P.Newhouse (eds.): Handbook of health economics, Elsevier: Amsterdam, 1141-1174.

Staat, Matthias and Maik Hammerschmidt (2000): Benchmark-ing the health sector in Germany – an application of Data Envel-opment Analysis, Research Paper, Institute for Market-Oriented Management, University Mannheim.

Staat, Matthias (2006): Efficiency of hospitals in Germany: a DEA-bootstrap approach, Applied Economics, 38 (19): 2255-2263.

Steinberg, Richard (2006): Economic theories of nonprofit or-ganizations, in: Walter W. Powell and Richard Steinberg (eds.): The nonprofit sector: a research handbook, Yale University Press: New Haven: 117-139.

Steinmann, Lukas, Gunnar Dittrich, Alexander Karmann, and Peter Zweifel (2004): Measuring and comparing the (in)efficiency of German and Swiss hospitals, The European Journal of Health Economics, 5 (3): 216-226.

Sundararajan, Vijaya, Toni Henderson, Catherine Perry, Amanda Muggivan, Hude Quan, and William A. Ghali (2004): New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality, Journal of Clinical Epidemiology, 57 (12): 1288-1294.

Syrjänen, Mikko J. (2004): Non-discretionary and discretionary factors and scale in data envelopment analysis, European Journal of Operational Research, 158 (1): 20–33.

Villalonga, Belen (2000): Privatization and efficiency: differentiat-ing ownership effects from political, organizational, and dynamic effects, Journal of Economic Behavior and Organization, 42 (1): 43-74.

Weisbrod, Burton A. (1988): The Nonprofit Economy, Harvard University Press: Cambridge, MA.

Wörz, Markus (2008): Erlöse - Kosten - Qualität: Macht die Krankenhausträgerschaft einen Unterschied?, VS-Verlag: Wies-baden.

Zismer, Daniel K. (1999): Physician compensation arrange-ments: management & legal trends, Aspen Publishers: Gaithers-burg.

Zuckerman, Stephen, Jack Hadley, and Lisa Iezzoni (1994): Mea-suring hospital efficiency with frontier cost functions, Journal of Health Economics, 13 (3): 255–280.

Biographies

Oliver Tiemann, MA, PhD, is a senior research fellow at the Munich School of Management, Ludwig-Maximilians-Universität München and Helmholtz Zentrum München, Germany. He re-ceived his doctoral degree in Business Administration in 2008 from Berlin University of Technology. His research interests include health care management, performance measurement and organizational behavior.

Jonas Schreyögg, MA, PhD, is professor for Health Services Management at the Munich School of Management, Ludwig-Maximilians-Universität München and Head of the Department for Health Services Management at Helmholtz Zentrum Mün-chen, Germany. At the same time he is an affiliated researcher at the Stanford University’s Center for Primary care and Outcomes Research. He received his doctoral degree in Economics in 2003 and his Habilitation in Business Administration and Health Economics in 2008, both from Berlin University of Technology.


Thin Capitalization Rules and Entrepreneural Capital StructureDecisions Alexandra Maßbaum, Caren Sureth


BuR -- Business ResearchOfficial Open Access Journal of VHBVerband der Hochschullehrer für Betriebswirtschaft e.V.Volume 2 | Issue 2 | December 2009 | 147--169

Thin Capitalization Rules andEntrepreneurial Capital Structure DecisionsAlexandra Maßbaum, Faculty of Business Administration and Economics, University of Paderborn, Germany, E-mail:[email protected]

Caren Sureth, Faculty of Business Administration and Economics, University of Paderborn, Germany, E-mail: [email protected]

AbstractTax planners often choose debt over equity financing. As this has led to increased corporate debtfinancing, many countries have introduced thin capitalization rules to secure their tax revenues. In ageneral capital structure model we analyze if thin capitalization rules affect dividend and financingdecisions, and whether they can partially explain why corporations receive both debt and equitycapital. We model the Belgian, German and Italian rules as examples. We find that the so-called Millerequilibrium and definite financing effects depend significantly on the underlying tax system. Further,our results are useful for the treasury to decide what thin capitalization type to implement.

Keywords: business taxation, capital structure, critical income tax rate, financing decision, Millerequilibrium, tax planning, thin capitalization

Manuscript received March 19, 2009, accepted by Rainer Niemann (Accounting) November 2,2009.

1 IntroductionSeen from a tax perspective it is often attractivefor shareholders of corporations to provide capi-tal as debt instead of equity capital. This has ledto an extensive enlargement of debt financing ofcorporations (OECD 1987: 8 et seq.). To protecttax revenues, the legislators of many countrieshave reacted to this development by increasinglyimplementing so-called thin capitalization rules(e.g., in Italy and the Netherlands, see Gouthière2005 for an overview of different thin capitaliza-tion rules) or by tightening existing rules (e.g., inGermany, Denmark, the UK, Spain, and France).Thin capitalization rules are regulations that limitthe corporate tax deductibility of interest paid toshareholders.Moreover, we observe that corporations issueshares as well as debt. This raises the questionof whether thin capitalization rules cancel out thetax shield of debt financing, which may explainthe attractiveness of one option over the other.Against this background we investigate how suchregulations affect the capital structure decisionsof corporate stockholders. We neglect problemsof information asymmetry between managers and

shareholders in the following.Hence, for simplicitywe abstract from principal-agent conflicts.During the last few decades many authors havecontributed to the field of corporate capital struc-ture decisions (for an overview of different capitalstructure models see Myers 2001 and Graham2006). In line with neoinstitutional theory, thetax based trade-off theory (e.g., Myers 1984) andthe non-tax-oriented pecking-order theory (Don-aldson 1961) have gained attention. Here, the the-ory of Modigliani and Miller (1958) represents aseminal work for many contributions based onneoclassical theory. In this early work, taxes werenot included. Many extensions to Modigliani andMiller (1958) take many different aspects into ac-count. Modigliani and Miller (1963) extend theirmodel themselves and integrate, among other as-pects, a corporate tax. Miller (1977) completes themodel anew and implements an income tax on theshareholder level. In both approaches a classicalcorporate tax system is assumed.Furthermore, in some contributions the incometax effects on the capital structure are modeled ina more refined manner. For instance, capital gainstaxes and, in turn, the asymmetric taxation of div-

147


idends and capital gains are highlighted by Farrarand Selwyn (1967), Brennan (1970), and Schneller(1980). Brennan (1970) and Zechner (1990) as-sume a progressive tax scale and allowances forinterest paid.Some papers take account of tax policy detailsin various countries. Swoboda (1991) examinesGermany’s and Austria’s 1991 tax laws in a Millerframework. In this context he considers Germany’scorporate tax, income tax and also property tax, lo-cal business tax and church tax. FungandTheobald(1984) consider the 1984 tax laws of France, Ger-many, the United Kingdom, and the USA. UnlikeSwoboda, they restrict themselves to corporate andincome taxes. Holland and Steiner (1996) extendSwoboda’s approach and investigate the influenceof Germany’s solidarity surcharge on the capitalstructure. Laß (1999) models US tax law between1977 and 1994 as well as German tax law be-tween 1976 and 1998, also using the Miller model.Kruschwitz (2007) models the German tax regimein 2007.Beyond detailed investigations of different taxesthere are many other extensions to the Modiglianiand Miller (1958) model. Hodder and Senbet(1990), Graham (2003), and Desai, Foley, andHines (2004) consider international tax aspects,DeAngelo andMasulis (1980a), DeAngelo andMa-sulis (1980b), and Graham (2000) take account ofnon-debt-tax shields, e.g., Kraus and Litzenberger(1973), Haugen and Senbet (1978), Kim (1978)model bankruptcy costs, and Jensen and Meck-ling (1976) and Leland (1998) integrate agencycosts (see Harris and Raviv 1991). Capital struc-ture models under uncertainty are examined byKim (1982), Zechner (1990), and Swoboda (1991).Miller andScholes (1978)andLitzenbergerandvanHorne (1978) consider dividend policy. Schneller(1980) andGraham (2003) offer amodel that takesaccount of different systems to prevent double tax-ation of dividends. A dynamic model is presentedby Fischer, Heinkel, and Zechner (1989).Beyond this finance-theory-oriented stream of lit-erature we refer to the discussion on intertem-porarily neutral capital income taxation and con-sumption-based tax systems in public finance. Theso-called ACE tax has become particularly popularas an investment and financing-neutral tax systemwith an allowance for corporate equity (ACE) de-ductible from taxable profits, which represents theopportunity costs of equity capital (Wenger 1983;

Boadway and Bruce 1984; Devereux and Freeman1991). This system is supposed to be close to the taxpoliticians’ idea of a fair and efficient tax system(Devereux and Freeman 1991: 4-6; Fehr and Wie-gard 2003: 298). As Belgium is one of the countriesto have introduced an ACE tax and simultaneouslytightened the thin capitalization rules (e.g., Gerard2006) it is interesting to analyze its interdepen-dencies and possible capital structure effects.Although the literatureprovidesdetailed investiga-tions of tax rules, thin capitalization rules have todate only been analyzed by Buettner, Overesch,Schreiber, and Wamser (2006), Overesch andWamser (2006)andOvereschandWamser (2009).The authors empirically investigate financing deci-sions in amultinational firm under a restrictive taxrule for stockholders’ debt financing, which is com-parable to the German thin capitalization rules. Inthis work, the main elements of a thin capital-ization tax rule are considered without modelingcountry-specific details of the regulation. Here,neither a specific debt-capital/equity-capital ratio,safe haven, nor a differentiation between profit-dependent and profit-independent loans are car-ried out. Amore sophisticated analysis of the influ-ence of thin capitalization rules on entrepreneurialcapital structure decisions has not yet been con-ducted.To fill this void, we integrate thin capitalizationrules into a capital structure decision model inthe following analysis. We investigate its influ-ence on corporate financing decisions analytically.It also helps the treasury to decide what typeof thin capitalization rule to implement under agiven tax setting. Our study can be regarded asa first step towards an empirical study on theinfluence of specific thin capitalization rules oncorporate capital structure and thus provides im-portant information to extend the work of Buett-ner, Overesch, Schreiber, and Wamser (2006),Overesch and Wamser (2006) and Overesch andWamser (2009). In contrast to empirical pub-lic finance papers that are often lacking a suffi-cient theoretical foundation or do not account forall relevant tax details (Desai, Foley, and Hines2004; Mintz and Weichenrieder 2005; Huizinga,Laeven, and Nicodème 2006; Weichenrieder andWindischbauer 2008) we provide a detailed the-oretical model. Moreover, future empirical paperscould verify whether the optimal financing anddividend policies we derive are actually practised.

148


We consider thin capitalization rules that are char-acterized by a given permitted debt-equity capitalratio because many countries limit debt financingreferring to such a permitted ratio. We take theItalian, German, and Belgian rules as examples tostudy their impact on capital structure decisions.The Italian and the German thin capitalizationrules are no longer in force. They have been substi-tuted by rules that restrict interest deduction irre-spective of apermitteddebt-to-equity ratio in thosecountries. Nevertheless, there are many countriesthat apply this type of thin capitalization rule,e.g., Bulgaria, Czech Republic, Denmark, Hun-gary, Latvia, Lithuania, the Netherlands, Poland,Portugal, Romania, Slovakia, and Slovenia. We re-fer to Belgium, Germany, and Italy as examplesto model all major characteristics of this type ofthin capitalization rule. These examples allow usto elaborate the mechanisms at work for differentimplementations of such rules that are commonor discussed in several countries and to show howthey interact with different tax systems.Our analysis can help investors not only to maketheir dividend and financing decisions under thincapitalization rules with a given permitted debt-equity ratio but also to adjust their decisions whentax laws change. Our analysis is representative ofall thin capitalization rules that refer to such apermitted debt-equity ratio.As the financing behavior of investors is crucial fordesigning thin capitalization rules and as the influ-ence of thin capitalization rules depends on severalother tax parameters (corporate tax rate, taxablefraction of dividends and capital gains, permit-ted ratio of debt to equity capital) it is importantfor the treasury to know how these rules inter-act. Analyzing different types of thin capitalizationrules and performing sensitivity analyses help thetreasury to implement a rule that contributes toa given political aim. Thus, the following investi-gation of different settings of thin capitalizationrules and tax systems is of general relevance fora comprehensive understanding of tax effects oncapital structure decisions.After developing the general model we apply ourmodel to selected countries. We refer to the capitalstructure model by Miller (1977). Miller investi-gates the market for debt capital and shows thatthismarket always leads to an equilibrium inwhichfor every company the capital structure is irrele-vant to the value of the firm. The same after-tax

return arises irrespective of the form of financingbecause this balances out the advantage that debtcapital incurs interest payments that aredeductiblefrom the corporate tax base (tax shield) and theadvantage that equity capital typically leads to rela-tively lower income tax burden (Miller 1977: 269 etseq.).In section 2 we develop a framework for a gen-eral capital structure model on the basis of Miller(1977) that includes thin capitalization rules. Weintegrate the Italian, German and Belgian thincapitalization rules into this model in sections 3to 5 and analyze whether a Miller equilibrium canemerge. If not, we determine the optimal capitalstructure and identify the most important value-driving factors. On the basis of Miller (1977), werefer to a single investor’s indifference towardsproviding capital as a loan or as equity capital toa corporation as a ‘‘Miller equilibrium’’. Note, thatin the following, this ‘‘equilibrium’’ is not a generalmarket equilibrium. In section 6 we summarizeand draw conclusions.This article is supplemented with Excel spread-sheets that provide the calculations that have beenperformed for Belgium, Italy, and Germany.1

2 Capital structure underrestricted shareholder debtfinancing

2.1 General assumptions

In the following we integrate thin capitalizationfor shareholder debt financing into Miller’s cap-ital structure model to derive conclusions aboutthe effectiveness of this regulation for financingdecisions in corporations. Our model relies on thefollowing set of assumptions.Weassumeaperfect capitalmarket under certaintyand identical debit and credit interest rates. Theborrower is a domestic corporation. The investorsare assumed to be domestic individuals who holdtheir investment and accordingly the provided cap-ital in private means. On the corporate level taxeson profits are considered. On the shareholder levelthe individual income tax is taken into account.All investors who have to decide to provide eitherequity or debt capital are assumed to be share-holders of the underlying corporation. Further we

1 These Excel files can be downloaded from www.business-research.org.

149


assume that the thin capitalization rule appliesto all shareholders and thus to all investors (seeAppendix A2).The interest payment Z amounts to iC, wherei denotes the market rate of return that can beearned on the capital market and C denotes theamount of debt capital provided by an investor.On the corporate level dividends are not tax-deductible, whereas interest payments on debt aregenerally tax-deductible as they are operating ex-penses. In line with Miller, the case of limitedinterest deductibility, particularly due to losses, isnot considered (Miller 1977: 262). We assume thata fraction α of debt is subject to the thin capi-talization rule (see Appendix A2). The permitteddebt-to-equity ratio will be denoted by µ and isassumed to be exceeded. Consequently, not thetotal amount of interest paid to a shareholder isdeductible, but only the interest on the permittedfraction of debt capital multiplied with the corpo-rate equity, thus iµE. As the thin capitalization ruleapplies to all shareholders, the shareholder’s sharein corporate equity is irrelevant. In the followingthe deductible amount of interest is denoted as theutilized safe havenH. Hence,

(1) H =

iµE, if αC > µE;

αiC, if 0 ≤ αC ≤ µE.

Equity capital E provided by the shareholder de-notes the equity capital that has already been pro-vided prior to the point of time when the financ-ing decision is made. The investor’s individualamount of equity is exogenously given. Further, itis assumed that all investors offer the same mixof equity and loans to the corporation and con-sequently all shareholders have an identical safehaven. It is therefore possible to determine theoverall safe haven for all shareholders jointly. It isnot necessary to refer to the single shareholder.Some countries raise additional taxes on profitson the corporate level. National tax rates and taxbases vary significantly. In the following the taxrate of these taxes is denoted by τa, the tax base byF .Under income tax law, earned interest is taxable ata fraction εwith ε ∈ [0,1]. The capital income fromshareholders’ invested equity capital consists ofdistributed dividends of the corporation and real-ized capital gains. The taxable fraction of dividendsD is denoted as γ with γ ∈ [0,1].

In the following the sum of retained earnings isgiven by G (Miller 1977: 268). G is not equal tocapital gains as this variable neglects the internalgrowth caused by reinvesting retained earnings atthe internal after-tax yield i(1 − τc − τa). Further,capital gains are not liable to income tax until theyare realized at shareholder level. We have to takeaccount of the time effect arising from the delayedtaxation of retained earnings. In the following,the time lag of capital gains taxation compared todividend taxation and the internal growth effectfrom retained earnings are both captured in thefactor θ > 0. It is necessary to implement θ into ourstatic model to highlight the difference in taxationof dividends and capital gains and the internalgrowth of retained earnings in present value terms.We obtain

(2) θ =(1 + i[1 − τc − τa])n

(1 + iτ)n,

withθGdenoting thepresent value of capital gains.We abstract from increases in value that are causedby speculative developments. Here, iτ is the after-tax market rate of return that the shareholder isable to earn alternatively on the capital market andis given by iτ = (1 − τiI)i. τiI denotes the investor’spersonal tax rate on interest income. If interestincome is included in the investor’s individualtax assessment, the tax rate τiI is equal to theinvestor’s personal income tax rate τi. If there isa withholding tax on interest income, τiI is thewithholding tax rate. The variable n denotes theperiod in years after which the capital gains arerealized. We assume that n is exogenously givenand hence the investor is not able to decide on theholding period n. Furthermore, we assume thatonly a fraction λ of the capital gain is taxable. Intotal the present value of assessable and taxablecapital gains G is λ θG (see Swoboda 1991: 857,and Laß 1999: 119, who assume that capital gainsare tax-exempt).We abstract from personal allowances, income-related expenses, standardized deductions, per-sonal exemptions, special expenses and extraordi-nary charges in themodel when calculating taxableincome. We assume there is no income from othersources apart from interest income, dividends, andcapital gains. The fractionof interest that is not tax-deductible on the corporate level is requalified ashidden profit distribution and therefore treatedas a dividend. Dividends and hidden distributions

150


are subject to an income tax rate τiD, interest pay-ments to an income tax rate τiI and capital gainsto τiG. Although the income tax schedule is pro-gressive in many countries, from the shareholders’perspective the income tax rate can be regarded asexogenously given.

2.2 Integrating taxes

Now we will formally describe all relevant fiscalrules. In a first step we have to determine theamount of interest that can be deducted when de-termining the corporate tax base. In total, interestpayments amounting to I are paid to the investor:

(3) I = iC.

The interest payments can be deducted from thecorporate tax base up to an amount of Id. Interestpayments on a fraction (1 − α) of the loans arecompletely deductible, i.e., I(1 − α), whereas in-terest payments on a fraction α of debt are onlydeductible in the amount of the safe haven H. Aswe assume the safe haven to be exceeded, it is equalto iµE.

(4) Id = I(1 − α) +H = iC(1 − α) + iµE.

The amount of non-deductible interest expensesInd which is requalified as a hidden distribution ofprofits is

(5) Ind = I − Id = iCα − iµE.

The taxable income in terms of corporate tax re-sults from the deduction of debt-capital interest Idfrom gross profit Π. Corporate tax amounts to:

(6) Tc = τc(Π − Id

)= τc

(Π − iC[1 − α] − iµE

).

Additional taxes on profits Ta are levied on thecorporate level. The tax base is equal to F; the taxrate amounts to τa.

(7) Ta = τa F .

The taxable interest Itaxable, dividendsDtaxable, andcapital gains Gtaxable are subject to income tax.Additionally, interest that is considered to be ahidden distribution of profits is subject to incometax. The taxable fraction of the hidden distributionof profits is denotedwith Btaxable. Thus, the burdenresulting from income tax is

Ti = τiI Itaxable + τiD(Dtaxable + Btaxable

)(8)

+ τiG Gtaxable.

The interest payments that are subject to incometax are identical to the deductible interest pay-ments Id on the corporate level. A fraction ε of Idis taxable. Hence, the interest payments that areliable to tax amount to

(9) Itaxable = ε Id = ε(iC[1 − α] + iµE

).

Only the fraction γ of the dividends is subject toincome tax

(10) Dtaxable = γD.

Interest that is interpreted as hidden distributionof profits is treated as dividends and thus likewisetaxable by the fraction γ. The hidden distributionof profits corresponds to non-deductible intereston the corporate level Ind. This amount denotesthe interest payments that exceed the safe haven.

(11) Btaxable = γB = γ Ind = γ(iCα − iµE

).

Retained earnings G result from the differencebetween the profit Π, the corporate tax Tc and theadditional taxes Ta, as well as the dividends D andthe interest I. Thereby, the present value of thecapital gains θG is taxable at the fraction λ. Inpresent value terms we obtain

Gtaxable = λθ(Π − Tc − Ta −D − I

)(12)

= λθ(Π − τc

[Π − iC(1 − α) − iµE

]︸︷︷︸Tc

− τa F︸︷︷︸Ta

−D − iC︸︷︷︸I

).

Inserting equations (9), (10), (11), and (12) into eq.(8) leads to

Ti = τiI ε(iC [1 − α] + iµE

)︸︷︷︸Itaxable

(13)

+ τiD γ(D + iCα − iµE

)︸︷︷︸Dtaxable+Btaxable

+ τiGλθ(Π − τc

[Π − iC(1 − α) − iµE

]−τa F −D − iC

).︸︷︷︸

Gtaxable

The total after-tax income of all investors Πτ iscomposed of the difference of the gross profit Π,corporate tax Tc, additional taxes on the corporatelevel Ta and income tax Ti,

(14) Πτ = Π − Tc − Ta − Ti

151


Πτ = Π − τc(Π − iC[1 − α] − iµE

)︸︷︷︸Tc


(15)

− τiIε(iC [1 − α] + iµE

)− τiDγ

(D + iCα − iµE

)− τiGλθ

(Π − τc

[Π − iC(1 − α)

−iµE]− τa F −D − iC

).︸︷︷︸

Ti

Similar to Miller (1977), to identify the optimalcapital structure based on the total after-tax in-come of all investors, we first have to determinethe optimal dividend policy. In a second step weanalyze the optimal capital structure. Against thebackground of the set of equations outlined above,we are able to investigate both an optimal dividendpolicy and an optimal capital structure under thincapitalization rules in a general capital structuremodel. Furthermore, analyzing the general settingenables us to draw conclusions about a real-worldexample, i.e. Italy. This is possible as the Italianthin capitalization rules are fairly simple and thuscorrespond to those of our general model. Later,we specify the model in more detail to study theimplications of modified and simultaneously morecomplex thin capitalizations rules. Such rules havebeen introduced in several countries. We take Ger-many and Belgium as examples of such modifiedrules and investigate whether the results of thegeneral model are robust under more complexshareholder financing rules.

3 General thin capitalization rules(Italy)

3.1 Critical income tax rate

In the following, we integrate general thin cap-italization rules into our capital structure model.Looking for corresponding real-world ruleswe findthat the rules in section 98 of the Italian IncomeTax Code fully comply with our general model.These rules, which are characterized by a smallnumber of attributes, no complicating exceptions,and simple parameters, can be regarded as anillustrating example of our general setting. Weimplement these rules into our framework to in-vestigate the influence of thin capitalization ruleson a general basis on investors’ financing decisions(see Appendix A1).Under section 98 of the Income Tax Code intereston debt capital that a corporation receives from a

substantial shareholder, i.e. a shareholder with ashareholding of at least 25%, has to be requalifiedas hidden distribution of profits if the permittedratio of debt to equity capital of currently 4:1 is ex-ceeded (Romanelli 2006: 372). Therefore, interestis not considered to be a hidden distribution untilthe debt of a substantial shareholder exceeds fourtimes their equity share.Additionally, a regional tax on productive activities(IRAP) is levied on the net value of the productionF . In line with IRAP interest expenses are nottax-deductible. The tax rate τa currently stands at4.25% (Romanelli 2006: 367 et seq.).Dividends and capital gains are 60% tax-exempt ifthe shareholder holds either 2% of voting poweror 5% of the capital of listed companies, or ifthey hold 20% of voting power or 25% of thecapital of other companies. Otherwise dividendsand capital gains are liable to a final withholdingtax of 12.5% (Romanelli 2006: 377). Aswe considersubstantial shareholders only, a withholding tax isnot applicable.Interest income arising from loans is subject to awithholding tax of 12.5% that is creditable againstthe shareholder’s income tax liability. A final with-holding tax applies to interest on current accountswith bank offices and bonds, but not on loans(Romanelli 2006: 378).Hence, interest payments, dividends, and capitalgains are subject to the same income tax rateτi = τiI = τiD = τiG.This reduces the net profit given by equation (15)to

Πτ = Π − τc(Π − iC[1 − α] − iµE

)︸︷︷︸Tc


(16)

− τiε(iC [1 − α] + iµE

)− τiγ

(D + iCα − iµE

)− τiλθ

(Π − τc

[Π − iC(1 − α)

−iµE]− τa F −D − iC

).︸︷︷︸

Ti

To identify the optimal dividend policy, equation(16) has to be differentiated with respect to divi-dends D and set equal to zero:

(17)∂Πτ

∂D= τi(λθ − γ) = 0.

We find an investor to be indifferent towards div-idend policy only in two cases, namely if their

152


marginal income tax rate τi amounts to τi = 0 or ifthe factor θ amounts to θ = γ

λ .The condition τi = 0 is only fulfilled if the investorhas no taxable income. Since we assume that theinvestor obtains interest payments and dividendsand/or capital gains, this condition is not fulfilledin our model. Hence, dividend policy is irrelevantonly if θ = γ

λ . Under current legislation we haveγ = λ which leads to θ = 1. The time and growth ef-fects from capital gains are cancelled out perfectlyand thus θ = 1. This is true only if the corporateinternal after-tax yield i(1 − τa − τa) is identical tothe after-tax market rate of return the shareholderis able to earn on the capital market iτ . Beyond thissetting, dividend policymay not be irrelevant. Thatsaid, full retention is always the optimal dividendpolicy. As long as retained earnings are taxed ata lower effective rate than distributed earnings,full retention of profits always leads to the highestnet profit. The same result is achieved by Milleralthough he does not outline it explicitly. Miller’sresults are explained by Swoboda (1991: 853) andLaß (1999: 45-46), who find a different result forthe German tax code. Both base their assumptionson a split corporate tax scale in connectionwith thefull imputation system. From the authors’ point ofview the optimal dividend policy depends on theratio between the personal income tax rate and thecorporate tax rate that is applied when profits areretained. If the personal tax rate is significantlylower than the corporate tax rate, a distribution ofprofits is beneficial to reduce the tax burden fromthe corporate tax rate applied to retained profitsto the lower personal income tax rate (Laß 1999:139-140).To identify the optimal capital structure based onthe optimal dividend policy, equation (16) has to bedifferentiated considering D* = 0 with respect toC. Rearranging finally leads to the critical incometax rate τ*i :

τ*i =[τc(C[1 − α] + µE

)](18)

·[ε(C[1 − α] + µE

)+ γ(Cα − µE)

+ λθ(τc[C[1 − α] + µE

]− C

)]-1.

Note that the factor θ is influenced by the after-taxinterest rate iτ and therefore by τ*i . Nevertheless,we assume that θ is exogenously given. To justifythis simplification we have analyzed the interde-pendency of θ and τ*i using an iterative simulation

(see Appendix A2). By means of equation (18) theoptimal financing decision can be determined forevery investor. All investors whose marginal taxrate τi equals the critical income tax rate τ*i are in-different towards the allocation of debt and equitycapital. They are referred to as marginal investors.Investors who have lower tax rates will offer a loanto the corporation, while investors with higher taxrates will offer equity capital (Swoboda 1991: 853).Due to the progressive income tax scale in manycountries, e.g., in Italy (Romanelli 2006: 378),a general irrelevance of the financing policy cannever be achieved for all taxpayers. Irrelevancecan only be achieved for the taxpayers who havemarginal tax rates that are identical to the criticalincome tax rate. Therefore, within this analysis it isonly possible to investigate whether or not aMillerequilibrium can be reached for specific taxpayers.Furthermore, the after-tax profit of a firm cannever be maximized through mixed financing. De-pending on the single parameters, either an equity-only or debt-only financing maximizes after-taxprofit, or themeans of financing is irrelevant to theafter-tax profit of the firm.

3.2 Sensitivity analysis

In the following we analyze the implications of avariation of different model parameters. Changesin gross profit Π and interest rate i do not influ-ence the critical income tax rate and therefore thefinancing decision. Hence, we concentrate on thefactor θ, the tax rates, the taxable fraction of divi-dends γ and capital gains λ, the fraction α of debt,the taxable fraction of interest ε, and the permittedratio of debt to equity µ.

3.2.1 Time and growth factor for capitalgains

To analyze how the different parameters affectthe investor’s financing decisions, all variables areassumed to be constant, except for the factor θ.We assume

τc = 33%; τmaxi = 43%;C = e 100,000; E = e 1,000;

γ = λ = 0.4; µ = 4; ε = 1; i = 6%; α = 1.

Interest payments are fully tax-liable, hence ε = 1.Dividends, hidden distributions, and capital gainsare subject to a shareholder relief system, i.e. theyare all subject to tax at a fraction of 40% (Romanelli

153


2006: 377). Hence, in our model we obtain γ = λ =0.4. As the permitted ratio of debt to equity capitalis 4:1, in our model the debt-equity parameter µis equal to 4. All interest payments to substantialshareholders are subject to the thin capitalizationrules (Romanelli 2006: 372). These regulationsimply α = 1.Then, the investor receives interest payments ofe6,000 and the safe havenH amounts to e240.We receive the critical income tax rate:

(19) τ*i =79.2

2,544 − 2,368.32θ.

Table 1: Critical income tax rates forvarious θ in the general model (e.g., Italy)

θ τ*i0.01 3.14%

0.25 4.06%

0.5 5.82%

0.75 10.32%

0.85 14.92%

0.9288 23.00%

0.95 26.93%

0.9964 43.00%

1 45.08%

1.07 800.19%

Table 1 shows that the values for the critical in-come tax rate τ*i vary between 3.14% and 800.19%(see Appendix A2). For θ = 1.07 we receive anextremely high critical income tax rate that againindicates that debt financing is always beneficial.If we determine critical income tax rates for highervalues of θ the critical tax rate may change its signand even become negative. A negative critical in-come tax rate is not easy to interpret. A closer lookat these scenarios clarifies that the basic mech-anisms at work do not change. In line with theresult for θ = 1.07 the resulting (negative) criti-cal tax rate again implies that debt financing isgenerally favorable.If θ > 0.9964, the critical tax rate is higher thanthe top tax rate of 43%. In these cases an equilib-rium is not possible for the underlying tax system.Then, investors will always prefer to provide a loaninstead of equity to the corporation leading to thehighest possible net profit Πτ .If θ < 0.9288 the critical income tax rate is lowerthan the minimum tax rate of 23%. In these casesall investors provide equity capital.

Only if 0.9288≤ θ ≤0.9964 will it depend on theinvestor’s individual income tax rate whether debtor equity financing is optimal. If the personal in-come tax rate is lower than the critical tax rate, theinvestor will offer a loan. If it is larger, they willprovide equity capital. If the marginal tax rate isequal to the critical income tax rate, the investoris indifferent to either option. In this case we findthe so-called Miller equilibrium for this investor.Although at first sight the after-tax internal rateof return falls short of the after-tax external rateof return (θ < 1) shareholders are still willing toprovide equity capital as they benefit from prefer-ential capital gains taxation. The time and growthfactorθdoes not reflect effects arising from the tax-able fraction of capital gains λ < 1. Consequently,providing equity capital can be optimal even forθ < 1.The specific outcome of the factor θ mainly de-pends on the ratio of the internal after-tax yieldi(1− τc − τa) and the after-tax market rate of returniτ . The holding period n exerts only little influenceon θ. Assuming that i(1 − τc − τa) and iτ are al-most equal, leads to θ ≈ 1. Then we see that debtfinancing will be beneficial for most investors.Income from equity capital, namely dividends andcapital gains, is subject to income tax at a fractionof γ = λ =0.4. By contrast, interest income is fullytaxed. Interest income amounting to the safe havenis also fully taxable, while the excess is subject totax to a fraction of γ =0.4. The lower taxationof dividends and capital gains on the shareholderlevel compared to interest income usually over-compensates their non-deductibility on the corpo-rate level. Nevertheless, debt capital often is moreadvantageous than equity capital because incomefrom equity capital, i.e. accumulated capital gains,is usually higher than interest income. Differenti-ating equation (19) with respect to θ we obtain the

derivative ∂τ*i

∂θ which is always positive:

(20)∂τ*i∂θ=

187,571(2,544 − 2,268θ)2

> 0.

An increase in θ causes a higher taxation of capitalgains and consequently a higher taxation of equi-ty capital. Hence, the relative advantage of debtcapital increases compared to equity capital.

154


3.2.2 Tax rates

A rise in the corporate tax rate has two counteract-ing effects. On the one hand, the tax advantage ofdebt capital (tax shield) at the corporate level in-creases and so does the relative advantage of debtcapital (see Buettner, Overesch, Schreiber, andWamser 2006: 11, who come to the same conclu-sion in the case of financing a German corporationowned by foreign shareholders). Equity capital ismore burdened by a higher tax rate, as dividendsare not tax-deductible on the corporate level. Onthe other hand, the factor θ decreases because thecorporate internal after-tax yield i(1 − τc − τa) de-clines. A decrease in θ leads, as shown above, to anincreasing relative advantage of equity capital. InItaly the tax-shield effect always dominates, caus-ing an overall increasing advantage of debt capital.Mathematically this result becomes obvious in that

the derivative ∂τ*i

∂τcis always positive.

A change in the corporate tax rate has a relativelylow impact on the investor’s financing decision.Assuming a holding period of, e.g., n = 10 years,a decrease in the tax rate from 33% to 10% leadsto changes in the critical income tax rate of 9percentage points. By contrast, an increase in thecorporate tax rate to 50% raises the critical taxrates by 6 percentage points.Varying the IRAP tax rate has no tax-shield effectbecause interest payments are not deductible un-der this tax. A change in the IRAP tax rate onlyamends the factor θ such that the relative advan-tage of equity capital increases. The derivative ∂τ

*i

∂τais always negative. E.g., assuming a holding periodof 10 years a rise in the tax rate from currently4.25% to 10% decreases the critical income tax ratefrom 12.57% to 3.11%.

3.2.3 Taxable fraction of dividends andcapital gains

In Italy dividends and capital gains are subject totax at the same fraction γ = λ =0.4. In this sectionwe analyze the effects of an isolated change inthe taxable fraction of dividends γ and the taxablefraction of capital gains λ, respectively. We alsoinvestigate the influence of a change in the taxablefraction of both dividends and capital gains. A risein the taxable fraction of dividends γ leads to adecrease in critical income tax rate and thereforean increase in the relative advantage of equity

capital:

(21)∂τ*i∂γ=

−1.27 · 108(4,000 + 96,000γ − 39,472θ

)2 < 0.This result, which seems surprising at first glance,can be explained as follows:we have already shownthat companies do not distribute profits in thismodel framework. Instead, payments for equitycapital are always realized as capital gains. If thetaxable fractionof dividends is increased, this leadsto a higher taxation of interest payments from thincapitalization that have to be requalified as hiddendistribution of profits. Therefore, an increase in γimplies a higher taxation of debt capital.Correspondingly, an isolated increase in the tax-able fraction of capital gains λ leads to a highertaxation of equity capital and an increasing rela-tive advantage of debt capital. The critical incometax rate rises when increasing λ:

(22)∂τ*i∂λ=

1.30 · 108θ(42,400 − 98,680λθ)2

> 0.

Focussing on a change in both the taxable fractionof dividends and capital gains (γ = λ), we obtain

(23)∂τ*i∂γ=

−1.27 · 108 + 1.3 · 108θ(4,000 + 1,000γ[9.6 − 9.9θ]

)2 R 0.

The derivative is positive or negative depending on

the factor θ. Figure 1 shows that the derivative ∂τ*i

∂γis zero if θ =0.9728. A negative derivative arisesif θ is smaller than 0.9728. Then, equity capitalbecomes more attractive. The effect from highertaxation of dividends that leads to an increasedadvantage of equity capital is higher than the effectfrom higher taxation of capital gains that causesan increased advantage of debt capital.

The derivative ∂τ*i

∂γ is positive if θ >0.9728. In thesecases the relative advantage of debt capital in-creases. The effect of the higher taxation of capitalgains overcompensates the effect of the higher div-idend taxation because a high value of the factorθ additionally causes a higher taxation of equitycapital.

3.2.4 Fraction of considered debt

At present all interest payments to substantialshareholders are subject to Italian thin capitaliza-tion rules. Therefore, the parameter α has been setequal to 1. Lowering the fraction α leads to a rise inthe critical income tax rate and hence an increas-

155


Figure 1: Derivative ∂τ*i∂γ for various θ in the

general model (Italy)

____ Derivative γτ∂

∂ *i for various θ in the general model (Italy)

-2

0.15

0.25

0 0.2 0.4 0.6 0.8 1

-4

-6

γτ∂∂ *

i

θ

ing relative advantage of debt capital compared toequity capital:

∂τ*i∂α=[−5.4 · 109 + 1.3 · 109θ

](24)

·[102,400 − 60,000α − 26,272θ

− 13,200θα]-2< 0.

This effect is true because the interest paymentswhich are subject to thin capitalization and requal-ified as hidden distributions of profit decrease, asdoes the corporate tax. A reduction in the fractionof considered debt α has a relatively high influenceon the critical income tax rate. A reduction fromα = 1 to α = 0.01 would lead to a rise in the criticalincome tax rate of up to 32 percentage points.

3.2.5 Other model parameters

Assuming a decrease in the taxable fraction ofinterest of ε = 1 to values with ε < 1, we obtain arise in the critical income tax rate. The derivative∂τ*i∂ε is always negative. This effect in turn increasesthe relative advantage of debt capital because alower taxable fraction of interest implies a lowertaxation of debt capital.The influence of the permitted ratio of debt toequity capital µ and of the amount of equity cap-ital E can be analyzed simultaneously since bothvariables only affect the safe haven H = iµE. Themultiple µE denotes themaximumamount of debtcapital that a substantial shareholder can provideto the corporation while tapping the full potential

of the safe haven. An increasing multiple µE andconsequently an increasing safe haven H leads toa decreasing corporate tax because the amountof deductible interest payments increases. At thesame time the income tax increases because the ad-ditionally deductible interest payments are taxedto the full amount whereas hidden distributionsare only subject to tax at the fraction γ =0.4. Theeffect on the corporate level always dominates.Therefore, the relative advantage of debt capitalcompared to equity capital increases with risingµE. If the permitted ratio of debt to equity capi-tal µ is increased, e.g., from 4 to 20, debt capitalmay be twenty times the equity capital contributedby one shareholder while a change in the criti-cal income tax rate amounts to a maximum of 17percentage points.

3.2.6 Results

We are able to show that except for the gross profitΠ and the interest rate i all model parameters havean influence on the investors’ financing decisions.Whereas the influence of the corporate tax rate,the IRAP tax rate, the taxable fraction of dividends,capital gains, and interest is low, the factor θ, thefraction of considered debt, and the multiple µEexert a high influence on financing decisions. As θismainly driven by the relation of the corporate taxrate to the individual income tax rate, obviouslythe tax rate difference determines the financingdecision significantly if capital gains are taxable. Acloser look clarifies that θ can significantly changethe critical income tax rate whereas a change in thecorporate tax rate cannot. E.g., if the corporate taxrate rises, two partial effects on the critical incometax canoccur.Oneeffect is an increase in the criticalincome tax rate because of an increasing tax shieldfrom debt financing. Simultaneously the criticalincome tax rate is reduced because of a decrease inthe factor θ. Consequently, the overall effect of arise in the corporate tax rate on the critical incometax rate determined by these two opposing partialeffects is small. Only the influence of the taxablefraction of dividends seems counterintuitive as arise in the taxable fraction of dividends γ causes arise in the attractiveness of equity capital.Given the Italian tax policy with capital gains tax-ation, the underlying thin capitalization rule witha given permitted debt-to-equity ratio is typicallynot irrelevant with respect to capital structure de-

156


cisions. Here, financing decisions mainly dependon the corporate-income tax rate differential andtherefore the parameter θ. Consequently, the taxscale is an important driver of the financing deci-sion. The treasury needs to know the distributionof investors across tax classes to forecast whetherit is likely that the majority of investors will tendto provide debt instead of equity capital under thethin capitalization rules.We find that changes in some model parametershave a higher impact on the critical income taxrates than changes in other parameters. This dif-ference in sensitivity is due to the functional formof the net profit.Several countrieshave implementedmore complexthin capitalizations rules. Against this backgroundand the results mentioned above, it is worthwhileto find out whether we obtain corresponding ordeviating results under more specified rules and towhat extent the characteristics of the underlyingtax systemdetermine the effects.We takeGermanyand Belgium as examples and perform an analysisin the following sections.

4 Complex thin capitalizationrules (Germany)

4.1 Critical income tax rate

In this section we integrate the German thin cap-italization rules according to section 8a of theGerman Corporate Tax Code as amended by theKorb II tax reform act (section 8a of the GermanCorporate Tax Code as amended by the Korb IItax reform act dated Dec. 22, 2003, BGBl. I 2003:2841 et seq.) into the model. This rule that wasintroduced in 2004 has been reformed by the Ger-man business tax reform act 2008 where it wassubstituted by an interest barrier (section 4h of theIncome Tax Code and section 8a of the GermanCorporate Tax Code as amended by the GermanBusiness Reform dated Aug. 14, 2007, BGBl. I2007: 1913 et seq., 1927 et seq.). Section 8a is anexample of a complex thin capitalization rule char-acterized by a given permitted debt-equity capitalratio.Under this debt-equity-ratio-based thin capitaliza-tion rule, interest on debt capital that a corporationreceives from a shareholder with a shareholdingof more than 25%, under certain circumstances,has to be requalified as a hidden distribution of

profits. This regulation only holds if the share-holder’s capital commitment is long term and ifthe tax allowance of e250,000 of interest on debtis exceeded. The permitted ratio of debt to eq-uity capital is 1.5:1 (section 8a para 1 no. 2 ofthe German Corporate Tax Code). As only interestpayments for long-term debt are subject to thisregulation, the parameter α represents the fractionof long-term debt in Germany. The tax allowance(section 8a para 1 of the German Corporate TaxCode) of e250,000 of interest is assumed to beexceeded.In addition to the corporate tax the German localbusiness tax has to be taken into account on thecorporate level. This tax treats equity and debt dif-ferently. To integrate effects of the local businesstax into the model further assumptions are neces-sary. Corporate income as defined in the GermanCorporate Tax Code is the basis for determiningthe local business tax base F (section 7 sentence 1of the German Local Business Tax Code). Since thelocal business tax itself is an operating expense, itis deductible from its own tax base. We considerthis deductibility by introducing an effective localbusiness tax rate τa (see Appendix A1). Further-more, amongst the various tax base adjustmentslisted in section 8 of the German Local BusinessTax Code, only the addition of 50% of the interestpayments for long-term debt (section 8 no. 1 ofthe German Local Business Tax Code) is consid-ered in the model. Splitting debt into long-termand short-term debt indirectly introduces a timedimension into the model. Although we develop astaticmodel that by definition does not account fortiming effects, a differentiation between long-termand short-term debt is necessary to distinguish be-tween different types of interest for local businesstax and thin capitalization purposes. Nevertheless,the model remains static. Reductions according tosection 9 of the German Local Business Tax Codeare not considered at all. The fraction of the in-terest that is long-term debt (section 8 no. 1 ofthe German Local Business Tax Code) is denotedby α ∈ (0,1]. If α = 0, section 8a of the GermanCorporate Tax Code does not apply. Hence, theaddition of interest payments for long-term debt isequal to 0.5 Id.Dividends, hidden distributions, and capital gainsare subject to the same income tax rate τi = τiI =τiD = τiG.Taking into account the local business tax, the net

157


profit Πτ is equal to (see Maßbaum and Sureth2008 for a more detailed analysis of the influenceof the German section 8a thin capitalization ruleon capital structure decisions)

Πτ = Π−τa[Π − iC(1 − α) − 0.5 iµE

]︸︷︷︸

Ta

(25)

− τc[Π − iC(1 − α) − iµE

−τa(Π − iC(1 − α) − 0.5 iµE

)]︸︷︷︸

Tc

− τiε[iC (1 − α) + iµE

]− τiγ

[D + iCα − iµE

]− τiλθ

[Π − τa

(Π − iC(1 − α)

− 0.5 iµE)

− τc(Π − iC(1 − α) − iµE

− τa[Π − iDC(1 − α) − 0.5 iµE

])−D − iC

].︸︷︷︸

Ti

To identify the optimal dividend policy, equation(25) has to be differentiated with respect to D:

(26)∂Πτ

∂D= τi(λθ − γ) = 0.

We see that the investor is only indifferent towardsdividend policy in two cases, namely if the incometax rate τi = 0 or if the factor θ =

γλ . This is the same

result we obtained for Italy (see section 3.1 andAppendix A2). Beyond these cases full retention isalways the optimal dividend policy.To obtain the optimal capital structure equation(25) has to be differentiated with respect toD* = 0to C and it has to be set to zero. As a result we getthe critical income tax rate τ*i ,

τ*i =

[τa[C(1 − α) + 0.5µE

](27)

+ τc[C(1 − α) + µE

− τa(C(1 − α) + 0.5µE

)]]·[

ε[C(1 − α) + µE

]+ γ(Cα − µE)

+ λθ(τa[C(1 − α) + 0.5µE

]+ τc

[C(1 − α) + µE

− τa(C(1 − α) + 0.5µE

)]− C

)]-1.

4.2 Sensitivity analysis

As in the general model, the gross profit Π andthe interest rate i have no influence on the criticalincome tax rate and therefore on investors’ finan-cing decisions. The other model parameters exertan influence on the critical income tax rate. Thedegree of sensitivity of the critical tax rate towardsvariation in the different parameters varies signifi-cantly. Therefore, we highlight the most importantvalue drivers.

4.2.1 Time and growth factor for capitalgains

To analyze the influence of the factor θ all othermodel parameters remain constant. We assume

τc = 25%; τmaxi = 42%; τa = 16.28%;C = e 100,000; E = e 1,000;

γ = λ = 0.5; µ = 1.5; ε = 1; i = 6%; α = 1.

We obtain the effective local business tax rateτa by considering the average collection rate ofh = 389% (Federal Statistical Office 2007) and ataxable business value ofm = 5%. As the permittedratio of debt to equity capital is 1.5:1, the parameterµ is equal to 1.5. The fraction of long-term debt isrepresented by the parameter α. Under incometax law earned interest is fully taxable (section 20para 1 no. 7 of the Income Tax Code), thereforeε is equal to 1. The distributed profits D, hiddendistributions, and capital gains are subject to thehalf-income system and thus only 50% of thistype of capital income is taxable, i.e. γ = λ =0.5 (section 20 para 1 no. 1 in conjunction withsection 3 no. 40 d) of the Income Tax Code, seeAppendix A1).The investor receives total interest I of e6,000.Within the safe haven interest of Id = H = e90 istax-deductible at the corporate level.Inserting the assumptions the critical income taxrate as a function of θ is

(28) τ*i =466.575

50,750 − 49,766.71θ.

158


Table 2 shows the critical income tax rates fordifferent values of the factor θ (see Appendix A2).

Table 2: Critical income tax rates forvarious θ in Germany

θ τ*i0.01 0.93%

0.25 1.22%

0.5 1.80%

0.75 3.48%

0.95725 15.00%

0.99845 42.00%

1 47.45%

1.015 197.04%

We see that the values for the critical incometax rate vary between 0.93% and 197.04%. If thefactor θ is higher than 0.99845, the critical incometax rate is higher than the top tax rate of 42%.In these cases, debt capital is advantageous for allshareholders. If θ is lower than0.95725, the criticalincome tax rate is lower than theminimum incometax rate of 15% and all investors will provide equitycapital. Only if 0.95725≤ θ ≤ 0.99845, it dependson the investors’ marginal tax rate as to whetherdebt or equity capital is favorable (section 3.2.1).For θ = 1.015 we receive an extremely high criti-cal income tax rate that again indicates that debtfinancing is always beneficial. If we determine crit-ical income tax rates for higher values of θ the crit-ical tax rate may change its sign and even becomenegative (see section 3.2.1 for an interpretation ofthese effects).

We obtain the derivative ∂τ*i∂θ by differentiating

equation (27)with respect toθ and inserting the as-sumptions. As in the general model this derivativeis also positive in Germany. Therefore, an increas-ing θ leads to an increasing relative advantage ofdebt capital compared to equity capital (section3.2.1).

(29)∂τ*i∂θ=

23,220(50,750 − 49,766.71θ)2

> 0.

In Figure 2 we compare the German and Italiancritical income tax rates. We see that the Germantax rates areusually lower than the Italian tax rates.Only if θ is higher than 0.9986 are the German taxrates higher than the Italian. This implies that inGermany equity capital is usually more attractive

in comparison to debt capital than in Italy. Onlyif θ > 0.9986 is debt capital more attractive inGermany than in Italy.

Figure 2: Critical income tax rates forvarious θ in Germany and Italy

____ Critical income tax rates for various θ under general thin capitalization rules (Italy)

----- Critical income tax rates for various θ under specified thin capitalization rules (Germany)

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

0 0.2 0.4 0.6 0.8 1 θ

*iτ

10%

20%

30%

50%

40%

Debt capital is usually less attractive in Germanythan in Italy, because debt capital is more severelyburdened due to the addition of a fraction of theinterest payments for long-termdebt for local busi-ness tax purposes and the lower permitted ratio ofdebt to equity capital. If the factor θ increasesthis implies a higher effective taxation of capitalgains and consequently a reduced attractivenessof equity capital in both Italy and Germany. Asthe nominal taxation of capital gains λ is lower inItaly than in Germany, equity capital becomes lessunattractive in Italy than in Germany.

4.2.2 Tax rates

A rise in the corporate tax rate τc incurs an in-creasing critical income tax rate and therefore anincreasing advantage of debt capital over equitycapital. This result is in line with the result weobtained for Italy (section 3.2.2).By contrast, an increase in the local business taxrate τa decreases the critical income tax rate andtherefore leads to an increasing relative tax advan-tage of equity capital. In this case the tax shieldeffect invoking a higher advantage of debt cap-ital is lower than the opposing effect caused bythe negative impact of rising τa on θ, because inGermany the fraction of interest payments that is

159


tax-deductible with respect to the local businesstax is lower than for corporate tax purposes.Assuming a holding period of n = 10 years ex-emplifies that a decrease in the corporate tax ratefrom 25% to 10% leads to a reduction in the criticalincome tax rate of only 0.7 percentage points, acorresponding increase in the corporate tax rateto an increase of only 0.5 percentage points. Sim-ilarly, changes in the local business tax rate alsoonly have a low influence on the critical incometax rate. Assuming a decrease in the local businesstax to 10% raises the critical income tax rate by0.2 percentage points, an increase to 30% resultsin a change in the critical income tax rate of 0.3percentage points.

4.2.3 Taxable fraction of dividends andcapital gains

As in Italy, in Germany dividends and capital gainsare subject to the same taxable fraction γ = λ.An exclusive rise in the taxable fraction of divi-dends γ, as in Italy, lowers the critical income taxrate and therefore increases the relative advantageof equity capital (section 3.2.3):

(30)∂τ*i∂γ=

-4.60 · 107

(1,500 + 98,500γ − 49,767θ)2< 0.

Analyzing c.p. an increased capital gains taxation,i.e. an increase in λ, we obtain the following deriva-tive that is, as in Italy, always positive (section3.2.3):

(31)∂τ*i∂λ=

4.64 · 107θ(50,750 − 99,533λθ)2

> 0.

A change in the taxable fraction of both dividendsand capital gains γ leads to a derivative that is,depending on the factor θ, positive, negative orzero:

(32)∂τ*i∂γ=

-4.6 · 107 + 4.64 · 107θ(1,500 + 98,500γ − 99,533γθ)2

R 0.

The derivative is positive if θ is higher than 0.9896,negative if θ is lower than 0.9896, and equal tozero if θ is 0.9896. For Italy we obtain a similarvalue of θ = 0.9728. Hence, the integration of thelocal business tax and the different values of themodel parameters do not substantially affect the

influence of the taxable fraction of dividends andcapital gains on the critical income tax rate andtherefore on investors’ financing decisions.

4.2.4 Fraction of considered debt

Assuming a change in the fraction of considereddebt α we obtain the following derivative:

∂τ*i∂α=[-1.86508 · 109 + 1.8605 · 109θ

](33)

·[100,750 − 50,000α

− 31,162θ − 18,605αθ]-2< 0.

In Germany only long-term debt capital is subjectto the thin capitalization rules of section 8a ofthe German Corporate Tax Code. Therefore, theinvestor can influence the parameter α by pro-viding long-term or short-term debt, respectively.Lowering the fraction of long-term debt leads toan increasing relative advantage of debt capitalcompared to equity capital.

4.2.5 Other model parameters

Considering changes in the taxable fraction of in-terest ε and the multiple of the permitted ratio ofdebt to equity capital and the amount of equitycapital µE we obtain the same influences as inItaly (section 3.2.5).A decrease in the taxable fraction ε of interestleads to an increasing relative advantage of debt

capital. The first derivative ∂τ*i∂ε is always negative.

Considering a rise in the multiple µE we obtain ahigher relative attractiveness of debt capital.

4.2.6 Results

To study the influence of the German thin capi-talization rules on financing decisions we had toextend the general model to include the local busi-ness tax. This tax is quite different from the ItalianIRAP as the resulting tax burden depends on theinvestor’s financing decision. In spite of the intro-duction of this tax we obtain very similar resultsfor both Germany and Italy. We can show that inboth countries the gross profit and the interest ratedo not influence the investors’ financing decisions,whereas changes in the other model parametersexert the same influences in both countries. Achange in the German local business tax rate has a

160


less severe influence on the critical income tax ratethan a change in the Italian IRAP tax rate becauseinterest payments are not tax-deductible under anIRAP. For this reason a variation of the IRAP taxrate does not invoke a tax shield.

5 Complex thin capitalizationrules (Belgium)

After having considered the Italian and Germanthin capitalization rules we now analyze the in-fluence of another debt-equity-ratio-based rule,namely the Belgian regulation on shareholders’financing decisions which was introduced in 1997.In Belgium interest payments are requalified ashidden distribution of profits and therefore treatedas dividends if the permitted ratio of debt to equitycapital of 1:1 is exceeded. The Belgian thin capi-talization rules are applicable to all shareholders.A minimum shareholding is not necessary. Thethin capitalization rules include all loans that ashareholder provides the corporation (Offermanns2006: 73).A notional interest deduction for equity capitalis granted to corporations. Thus, the Belgian sys-tem applies an Allowance for Corporate Equity(ACE) device which was intended to account forthe distinction of interest payments to externalproviders of capital compared to those from as-sociated members of corporations. Introducing anACE was regarded as a step towards a more neu-tral tax system and towards improving Belgium asan investment location (Gerard 2006: 156). Thededuction is based on the book value of the com-pany’s equity. It is calculated by multiplying theequity by a fixed percentage determined by thegovernment. For 2007 the percentage is equal toin = 3.442%. The notional interest rate is set bythe government on the basis of the interest rateon 10-year government bonds (Cowley, Gutiérrez,Kesti, and Soo 2008: 91).In contrast to Italy and Germany, additional taxesτa are not levied on the corporate level.Dividends and interest income are subject to awithholding tax of τiD = 25% and τiI = 15% re-spectively (Offermanns 2006: 78). Alternatively,the investor can choose to include dividends andinterest income into their individual assessment ifthis leads to a lower taxation. In this case τiD = τiI .In the following we analyze both cases, the sce-nario with withholding tax in section 5.1 and with

individual assessment in section 5.2. We only con-sider the case that the investor chooses to includeboth interest and dividends into assessment. Forreasons of transparency we do not model mixedscenarios.Capital gains are not subject to tax, thereforeτiG = 0 (Offermanns 2006: 76). In a first stepwe assume that capital gains are also subject tothe withholding tax of 25%. This assumption en-ables us to isolate the effects of the introductionof a capital gains tax. Hence, we assume that τiG =25%.

5.1 Withholding tax

5.1.1 Critical income tax rate

Considering the withholding tax we assume

τiD = τiG = 25%; τiI = 15%.

We obtain the following net profit Πτ

Πτ = Π − τc(Π − iC(1 − α) − iµE − in E

)︸︷︷︸

Tc

(34)

− 0.15 ε(iC (1 − α) + iµE

)− 0.25 γ

[D + iCα − iµE

]− 0.25 λθ

(Π − τc

[Π − iC(1 − α)

−iµE − in E]−D − iC

).︸︷︷︸

Ti

By differentiating equation (34) with respect to thedividends D the following dividend policy turnsout to be optimal:

(35)∂Πτ

∂D= λθ − γ = 0.

The investor is only indifferent towards the div-idend policy if λθ = γ. As capital gains are notsubject to tax in Belgium at present, we receiveλθ = 0. The condition only holds if γ = 0, namelyif dividends are not subject to tax. Since dividendsare fully taxable in Belgium, hence γ = 1 and in-difference to the dividend policy is not possible atpresent. Instead full retention is always the opti-mal dividend policy. Because capital gains are notsubject to tax whereas dividends are, full retentionalways leads to the highest net profit.

161


Taking account of the withholding tax the incometax rate is no longer a function of the tax basebut exogenously given, because τiI = 15% andτiD = τiG = 25%. Thus, a critical income tax rate τ*icannot be determined. The optimal capital struc-ture decision does not depend on the investor’smarginal income tax rate but only on the values ofthe other model parameters.

5.1.2 Sensitivity analysis

In the followingweanalyze the influenceof selectedmodel parameters on the investor’s financing de-cisions.For the basic scenario we assume

τc = 33%; τmaxi = 50%; in = 3.442%; i = 6%;C = e 100,000; E = e 1,000; Π = e 10,000;

γ = 1; λ = 0; µ = 1; ε = 1; α = 1.

Corporations with a tax base up to e 322,500 aresubject to a progressive tax scale. Corporationswith a higher tax base are subject to a tax rate of33% (Offermanns 2006: 65, see Appendix A1). Weassume that the tax base exceeds e 322,500. Theshareholders are subject to a progressive incometax scale. The top tax rate τmaxi to the amount of50% is imposed if the tax base exceeds e 31,700(Offermanns 2006: 77, see Appendix A1).Dividends, interest payments, andhiddendistribu-tions are subject to income tax to the full amount,hence ε = γ = 1. Capital gains are tax-exempt,hence λ = 0. The thin capitalization rules are ap-plicable for all loans provided by shareholders, i.e.α = 1. The permitted ratio of debt capital to equitycapital is 1:1, i.e. µ = 1.Inserting the assumptions into equation (34) weobtain a net profit of e6,711 in the case of equityfinancing and a net profit of e 5,237 in the case ofdebt financing. The investors will therefore decideto provide equity capital instead of debt capital.Payments for equity financing are only taxed withcorporate tax at 33%. In line with notional interestdeduction, a notional amount of interest can bededucted from the corporate tax base. The pay-ments for equity are not taxed on the shareholderlevel because full retention is the optimal dividendpolicy and capital gains are tax-exempt.Interest payments are tax-deductible on the corpo-rate level up to the safe haven. Those that exceedthe safe haven are non-deductible and thereforesubject to corporate tax of 33%. On the share-

holder level interest payments up to the safe havenare subject to the withholding tax rate of 15%whilethe excess is subject to the withholding tax of 25%.Consequently, interest payments amounting to thesafe haven are possibly taxed lower than paymentsfor equity capital. But interest payments that ex-ceed the safe haven are taxed considerably higher(33% corporate tax plus 25% withholding tax).Since we assume the safe haven always to be ex-ceeded and since Belgium has a very low safehaven, interest payments are always taxed higherthan capital gains.As in Italy and Germany, a variation of the grossprofit and the interest rate has no influence on thefinancing decisions. The same is true for changesin the notional interest rate.As long as capital gains are not subject to tax, i.e.λ = 0, a change in the time and growth factor forcapital gains θ has no influence on the investor’sfinancing decisions. If capital gains are not tax-exempt, i.e. λ > 0, a variation of θ influences theinvestor’s financing decisions. Since capital gainsare subject to tax at λθ, a change in the factor θ hasthe same influence on the financing decisions as achange in the taxable fraction λ.As in Italy and Germany, an increasing corporatetax rate leads basically to an increasing relative ad-vantage of debt capital compared to equity capital(section 3.2.2). The increasing relative advantageof debt capital is very low, due to the notional in-terest deduction for payments for equity capital. Asτc ∈ [0,1] an irrelevance of the financing decisionor even an advantage of debt capital cannot occurin Belgium.Investigating the effects of an isolated decreasingtaxable fraction of dividendswe find a rising attrac-tiveness of debt capital (section3.2.3). The investoris only indifferent towards debt and equity capitalif γ = -0.00038. As γ ∈ [0,1] a change in the tax-ation of dividends does not change the financingdecision. Equity capital is always advantageous.The introduction of a capital gains taxation, whileassuming that dividends are subject to tax to thefull amount, i.e. γ = 1, leads to an increased relativeadvantage of debt capital (section 3.2.3).Neverthe-less, debt capital would become advantageous onlyfor negative λ-values. As λ ∈ [0,1], equity capital isalways advantageous.In Italy and Germany, a change in the taxablefraction of both dividends and capital gains (γ =λ) leads, depending on the value of the factor

162


θ, to a change in the relative advantage of debtover equity capital (section 3.2.3). In Belgium, achange in these parameters does not change theshareholder’s financing decision. A decreasing γand λ leads to an increasing relative advantageof debt capital, but an advantage of debt capitaltowards equity capital can only occur for negativevalues of γ and λ.A decrease in the fraction of considered debt frompresently α = 1 to α < 1 reduces the drawback ofdebt over equity caused by the thin capitalizationrule (section 3.2.5). If α = 0.4242 the investor isindifferent towards debt and equity financing. Ifthe tax law is modified accordingly, which is ratherunlikely, the net profit is identical for bothdebt andequity financing. Only if the fraction of considereddebt is lower than 0.4242 do we find an advantageof debt financing because then, net profit increaseswith an increasing gearing rate.A tax reform changing α directly to alleviate in-terest deduction restrictions is very unlikely to bediscussed. It is more likely that thin capitalizationrules are alleviated by restricting their applica-tion to long-term debt as is the case in Germany,for instance. If the Belgian government only re-ferred to interest payments on long-term debt intheir modified thin capitalization rules, a fractionof long-term debt of 42.42% would be equal to afraction of α = 0.4242. In this case, the investorcan benefit from debt financing if their intereston long-term debt does not exceed 42.42% of theinterest on short-term debt.Given that interest payments are subject to tax tothe full amount, namely ε = 1, a decrease in thetaxable fraction ε leads to an increasing relativeadvantage of debt capital. Concentrating on theinfluence of a change in ε on the financing decision,we find that the investor is indifferent towards debtand equity financing if ε = -164. Since ε ∈ [0,1]in fact an indifference cannot arise. Instead, theinvestorwill alwaysprefer toprovideequity capital.In contrast to Italy and Germany a change in thepermitted ratio of debt to equity capital µ and inthe amount of equity capital E cannot be analyzedjointly, because a change in equity capital notonly influences the safe haven H = iµE, but alsothe notional interest deduction in E. Assuming anincreasing µ or an increasing E we obtain a risingrelative advantage of debt capital (section 3.2.5).Capital structure irrelevance arises if µ is increasedto 58or if debt capital is increased to 104,000.Debt

capital will be preferred if µ > 58 or E > 104,000.

5.1.3 Results

Considering the Belgian thin capitalization rules inthe light of the withholding tax option, the finan-cing decision, unlike in Italy and Germany, doesnot depend on the investor’s marginal income taxrate but only on the othermodel parameters. Vary-ing different model parameters leads to similarfinancing decision patterns as under the Italian-type and German-type rules. Whereas in Italy andGermany capital structure irrelevance is gener-ally possible when assuming currently codifiedtax rates and debt-to-equity ratio, providing eq-uity capital is always favorable in Belgium becausecapital gains are exempt from taxation, thin cap-italization rules discriminate debt financing anda notional interest deduction on equity capital isgranted. Although an ACE often is regarded as ameans to provide tax neutrality, in Belgium thisneutrality property has been undermined by intro-ducing a thin capitalization rule, which obviouslyexacerbates thediscriminationofdebt capital (Ger-ard 2006: 156).

5.2 Assessment

5.2.1 Critical income tax rate

The shareholder can optionally include dividendsinto the individual tax assessment. In this casedividends and interest are not subject to the with-holding tax of 25% or 15% but to the shareholder’sindividual income tax rate. Hence, the option toinclude dividends or interest in the assessment isonly reasonable for shareholders with an incometax rate less than 25% and 15% respectively. Forsimplicity we only consider the case that the choiceto include income in the assessment is beneficialfor both dividends and interest.If dividends and interest income are included inthe assessment, all sources of income are subjectto the same income tax rate. Therefore, we have

τiD = τiI = τiG = τi.

The net profit is thus equal to

Πτ = Π − τc(Π − iC(1 − α) − iµE − in E

)︸︷︷︸

Tc

(36)

163


− τiε(iC (1 − α) + iµE

)− τiγ

[D + iCα − iµE

]− τiλθ

(Π − τc

[Π − iC(1 − α)

−iµE − in E]−D − iC

)︸︷︷︸

Ti

.

We obtain the same optimal dividend policy asfor Italy and for Germany. The investor is onlyindifferent towards debt and equity financing ifτi = 0, namely if they have no taxable income, orif λθ = γ (section 3.1). Otherwise full retention isalways the optimal dividend policy.We obtain the following critical income tax rate bydifferentiating equation (36) with respect to C:

τ*i =[τc[C(1 − α) + µE

]](37)

·[ε[C(1 − α) + µE

]+ γ(Cα − µE)

+ λθ(τc[C(1 − α) + µE

]− C

)]-1.

5.2.2 Sensitivity Analysis

Just like in the other countries the gross profit andthe interest rates do not influence the investor’sfinancing decisions. The other model parametersinfluence the critical income tax rate and thereforethe financing decisions.We assume

τc =33%; τmaxi =15%; in =3.442%; i=6%;C = e 100,000; E = e 1,000; γ =1;

λ =0; ε=1; µ =1; α =1.

The top tax rate τmaxi is equal to 15% because onlyshareholders with an income tax rate of less than15% will decide to include dividends and interestincome in their individual assessment. Insertingthe assumptions into equation (37) we obtain

(38) τ*i = 0.33%.

All investors have a marginal income tax rate thatis higher than the critical income tax rate. To con-clude for all investors we have to abstract fromthe typically rather small group of investors with amarginal income tax rate between zero and 0.33%.Therefore, all investors prefer to provide equitycapital instead of debt capital. This result corre-

sponds to that obtained for the withholding tax(section 5.1.2).A variation of the factor θ has no influence onthe critical income tax rate because capital gainsare tax-exempt. Only if λ > 0 will a change in thefactor θ affect the investor’s financing decisions. Inthese cases a change in this factor influences thefinancing decisions in the same way as a change inthe taxable fraction of capital gains λ does.In line with the results for the other countries,we obtain a rising critical income tax rate and anincreased relative advantage of debt capital if weassume a rising corporate tax rate (section 3.2.2).Varying the taxable fraction of dividends γ whileassuming that capital gains are tax-free leads to anegative derivative. We find that lowering γ causesa rising advantage of debt capital (section 3.2.3):

(39)∂τ*i∂γ=

-117,612(60 + 5,940γ

)2 < 0.We obtain critical income tax rates higher than15% and therefore an attractiveness of the with-holding tax for interest income over the individualassessment for γ < 0.0121.Assuming a taxation of capital gains, i.e. λ > 0, anda taxation of dividends to the full amount, i.e. γ =1,we obtain the following positive first derivative:

(40)∂τ*i∂λ=

118,408θ(6,000 − 5,980λθ

)2 > 0.As expected, implementing a capital gains tax in-vokes a higher relative advantage of debt capital(section 3.2.3). The effect of a rising λ on the criticalincome tax rate is very low. We obtain critical in-come tax rates of over 15% only for very high capitalgains taxes with only hypothetical relevance (λθ >0.98124).A change in the taxable fraction of both dividendsand capital gains (γ = λ) leads to the followingderivative that is a function of the value of thefactor θ and may take positive or negative values:

(41)∂τ*i∂γ=

-117,612 + 118,408θ(60 + 5,940γ − 5,980γθ

)2 R 0.

If θ > 0.993, the partial derivative is positive. Ifθ < 0.993, it is negative.

164


As in the previously analyzed countries a decreas-ing fraction of considered debt leads to an in-creased critical income tax rate that implies an in-creasing advantage of debt capital (section 3.2.5).

The derivative ∂τ*i

∂α is always negative. Lowering thefraction of considered debt α from 1 to 0.5 leads toa rise in the critical income tax rate from 0.33% to16.83%. Lowering α to 0.25 leads to τ*i = 25.08%.If α = 0.5, the investor therefore will decide infavor of the withholding tax for interest income. Ifα = 0.25, they will prefer the withholding tax fordividends.Decreasing the taxable fraction of interest εwe finda rise in the relative advantage of debt capital. The

partial derivative ∂τ*i

∂ε is always negative.A rise in the permitted ratio of debt to equitycapital µ or the amount of equity capital E leadsto an increase in the relative advantage of debt

capital. The first derivatives ∂τ*i

∂µ and∂τ*i∂E are always

positive (section 3.2.5). To obtain a critical tax rateof more than 15% the permitted ratio of debt toequity capital has to be increased to 46:1 or debtcapital has to be increased to e46,000.

5.2.3 Results

Assuming assessment is attractive for investors,their financing decisions depend, as in Italy andGermany, on their marginal income tax rate. Thedifferent model parameters exert an influence onthe decision that corresponds to the one identi-fied for the other countries. As in the case of thewithholding tax, equity capital is always beneficialwhen considering currently codified tax rates anddebt-to-equity-capital ratio because capital gainsare exempt from taxation and a notional inter-est deduction is granted for equity capital. If thethin capitalization rule was abolished we wouldobtain a critical income tax rate that is equal to thecorporate tax rate. This result is in line with theneutrality property of an ACE tax and implies that,given the assumption of a perfect capital market,we lose capital structure irrelevance as soon as theincome tax rate differs from the corporate tax rate.The analysis of theBelgian tax systemclarifies that,given a neutral corporate tax, the political aim tofoster equity financing and prevent a drain of taxrevenues due to debt financing, a thin capitaliza-tion rule with a given permitted debt-to-equityratio may help to achieve this aim irrespective ofthe investors’ income tax rates.

6 Conclusions and future researchFrom a tax planner’s point of view, it is oftenattractive to choose debt over equity financing.Nevertheless, corporations issue shares as well asdebt, raising the question on possible causes. Asindividual tax planning has led to an increase indebt financing of corporations, many countrieshave introduced thin capitalization rules to securetheir tax revenues. These thin capitalization rulesmight explain why corporations still issue bothdebt and equity.Against this background we investigate how suchregulations affect the capital structure decisionsof stockholders of corporations. We examine gen-eral thin capitalization rules and more specificand complex examples of regulations to restrictshareholder financing as to their impact on financ-ing decisions of shareholders of corporations. Thegeneral case is exemplified by the Italian rules ofthe 2007 tax law and the more specific regulationsare illustrated by the examples of the German andBelgian thin capitalization rules of the 2007 taxlaw. The selected rules can be regarded as repre-sentative examples ofmany countries’ thin capital-ization rules that are characterized by a permitteddebt-to-equity ratio. As a corresponding regulationexists in several countries, the knowledge gainedin our study can be used as the basis for follow-upempirical studies on the impact of thin capitaliza-tion rules on capital structure decisions in thesecountries.Furthermore, as the financing behavior of in-vestors is crucial for designing thin capitalizationrules and as the influence of thin capitalizationrules depends heavily on a lot of other tax param-eters (corporate tax rate, taxable fraction of div-idends and capital gains, permitted ratio of debtto equity capital) it is important for the treasuryto know how these rules interact. Our results helpthe treasury to implement a rule that contributesto their given political aim in a given tax setting.Taking into account the relevant tax rates and thecodified debt-equity ratio we find similar resultsfor Italy and Germany. In particular, a so-calledMiller equilibrium is possible when the differenceof the corporate tax rate and the individual in-come tax rate is relatively low and when the pro-videddebt comprises a small fraction of considereddebt. The fraction of considered debt is presentlyonly limited under the German-type rule because

165


there only long-term debt is taken into account.In Belgium equity capital always incorporates anadvantage considering current tax rates and thecurrent debt-to-equity-capital ratio because eventhough Belgium has introduced a supposedly neu-tral ACE tax debt capital is discriminated by a thincapitalization rule. This result holds in the case ofboth withholding tax and assessment.Investigating changes in the model parameters weobtain similar results for all underlying tax sys-tems. In particular, varying the fraction of consid-ered debt as well as the ratio of corporate tax rateand income tax rate and the permitted ratio of debtto equity capital have a substantial impact on thecritical income tax rate and make a Miller equi-librium possible. Among these parameters, onlythe permitted ratio of debt to equity capital andthe tax rates can be influenced by the legislator.The remaining parameters can be changed by thetaxpayers themselves.There is a broad range of issues that should beaddressed in future research. Although the presentanalysis only focuses on individual shareholders,corporate shareholders can easily be integratedinto the model by changing the denotation ‘‘in-come tax’’ to ‘‘corporate tax’’ on the shareholderlevel and changing the values of the taxable fractionof dividends γ and capital gains λ in the sensitiv-ity analysis. As thin capitalization rules, especiallyin Europe, are applicable to both domestic andforeign investors, for simplicity we focus only ondomestic shareholders. An extension of the modelwith respect to international investors is possible.Effects on financingdecisions arising from interna-tional tax base or tax rate differentials are capturedin our sensitivity analysis in sections 3.2, 4.2, 5.1.2and 5.2.2. The basic results remain the same inthe international context. In the case of differentinterpretations of income (dividend or interest in-come) that has been requalified as dividends undera thin capitalization rule we have to fall back on asingle-case analysis. General results can no longerbe deduced.In 2008 both Germany and Italy (see Marino andRusso 2008 for an overview of the new Italianregulation) changed their thin capitalization rulesand implemented an interest barrier. Under Ger-many’s 2008 business tax reform (section 4h of theIncome Tax Code and section 8a of the GermanCorporate Tax Code as amended by the GermanBusiness Reform dated Aug. 14, 2007, BGBl. I

2007: 1913 et seq., 1927 et seq.), in principle in-terest payable is deductible up to an amount thatcorresponds to the interest earnedby the company.The residual interest payable can be deducted fromthe tax base up to an amount of 30% of the residualprofit, plus depreciation and interest expenses andless interest income. Interest expenses that remainnon-deductible according to this regulation haveto be carried forward and, subject to the same con-ditions, increase interest expenses in subsequentfinancial years. In this regard, the non-deductibleinterest expenses are not classified as hidden dis-tribution of profit. Hence, on the shareholder levelthey are treated as interest income. The tax al-lowance up to which the regulation is not appliedamounts to e 3,000,000. With respect to the in-terest barrier, initially only statements about thetendency can be made.On the corporate level debt capital is advantageous,because interest payments are at least partially tax-deductible, whereas dividends and capital gainsare not. In contrast, on the shareholder level in-terest income is fully taxable and subject to theshareholder’s individual income tax rate (see Ap-pendix A1). A withholding tax amounting to 25%is on principle levied on dividends and capitalgains (see Kiesewetter and Lachmund 2004 whoinvestigate and demonstrate the effects of a finalwithholding tax on the capital structure of enter-prises with the help of investment appraisal andwho design, based on the achieved results, a with-holding tax that is independent of the financingform). The shareholder can also choose to includedividends and capital gains in their assessment. Asubstantial shareholder with a holding of at least25% can additionally choose a shareholder reliefsystem for dividends and capital gains. In thiscase they are subject to the shareholder’s individ-ual income tax rate, but only taxed at a fractionof 60%. The shareholder relief system is alwaysadvantageous unless the shareholder exhibits anindividual income tax rate of 42%. To sum up, div-idends and capital gains are usually lower taxed onthe shareholder level than interest income. Hence,capital structure irrelevance is possible since thelower taxation of income from equity on the share-holder level counteracts the lower taxation of in-come from interest on the company level causedby the (partial) deductibility of interest payments.If the capital company incurs losses over a longerperiod and interest payments hence cannot be de-

166


ducted in the same period, the attractiveness ofequity financing compared to debt financing in-creases. In the extreme case, if interest on debtcannot be deducted at all, equity financing alwaysgenerates the highest net profit.To gain more precise results for Germany a quan-titative analysis including all items of Germany’s2008 Tax Reform Act is necessary. Furthermore,an analysis of the new Italian thin capitalizationrules anda comparisonof the results of theGermanand Italian regulation is reasonable. We relegatethis issue to future research.For now, our analysis highlights that for com-panies in countries with thin capitalization rulesand a permitted debt-equity ratio, debt financingcan typically be tax-optimized by shareholders.However, under certain circumstances the capitalstructure can become irrelevant.

AppendixA1: Supplementary information on thenational tax laws

Belgium

TheBelgian thin capitalization rulewas introducedin 1997 and is still in force.

Belgium raises a surcharge of 3% of the corporatetax. We neglect this surcharge.

Belgium raises local surcharges on the income tax.We neglect these local surcharges.

Germany

The German thin capitalization rule was intro-duced in2004. It hasbeen reformedby theGermanbusiness tax reform act 2008 where it was sub-stituted by an interest barrier.

From 2008 on the local business tax is not de-ductible as an operating expense anymore.

The half-income system was abolished in 2009and replaced by a final flat tax on capital income of25%. A modified shareholder relief system is onlyapplicable to capital income from corporate sharesheld as business assets. Substantial shareholderswith a shareholding of at least 25% can choosebetween the withholding tax and the shareholder

relief system.

From 2009 on a withholding tax amounting to25% is on principle levied on dividends and capitalgains. The withholding tax is not applicable to ashareholding of at least 10% (see section 32d of theIncome Tax Code for an overview of the taxationof income from equity capital).

Italy

Weconsider Italy’s thin capitalization rule in 2007.For an overview of this regulation see, for example,Gusmeroli and Russo (2004). This regulation wasintroduced in 2004. In 2008 Italy replaced itsformer rules with a regulation similar to the newGerman interest barrier. See Marino and Russo(2008) for an overview of this new regime.

A2: Supplementary information on theanalysis

We assume that the thin capitalization rule ap-plies to all shareholders and thus to all investors.The thin capitalization rule is for example appli-cable to all shareholders in Belgium and Spain. Inmany other countries, including Germany, Italy,the Netherlands and Portugal, thin capitalizationrules only apply to what are known as substantialshareholders. These are shareholders with a givenminimum shareholding. Loans from third partiesand non-substantial shareholders can easily be in-tegrated into the model. As interest payments tothird parties and to non-substantial shareholdersdo not induce a thin capitalization treatment, theseextensions do not yield any new insights for ouranalysis.

We assume that a fractionα of debt is subject to thethin capitalization rule. In Germany the regulationonly applies to interest paid on long-term debt.Other countries apply the regulation to all interestpayments to specific shareholders.

The values in Table 1 neglect that the factor θshould be endogenously determined. As the influ-ence of τ*i on θ cannot be derived analytically wehave to fall back on simulation. To point out thatthe resulting changes in our results are very smallwe have performed an iterative simulation. By it-eration we can account for the interdependencyof τ*i and θ. We find that the resulting critical tax

167


rates are slightly higher. At a maximum we iden-tify a deviation of 3.75 percentage points from therates determined for an exogenously given θ. Wefind similar deviations if we vary other model pa-rameters. Against this background, it is acceptableto conduct the following investigation abstractingfrom this interdependency. We hence do not haveto fall back on numerical results on optimal cap-ital structure; instead, analytical and thus moregeneral results can be obtained.

In Germany the investor is indifferent towardsdividend policy if the income tax rate τi = 0 or ifthe factor θ = γ

λ . The condition τi = 0 is true if thetaxable income of the investor is lower than thebasic allowance ofe 7,664 for singles ande 15,328for married couples (see section 32a of the IncomeTax Code).

The values in Table 2 neglect the influence of τ*ion θ. By iteration we can show that the criticaltax rates are slightly higher. At a maximum weidentify a deviation of 0.6 percentage points fromthe rates determined for an exogenously given θ.We find similar deviations if we vary other modelparameters.

Acknowledgements

We are indebted to two anonymous referees andthe department editor, Rainer Niemann, for theirvery valuable suggestions,whichhelped to improvethe paper considerably. We thank the participantsof the European Accounting Association AnnualCongress 2008 in Rotterdam and 2009 in Tam-pere for fruitful discussions. Furthermore, we areparticularly grateful for very helpful commentsfrom Rolf König and Lutz Kruschwitz. The usualdisclaimer applies.

References

Boadway, Robin and Neil Bruce (1979): Depreciation and in-terest deductions and the effect of the corporation income taxon investment, Journal of Public Economics, 11 (1): 93-105.

Brennan, Michael J. (1970): Taxes, Market Valuation and Cor-porate Financial Policy,National Tax Journal, 23 (4): 417-427.

Buettner, Thies,Michael Overesch, Ulrich Schreiber, andGeorgWamser (2006): The Impact of Thin-Capitalization Rules onMultinationals’ Financing and Investment Decisions, ZEWDis-cussion Paper No. 06-068: Mannheim.

Cowley, Nick, Carlos Gutiérrez, Juhani Kesti, and Mei-JuneSoo (eds.) (2008): Global Corporate Tax Handbook 2008,International Bureau of Fiscal Documentation: Amsterdam.

DeAngelo,Harry andRonaldW.Masulis (1980a):Optimal Cap-ital Structure under Corporate and Personal Taxation, Journalof Financial Economics, 8 (1): 3-29.

DeAngelo,Harry andRonaldW.Masulis (1980b): Leverage andDividend Irrelevancy under Corporate and Personal Taxation,Journal of Finance, 35 (2): 453-464.

Desai, Mihir A., C. Fritz Foley, and James R. Hines (2004):A Multinational Perspective on Capital Structure Choice andInternal Capital Markets, Journal of Finance, 59 (6): 2451-2487.

Devereux, Michael and Harold Freeman (1991): A GeneralNeutral Profits Tax, Fiscal Studies, 12 (3): 1-15.

Donaldson, Gordon (1961): Corporate Debt Capacity: A Studyof Corporate Debt Policy and the Determination of Corpo-rate Debt Capacity, Division of Research, Graduate School ofBusiness Administration, Harvard University: Boston.

Farrar, Donald E. and Lee L. Selwyn (1967): Taxes, Corpo-rate Financial Policy, and Return to Investors, National TaxJournal, 20 (4): 444-454.

Federal Statistical Office (2007): Trade tax rates rose slightlyon a federal average in 2006, press release dated 08/22/2007,http://www.destatis.de/jetspeed/portal/cms/Sites/destatis/Internet/EN/press/pr/2007/08/PE07__329__735.psml.(Access Date: 2009-01-29).

Fehr, Hans and Wolfgang Wiegard (2003): ACE for Germany?Fighting for a Better Tax System, in: Michael Ahlheim, Heinz-Dieter Wenzel, and Wolfgang Wiegard (eds.), Steuerpolitik -Von der Theorie zur Praxis: Festschrift für Manfred Rose,Springer: Berlin et al., 297-324.

Fischer, Edwin O., Robert Heinkel, and Josef Zechner (1989):Dynamic Capital Structure Choice: Theory and Tests, Journalof Finance, 44 (1): 19-40.

Fung,WilliamK.H. andMichael F. Theobald (1984): Dividendsand Debt under Alternative Tax Systems, Journal of Financialand Quantitative Analysis, 19 (1): 59-72.

Gerard, Marcel (2006): Belgium Moves to Dual Allowance forCorporate Equity, European Taxation, 46 (4): 156-162.

Gouthière, Bruno (2005): A Comparative Study of the ThinCapitalization Rules in the Member States of the EuropeanUnion and Certain Other Countries, European Taxation, 45(9-10): 367-451.

Graham, John R. (2000): How Big are the Tax Benefits ofDebt?, Journal of Finance, 55 (5): 1901-1941.

Graham, John R. (2003): Taxes and Corporate Finance: AReview, The Review of Financial Studies, 16 (4): 1075-1129.

Graham, John R. (2006): A Review of Taxes and CorporateFinance, Foundations and Trends in Finance, 1 (7): 573-691.

Gusmeroli, Michele and Massimiliano Russo (2004): ItalianThin Capitalization Rules, Tax Treaties and EC Law: Much AdoAbout Something, Intertax, 32 (10): 493-519.

Harris, Milton and Artur Raviv (1991): The Theory of CapitalStructure, Journal of Finance, 46 (1): 297-355.

Haugen, Robert A. and LemmaW. Senbet (1978): The Insignif-icance of Bankruptcy Costs to the Theory of Optimal CapitalStructure, Journal of Finance, 33 (2): 383-393.

Hodder, James E. and LemmaW. Senbet (1990): InternationalCapital Structure Equilibrium, Journal of Finance, 45 (5):1495-1516.

Holland, Susanne and Peter Steiner (1996): Zum Einfluß desSolidaritätszuschlags auf die Kapitalstruktur deutscher Kapi-talgesellschaften, Journal fürBetriebswirtschaft, 46 (2): 67-75.

168


Huizinga, Harry, Luc Laeven, and Gaëtan Nicodème (2006):Capital Structure and International Debt Shifting in Europe,ETPF Research Paper, European Tax Policy Forum: London.

International Bureau of Fiscal Documentation (2006): Euro-pean Tax Handbook 2006, C.F. IBFD: Amsterdam.

Jensen, Michael C. and William H. Meckling (1976): Theory ofthe Firm: Managerial Behavior, Agency Costs and OwnershipStructure, Journal of Financial Economics, 3 (4): 305-360.

Kiesewetter, Dirk and Andreas Lachmund (2004): Wirkungeneiner Abgeltungssteuer auf Investitionsentscheidungen undKapitalstruktur von Unternehmen, Die Betriebswirtschaft, 64(4): 395-411.

Kim, E. Han (1978): A Mean-Variance Theory of Optimal Capi-tal Structure and Corporate Debt Capacity, Journal of Finance,33 (1): 45-63.

Kim,E.Han (1982):Miller’sEquilibrium,ShareholderLeverageClienteles, and Optimal Capital Structure, Journal of Finance,37 (2): 301-319.

Kraus, Alan and Robert H. Litzenberger (1973): A State-Preference Model of Optimal Financial Leverage, Journal ofFinance, 28 (4): 911-922.

Kruschwitz, Lutz (2007): Finanzierung und Investition, 5thed., Oldenbourg: München et al.

Laß, Frank (1999): Der Einfluß der Besteuerung aufunternehmerische Finanzierungsentscheidungen, Shaker:Aachen.

Leland, Hayne E. (1998): Agency Costs, RiskManagement, andCapital Structure, Journal of Finance, 53 (4): 1213-1243.

Litzenberger, Robert H. and James C. van Horne (1978): Eli-mination of the Double Taxation of Dividends and CorporateFinancial Policy, Journal of Finance, 33 (3): 737-750.

Marino, Tancredi and Massimiliano Russo (2008): ItalianRestyling of Interest Deduction Rules: the Amendments ofthe Italian Finance Bill for 2008, Intertax, 36 (5): 204-210.

Maßbaum, Alexandra and Caren Sureth (2008): The Im-pact of Thin Capitalization Rules on Shareholder Financ-ing, http://www.arqus.info/paper/arqus_39.pdf (Access Date:2009-11-10).

Miller, Merton H. and Myron S. Scholes (1978): Dividends andTaxes, Journal of Financial Economics, 6 (4): 333-364.

Mintz, Jack and Alfons J. Weichenrieder (2005): Taxationand the Financial Structure of German Outbound FDI, CESifoWorking Paper 1612, Center for Economic Studies: Munich.

Modigliani, Franco and Merton H. Miller (1958): The Cost ofCapital, Corporation Finance and the Theory of Investment,American Economic Review 48 (3): 261-297.

Modigliani, Franco and Merton H. Miller (1963): CorporateIncome Taxes and the Cost of Capital: a Correction, AmericanEconomic Review, 53 (3): 433-443.

Myers, Stewart C. (1984): The Capital Structure Puzzle, Journalof Finance, 39 (3): 575-592.

Myers, Stewart C. (2001): Capital Structure, Journal of Eco-nomic Perspectives, 15 (2): 81-102.

OECD (1987): Issues in International Taxation, No 2, ThinCapitalization: Organisation for Economic Co-Operation andDevelopment: Paris.

Offermanns, René (2006): Belgium, in: Juhani Kesti (ed.):European Tax Handbook 2006, International Bureau of FiscalDocumentation, Amsterdam: 63-82.

Overesch, Michael and Georg Wamser (2006): GermanInbound Investment, Corporate Tax Planning, and Thin-Capitalization Rules - a Difference-in-Differences Approach,Discussion Paper No. 06-075, Zentrum für EuropäischeWirtschaftsforschung: Mannheim.

Overesch, Michael and Georg Wamser (2009): CorporateTax Planning and Thin-Capitalization Rules: Evidence froma Quasi-Experiment, Applied Economics, forthcoming.

Romanelli, Luca (2006): Italy, in: Juhani Kesti (ed.): EuropeanTaxHandbook 2006, International Bureau of Fiscal Documen-tation, Amsterdam: 363-382.

Schneller, Meir I. (1980): Taxes and the Optimal Capital Struc-ture of the Firm, Journal of Finance, 35 (1): 119-127.

Swoboda, Peter (1991): Irrelevanz oder Relevanz der Kapital-struktur und Dividendenpolitik von Kapitalgesellschaften inDeutschland und Österreich nach der Steuerreform 1990 bzw.1989?, Zeitschrift für betriebswirtschaftliche Forschung, 43(10): 851-866.

Weichenrieder, Alfons J. and Helen Windischbauer (2008):Thin-CapitalizationRules andCompanyResponses.Experiencefrom German Legislation, CESifo Working Paper 2456, Centerfor Economic Studies: Munich.

Wenger, Ekkehard (1983): Gleichmäßigkeit der Besteuerungvon Arbeits- und Vermögenseinkünften, Finanzarchiv, 41 (2):207-252.

Zechner, Josef (1990): Tax Clienteles and Optimal CapitalStructure under Uncertainty, The Journal of Business, 63 (4):465-491.

Biographies

Alexandra Maßbaum is doctoral student and re-search assistant at the Chair of Business Administration,esp. Business Taxation at the University of Paderborn.She received her diploma degree in Business Adminis-tration with specialization in accounting, taxation, au-diting, andmanagement accounting, aswell as corporateand tax law from the University of Osnabrück.

Caren Sureth is Professor and Chair of Business Ad-ministration, esp. Business Taxation since 2004 and alsoHead of the Center for Tax Research at the Unversity ofPaderborn. She received her doctorate from the Univer-sity of Bielefeld, where she also reached her habilitation.She studied Business Administration, English, French,and Chinese and received her Diploma Degree at theUniversity of Passau. Her main fields of research arethe impact of taxation on entrepreneurial decisions, taxplanning, and taxation under uncertainty.

169


Do Taxes Matter in the CAPM? Lutz Kruschwitz, Andreas Löffler



Do Taxes Matter in the CAPM?Lutz Kruschwitz, Department of Banking and Finance, Freie Universität Berlin, E-Mail: [email protected]

Andreas Löffler, Department of Finance and Investment, Universität Paderborn, E-Mail: [email protected]

AbstractThe traditional literature on the CAPM assumes that investor’s tax payments simply vanish from themodel. This assumption is not at all consistent with the actual behavior of the Treasury. The theoryof general equilibrium states that an interest rate rf = 0 will not affect prices if taxes are introduced.We show that this result can be extended to the CAPM if the tax payments are redistributed amonginvestors.

Keywords: CAPM, Tax-CAPM, equilibrium, taxes, CARA utility

Manuscript received June 13, 2009, accepted by Rainer Niemann (Accounting) November 10,2009.

1 Presentation of the ProblemThe standard CAPM is a neoclassical equilibriummodel basedon some simplifying assumptions (seeSharpe 1964, Lintner 1965, and Mossin 1966). Inparticular, it relies on the assumption that in-vestors base their decisions on the expected valueand variance of cash flows with no taxes beingcollected.Brennan (1970) was the first to model taxes withina CAPM.Hisworkwas the foundation of numerousfurther contributions. Whereas Brennan analyzeda proportional tax, Litzenberger and Ramaswamy(1979) and Litzenberger and Ramaswamy (1980)looked at a progressive tax. As early as in the 1990sKoenig (1990) applied Brennan’s model to the taxsystem that was valid in Germany at that time.More recent analyses taking current German taxregulations into account have been provided byWiese (2004), Jonas, Loeffler, and Wiese (2004),Wiese (2006a),Wiese (2006b),RappandSchwetz-ler (2007) and Wiese (2007).It should be emphasized that all thesemodels weresingle period models in which the investors basetheir decisions on the µ-σ criterion while assumingrisk aversion. When taking a closer look at howtaxes are modeled, a circumstance emerges that sofar has gone unnoticed. At the end of the periodthe Treasury collects taxes which are not returnedto the taxpayers. For a drastic illustration of thismodel’s peculiarity let us imagine that investors

live on an island which once a year is ravaged bypirates who relieve them of taxes, which they willnever see again.This circumstance is hardly compatible with anequilibrium model. Rather, it would appear moreappropriate to assume that the tax authorities re-invest their revenue in the form of transfer pay-ments. These include unemployment insurance,pension funds, compulsory health insurance, de-velopment aid etc.In a recent paper Eikseth and Lindset (2009) con-sider the redistribution of taxes. Surprisingly, theydo not report on any effects on the equilibrium se-curity prices. Authors dealing with scientific topicsof public finance have also made such assump-tions (see, e.g., Konrad 1991 p. 167, Buchholz andKonrad 2000 p. 87). or at least point out that themodel analysis responds very sensitively depend-ing on which redistribution system is assumed(see, e.g., Stiglitz 1972 or Rapp and Schwetzler2008). Konrad analyzes a model where the dif-ference between cash flows and opportunity costsis taxed, whereas in our model cash flow minusdepreciation forms the tax base. Konrad de factolooks at a Johansson-Samuelson tax. Seen in thislight, our model is closer to the existing nationaltax codes. Rapp and Schwetzler look at a binomialmodel where at t = 1 exactly two states are pos-sible. In this model investors trade on two stockmarkets that are not connected to each other. Theredistribution is given by a random variable. It

171


turns out that the specific form of redistributionhas an important impact on the asset’s prices. Inour model the uncertainty is much more generalsince we allow for multiple states of nature.If one then takes into account that the revenue ofthe tax authorities is allocated to investors accord-ing to a specific rule, then this will impact on theoptimal portfolio and hence on equilibrium prices.We must assume that the tax CAPM equationsderived by Brennan and others are questionable,at least to the extent that they are based on theunrealistic assumption that the tax revenue van-ishes. In the next two sections we wish to showthat the equilibrium prices in the CAPM remainunchanged, provided one of the following two re-quirements is met:

• The risk-free interest rate vanishes.

• Investors have µ-σ utility functions with con-stant absolute risk aversion.

We are able to provide convincing intuitions forboth conclusions.

2 The ModelThe CAPM is a two-period model, represented bytoday (time 0) and tomorrow (time 1). The futureis uncertain. We examine I investors i = 1, . . . , Iwho trade on a capital market and have individualµ-σ utility functions

(1) U i(µ,σ2)

depending on expectation and variance of securi-ties’ payments. Utility functions like this can be ex-plained through independent axioms for the pref-erences (see Loeffler 1996b). These utility func-tions are strictly monotonically increasing in thefirst variable (expected value µ) and strictly mono-tonically decreasing in the second variable (vari-ance σ2). Moreover, they are quasi-concave, whichmeans that the answers of the individual maxi-mization problems are always unique.There are n = 0, . . . ,N assets with cash flows of Ynat a specific time t = 1. These cash flows consistboth of dividends and capital gains. The 0-th assetis the risk-free asset and pays a unit in the nextperiod; all other assets are uncertain. Portfolio Xconsists of these N + 1 assets, whereby the n-thentry Xn designates the number of the n-th asset

which an investor holds at a specific time t = 0. Thepayment of the portfolio X then reads as follows:

(2) X0 +N∑n=1

XnYn.

The amount of the optimal risk-free asset isX i0. Ac-

cordingly, X i designates the (optimal) risky port-folio of investor i, whereby this portfolio is a N-dimensional vector with entries X i

1to X iN .

The vector E includes theN expected values of thepayments of the risky assets. The correspondingcovariance matrix isΩwith the dimensionsN ×N .Hence the expected value of the portfolio (X i

0,X i)

is X i0+ X i · E. The variance is (X i)T · Ω · X i.

The investors’ initial endowment (X i0, X i) consists

of the risky asset and a share of the riskless asset.For every risky share there is one unit in total; theriskless asset is given in zero net supply, hence

(3)I∑i=1

X i0 = 0,I∑i=1

X in = 1 (n > 0).

We designate the sum of the risky initial endow-ments as the market portfolioM.Theprice of aportfolioX is describedby the symbolp(X). Hence the budget restriction of an investorreads as follows: p(X i

0) + p(X i) = p(X i

0) + p(X i).

There is no arbitrage in the equilibrium. Therefore,theprices are linear. Consequently, there is a vectorp which implies that

(4) ∀X p(X) = p · X.

Vectorp is notnecessarilypositive aswearedealingwith prices of the basic assets rather than stateprices.We assume that the tax authorities collect a taxon the investors’ income. The valuation basis isthe difference resulting from the return flows ofthe optimal portfolio and the capital endowment.Risky and riskless entries are subject to the samelinear tax rate τ. Therefore, the i-th investor’ taxdebts amount to

(5) T i = τ

(X i0 − p(X

i0) +

N∑n=1

X inYn − p(Xi)

)=

= τ

(X i0 − p(X

i0) +

N∑n=1

X inYn − p(Xi)

).

As opposed to Brennan (1970) in our model divi-dends, capital gains and interest are taxed differ-

172


ently.Whether our results will survivewith a richertax system remains to be seen.The total tax revenues

∑Ii=1 T

i are redistributed tothe investors according to a certain rule. The i-thinvestor receives the shareωi of the entire revenuefrom taxes, whereby the equation

(6)I∑i=1

ωi = 1

applies. Evidently the sharesωi are then determin-istic. If the distribution parameters are acceptedas random variables, then any equilibrium couldbe explained. We know that a certain total amountcan be distributed to the investors. If we then wishto attain a special income distribution, we selecttheωi in precisely such a way that in each environ-mental condition and depending on the paymentof the initial endowment, the required financialmeans can be either allocated to the investorsor taken away. This counteracts the distributionmechanism, however.An equilibrium is characterized by two conditions:Each investor maximizes their individual utilityand the markets are cleared. In our model there isanother problemwhichwewish to point out: Basedon our assumptions the activity of one investorhas, due to the tax returns, an impact on theoptimization problem of all other investors.It would, therefore, be conceivable that investorstake the initiative and form coalitions which mightinitially lead to fictitious equilibria that could at-tain stability only on the basis of these tax re-turns. Such equilibria are discussed in the liter-ature as ‘‘sunspot equilibrium’’ (see, for instanceShell (forthcoming)). We wish to rule out suchequilibria and therefore assume that -- providedthe investor has the profit of his utility -- he willtake a decision that is optimal for the other in-vestors, hence avoiding sunspot equilibria.

Definition 2.1 An equilibrium is described bya price vector p and a demand quantity(X i0,X i)i=1,...,I with the following two character-

istics:

1. Demand (X i0,X i) maximizes the utility for in-

vestor i, provided the allocation of all otherinvestors is optimal (see equation (13)).

2. Total demand corresponds to the completeinitial endowment of all investors (marketclearance).

Due to the law of Walras it is sufficient undermarket clearance to take into account only therisky assets. If an equilibrium exists there, then anequilibrium also exists for riskless assets.Let us now take a glance at the individual util-ity maximization problem. We know the expectedvalue and the variance in the assets’ cash flows.The typical maximization problem appears in thefollowing form:

(7) maxX i0,X iU i

(µ

[X i0 +

N∑n=1

X iYn−

− τ

(X i0 − p(X

i0) +

N∑n=1

X inYn − p(Xi)

)

+ωiτI∑j=1

(X j0− p(X j

0) +

N∑n=1

X jnYn − p(Xj)

) ,σ2

(1 − τ) N∑n=1

X inYn + ωiτ

I∑j=1

N∑n=1

X jnYn

.The argument of the variance function can be sim-plified because the risk-free asset and the price ofthe portfolio are irrelevant. Taking into accountthat the return flow of the riskless asset has beennormalized to the value 1 and based on equation(4), the budget constraint of the i-th market par-ticipant can be written out as follows:

(8)X i0

1 + rf= p(X i0) + p(X

i) − p · X i .

We can simplify the problem since we know theexpected value and the covariance matrix. If weadditionally take into account that both

(9)I∑j=1

X j = 1 andI∑j=1

X j0= 0

must apply for the equilibrium, the optimizationproblem can be rewritten in the form:

(10) maxX i0,X iU i(X i0 + X

i · E −

−τ(X i0 − p(X

i0) + (X

i · E − p(X i)))+

+ωiτ(E − p) · 1 ,((1 − τ)X i + ωiτ

)T· Ω ·

((1 − τ)X i + ωiτ

)).

173


Finally we will proceed to solve the side condition(8) for X i

0and plug it into the function to be

maximized. This will result in

(11) maxX i

U i((1 + rf )

(p(X i0) + p(X

i) − X i · p)+

+ X i · E − τ[(1 + rf )

(p(X i0) + p(X

i) − X i · p)−

−p(X i0) + Xi · E − p(X i)

]+ ωiτ(E − p) · 1,(

(1 − τ)X i + ωiτ)T· Ω ·

((1 − τ)X i + ωiτ

)).

The first-order conditions read as follows:

(12) U iµ

(E − (1 + rf ) · p − τ

(E − (1 + rf ) · p

))+

+ 2(1 − τ)U iσ2Ω ·

((1 − τ)X i + ωiτ

)= 0 .

After cancellations of 1 − τ the result will be

(13) U iµ

(E − (1 + rf ) · p

)+

+ 2U iσ2 Ω ·

((1 − τ)X i + ωiτ

)= 0 ,

whereby the arguments of the utility functions(expected value and variance) are provided by

(14) µ(X) = (1 + rf )(p(X i0) + p(X

i) − X i · p)+

+X i ·E−τ[(1 + rf )

(p(X i0) + p(X

i) − X i · p)

−p(X i0) + Xi · E − p(X i)

]+ ωiτ(E − p) · 1

and

(15) σ2(X) =((1 − τ)X i + ωiτ

)T· Ω ·

·((1 − τ)X i + ωiτ

).

Equation (13) is the starting point for our nextconsiderations.

2.1 Vanishing Risk-free Rate

Proposition 2.1 (vanishing risk-free inter-est rate) The risk-free interest rate is zero.We look at an equilibrium p and (X i

0,X i)i=1...,I

at the tax rate of τ > 0. We then deal with anallocation (Z i

0,Z i)i=1,...,I from (20) whereby p and

this allocation also constitute an equilibrium at atax rate of τ = 0.

Please note that we do not show the uniquenessof equilibrium prices. Statements on uniqueness

in the CAPM can be found in Hens, Laitenberger,and Loeffler (2002). Bottazzi, Hens, and Loeffler(1998) show that the CAPM typically possessesmany equilibria. Hence it is possible that variousequilibriawith different prices can exist at tax ratesof τ > 0 or even τ = 0. We will show that at leastone of the possible equilibria must have the pricevector p.The result of the theorem is intuitive. A risk-less interest rate of zero is effectively equivalentto interest-corrected profit taxation (see Wenger1983). For a considerable time it has been knownthat this tax is investment-neutral and thereforehas no influence on the prices of goods in theequilibrium. Thus this result does not come as asurprise, and it is independent of the particularredistribution system.

Proof:We take the first-order conditions according to (13)and set the interest rate to zero. As a result we have

(16) U iµ (E − p) +2U

iσ2 Ω ·

((1 − τ)X i + ωiτ

)= 0 ,

and the utility function arguments are simplifiedto

µ(X) = p(X i0) + p(Xi)+(17)

+((1 − τ)X i + ωiτ

)T· (E − p),

σ2(X) =((1 − τ)X i + ωiτ

)T· Ω ·(18)

·((1 − τ)X i + ωiτ

).

On the basis of this maximization condition itbecomes evident that every equilibrium with a taxrate τ > 0 has a corresponding equilibrium withthe tax rate τ = 0 possessing the same prices forrisky assets.We assume that the allocation (X i

0,X i)

and the price vector p describe an equilibrium forτ > 0. Moreover, if we define

(19) ∀n > 0 Z in := Xin(1 − τ) + ω

iτ and

Z i0 := p(Xi0) + p(X

i) − Z i · p

we can show that this allocation leads to an equilib-rium when the price vector p remains unchanged,provided τ = 0. In particular we must verify thefollowing:

1. The aggregate demand is identical to the ag-gregate supply (market clearance).

174


2. Investor i is able to acquire theportfolio (Z i0,Z i)

with an optimal utility for themselves.

Based on the law of Walras we can restrict ourconsiderations to the market for risky assets. Allmarkets for risky financial assets are cleared be-cause we assume

(20) ∀n > 0I∑j=1

Z jn =I∑j=1

((1 − τ)X in + ωiτ) = 1.

It is obvious that investor i abides by their budgetconstraint. We therefore only need to verify thatthis portfolio maximizes their individual utility.For this purpose we assume that portfolio X i isoptimal. It then fulfills the first-order conditionat τ > 0 according to equation (16). To validatewhether portfolio Z i is optimal -- in the event thatthe tax rate vanishes -- we let this portfolio take theplace of X i in the equation (16) and also employτ = 0. As a result we receive

(21) U iµ (E − p) + 2U

iσ2 Ω · (Z i) = 0

with the new arguments

µ(Z) = p(X i0) + p(Xi) +

(Z i)T· (E − p),(22)

σ2(Z) =(Z i)T· Ω ·

(Z i).(23)

This equation is in fact fulfilled for Z i = (1− τ)X i +ωiτ by means of construction. This is what weintended to show.Konrad (1991) found a neutrality result that seemscomparable toour theorem. In fact his result differsin an important aspect from our model. Konradallows opportunity costs to be deducted from thetax base. This is a tax first advocated by Johanssonand Samuelson (taxation of economic gain). Ourmodel is much closer to the tax systems currentlyin place in most countries.

2.2 CARA Utility Functions

The literature typically discusses the concept ofabsolute risk aversion within the scope of the ex-pected utility theory. However, it can also be de-fined within µ-σ utility functions. For this purposelet us take a look at the quotient

(24) Si :=U iµ

U iσ2.

The function features constant absolute risk aver-sion when the quotient Si is not dependent onthe expected value µ, but rather dependent on thevariance σ2 (see proposition 3 in Lajeri-Chaherliand Nielsen 1993 or property 5 in Meyer 1987).

Proposition 2.2 (CARA utility function) Theinvestors have functions with constant absoluterisk aversion. Moreover, p and (X i

0,X i)i=1...,I con-

stitute an equilibrium for the tax rate τ > 0. Thenwe have an allocation (Z i

0,Z i)i=1,...,I as in (29) in

such a way that p and this allocation constitutean equilibrium for the tax rate τ = 0.

This result, too, is easily explained intuitively. Inthe case of constant absolute risk aversion theoptimal supply of risky assets remains constantwhen the investor’s income is subject to changes(no income effects). Hence the income tax has noinfluenceon the riskyportionof theoptimal portfo-lio, since the risky assets are taxed proportionallyand the redistribution affects the endowment ofthe risk-free assets. If, however, the optimal riskyassets are identical to a situation without taxes,we conclude from the law of Walras that also theoptimal risk-free assets must be in an equilibrium.

Proof:Due to rf > 0 we now have p0 < 1. The adequateconversion of the first-order condition (13) givesus

(25)U iµ

U iσ2

(E −

pp0

)−2Ω ·

((1 − τ)X i + ωiτ

)= 0 .

After aggregation through all investors and by tak-ing the market clearance into account we will ob-tain

(26)

(E −

pp0

) I∑i=1

U iµ

U iσ2= 2 Ω · 1.

At a constant risk aversion quotients are only de-pendent on the variance

(27)((1 − τ)X i + ωiτ

)T· Ω ·

((1 − τ)X i + ωiτ

).

To prove that the tax rate has no influence on theprices of risky assets, we will -- under Walras’ law-- concentrate on the risky assets and assume thatthe allocation (X i)i=1,...,I and the price vector p con-stitute an equilibrium. Then for these allocationsand a tax rate of τ > 0 the first-order conditions

175


in compliance with (25) have been met for everyinvestor. Hereby we are dealing with the quotientsU iµ

U iσ2at the position

(28) σ2(X) =((1 − τ)X i + ωiτ

)T· Ω ·

·((1 − τ)X i + ωiτ

).

Now we deal with the case without taxes. Let theequilibrated allocation be designated as (Z i)i=1,...,I .The first-order condition must be fulfilled for thisallocation as well and this time τ = 0 applies.However, the tax rate only appears in the argumentof the condition. We now select

(29) Z i := (1 − τ)X i + ωiτ,

and as a result equation (25) applies trivially. Thebudget constraints are met, because the risk-freeasset is selected precisely in such a way that onecan purchase the risk-free portfolio. The marketwill continue to be cleared since

(30)I∑i=1

Z i =I∑i=1

(1 − τ)X i + ωiτ = 1.

We have thus attained an equilibrium at the pricep.At the beginning of this section we referred to theconcept of absolute risk aversion (CARA) that istypically used with expected utility. The literatureon CARA is manifold (see, e.g., Meyer 1987). Char-acteristics of relative risk aversion within expectedutility are discussed, e.g., in Katz (1983), Briys andEeckhoudt (1985), and Hey (1985). The case ofrelative risk aversion is also intensely discussed.Going back to Mehra and Prescott (1985) it isusually stated that constant relative risk aversion(CRRA) is a reasonable assumption. Sinn (1989,p. 162) has concluded, that CRRA are the only util-ity functions that are compatible with empiricalresults of psychological experiments. Therefore,one could justifiably ask why we do not applyCRRA-functions here. The answer is simply thatwithin the µ-σ2 calculus there are no functionswith a constant relative risk aversion (for a proofsee Loeffler 1996a p. 32 f.).

2.3 Two Further Examples

Ifwewant to generate the result according towhichprices are dependent on the tax rate, then the risk-free interest rate may not vanish and there may beno CARA utility functions.

For such an example we examine just one sin-gle risky asset and two investors. One of the twodecision-makers owns a risky asset; the other ownsnothing. The matrix Ω degenerates to the varianceof the risky asset; for simplicity the matrix is setto 1. As a consequence E and p are numbers.The utility function for both investors is identicaland should be

(31) U(µ,σ2) =µ

1 + σ2.

It is strictly quasi-concave and thus satisfies therequirements needed for a unique solution of themaximization calculation. Starting from (13) theFOC for every investor is as follows:

(32) 0 =E − (1 + rf )p1 + σ2

−

− 2((1 − τ)X i + ωiτ

) µ(1 + σ2)2

.

Simplifying and inserting the arguments for µ andσ (considering that the initial endowment of risk-free asset is zero) leads us to

(33)(E − (1 + rf )p

)(1 +

((1 − τ)X i + ωiτ

)2)=

= 2((1 − τ)X i + ωiτ

) [(1 + rf (1 − τ))p(X

i) +

+(1 − τ)X i · (E − (1 + rf )p) + ωiτ(E − p)

].

We substitute Z i := (1 − τ)X i + ωiτ and formulatethe equation as follows:

(34) 0 = (Z i)2 + 2Z i(1 + rf (1 − τ)E − (1 + rf )p

p(X i) +

+ωiτrfp

E − (1 + rf )p

)− 1 .

This is a quadratic equation with the unknown Z i

which can be solved. The solution is dependent onthe initial endowment. One of the investors owns ashare (i.e. p(X i) = p), the other owns nothing (i.e.p(X i) = 0).

Investor with one share as initial endow-mentThe solution of the equation reads

(35) Z1 = −1 + rf (1 − τ) + τrfω1

E − (1 + rf )pp±

±

√(1 + rf (1 − τ) + ω1τrf

E − (1 + rf )pp)2+ 1.

176


Investor without a share as initial endow-mentThe solution of the equation reads

(36) Z2 = −1 + rf (1 − τ) + pτrfω2

E − (1 + rf )p±

±

√(1 + rf (1 − τ) + pτrfω2

E − (1 + rf )p

)2+ 1.

As a consequence of the market clearance andω1+ω2 = 1 andafter the additionof bothportfoliosand some rearrangement, the following results:

(37) E + 1 + rf (1 − τ) =

= ±√((1 + rf (1 − τ) + ω1τrf )p

)2+ (E − (1 + rf )p)2

±√(1 + rf (1 − τ) + pτrfω2

)2+ (E − (1 + rf )p)2 .

For the event of a uniform distribution of taxrevenue ωi = 1

2we have determined a numerical

solution, as shown in the table below. The otherparameters are rf = 5%, E = 2. Although twosolutions result, one of them leads straight to areturn of 5% for the risky assets and is thereforeeconomically meaningless. It is evident that theprice is contingent on the tax rate. In our model all

Table 1: Contingency of equilibrium price pon the tax rate τ

τ p

0% 0.7497

10% 0.7515

20% 0.7532

30% 0.7550

40% 0.7567

50% 0.7585

60% 0.7603

70% 0.7620

80% 0.7638

90% 0.7656

investors had the same tax rate. This assumption isof particular importance for our results. It is easyto demonstrate that with a heterogenous taxationprices will depend on the tax rate. To this end weassume that there is only one risky asset. Ω, p, andE degenerate to real numbers. Furthermore, weassume that investor 1 bears no tax (τ1 = 0) andinvestor 2 is fully taxed (τ2 = τ). Both investorshave utility functions of the form U = µ − 1

2σ2.

Then we have

(38) 0 = E − (1 + rf )p + ΩX1

for the first investor and

(39) 0 = E − (1 + rf )p + Ω((1 − τ)X2 + ω2τ

)for the second one. Additionally,X1+X2 = 1musthold in equilibrium. From these three equations

(40)(1 + rf )p − E

Ω= ω2

τ2 − τ

can be concluded by some simple rearrangement.Obviously, the price depends on the tax rate.

3 ConclusionAuthors contributing to the tax CAPM discussionusually neglect the fact that the Treasury typicallyredistributes the tax revenues that were collectedfrom investors. Whereas this may be acceptablefor a partial model, it is certainly unacceptable fora total model. Surprisingly, this aspect has beenoverlooked by a large number of authors. Basedon a total model that contained the redistribu-tion we can show that equilibrium security pricesare not affected by taxation, if either the risklessrate vanishes or investors show constant absoluterisk aversion. Both results can be explained byeconomic reasoning. However, neither vanishinginterest rates nor CARA utility functions are re-alistic cases. Since there seems to be no way toextend our results beyond those two special cases,we conclude as a rule that taxes do have an impacton security prices in the CAPM as a rule. Furtherresearch should try to reveal whether this impactis substantial or negligible.

References

Bottazzi, Jean-Marc, Thorsten Hens, and Andreas Löffler(1998): Market Demand Functions in the Capital Asset PricingModel, Journal of Economic Theory, 79 (2): 192-206.

Brennan, Michael J. (1970): Taxes, Market Valuation and Cor-porate Financial Policy, National Tax Journal, 23: 417-427.

Briys, Eric and Louis Eeckhoudt (1985): Relative Risk Aver-sion in Comparative Statics: Comment, American EconomicReview, 75 (1): 281-283.

Buchholz, Wolfgang and Kai A. Konrad (2000): Risiko undSteuern, in: Wolfgang Buchholz and Alois Oberhauser (eds.):Probleme der Besteuerung, 3, Duncker & Humblot: Berlin,63-139.

Eikseth, HansM. and Snorre Lindset (2009): A Note on CapitalAsset Pricing and Heterogeneous Taxes, Journal of Bankingand Finance, 33 (3): 573-577.

Hens, Thorsten, JörgLaitenberger, andAndreasLöffler (2002):Two Remarks on the Uniqueness of Equilibria in the CAPM,Journal of Mathematical Economics, 37 (2): 123-132.

177


Hey, John D. (1985): Relative Risk Aversion in ComparativeStatics: Comment, American Economic Review, 75 (1): 284-285.

Jonas, Martin, Andreas Löffler, and Jörg Wiese (2004): DasCAPM mit deutscher Einkommensteuer, Die Wirtschaftprü-fung, 57 (17): 898-906.

Katz, Eliakim (1983): Relative Risk Aversion in ComparativeStatics, American Economic Review, 73 (3): 452-453.

König, Rolf J. (1990): Ausschüttungsverhalten von Aktienge-sellschaften, Besteuerung und Kapitalmarktgleichgewicht,Steuer- und Wirtschaftsverlag: Hamburg.

Konrad, Kai A. (1991): Risk Taking and Taxation in CompleteCapital Markets, The Geneva Papers on Risk and InsuranceTheory, 16 (2): 167-177.

Lajeri-Chaherli, Fatma and Lars T. Nielsen (1993): Risk Aver-sion and Prudence, Working Paper, INSEAD.

Lintner, John (1965): The Valuation of Risky Assets and theSelection of Risky Investments in Stock Portfolios and CapitalBudgets, The Review of Economics and Statistics, 47 (1): 13-37.

Litzenberger, Robert H. and Krishna Ramaswamy (1979): TheEffect of Personal Taxes and Dividends on Capital Asset Prices:Theory and Empirical Evidence, Journal of Financial Eco-nomics, 7 (2): 163-195.

Litzenberger, Robert H. and Krishna Ramaswamy (1980): Div-idends, Short Selling Restrictions, Tax-Induced Investor Clien-teles, and Market Equilibrium, The Journal of Finance, 35 (2):469-482.

Löffler, Andreas (1996a): Capital Asset Pricing Model mitKonsumtion: Eine gleichgewichtstheoretische Untersuchung,Deutscher Universitäts-Verlag: Wiesbaden.

Löffler, Andreas (1996b): Variance Aversion Implies µ-σ2-Criterion, Journal of Economig Theory, 69 (2): 532-539.

Mehra, Rajnish and Edward C. Prescott (1985): The EquityPremium: A Puzzle, Journal of Monetary Economics, 15 (2):145-161.

Meyer, Jack (1987): Two-Moment Decision Models and Ex-pected Utility Maximization, American Economic Review, 77(3): 421-430.

Mossin, Jan (1996): Equilibrium in a Capital Asset Market,Econometrica, 34 (4): 768-783.

Rapp, Marc S. and Bernhard Schwetzler (2007): DasNachsteuer-CAPM im Mehrperiodenkontext: Stellungnahmezum Beitrag von Dr. Jörg Wiese, FinanzBetrieb, 2007 (2):108-116.

Rapp, Marc S. and Bernhard Schwetzler (2008): EquilibriumSecurity Prices with Capital Income Taxes and an ExogenousInterest Rate, Finanzarchiv, 64 (3): 425-442.

Sharpe, William F. (1964): Capital Asset Prices: A Theory ofMarket Equilibrium under Conditions of Risk, The Journal ofFinance, 19 (3): 425-442.

Shell, Karl (forthcoming): Sunspot Equilibria, in: StevenDurlauf and Lawrence Blume (eds.): The New Palgrave: ADictionary of Economics, Palgrave Macmillan: New York.

Sinn, Hans-Werner (1989): Economic Decision under Uncer-tainty, 2nd ed. , Physica: Heidelberg.

Stiglitz, Joseph E. (1972): Taxation, Risk Taking, and the Allo-cation of Investment in a Competitive Economy, in: Michael C.Jensen (ed.):Studies in theTheory of CapitalMarkets, Praeger:New York et al. , 294-374.

Wenger, Ekkehard (1983): Gleichmäßigkeit der Besteuerungvon Arbeits- und Vermögenseinkünften, Finanzarchiv, 41 (2):207-252.

Wiese, Jörg (2004): Unternehmensbewertung mit demNachsteuer-CAPM?, Working Paper, Ludwig-Maximilians-Universität: München.

Wiese, Jörg (2006a): Das Nachsteuer-CAPM imMehrprodukt-kontext, FinanzBetrieb, 2006 (8): 242-248.

Wiese, Jörg (2006b): Komponenten des Zinsfußes in Un-ternehmensbewertungskalkülen: TheoretischeGrundlage undKonsistenz, Lang: Frankfurt/Main.

Wiese, Jörg (2007): Unternehmensbewertung und Abgel-tungssteuer, Die Wirtschaftsprüfung, 60 (9): 368-375.

Biographies

Lutz Kruschwitz is chair of Banking and Fi-nance at the Freie Universität Berlin since 1991.He was born in 1943 in Berlin, Germany, and isa graduate in business administration from theFreie Universität Berlin. Kruschwitz received hisdoctoral degree in 1970 and his Habilitation (post-doctoral lecture qualification) in 1975.His researchinterests include corporate finance and valuationtheory in particular.He holds anHonoraryDoctor-ate from theUniversity of Tübingen, Germany, andis serving as Honorary Professor at the Universityof Vienna, Austria.

Andreas Löffler is chair of Finance and Invest-ment at the University of Paderborn since 2008.He was born in 1964 in Szeged, Hungary, and isa graduate in mathematics from the University ofLeipzig. Löffler received his first doctoral degree inmathematics in 1992, his second doctoral degree in1995 and his Habilitation in 1999. His research in-terests include decision theory, equilibrium theoryand valuation theory.

178


VHB-JOURQUAL2: Method, Results, and Implications of the GermanAcademic Association for Business Research's Journal Ranking Ulf Schrader, Thorsten Hennig-Thurau


BuR - Business Research

Official Open Access Journal of VHB

Verband der Hochschullehrer für Betriebswirtschaft e.V.

Volume 2 | Issue 2 | December 2009 | 180-204

180

1 Introduction VHB-JOURQUAL represents the official journal

ranking of the German Academic Association for

Business Research (Verband der Hochschullehrer

für Betriebswirtschaftslehre – VHB). It rates and

ranks international and German-language academic

journals which are considered relevant for German-

speaking business researchers based on the quality

assessments of VHB members who comprise more

than 90% of all professors and researchers with a

Ph.D. in business administration at German, Aus-

trian, and Swiss universities. Since its initial publi-

cation in 2003 (Hennig-Thurau, Walsh, and

Schrader 2004), the ranking has become the most

prominent business research-journal ranking in

German-speaking countries.

The results of an online survey of VHB members in

2007 (n = 489) give evidence that VHB-JOUR-

QUAL is widely accepted and broadly used for

evaluating the scientific performance of business

scholars in Austria, Germany and the German-

speaking part of Switzerland (Schrader and Hennig-

Thurau 2007). 59 % of the respondents assessed

VHB-JOURQUAL as “good” or “very good”, while

only 9 % held a negative attitude toward the rank-

ing. VHB-JOURQUAL is of special importance for

the formal post-doctoral assessment in German-

speaking countries – the so-called “Habilitation,” a

traditional requirement for obtaining a full profes-

sorship – and the appointment of full professors; in

each case, 54 % of the respondents judged the rank-

ing being of “high” or “very high” relevance for their

institutions. Other areas for which VHB-

JOURQUAL is considered to have substantial rele-

vance include the hiring and evaluation of assistant

professors (“Junior-Professoren”), research per-

formance evaluations (with a possible impact on

budget allocation), and the evaluation of the rapidly

VHB-JOURQUAL2: Method, Results, and Implications of the German Academic Asso-ciation for Business Research's Journal Ranking

Ulf Schrader, Institute of Vocational Education and Work Studies, Division of Work Studies: Economy and Sustainable Consumption, Technical

University of Berlin, Germany, E-Mail: [email protected]

Thorsten Hennig-Thurau, Department of Marketing and Media Research, Bauhaus-University of Weimar, Germany, and Faculty of Management,

Cass Business School, City University London, E-Mail: [email protected]

Abstract

VHB-JOURQUAL represents the official journal ranking of the German Academic Association for Business

Research. Since its introduction in 2003, the ranking has become the most influential journal evaluation

approach in German-speaking countries, impacting several key managerial decisions of German, Aus-

trian, and Swiss business schools. This article reports the methodological approach of the ranking’s second

edition. It also presents the main results and additional analyses on the validity of the rating and the un-

derlying decision processes of the respondents. Selected implications for researchers and higher-education

institutions are discussed.

Keywords: Journal ranking, rating, impact factors, VHB-JOURQUAL

This article was commissioned by the Editor-in-Chief and went through the review process. Manuscript

received September 10, 2009, accepted by Sönke Albers (Editor-in-chief) November 9, 2009.





181

growing number of cumulative doctoral disserta-

tions.

VHB-JOURQUAL is considered a major driver of

the radical transformation of the German business-

administration community into a much more re-

search-focused and internationally active academic

discipline (Homburg 2008). In an environment

which Simon (1993) described as a “black hole” –

with German-speaking business scholars absorbing

their international colleagues’ findings, but giving

nothing back to the international community – ,

VHB-JOURQUAL offered for the first time a com-

parison of the scientific quality of articles published

in German-language journals with those published

in international journals. As the ranking provided

strong evidence that the quality of even the best

German-language journals is perceived to be sub-

stantially lower than the quality of leading interna-

tional journals, VHB-JOURQUAL has since then

become a strong motivating force for German schol-

ars to publish their best work internationally and

enter “global competition”.

This article describes the methodology of VHB-

JOURQUAL and reports specific features and key

results of the second edition of the ranking (VHB-

JOURQUAL2), which is based on a survey con-

ducted among VHB members in 2008. We also

investigate the ranking’s validity and provide con-

cluding remarks on the benefits and limitations of

VHB-JOURQUAL.

2 Background: Survey-based vs.

citation based journal-rankings There are two basic approaches for conducting aca-

demic-journal rankings: citation-based rankings

(e.g., Azar and Brock 2008; Ritzberger 2008; Vieira

2008) and survey-based rankings (e.g., Bräuninger

and Haucap 2002; Hennig-Thurau, Walsh, and

Schrader 2004). Hybrid rankings, as a third ranking

type, combine both approaches (e.g., Franke and

Schreier 2008; Schulze, Warning, and Wiermann

2008). We will compare the different approaches in

terms of strengths and weaknesses and explain how

VHB-JOURQUAL addresses them.

Citation-based journal rankings such as the ISI

Journal Citation Reports are often considered “ob-

jective” (e.g., Ritzberger 2008). Here, the ranking

position of a journal depends on the number of

citations the papers in the journal receive. The idea

behind it is that citations are “the scientific commu-

nity’s version of dollar voting by consumers for

goods and services” (Laband and Piette 1994a: 641).

Similar to the way economic theory considers pay-

ment at the cash desk as the result of a quality as-

sessment by consumers, a citation-based ranking

treats a citation as a proof of perceived quality.

However, there are two major caveats to this ap-

proach, particularly when audiences are interested

in the scientific quality of a journal (e.g., Schulze,

Warning, and Wiermann 2008):

• Quality vs. impact. Citations are not objective

indicators for an article’s scientific quality. The

decision to accept a submitted article, making it

available for citations, is subjective and not only

determined by scientific quality (Blank 1991;

Frey and Rost 2008; Laband and Piette 1994b;

Starbuck 2004), and authors’ decision to cite an

article is, in addition to quality, also influenced

by the article’s type and topic (e.g., state-of-the-

art review), its shortcomings as well as citation

cartel memberships (e.g., Fabel and Heße

1999). Thus, citation indexes do not measure

the scientific quality but rather the impact of a

journal. The ISI Journal Citation Reports ranks

a journal according to its “impact factor”, not to

a quality index. Impact and scientific quality

should not be treated as synonyms since the

empirical correlations between them can be

weak or even negative in some cases (e.g., Maier

2006; Schlinghoff and Backes-Gellner 2002).

• Data availability. Citation data can only be

obtained from a limited number of sources and

is not available for a substantial number of

journals. The market-dominating provider of

interdisciplinary citation indexes is Thomson

Reuters with the Social Science Citation Index

(SSCI), the Science Citation Index-Expanded

(SCI-X), the Arts & Humanities Citation Index

(A&HCI), the Conference Proceedings Citation

Index - Science (CPCI-S), and the Conference

Proceedings Citation Index Social - Science &

Humanities (CPCI-SSH), constituting the “ISI

Web of Knowledge.” Elsevier introduced the

competing database Scopus in 2004, with

limited market penetration so far. For business

researchers, the most important index is the

SSCI with its business, business/finance, and

management categories, which – as reported

later in the section on ranking validation –

cover only about 20 % of the journals





182

considered relevant for business scholars and

included in VHB-JOURQUAL2, with only one

German-language journal being represented

there (BFuP – Betriebswirtschaftliche For-

schung und Praxis). Considering all SSCI cate-

gories and other indexes mentioned above,

these numbers will be only slightly higher

(Clermont and Schmitz 2008; Dyckhoff and

Schmitz 2007). In 2009, the ISI covered in total

nine and Scopus 12 German-language journals

ranked in VHB-JOURQUAL2, which have

predominantly low ratings (Clermont 2009).

Consequently, in German-speaking countries

citation-based rankings are hardly able to

measure publication productivity of business

scholars (Albers 2009; Dilger 2000).

Survey-based rankings also face limitations. How-

ever, these obstacles appear less systematic and

strongly depend on the mode of data collection and

analysis when compared to those of citation-based

rankings. Major threats to the validity of survey-

based rankings include the following issues:1

• Sample. Respondents from whose assessments

survey-based rankings are constructed are not

always suited for assessing the scientific quality

of academic journals. VHB-JOURQUAL aims to

overcome this potential threat by including only

VHB members who, at a minimum, hold a post-

doctoral position, with the majority being full

professors. VHB-JOURQUAL also requires

active readership of a journal to rate its quality

(it first asks the respondents to indicate which

journals they have recently read and only in a

second step asks respondents to evaluate those

journals only) and uses an expertise factor to

account for the respondents’ differing levels of

expertise (see also Heischmidt and Gordon

1993; Extejt and Smith 1990).

• Too few or too many journals. Survey-based

rankings are often forced to limit the number of

included journals to avoid overstraining

respondents. VHB-JOURQUAL assures com-

pleteness of relevant journals by using a

multistage process starting with the Journal

Quality List by Anne-Wil Harzing (for VHB-

JOURQUAL2: Harzing 2007), limiting the

1 Please note that additional information on the VHB-JOUR-

QUAL2 methodology is provided later in this article.

danger of overstraining by a highly customized

online survey design. Specifically, all journals

are assigned to business sub-disciplines (e.g.,

accounting, finance), and respondents are asked

to evaluate only journal titles relevant for their

specific sub-discipline.

• Strategic answers. Opponents of the survey-

based approach argue that researchers not

always evaluate journals according to their

actual quality perception, but rather in a way

that is best for them (e.g., Schulze, Warning,

and Wiermann 2008). Specifically, given that

journal rankings can influence careers, scholars

will have an interest that the journals they

publish in or serve for as editors or reviewers

are highly ranked; something which they can

influence through their own ratings. As the

incentive for such strategic answers strongly

depends on the individual researcher’s impact

on a journal’s rating, VHB-JOURQUAL only

considers journals with at least ten individual

ratings. Moreover, ratings in VHB-JOUR-

QUAL2 are not anonymous – each respondent

had to use a unique ID and agree that his data

could be linked to his personal information by

the authors of this study. This should reduce the

motivation for strategic answers, since evident

over- or underrating may become overt to the

VHB-JOURQUAL authors. In addition, outlier

judgments were systematically removed as will

be explained later in more detail.

Finally, hybrid rankings combine data from expert

surveys and citation indices. This extends the num-

ber of journals and bases the evaluation on a

broader foundation. However, the hybrid approach

does not heal limitations inherent in the sources

which serve as necessary inputs such as missing

citation data for German-language journals.

3 Measuring journal quality in

VHB-JOURQUAL Survey-based rankings usually solely focus on the

overall quality of the articles that are published by a

certain journal. In VHB-JOURQUAL, we conceptu-

alize overall scientific quality of a journal as being

defined by two quality dimensions which are meas-

ured separately, namely the quality of the articles

published in a journal (article quality) and the qual-

ity of the review process of the journal (review qual-

ity). Both quality dimensions are treated as forma-





183

tive indicators of overall journal quality (for a simi-

lar approach, see Rossiter 2002). Our distinction

between article and review quality draws from qual-

ity literature in related fields such as service man-

agement, which has often been used as a template

for higher-education research (e.g., Hennig-Thurau,

Langer, and Hansen 2001).

Service research distinguishes between outcome-

related quality aspects and those aspects which are

related to the process of the service production as

dimensions of overall (perceived) service quality

(e.g., Brady and Cronin 2001). While article quality

serves as the equivalent of outcome quality in the

context of academic-knowledge generation, review

quality refers to the process of how articles (and

their authors) are treated by the respective journal.

In that sense, article and review quality measure

different dimensions of academic-knowledge gen-

eration. Not all academics (as “customers”) experi-

ence both quality dimensions; while article quality

can be assessed by all readers of a journal’s articles

(including those who had only limited contact with a

journal), the assessment of review quality requires

deeper experiences and insight into a journal’s in-

ternal processes. The inclusion of review quality not

only helps to capture scientific quality in a more

comprehensive way, but also adds dynamics and

flexibility, as changes in the review process will be

experienced much faster than their manifestation in

printed articles.

The VHB-JOURQUAL index, on which the rating

and ranking of journals is based, combines the two

quality dimensions of article quality and review

quality in a weighted additive composition (see

equation 1). Both quality dimensions are measured

with single items which is an adequate procedure

for expert surveys with formative measures, as is the

case with VHB-JOURQUAL (Rossiter 2002).

(1)

with JQIJ : VHB-JOURQUAL index value of journal J (10-point scale from 1 = ‘very low’ to 10 = ‘very high’),

AQJ,i : Scientific quality of articles in journal J as perceived by respondent i (on a 10-point

scale from 1 = ‘very low’ to 10 = ‘very high’),

RQJ,i : Review quality of journal J as perceived by respondent i (on a 10-point scale from 1 = ‘very low’ to 10 = ‘very high’),

nJ,AQ : Number of respondents who have evalu-ated the article quality of journal J,

nJ,RQ : Number of respondents who have evalu-ated the review quality of journal J,

Ei : Weighting factor for the expertise of re-spondent i,

aJ, bJ : Weighting factors for journal J, with aJ + bJ = 1.

Article quality is measured with the item “I consider

the scientific standard of articles published in this

journal to be … [Please indicate a number between 1

= ‘extremely low’ und 10 = ‘extremely high’]”. The

original German wording of this item is: “Das wis-

senschaftliche Niveau der in dieser Zeitschrift veröf-

fentlichten Artikel halte ich für…[Bitte Zahl zwi-

schen 1 = ‘extrem gering’ und 10 = ‘extrem hoch’

angeben].“ The item is evaluated by respondents

who have read at least one new article in the respec-

tive journal in the last five years (VHB-JOUR

QUAL2: 2003-2007).

Review quality is measured with the item “The sci-

entific quality of the review process is … [Please

indicate a number between 1 = ‘extremely low’ und

10 = ‘extremely high’]”. The original German wor-

ding of the item is: “Die wissenschaftliche Qualität

des Review-Prozesses ist…[Bitte Zahl zwischen 1 =

‘extrem gering’ und 10 = ‘extrem hoch’ angeben].“

The scientific quality of the review process is defined

as “standards for submissions requested by review-

ers and/or editors” (in German: “die Ansprüche, die

Gutachter und/oder Schriftleiter bzw. Editoren an

eingereichte Beiträge stellen”). Review quality is

rated only by those who have either submitted at

least one paper within the last five years or know the

review process as reviewers or editors of the journal.

In VHB-JOURQUAL1 only authors who had sub-

mitted papers were allowed to evaluate the review

process. Also letting reviewers and editors evaluate

review quality leads to a higher number of review-

process evaluations and reduces the likelihood that

ratings and rankings are mainly or only based on

article-quality assessments.

We weight the formative indicators of article quality

and review quality equal (both aJ and bJ = 0.5),

assuming both play the same role for determining

overall scientific quality of a journal. However, to

account for the limited number of review evalua-





184

,

,

,

,,

,

,

0 for 0

0.1 for 1

0.2 for 2( ),

0.3 for 3 5

0.4 for 5 10

0.5 for 10

J J RQ

J J RQ

J J RQ

J J RQJ J RQ

J J RQ

J J RQ

b n

b n

b nb f n with

b n

b n

b n

= = = = = = = = ≤ <

= ≤ < = ≥

tions for several journals and the loss in reliability

associated with a small sample size for the review-

quality indicator, we adjusted the weights for article

and review quality if the number of review quality

assessments fell short of a certain threshold. Equa-

tion 2 lists the adjustments in weighting the two

quality criteria for different numbers of ratings of a

journal’s review quality. If no respondent has rated

a journal’s review process, the weight for b becomes

0 and (since aJ = 1 – bJ) the overall JOURQUAL

rating is solely based on the article-quality indicator

(with aJ = 1).

(2)

Expertise is included in the calculation of the VHB-

JOURQUAL2 overall quality score as a weighting

factor since we expect a positive correlation between

the expertise of a respondent and the validity of her

or his assessment. We argue that scholars who have

published in a variety of journals, and particularly in

leading international journals, will be better able to

judge the quality of articles and review processes

than colleagues who lack that experience. Specifi-

cally, we operationalize expertise by three indica-

tors: (a) the number of different journals a respon-

dent has published in during the last five years; (b)

the number of publications in high-quality journals

(defined as journals with an unweighted VHB-

JOURQUAL index value ≥ 7) during the last five

years; and (c) the international experience gathered

by publications in English-language high-quality

journals during the last five years. For the first two

indicators the raw values are transformed to scales

ranging from 1 to 2; for indicator (c) respondents

with at least one international high-quality journal

publication receive a 2, all others a 1. The three indi-

cators are multiplied and then rescaled to an exper-

tise scale ranging from 1 to 5. As a consequence, the

evaluations of the respondents scoring with the

highest expertise are weighted five times higher

than the assessments of the respondents with the

lowest expertise. We prefer a multiplicative function

of expertise over an additive one, as we argue that

academic expertise builds up in a non-linear way,

with the three indicators’ impact on overall exper-

tise not being independent.

While the journal ranking results directly from the

different VHB-JOURQUAL index values, the rating

assigned to an index value requires a categorization.

VHB-JOURQUAL sorts journals into rating catego-

ries based on their index values, using absolute

numbers as thresholds for defining the categories

for the lack of a more objective categorization. Table

1 lists the six rating categories ranging for A+ to E

and the respective VHB-JOURQUAL scores.

Table 1: Thresholds for Rating Categories

VHB-JOURQUAL

Rating Category

VHB-JOURQUAL

Index Value

A+ 9 ≤ JQIJ

A 8 ≤ JQIJ < 9

B 7 ≤ JQIJ < 8

C 6 ≤ JQIJ < 7

D 5 ≤ JQIJ < 6

E JQIJ < 5

4 Survey and sample of VHB-

JOURQUAL2 The VHB-JOURQUAL2 survey was conducted from

February to April 2008. Every VHB member with a

registered email address (N = 1,555) received an

invitation email from the VHB chairmen and the

authors of the study which contained a unique link

to an online questionnaire. Every participant went

through the highly individualized survey procedure

depending on his or her specific research fields,

reading habits, submission activities and reviewer

or editor positions. 1,011 respondents started the

evaluation process (response rate of 65 %), about

600 fully completed the questionnaire. Many re-

spondents who “dropped out” had actually pro-

ceeded through major parts of the survey. We in-

cluded all responses in our calculations regardless of

technical completion.

The questionnaire included a total of 1,633 journals.

In addition to academic journals in a narrow sense,

the list also comprises yearbooks and proceedings

with a homogeneous review process. This is due to

the observation that in some business sub-

disciplines (such as information systems) refereed

conference proceedings fulfill similar tasks than

actual journals. The list of journals for VHB-





185

JOURQUAL2 was selected in a multi-step process.

First, we synchronized the VHB-JOURQUAL1 jour-

nal list with the established Harzing-List of busi-

ness-administration journals (Harzing 2007). Sec-

ond, we assigned the journals to different business

sub-disciplines which strongly overlapped with the

16 sections of the VHB (“Wissenschaftliche Kom-

missionen”; e.g., finance and banking, taxation,

international management). Third, in a joint effort

with all chairpersons of the VHB sections, we added

new or previously overlooked journals, eliminated

those which have ceased publication and adjusted

the categorization. Forth, we invited all VHB mem-

bers to complete the journal list and asked them to

indicate which journals they actually read in an

online pre-survey (n = 489). The final list then in-

cluded all journals which were either indicated by at

least two respondents in the pre-survey or evaluated

by at least five respondents in VHB-JOURQUAL1.

Journals which did not meet one of these require-

ments were excluded as we expected them not to

receive the minimum number of ten evaluations in

the main survey required for the inclusion in the

final ranking.

To reduce the number of strategic answers, we lim-

ited the anonymity of the study by informing the

participants in the invitation mail that the authors

of the study would be able to assign every rating to

the individual respondent. Also, indicating the

number of submissions to and publications in a

journal was not sufficient; respondents also had to

name respective papers’ short titles. This informa-

tion was used to assure a valid calculation of expert

factors and that review processes are only evaluated

by respondents who have actually experienced them

within the period of observation. While we are

aware that these measures cannot completely re-

move all kinds of strategic behavior, they certainly

increased the psychological barriers to behave in

such a way.

To further improve the quality of our data, we ex-

cluded outlier ratings when calculating the VHB-

JOURQUAL2 ratings, as we assumed outliers to be

based on misunderstandings or strategic misevalu-

ations. Specifically, for each journal, we kept all

responses within the 99 % confidence interval (two-

sided) for both dimensions of quality (i.e. article and

review quality) and deleted those responses outside

the confidence interval. This procedure resulted in

the removal of a total of 315 individual ratings.

About 90 % of these deleted ratings were below the

confidence interval; thus, the risk of overrating

seems to be lower than the risk of underrating.

5 VHB-JOURQUAL2 results

5.1 General Results

742 journals received more than 10 ratings in VHB-

JOURQUAL2; journals which received less than 10

ratings were excluded to assure a sufficient level of

reliability (Web-Appendix 1 contains the alphabeti-

cal list of all 742 journals).2 As VHB-JOURQUAL2

aims at business scholars, we only included journals

in the ranking which – in addition to being read by

at least 10 respondents – could be assigned to at

least one sub-discipline of business administration

(e.g., accounting, marketing) or whose review proc-

ess was evaluated by a minimum of 5 respondents.

666 journals met at least one of these criteria and

were subsequently included in the VHB-

JOURQUAL2 journal ranking.

Figure 1 shows the distribution of VHB-JOUR-

QUAL2 (JQ2) ratings for all 666 journals for the

overall VHB-JOURQUAL2 index value as well as

the two quality dimensions and lists descriptive

information. The mean of the overall JQ2 score is

6.22, equal to a C-rating.

Figure 2 displays the distributions for the quality

weighting parameters and the respondent expertise

parameter. The weighting factor for the two quality

criteria varied between 0 and 0.5, with an average

weighting for the review process of .24 (standard

deviation = .18).

As can be seen in Figure 3, the expertise factor for

the respondents has a mean value of 1.36 (standard

deviation = .58) and is positively skewed (the forth

quartile ranges from 1.74 to 5), that is, relatively few

researchers have very high expertise scores. This is

consistent with the finding of Dyckhoff and Schmitz

2 Please note that this number differs from the initial results pub-

lished on the VHB website as we deleted journals which were duplications of other journals with slightly different names or had ceased publication before the time frame considered in this rank-ing; we thank Robert Hofmeister from the Thurgau Institute of Economics at University of Konstanz for his valuable input. For duplicated journals, the evaluation scores were merged on an in-dividual respondent level; if more than one journal version was rated by a respondent, the mean of his or her ratings for the re-spective journal was considered. As a result of this merging proc-ess, the scores for these journals can differ from those originally reported on the VHB website.





186

Figure 1: Distribution of Journal-Quality Measures in the Sample and Descriptive Statistics

(2007) that about two out of three German profes-

sors had no international refereed publication be-

tween 1990 and 2004; something which has

changed only recently (Homburg 2008). Figure 3

also shows the distribution of the three expertise

indicators.

Figure 2: Distribution of Quality Weighting

Table 2 lists the A+ and A ranked journals according

to their overall quality index value and provides

additional information on the assessments of the

VHB-JOURQUAL quality dimensions and changes

between JQ1 and JQ2. The full list of all ranked

journals is reported in Web-Appendix 2. In addition

to the expertise-weighted index values, we also re-

port the unweighted quality assessments for each

journal.

Within the top ten journals are four marketing and

three finance journals, with the Journal of Finance

being listed as the number one journal. Administra-

tive Science Quarterly, ranked fourth, has the high-

est JQ2 index value of all general management

journals. 14 Journals (= 2 %) are rated A+, 50 (= 8

%) are rated A, 152 (= 23 %) are rated B, 186 (= 28

%) are rated C, 143 (= 22 %) are rated D, and 121 (=

18 %) are rated E. The best German-language jour-

nal is the B-rated Wirtschaftsinformatik (since

2009 also available in English as Business & Infor-

mation Systems Engineering) ranked 169, followed

by Schmalenbachs Zeitschrift für betriebswirt-

schaftliche Forschung (zfbf) ranked 177.

Considering the changes from JQ1 to JQ2, the rat-

ings on average have decreased. Taking into account

only those 326 Journals which were included in

both rankings, the mean score for the overall quality

has fallen from 6.86 to 6.22; 275 journals (or 84 %)

received a lower and only 51 a higher quality as-

sessment than five years ago. This indicates that

respondents have not systematically overrated their

preferred journals, but displayed a critical attitude

in general. This attitude might be the result of a

growing sensitivity among VHB members with re-

gard to academic journal quality; a trend which has

certainly been strengthened by the increasing num-

ber of VHB members which have experienced inter-

national journal review processes.

Overall JQ2 Index Value Article Quality Review Quality

Frequency

Mean = 6.22 SD = 1.50 n = 666 Mean = 6.16 SD = 1.48 n = 666 Mean = 6.51 SD = 1.91 n = 530

Fre

qu

en

cy

Weighting Review Process

Frequency

Mean = 0.24 SD = .18 n = 666

Fre

qu

en

cy





187

Figure 3: Distribution of Respondent Expertise Parameters in the Sample and Descriptive

Statistics

The results also reveal that the double blind review-

ing process has become a condition sine qua non for

a high-quality journal among VHB members, fol-

lowing international standards. The journals with

the highest decrease (by percentage) in quality ra-

ting did not employ a rigorous review process dur-

ing the last five years (Appendix 3 lists the journals

and their respective changes since JQ1). Most of the

journals with a high decrease in quality perception

had already a below average rating in JQ1.

Table 3 lists the best-rated journals for 16 business-

administration sub-disciplines; the disciplines were

selected based on the VHB sections structure. The

table also reveals to which extent the journal-quality

ratings are affected by judgments of researchers

who do not belong to a specific sub-discipline. In

other words: Does the “core audience” of a journal

judge its quality differently than other scholars?





188

Table 2: A+ and A ranked Journals in VHB-JOURQUAL2

Rank Journal ISSN

JQ2 rating cate-gory

JQ2 index value

Change in % JQ2 vs. JQ1

JQ2 index value uw

Change in % w vs uw JQ2

Mean AQ

SD AQ Mean RQ

SD RQ n AQ n RQ Weight of RQ

1 Journal of Finance 0022-1082 A+ 9.80 1.86 9.79 0.14 9.80 0.52 9.79 0.43 108 5 0.4

2 American Economic Review 0002-8282 A+ 9.75 1.42 9.68 0.68 9.67 0.83 9.87 0.50 121 8 0.4

3 Review of Financial Studies 0893-9454 A+ 9.48 2.32 9.46 0.16 9.38 1.05 9.71 0.49 41 4 0.3

4 Administrative Science Quarterly 0001-8392 A+ 9.48 1.74 9.44 0.38 9.21 1.07 9.75 0.44 249 11 0.5

5 Journal of Marketing 0022-2429 A+ 9.46 -0.85 9.43 0.27 9.49 0.77 9.43 0.97 122 22 0.5

6 Journal of Consumer Research 0093-5301 A+ 9.44 0.49 9.34 1.04 9.12 1.10 9.91 0.29 75 6 0.4

7 Journal of Financial Economics 0304-405X A+ 9.43 -1.10 9.39 0.39 9.67 0.64 8.47 0.58 60 2 0.2

8 Information Systems Research 1047-7047 A+ 9.42 6.74 9.41 0.14 9.28 0.96 9.75 0.47 36 4 0.3

9 Journal of Marketing Research 0022-2437 A+ 9.34 -4.10 9.28 0.62 9.49 0.89 9.18 1.28 99 13 0.5

10 Marketing Science 0732-2399 A+ 9.29 -4.60 9.30 -0.11 9.51 0.79 9.07 1.39 78 12 0.5

11 Management Science 0025-1909 A+ 9.20 -0.97 9.17 0.36 9.30 0.93 9.11 1.15 330 47 0.5

12 Operations Research 0030-364X A+ 9.19 4.59 9.16 0.24 8.79 1.43 9.77 0.43 61 9 0.4

13 Academy of Management Journal 0001-4273 A+ 9.08 -0.83 8.97 1.23 8.86 1.29 9.29 1.15 289 41 0.5

14 Academy of Management Review 0363-7425 A+ 9.07 5.96 8.99 0.85 8.65 1.33 9.48 1.05 266 17 0.5

15 Journal of Financial and Quantitative Analysis

0022-1090 A 8.95 0.72 8.70 2.77 8.98 0.86 8.88 1.37 58 4 0.3

16 RAND Journal of Economics (for-merly: Bell Journal of Economics)

0741-6261 A 8.93 -2.49 8.91 0.23 8.90 1.00 8.98 0.94 70 7 0.4

17 Mathematical Programming 0025-5610 A 8.92 3.96 8.88 0.52 8.80 1.62 10.00 16 1 0.1





189

Table 2 continued: A+ and A ranked Journals in VHB-JOURQUAL2

Rank Journal ISSN


JQ2 index value


JQ2 index value uw


Mean AQ

SD AQ Mean RQ


18 Organization Science 1047-7039 A 8.90 0.11 8.89 0.11 8.84 1.13 8.95 1.10 92 11 0.5

19 Journal of Accounting and Economics 0165-4101 A 8.89 0.83 8.66 2.49 9.16 1.03 7.80 2.17 62 2 0.2

20 MIS Quarterly 0276-7783 A 8.84 n.a. 8.78 0.64 8.62 1.56 9.34 0.78 73 4 0.3

21 Journal of International Business Studies JIBS

0047-2506 A 8.81 2.15 8.71 1.11 8.89 1.02 8.73 1.46 64 17 0.5

22 Review of Accounting Studies 1380-6653 A 8.79 -0.89 8.75 0.52 8.59 1.19 9.09 1.12 51 6 0.4

23 Accounting Review 0001-4826 A 8.78 1.27 8.68 1.19 8.82 1.35 8.69 1.47 79 3 0.3

24 Journal of Labor Economics 0734-306X A 8.71 n.a. 8.66 0.66 8.74 0.97 8.64 0.52 20 3 0.3

25 Journal of Risk and Insurance 0022-4367 A 8.62 n.a. 8.36 3.04 8.20 1.52 9.61 0.84 23 4 0.3

26 Transportation Science 0041-1655 A 8.60 0.94 8.51 1.05 8.40 1.09 8.90 0.77 37 7 0.4

27 Journal of the Academy of Marketing Science

0092-0703 A 8.50 -4.29 8.45 0.61 8.38 1.08 8.63 0.91 78 12 0.5

28 Proceedings of the International Conference on Information Systems (ICIS)

A 8.48 n.a. 8.41 0.86 8.39 1.12 8.57 0.95 53 31 0.5

29 Journal of Industrial Ecology 1088-1980 A 8.47 n.a. 8.52 -0.50 8.02 1.34 9.53 0.97 12 3 0.3

30 SIAM Journal on Computing (Society for Industrial and Applied Mathemat-ics)

0097-5397 A 8.46 n.a. 8.45 0.01 8.46 1.25 11 0 0.0

31 Strategic Management Journal 0143-2095 A 8.41 -5.64 8.37 0.54 8.67 1.31 8.16 1.86 233 26 0.5

32 Research Policy 0048-7333 A 8.41 10.46 8.40 0.13 8.37 1.12 8.46 1.01 76 20 0.5

33 Journal of Service Research 1094-6705 A 8.40 0.02 8.22 2.06 8.00 1.32 8.99 1.47 71 7 0.4





190


Rank Journal ISSN


JQ2 index value


JQ2 index value uw


Mean AQ

SD AQ Mean RQ


34 Journal of Business Venturing 0883-9026 A 8.38 5.13 8.30 0.86 8.17 1.26 8.58 1.38 69 16 0.5

35 Voluntas. International Journal of Voluntary and Nonprofit Organiza-tions

0957-8765 A 8.36 n.a. 8.16 2.46 8.30 1.05 8.52 0.56 12 3 0.3

36 Journal of Applied Psychology 0021-9010 A 8.33 -7.05 8.39 -0.67 8.30 1.44 8.41 1.38 94 4 0.3

37 Accounting, Organizations and Soci-ety

0361-3682 A 8.33 n.a. 8.46 -1.53 8.60 1.57 7.94 0.97 64 6 0.4

38 Journal of Management Information Systems

0742-1222 A 8.32 n.a. 8.29 0.36 8.04 1.21 9.42 0.59 29 2 0.2

39 Production and Operations Manage-ment

1059-1478 A 8.32 13.48 8.24 0.87 7.83 1.53 8.81 1.00 50 11 0.5

40 Economic Journal 0013-0133 A 8.29 -1.93 8.16 1.55 8.44 1.21 7.00 0.00 34 1 0.1

41 Journal of Industrial Economics 0022-1821 A 8.24 n.a. 8.21 0.38 8.04 1.35 8.69 1.36 54 3 0.3

42 Discrete Applied Mathematics 0166-218X A 8.24 n.a. 8.45 -2.61 7.51 1.48 9.33 0.93 11 5 0.4

43 Health Care Management Science 1386-9620 A 8.23 n.a. 8.31 -0.94 8.07 1.02 8.49 0.53 17 5 0.4

44 Journal of Economic Behavior and Organization

0167-2681 A 8.22 -8.56 8.20 0.30 8.19 1.13 8.29 0.50 39 3 0.3

45 Journal of the European Economic Association

1542-4766 A 8.20 n.a. 8.07 1.58 8.20 0.88 15 0 0.0

46 Entrepreneurship: Theory and Prac-tice

1042-2587 A 8.18 20.68 8.07 1.35 7.66 1.72 8.70 0.62 56 12 0.5

47 Journal of Economics and Manage-ment Strategy

1058-6407 A 8.17 -7.63 8.05 1.52 7.87 1.49 8.62 1.04 51 9 0.4

48 International Journal of Research in Marketing

0167-8116 A 8.17 -8.10 8.05 1.46 8.07 1.30 8.26 1.18 75 19 0.5

49 Philosophy of Science 0031-8248 A 8.16 n.a. 8.15 0.08 8.16 0.94 13 0 0.0





191


Rank Journal ISSN


JQ2 index value


JQ2 index value uw


Mean AQ

SD AQ Mean RQ


50 IIE Transactions 0740-817X A 8.12 -0.75 8.09 0.33 7.93 1.12 8.31 0.92 36 11 0.5

51 Organizational Behavior and Human Decision Processes

0749-5978 A 8.12 -4.97 7.98 1.70 8.14 1.17 8.00 0.00 27 1 0.1

52 Journal of Retailing 0022-4359 A 8.12 n.a. 8.08 0.49 8.20 1.24 8.00 1.76 75 5 0.4

53 Journal of Product Innovation Man-agement

0737-6782 A 8.12 2.45 8.11 0.11 7.76 1.23 8.47 0.73 74 16 0.5

54 OR Spectrum (formerly: OR Spek-trum)

0171-6468 A 8.10 -0.33 8.09 0.14 7.99 1.21 8.21 1.22 83 43 0.5

55 Journal of Health Economics 0167-6296 A 8.10 n.a. 8.13 -0.34 8.10 1.68 16 0 0.0

56 European Journal of Operational Research EJOR

0377-2217 A 8.09 -2.36 8.17 -0.90 8.21 1.27 7.98 1.52 79 46 0.5

57 Contemporary Accounting Research/ Recherche Comptable Contemporaine

0823-9150 A 8.08 n.a. 8.07 0.05 8.28 1.43 7.59 2.30 61 3 0.3

58 Management Accounting Research 1044-5005 A 8.07 -7.91 8.02 0.65 7.88 1.43 8.36 0.80 57 9 0.4

59 Review of Finance (formerly: Euro-pean Finance Review)

1572-3097 A 8.06 2.71 8.04 0.29 7.78 0.96 8.48 1.33 38 8 0.4

60 Journal of Scheduling 1094-6136 A 8.05 n.a. 8.15 -1.20 7.86 1.46 8.34 1.55 25 6 0.4

61 Journal of Banking and Finance 0378-4266 A 8.05 -1.68 7.89 1.98 8.08 1.18 8.01 1.23 72 14 0.5

62 Journal of Accounting Research 0021-8456 A 8.03 -12.15 7.96 0.93 9.03 1.12 4.03 1.15 77 2 0.2

63 International Journal of Game Theory 0020-7276 A 8.02 -4.03 8.00 0.21 8.02 1.82 12 0 0.0

64 Review of Derivatives Research 1380-6645 A 8.01 n.a. 7.95 0.80 7.87 0.73 8.34 0.85 11 4 0.3

Notes: uw = unweighted; w = weighted; AQ = article quality; RQ = review quality; n = number of evaluators; n.a. = not applicable because the journal was not ranked in VHB-JOURQUAL1.





192

Table 3: Top 10 Journals for Different Sub-disciplines

General Management*

Journal Rating catego-ry

JQ2 index value

Mean AQ

Mean RQ

Administrative Science Quarterly A+ 9.48 9.21 9.75

Management Science A+ 9.20 9.30 9.11

Academy of Management Journal A+ 9.08 8.86 9.29

Academy of Management Review A+ 9.07 8.65 9.48

Strategic Management Journal A 8.41 8.67 8.16

Journal of Economics and Management Strategy

A 8.17 7.87 8.62

Organization Studies B 7.99 7.70 8.28

Journal of Management B 7.85 7.33 8.37

Journal of Management Studies B 7.55 7.33 7.78

International Journal of Industrial Or-ganization

B 7.51 7.31 7.80

Accounting and Auditing


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Journal of Financial and Quantitative Analysis

A 8.95 8.98 8.88 8.64 n.a. 16 0

Journal of Accounting and Economics A 8.89 9.16 7.80 9.30 9.00 45 1

Review of Accounting Studies A 8.79 8.59 9.09 8.65 9.09 39 6

Accounting Review A 8.78 8.82 8.69 8.85 8.61 56 2

Accounting, Organizations and Society A 8.33 8.60 7.94 8.71 7.82 45 5

Contemporary Accounting Research/ Recherche Comptable Contemporaine

A 8.08 8.28 7.59 8.41 9.10 45 2

Management Accounting Research A 8.07 7.88 8.36 8.05 8.36 42 9

Journal of Accounting Research A 8.03 9.03 4.03 9.11 5.00 56 1

Journal of Business Finance and Account-ing

B 7.94 8.13 7.76 8.21 8.74 30 7

Auditing: A Journal of Practice and The-ory

B 7.93 7.70 10.00 7.63 10 19 1

Banking and Finance


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Journal of Finance A+ 9.80 9.80 9.79 9.92 10.00 42 3

Review of Financial Studies A+ 9.48 9.38 9.71 9.48 10.00 28 2

Journal of Financial Economics A+ 9.43 9.67 8.47 9.67 8.00 35 1

Journal of Financial and Quantitative Analysis

A 8.95 8.98 8.88 9.00 8.84 37 3

Review of Finance (formerly: European Finance Review)

A 8.06 7.78 8.48 7.83 8.86 25 7





193

Table 3 continued: Top 10 Journals for Different Sub-disciplines

Banking and Finance


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Journal of Banking and Finance A 8.05 8.08 8.01 8.06 8.01 39 11

Review of Derivatives Research A 8.01 7.87 8.34 7.84 8.14 9 3

Journal of Business Finance and Account-ing

B 7.94 8.13 7.76 8.53 7.39 18 8

Mathematical Finance B 7.90 7.90 n.a. 7.67 n.a. 15 0

Journal of Financial Markets B 7.73 7.28 9.51 7.07 9.51 20 2

Business Information Systems


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Information Systems Research A+ 9.42 9.28 9.75 9.18 10.00 28 2

Mathematical Programming A 8.92 8.80 10.00 9.47 10.00 4 1

MIS Quarterly A 8.84 8.62 9.34 8.88 9.48 44 2

Proceedings of the International Confer-ence on Information Systems (ICIS)

A 8.48 8.39 8.57 8.38 8.65 37 22

SIAM Journal on Computing (Society for Industrial and Applied Mathematics)

A 8.46 8.46 n.a. 8.06 n.a. 5 0

Journal of Management Information Systems

A 8.32 8.04 9.42 8.02 10.00 18 1

Information Systems Journal B 7.98 7.76 8.49 7.75 8.00 16 1

Journal of the Association for Information Systems (JAIS)

B 7.96 7.72 8.52 7.53 8.42 26 2

INFORMS Journal on Computing (for-merly: ORSA Journal on Computing)

B 7.91 7.54 8.46 7.61 8.66 16 3

Journal of Strategic Information Systems B 7.87 7.47 8.81 7.32 9.00 12 1

Corporate Taxation


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

European Accounting Review B 7.65 7.30 8.00 7.30 n.a. 15 0

National Tax Journal B 7.64 7.60 8.00 7.48 8.00 16 1

FinanzArchiv B 7.44 7.54 7.22 8.62 8.89 24 2

Journal of the American Taxation Associa-tion

B 7.28 6.98 10.00 7.08 10.00 10 1

Steuer und Wirtschaft B 7.20 7.20 7.20 8.29 8.12 27 12

Journal of International Accounting Auditing and Taxation

C 6.77 6.77 n.a. 7.39 n.a. 8 0

Journal of Taxation C 6.23 6.23 n.a. 6.27 n.a. 16 0

Fiscal Studies C 6.19 6.19 n.a. 7.18 n.a. 9 0

Canadian Tax Journal C 6.08 6.09 6.00 6.77 6.00 9 1

Internationales Steuerrecht D 5.99 5.56 6.65 5.88 6.65 28 8





194


Environmental Management


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Journal of Industrial Ecology A 8.47 8.02 9.53 7.91 10.00 11 2

Business Ethics Quarterly (BEQ) B 7.88 7.54 9.25 7.25 10.00 8 1

Ecological Economics B 7.63 8.04 4.00 8.18 4.00 12 1

Business Strategy and the Environment B 7.57 7.47 7.73 7.73 7.73 13 6

Zeitschrift für Umweltpolitik und Umwelt-recht

B 7.05 6.91 7.39 6.73 7.39 11 4

Journal of Business Ethics C 6.92 7.09 6.66 6.18 5.15 10 3

Journal of Environmental Economics and Management

C 6.67 6.67 n.a. 7.69 n.a. 5 0

International Journal of Innovation and Sustainable Development

C 6.50 6.45 7.00 7.06 7.00 4 1

Journal of Cleaner Production C 6.16 6.23 6.04 6.39 6.04 13 5

Journal of Macromarketing C 6.05 5.73 7.34 7.00 n.a. 2 0

Higher Education Management


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Management Learning B 7.05 6.95 7.48 7.00 n.a. 1 0

Academy of Management Learning and Education

C 6.92 6.30 9.40 4.00 n.a. 1 0

Higher Education C 6.05 5.84 8.00 6.52 8.00 3 1

Research in Higher Education D 5.99 5.99 n.a. 6.46 n.a. 4 0

Journal of Marketing Education D 5.35 5.39 5.00 6.00 n.a. 1 0

Chronicle of Higher Education E 4.90 4.90 n.a. 5.29 n.a. 6 0

Hochschulmanagement. Zeitschrift für die Leitung, Entwicklung und Selbstverwal-tung von Hochschulen und Wissen-schaftseinrichtungen

E 4.45 4.20 5.03 5.49 5.99 2 2

Das Hochschulwesen E 3.58 3.58 n.a. 3.09 n.a. 6 0

Forschung & Lehre E 3.29 3.02 3.94 2.61 1.00 12 1

DUZ. Deutsche Universitäts-Zeitung E 2.32 2.32 n.a. 2.35 n.a. 10 0

Human Resources and Organization


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Organization Science A 8.90 8.84 8.95 8.72 8.84 66 9

Journal of International Business Studies JIBS

A 8.81 8.89 8.73 9.18 9.39 31 9

Journal of Labor Economics A 8.71 8.74 8.64 8.62 8.49 13 2

Journal of Applied Psychology A 8.33 8.30 8.41 8.83 8.08 37 3





195


Human Resources and Organization


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Journal of Economic Behavior and Or-ganization

A 8.22 8.19 8.29 8.15 n.a. 20 0

Organizational Behavior and Human Decision Processes

A 8.12 8.14 8.00 8.17 9.00 14 1

Journal of Law, Economics, and Organiza-tion

B 7.93 8.01 7.62 7.82 9.00 16 1

Research in the Sociology of Organizations B 7.93 7.81 9.00 7.85 n.a. 18 0

Organizational Behaviour and Human Performance

B 7.85 7.85 n.a. 7.92 n.a. 14 0

Human Relations B 7.85 7.66 8.03 7.71 7.60 52 8

International Management


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Journal of International Business Studies JIBS

A 8.81 8.89 8.73 9.32 9.33 32 11

Journal of International Management B 7.59 7.50 7.72 7.60 7.85 16 5

Journal of International Marketing B 7.57 7.49 7.70 7.94 8.12 17 6

Journal of World Business (formerly: Columbia Journal of World Business)

B 7.39 7.28 7.54 7.13 7.44 22 4

International Economic Review B 7.35 7.27 8 5.12 n.a. 5 0

International Business Review B 7.09 6.78 7.57 7.17 7.84 25 6

Management International Review MIR C 6.86 6.46 7.25 7.27 7.95 41 22

European Journal of International Man-agement

C 6.44 6.27 8.00 6.15 8.00 18 1

International Journal of Cross Cultural Management

C 6.36 6.28 7.00 6.29 7.00 15 1

Cross-Cultural Research C 6.35 6.35 n.a. 6.53 n.a. 6 0

Logistics


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Transportation Science A 8.60 8.40 8.90 8.41 9.08 25 6

Discrete Applied Mathematics A 8.24 7.51 9.33 7.39 8.80 7 3

Naval Research Logistics B 7.75 7.96 7.43 7.86 7.43 33 5

Transportation Research Part B: Meth-odological

B 7.70 8.00 5.00 7.94 5.00 17 1

Journal of Business Logistics B 7.60 6.89 9.24 6.79 9.24 26 4

Journal of Supply Chain Management (formerly: International Journal of Pur-chasing and Materials Management)

B 7.49 6.67 8.71 6.53 8.71 24 6

International Journal of Physical Distribu-tion and Logistics Management

B 7.41 7.03 7.80 6.94 7.80 34 10





196


Logistics


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Transportation Research Part A: Policy and Practice

B 7.40 7.40 n.a. 7.10 n.a. 11 0

Transportation Research Part E: Logistics and Transportation Review (formerly: Logistics and Transportation Review)

B 7.33 7.42 7.00 7.24 7.00 16 2

International Journal of Logistics: Re-search and Applications

C 6.87 6.48 7.80 6.38 7.80 18 3

Management of Technology and Innovation


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Research Policy A 8.41 8.37 8.46 8.53 8.38 38 16

Journal of Business Venturing A 8.38 8.17 8.58 8.54 8.73 36 11

Entrepreneurship: Theory and Practice A 8.18 7.66 8.70 8.04 8.65 27 8

Journal of Product Innovation Manage-ment

A 8.12 7.76 8.47 8.34 8.50 37 15

IEEE Transactions on Engineering Man-agement

B 7.76 7.16 8.36 7.24 7.99 30 7

Journal of Small Business Management (JSBM)

B 7.30 7.27 7.34 7.47 n.a. 15 0

Strategic Entrepreneurship Journal B 7.15 7.15 n.a. 7.15 n.a. 15 0

Technological Forecasting and Social Change

B 7.04 7.38 4.00 7.07 n.a. 8 0

International Journal of Technology Management

C 6.96 6.76 7.16 6.95 7.19 32 14

Industrial and Corporate Change C 6.94 7.37 5.96 7.58 7.00 16 1

Marketing


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Journal of Marketing A+ 9.46 9.49 9.43 9.60 9.38 75 18

Journal of Consumer Research A+ 9.44 9.12 9.91 9.30 9.87 56 4

Journal of Marketing Research A+ 9.34 9.49 9.18 9.59 9.34 69 11

Marketing Science A+ 9.29 9.51 9.07 9.67 9.20 54 10

Journal of the Academy of Marketing Science

A 8.50 8.38 8.63 8.49 8.93 55 9

Journal of Service Research A 8.40 8.00 8.99 8.26 9.24 41 6

International Journal of Research in Marketing

A 8.17 8.07 8.26 8.37 8.32 55 16

Journal of Retailing A 8.12 8.20 8.00 8.33 8.00 61 5

Journal of Product Innovation Manage-ment

A 8.12 7.76 8.47 7.72 8.63 34 9

Marketing Letters B 7.85 7.73 8.04 8.04 7.92 49 8





197


Operations Research


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Information Systems Research A+ 9.42 9.28 9.75 8.29 n.a. 3 0

Operations Research A+ 9.19 8.79 9.77 8.94 9.75 37 8

Mathematical Programming A 8.92 8.80 10.00 8.93 10.00 11 1

MIS Quarterly A 8.84 8.62 9.34 7.28 n.a. 13 0

Transportation Science A 8.60 8.40 8.90 8.43 8.98 26 6

SIAM Journal on Computing (Society for Industrial and Applied Mathematics)

A 8.46 8.46 n.a. 9.05 n.a. 7 0


IIE Transactions A 8.12 7.93 8.31 8.22 8.37 24 10

OR Spectrum (formerly: OR Spektrum) A 8.10 7.99 8.21 8.34 8.47 43 31

European Journal of Operational Re-search EJOR

A 8.09 8.21 7.98 8.53 8.23 42 33

Philosophy of Science**


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Research Policy A 8.41 8.37 8.46 8.19 8.34 9 2

Philosophy of Science A 8.16 8.16 n.a. 8.36 n.a. 8 0

Theory and Decision B 7.75 8.23 6.64 8.56 4.00 5 1

Academy of Management Learning and Education

C 6.92 6.30 9.40 5.87 n.a. 12 0

Production Management


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

MIS Quarterly A 8.84 8.62 9.34 7.46 n.a. 13 0

Production and Operations Management A 8.32 7.83 8.81 7.87 8.48 41 8


IIE Transactions A 8.12 7.93 8.31 8.20 8.37 27 10

Journal of Operations Management B 7.84 7.61 8.17 7.48 7.71 30 6

Manufacturing and Service Operations Management

B 7.72 8.46 4.75 8.27 4.75 21 2

International Journal of Production Eco-nomics

B 7.55 7.62 7.47 7.76 7.67 45 25

International Journal of Production Re-search

B 7.54 7.59 7.49 7.82 7.68 41 19

Journal of Supply Chain Management (formerly: International Journal of Pur-chasing and Materials Management)

B 7.49 6.67 8.71 6.07 9.00 21 1

International Journal of Physical Distribu-tion and Logistics Management

B 7.41 7.03 7.80 6.75 7.65 25 4





198


Public- and Non-Profit Management


JQ2 index value

Mean AQ

Mean RQ

Mean AQ only SM

Mean RQ only SM

n SM for AQ

n SM for RQ

Voluntas. International Journal of Volun-tary and Nonprofit Organizations

A 8.36 8.30 8.52 8.04 8.69 8 2

Journal of Accounting and Public Policy B 7.77 7.81 7.60 7.70 n.a. 5 0

Nonprofit and Voluntary Sector Quarterly B 7.65 7.65 n.a. 7.95 n.a. 9 0

Journal of Public Policy and Marketing B 7.59 7.37 8.48 7.43 n.a. 4 0

Public Administration B 7.56 7.21 8.97 7.59 8.97 9 2

System Dynamics Review B 7.47 7.34 7.76 9.50 n.a. 2 0

Nonprofit Management and Leadership B 7.23 7.15 8.00 7.65 n.a. 10 0

Journal of Non Profit and Public Sector Marketing

B 7.00 7.00 n.a. 7.24 n.a. 7 0

International Journal of Nonprofit and Voluntary Sector Marketing

C 6.74 6.60 8.00 6.55 8.00 7 1

Zeitschrift für öffentliche und gemeinwirt-schaftliche Unternehmen

C 6.25 6.45 5.95 6.43 6.04 16 6

Notes: AQ = article quality; RQ = review quality; n = number of evaluators; SM = section members.

* No general management section exists within the VHB; ** Less than 10 journals with n ≥ 10 are considered as falling into this cate-

gory

As can be seen, differences between quality ratings

are usually very limited. Specifically, for the journals

listed in Table 3, total ratings and those based on

section members correlate with r = .91 (p < .01, n =

133) in the case of article quality and with r = .92 (p

< .01, n = 96) for review quality. The average ratings

are slightly higher for the section members (7.85 vs.

7.57 for article quality; 8.30 vs. 8.08 for review-

process quality), but the difference is not significant

for any of the two quality indicators.

In our model, we assume that article quality and

review quality define the overall scientific quality of

an academic journal. Both constitute different, but

related dimensions of quality, as a high-quality

process will usually go hand in hand with high out-

come quality; something which is also reflected by a

correlation of r = .75 between the two quality di-

mensions. The merit of measuring quality via the

two dimensions becomes apparent when studying

those journals for which both quality criteria differ

substantially. Table 4 lists those journals whose

articles and review standards are perceived most

differently by the respondents.

Among the journals whose articles receive better

quality ratings than the review process are some

which are explicitly positioned as “transfer” jour-

nals, linking scientific insights with managerial

audiences (e.g., Harvard Business Manager, Sloan

Management Review). Those journals which re-

ceive higher review ratings than article ratings,

however, seem to have difficulties to fully transfer

the quality of their review process into their final

product, the published articles. An alternative ex-

planation might be that these journals have in-

creased the quality of their review through proce-

dural changes only recently, but the change has not

reached the majority of the journals’ readership yet,

since article-related ratings can be expected to be

more resistant to change than review ratings.

5.2 Respondent-level Determinants of

Quality Assessments

To learn which variables explain the interpersonal

differences in journal-quality ratings and to shed

more light on the unobserved heterogeneity which

underlies the aggregated results, we conducted an

additional post-hoc analysis. We focused on the

three most prominent German-language business-

research journals, namely Die Betriebswirtschaft –

DBW, Zeitschrift für Betriebswirtschaft – ZfB,

Schmalenbachs Zeitschrift für betriebswirtschaftli-

che Forschung – zfbf (Macharzina, Wolf, and Rohn

2004; Schlinghoff and Backes-Gellner 2002). This

selection offers two main advantages: these journals

(a) are read by a large number of respondents, and

they (b) contain articles from different subdisci-





199

Table 4: Strongest Differences between Article and Review Quality

Journal Absolute Difference between Article Quality and Review Quality

Article-Quality Rating

Highest positive differences

Zeitschrift für angewandte Umweltforschung 2.63 5.22

European Journal of Information Systems 1.65 7.15

Zeitschrift für das gesamte Kreditwesen 1.55 4.01

Journal of Financial Intermediation 1.50 8.28

International Transactions in Operational Research 1.47 6.29

Sloan Management Review 1.47 6.03

Corporate Ownership and Control 1.47 5.96

Harvard Business Manager 1.38 4.08

Journal of Empirical Finance 1.18 7.44

Betriebswirtschaftliche Blätter 1.17 3.27

Highest negative differences

Journal of Supply Chain Management (formerly: Interna-tional Journal of Purchasing and Materials Management)

-2.04 6.67

Venture Capital: An International Journal of Entrepreneu-rial Finance

-1.98 5.61

Der Markt. Zeitschrift für Absatzwirtschaft und Marketing -1.91 3.74

NeuroPsychoEconomics -1.84 5.90

Discrete Applied Mathematics -1.82 7.51

Managing Service Quality -1.80 5.43

Academy of Management Perspectives (formerly: Academy of Management Executive)

-1.66 5.81

Decision Sciences -1.66 6.97

Tagungsbände der Konferenz Modellierung betrieblicher Informationssysteme (MOBIS)

-1.65 5.14

International Journal of Management Reviews IJMR -1.47 5.59

Note: Only journals with n ≥ 5 for review quality were considered for this analysis.

plines of business administration, attracting a highly

diverse readership.

We conducted OLS regressions for each of the three

journals, with the perceived article quality serving as

dependent variable. As independent variables, we

included the individual respondent’s expertise fac-

tor, his or her affiliation with the 16 VHB sections

which represent business-administration sub-

disciplines (scholars can be affiliated with multiple

sections), the respondent’s status as a board mem-

ber of the respective journal, his or her academic

status (i.e. full professor or not), as well as age and

gender as demographic characteristics. The regres-

sion results are reported in Table 5.

The results show a substantial amount of overlap

between the three journals; a correlation analysis

with the standardized regression coefficients as

cases exhibits correlations of r = .66 (DBW/zfbf), r =

.68 (DBW/ZfB), and r = .84 (zfbf/ZfB). Consistent-

ly, corporate taxation researchers tend to rate the

German business-administration journals higher

than scholars from other sub-disciplines; the same

is true for accounting scholars. These findings might

be attributed to the fact that in both sub-disciplines





200

Table 5: Determinants of Journal-Quality Ratings

Variable DBW ZfB zfbf

Expertise factor -.157** -.090* -.034

Membership in VHB sections:

Accounting and Auditing .106* .123** .160**

Banking and Finance .046 -.008 .050

Business Information Systems .049 .056 .002

Corporate Taxation .129** .185** .159**

Environmental Management .083 .008 .006

Higher Education .027 .017 .017

Human Resources .041 -.028 -.070

International Management .023 -.009 .010

Logistics -.080 -.033 -.051

Management of Technology and Innovation -.007 .002 .010

Marketing .125* .066 .087

Operations Research -.141** -.068 -.084

Organization .055 -.016 -.063

Philosophy of Science .026 -.003 .018

Production Management .013 .221** .105

Public Management .118** .063 .095*

Age .037 .081 .101*

Gender (1 = female, 2 = male) .004 -.015 -.069

Full professor rank -.126** -.109* -.072

Board membership .058 .047 -.031

R2 .148 .124 .126

R2 adjusted .111 .088 .088

Note: All parameter are standardized regression coefficients; ** p < .01, *p < .05.

the domestic legal framework plays a crucial role, so

that research in these fields will have a stronger

focus on national issues. Consequently, the relative

importance of the domestic scientific community

and of its German-language journals might be

higher than in other disciplines. As the top domestic

journals are the best possible publication outlets for

these researchers, competition for publication space

and article quality might also be relatively higher.

Furthermore, we see that scholars with higher levels

of research expertise tend to rate the German-

language general business-administration journals

lower; the effect is significant for two of the three

journals. Obviously, experience with review proces-

ses of international journals – a major facet of re-

search expertise – tends to heighten the researcher’s

comparison standard, resulting in lower evaluations

for German business-administration journals. Also,

established scholars tend to hold a more critical

attitude – ratings of full professors tend to be lower

for the German business-administration journals,

with the effect being significant again for two of the

three journals.

6 Validation of VHB-JOURQUAL2 To test the reliability and validity of VHB-

JOURQUAL2, we compare its results with a num-

ber of other international ratings: the first edition of





201

VHB-JOURQUAL from 2003 (i.e. VHB-JOUR-

QUAL1), the 2008 ISI Journal Citation Impact Fac-

tors (ISI Impact Factors), the British Association of

Business Schools Academic Journal Quality Guide

from 2009 (ABS09), the French Centre National de

la Recherche Scientifique ranking from 2008

(CNR08), and the Dutch Erasmus Research Insti-

tute of Management Journals Listing from 2006

(EJL06). With the exception of the ISI Impact Fac-

tors (due to copyright issues), all these rankings are

included in the Harzing list (Harzing 2009) and –

for the journals ranked in VHB-JOURQUAL2 – also

in Web-Appendix 2 of this paper.

The comparison with VHB-JOURQUAL1 allows us

to assess the reliability of the results, as the method

and population for both rankings is very similar.

Although differences will result from changes in

journal quality over time and perception changes,

they should be of a somewhat limited size. VHB-

JOURQUAL1 itself has been successfully validated

in comparison with leading international survey-

based journal rankings (Hennig-Thurau, Walsh,

and Schrader 2004). For the 666 business-

administration journals in VHB-JOURQUAL2, we

collected the VHB-JOURQUAL1 index values of

journals in the final ranking (nJ ≥ 10). Data was

available for 326 journals from VHB-JOURQUAL1.

We find that the correlation is significant and sub-

stantial, with r = .94 (p < .01).

Regarding the comparison between VHB-JOUR-

QUAL2 and the ISI Impact Factors, it is important

to see that both ratings measure related, but distinct

constructs. While VHB-JOURQUAL2 explicitly

focuses on the scientific quality of a journal, ISI

Impact Factors are an established measure which

reflects the degree to which a journal’s articles are

read and actively cited by researchers. Since high-

quality articles are on average more likely to be cited

by scholars than low-quality ones (e.g., Hult, Rei-

mann, and Schilke 2009), a significant correlation

between the two rankings can be expected. How-

ever, due to the conceptual differences between the

two constructs, the correlation between VHB-

JOURQUAL2 and the ISI Impact Factors should be

weaker than the correlation between the two ver-

sions of VHB-JOURQUAL. We collected the impact

factors from the ISI Journal Citation Report Edition

2008 (Thomson Reuters 2009) for the disciplines

business, business/finance, and management. Data

was available for 137 journals listed in VHB-JOUR-

QUAL2. We find that the correlation between VHB-

JOURQUAL2 and the ISI Impact Factors is r = .57

(p < .01); it is slightly higher (r = .59) when quad-

ratic scores are used to account for the skewed dis-

tribution of the ISI Impact Factors. In addition to

being significant and substantial, these correlations

are also substantially lower than the one between

VHB-JOURQUAL2 and VHB-JOURQUAL1, which

is in line with our theoretical arguments.

Finally, when comparing the VHB-JOURQUAL2

ratings with the international journal rankings listed

above, we ran pairwise comparisons and deter-

mined the correlation between VHB-JOUQUAL2

and each ranking. When doing this, we included all

journals which are considered in VHB-JOUR-

QUAL2 and the respective comparison ranking (n =

329 for ABS09; n = 233 for CNR08; n = 207 for

EJL06). In each case the correlation is strong and

significant (p < .01) with r = .64 for ABS09, r = .70

for CNR08, and r = .56 for EJL06. It is again consis-

tent with our arguments that the correlation be-

tween VHB-JOURQUAL2 and EJL06 is relatively

weaker than between VHB-JOURQUAL2 and the

two other rankings, as EJL06 is partly based on

citations.

In summary, we interpret these results as strong

support for VHB-JOURQUAL2’s reliability and vali-

dity.

7 Discussion, Implications, and

Future Research Perspectives

7.1 Discussion and Implications

This article reports the results of VHB-JOUR-

QUAL2, a survey-based ranking of 666 business-

administration journals, and details the underlying

methodology. In addition to presenting the results

for the different journals, we provide empirical evi-

dence for the ranking’s reliability and validity. Thus,

we have confidence that VHB-JOURQUAL2 is a

sound instrument to evaluate the journal-publishing

achievements of business researchers. As such, we

believe that the major contribution of this ranking is

its ability to reduce the level of arbitrariness and the

importance of non-performance-related network

characteristics (such as “academic provenance”)

from key decisions made at universities – some-

thing which will benefit both universities and good

scholars.

As its predecessor, VHB-JOURQUAL2 carries the

potential to stimulate business researchers in Ger-





202

many, Austria, and Switzerland to compete with

colleagues from around the world for publication

space in leading international journals, which are

highlighted in this ranking. By doing this, VHB-

JOURQUAL2 might further raise the level of global

competitiveness of the German-speaking business-

administration community, a trend which we al-

ready see as a result of the existence of VHB-

JOURQUAL1 (Homburg 2008). Furthermore, we

hope that the ranking will also help to make schol-

ars’ intellectual achievements much easier to com-

municate to colleagues, department heads, deans,

and rectors, a precondition for getting adequate

rewards for such achievements.

Nevertheless, we believe that the results reported

here have to be treated with great care. We have

ambiguous feelings when we read job postings for

full professorships which say that the „scientific

performance is mainly evaluated by number and

quality of scientific publications in international

journals according to the VHB-JOURQUAL Rank-

ing” (like the University of Siegen in 2008, own

translation). Although this indicates that VHB-

JOURQUAL indeed influences the community (and

that we have reached an objective we had when we

once initiated it), we see the danger that the impor-

tance of the VHB-JOURQUAL ranking might be

carried to excess. The scientific performance – not

to say the overall performance – of an academic

must not be solely judged on the basis of a single

criterion, that is, a scholar’s top journal publications

according to VHB-JOURQUAL. As Albers (2009:

361) states, “we should be aware that any ranking

can only provide a small piece of the overall per-

formance picture”. Business researchers have to be

careful not to over-emphasize the part of the picture

which is measurable by ratings and rankings today,

since the result would be counterproductive for the

whole profession (Adler and Harzing 2009). As a

consequence, we see a strong need for additional

rankings of scholarly performances. Alternative

ratings that measure business researchers’ contribu-

tions in journal articles might use VHB-JOURQUAL

as a comprehensive and powerful source for inte-

grating survey-based and citation-based approaches

into hybrid rankings; other rankings might want to

emphasize a journal’s reputation or its importance

for knowledge transfer. The discrepancy between

scientific quality and relevance for business manag-

ers has been shown by Oesterle (2006), who re-

ported a significant negative correlation between

VHB-JOURQUAL results and academic-journal use

by German managers with a PhD. In addition to

different journal evaluations, powerful measure-

ment tools for books, teaching, or university man-

agement achievements would be valuable to avoid

the threat of one-dimensional university professors

(Frey 2007).

However, VHB-JOURQUAL should not be blamed

for its occasional misuse. It undisputedly covers a

key facet of scholars’ professional performance by

providing a reliable and valid estimate of the scien-

tific quality of a business researcher’s journal arti-

cles. It should be treated as such, no more, but also

no less.

7.2 Future Research Perspectives

While the current state of VHB-JOURQUAL pro-

vides an established tool for research-performance

evaluation, it also raises questions which should be

considered as opportunities for future research.

Regarding the VHB-JOURQUAL methodology, it is

obvious that some elements are based on pragmatic

considerations and might be considered arbitrary, at

least to a certain extent. So we encourage future

research to identify more theoretically and/or em-

pirically justifiable approaches for the following

aspects of VHB-JOURQUAL and compare their

results with the current ranking:

Definition and labeling of rating categories: In-

stead of measuring quality on a non-labeled 10-

point scale and assigning the categories ex-post, the

respondent might be asked to use the category la-

bels themselves. While this was not possible in the

first edition of the ranking (and perhaps might have

raised substantial problems even when collecting

data for the second edition), the category labels are

now widely established among German-speaking

business scholars. Alternatively, an empirical ap-

proach using empirical distributions might be cho-

sen to assign rating categories, which reduces the

danger that journals fall directly below the threshold

between two categories.

Selection, weighting, and composition of indicators

for the expertise factor: Regarding respondents’

expertise, two relevant questions refer to the dimen-

sions of expertise and their composition: Should

other factors than those currently represented by

the expertise construct be considered when measur-

ing expertise? And is the current multiplicative

composition optimal – and how would a different

combination of expert dimensions affect the results?





203

However, the current results show that the impact

of the expertise factor should not be overrated, as

expertise-weighted results and unweighted results

do not differ substantially for most journals.

Weighting of article and review quality: The cur-

rent version of VHB-JOURQUAL posits that article

and review quality are of equal importance for con-

structing the overall quality score of a journal (with

adjustments if only a small number of judgments

exists for review quality). An alternative approach

would be to empirically determine the relevance (or

factor weights) of the two quality dimensions. For

example, conjoint measurement approaches can

support (or reject) our decision for weighting of

both determinants equal. Also, the weight correc-

tion for the review quality dimension in the case of

limited review-related judgments might be ques-

tioned and potentially improved. Especially the

potentially strong effect of low numbers of review

quality ratings which differs substantially from the

usually much higher number of article ratings would

deserve additional thought; maybe it would be ad-

vantageous to treat review assessments as outliers

in such a case. At the same time, theoretical or em-

pirical arguments would be valuable to demonstrate

the superiority of alternative approaches.

Minimum number of article-quality evaluations:

While the current version of VHB-JOURQUAL

considers a minimum number of 10 ratings as the

threshold for the inclusion of a journal, future re-

search might address whether this number is ade-

quate to guarantee sufficient reliability or if lower

numbers are possible – or a higher number re-

quired.

In addition, future studies would be welcome which

provide insight why most of the highest-ranked

journals come from marketing and finance. Both are

large and global disciplines with a long tradition in

journal ratings, which had also a large number of

respondents in VHB-JOURQUAL2. We tested for

an impact of the number of raters of a journal on the

journal’s quality assessment, but found none within

our sample (neither linear nor non-linear). We sus-

pect that global competition for publication space is

particularly strong in marketing and finance, which

would contribute to the establishment of highly

selective journals which are then perceived as of

outstanding academic quality. The analysis of de-

terminants for three German-language journals has

shown that marketing and finance scholars do not

have a general tendency to evaluate academic jour-

nals more positively than researchers from other

business areas.

It is inevitable that the need for justification and

adequate foundation of a journal ranking increases

with its importance. Thus, we see the success of

VHB-JOURQUAL as an obligation for its improve-

ment. However, the potential trade-off between

optimization and comparability of different JOUR-

QUAL editions needs careful considerations.

Acknowledgments

Both authors contributed equally to this article. We

thank the BuR reviewers; the board of the VHB,

particularly its former chairman Sönke Albers and

its current chairman Alfred Wagenhofer, for exten-

sive valuable input to our study; Paul Marx for pro-

gramming the online questionnaire and his help

with the data handling; and all participating schol-

ars for their investments in terms of time and effort,

which has been essential for the success of this pro-

ject. We also thank numerous colleagues for provid-

ing informational and motivational support. Last

but not least, Uschi Hansen mentored our initial

idea for this ranking several years ago – thank you!

Web-Appendix 1:

List of all Journals Rated in VHB-JOURQUAL2 in

Alphabetical Order

Web-Appendix 2:

VHB-JOURQUAL2 Ranking of Business Research

Journals

Web-Appendix 3:

Changes of JQ Index Values from VHB-JOUR-

QUAL1 to VHB-JOURQUAL2

References

Adler, Nancy J. and Anne-Wil Harzing (2009): When Knowledge

Wins: Transcending the Sense and Nonsense of Academic Rank-

ings, Academy of Management Learning and Education, 8 (1):

72–95.

Albers, Sönke (2009): Misleading Rankings of Research in Busi-

ness, German Economic Review, 10 (3): 352–363.

Azar, Ofer H. and David M. Brock (2008): A Citation-Based

Ranking of Strategic Management Journals, Journal of Econom-

ics & Management Strategy, 17 (3): 781–802.

Blank, Rebecca M. (1991): The Effects of Double-Blind versus

Single-Blind Reviewing: Experimental Evidence from the Ameri-

can Economic Review, American Economic Review, 81 (5): 1041–

1067.

Brady, Michael K. and J. Joseph Cronin Jr. (2001): Some New

Thoughts on Conceptualizing Perceived Service Quality: A Hierar-

chical Approach, Journal of Marketing, 65 (3): 34–49.





204

Bräuninger, Michael and Justus Haucap (2002): What Econo-

mists Think of Their Journals and How They Use Them: Reputa-

tion and Relevance of Economics Journals, Discussion Paper,

University of the Federal Armed Forces: Hamburg.

Clermont, Marcel (2009): Erfassung betriebswirtschaftlich rele-

vanter Zeitschriften in Literaturdatenbanken: Analyse hinsichtlich

Vollständigkeit und Schlussfolgerungen für die Literatursuche

und die Messung der Forschungsleistung, Working Paper,

RWTH: Aachen.

Clermont, Marcel and Christian Schmitz (2008): Erfassung be-

triebswirtschaftlich relevanter Zeitschriften in den ISI-Daten-

banken sowie der Scopus-Datenbank, Zeitschrift für Betriebs-

wirtschaft, 78 (10): 987–1010.

Dilger, Alexander (2000): Plädoyer für einen sozialwissenschaftli-

chen Zitationsindex, Die Betriebswirtschaft, 60 (4): 473–484.

Dyckhoff, Harald and Christian Schmitz (2007): Forschungsleis-

tungsmessung mittels SSCI oder SCI-X? Internationale Sichtbar-

keit und Wahrnehmung der deutschen Betriebswirtschaftslehre

von 1990 bis 2004, Die Betriebswirtschaft, 67 (6): 640–664.

Extejt, Marian M. and Jonathan E. Smith (1990): The Behavioral

Sciences and Management: An Evaluation of Relevant Journals,

Journal of Management, 16 (3): 539–551.

Fabel, Oliver and Frank Heße (1999): Befragungsstudie vs. Publi-

kationsanalyse: Zur Interpretation von Ranglisten der For-

schungsaktivitäten deutscher betriebswirtschaftlicher Fachberei-

che, Die Betriebswirtschaft, 59 (2): 196–204.

Franke, Nikolaus and Martin Schreier (2008): A Meta-Ranking of

Technology and Innovation Management/Entrepreneurship

Journals, Die Betriebswirtschaft, 68 (2): 185–216.

Frey, Bruno S. (2007): Evaluierungen, Evaluierungen … Evaluitis,

Perspektiven der Wirtschaftspolitik, 8 (3), 207–220.

Frey, Bruno S. and Katja Rost (2008): Do Rankings Reflect Re-

search Quality?, Working Paper, SSRN.

Harzing, Anne-Wil (ed.) (2007): Journal Quality List, 26th ed., 5

March 2007, www.harzing.com/jql.htm (Access Date: 2007-03-

10).

Harzing, Anne-Wil (ed.) (2009): Journal Quality List, 34th ed., 3

July 2009, www.harzing.com/jql.htm (Access Date: 2009-08-28).

Heischmidt, Kenneth A. and Peter Gordon (1993): Rating of

Marketing Publications: Impact of Accreditation and Publication

History, Journal of Education for Business, 68 (3): 152–158.

Hennig-Thurau, Thorsten, Markus F. Langer, and Ursula Hansen

(2001): Modeling and Managing Student Loyalty: An Approach

Based on Relationship Quality, Journal of Service Research, 3

(4): 331–344.

Hennig-Thurau, Thorsten, Gianfranco Walsh, and Ulf Schrader

(2004): VHB-JOURQUAL: Ein Ranking von betriebswirtschaft-

lich-relevanten Zeitschriften auf der Grundlage von Expertenur-

teilen, Schmalenbachs Zeitschrift für betriebswirtschaftliche

Forschung, 56 (6): 520–545.

Homburg, Christian (2008): Internationalität, Praxisnähe, Spit-

zenforschung: Deutsche Universitäten im Zielkonflikt?, in: Tech-

nische Universität Bergakademie Freiberg (ed.): Zukunft deut-

scher Universitäten: Standpunkte und Perspektiven, Gabler:

Wiesbaden, 31–60.

Hult, G. Tomas M., Martin Reimann, and Oliver Schilke (2009):

Worldwide Faculty Perceptions of Marketing Journals: Rankings,

Trends, Comparisons, and Segmentations, global EDGE Business

Review, 3 (3): 1-10.

Laband, David N. and Michael J. Piette (1994a): The Relative

Impacts of Economics Journals: 1970 – 1990, Journal of Eco-

nomic Literature, 32 (2): 640–666.

Laband, David N. and Michael J Piette (1994b): Does the "Blind-

ness" of Peer Review Influence Manuscript Selection Efficiency?,

Southern Economic Journal, 60 (4): 896–906.

Macharzina, Klaus, Joachim Wolf, and Anne Rohn (2004): Quan-

titative Evaluation of German Research Output in Business Ad-

ministration - 1992-2001, Management International Review, 44

(3): 335–359.

Maier, Gunther (2006): Impact factors and peer judgment: The

case of regional science journals, Scientometrics, 69 (3): 651–667

Oesterle, Michael-Jörg (2006): Wahrnehmung betriebswirt-

schaftlicher Fachzeitschriften durch Praktiker, Die Betriebswirt-

schaft, 66 (3): 307–325.

Ritzberger, Klaus (2008): A Ranking of Journals in Economics

and Related Fields, German Economic Review, 9 (4): 402–430.

Rossiter, John R. (2002): The C-OAR-SE Procedure for Scale

Development in Marketing, International Journal of Research in

Marketing, 19 (4): 305-335.

Schlinghoff, Axel and Uschi Backes-Gellner (2002): Publikations-

indikatoren und die Stabilität wirtschaftswissenschaftlicher Zeit-

schriftenrankings, Schmalenbachs Zeitschrift für betriebswirt-

schaftliche Forschung, 54 (6): 343–362.

Schrader, Ulf and Thorsten Hennig-Thurau (2007): VHB-

JOURQUAL2: Planungsstand und Ergebnisse der Vorbefragung.

Presentation at the 69th annual scientific conference of the VHB,

June 2nd, Paderborn.

Schulze, Günther G., Susanne Warning, and Christian Wiermann

(2008): Zeitschriftenrankings für die Wirtschaftswissenschaften:

Konstruktion eines umfassenden Metaindexes, Perspektiven der

Wirtschaftspolitik, 9 (3): 286–305.

Simon, Hermann (1993): Die deutsche Betriebswirtschaftslehre

im internationalen Wettebewerb – ein schwarzes Loch?, in: Horst

Albach and Klaus Brockhoff (eds.): Die Zukunft der Betriebswirt-

schaftslehre in Deutschland, Zeitschrift für Betriebswirtschaft,

Special Issue 9/1993: 73–84.

Starbuck, William H. (2004): Vita Contemplativa: Why I stopped

trying to understand the real world, Organization Studies, 25 (7):

1233–1254.

Thomson Reuters (2009): Journal Citation Reports, Edition

2008, http://admin-apps.isiknowledge.com/JCR/JCR (with li-

cence, Access Date: 2009-08-28).

Vieira, Pedro Cosme da Costa (2008): An Economics Journals’

Ranking that Takes into Account the Number of Pages and Co-

authors, Applied Economics, 40 (7): 853–861.

Biographies

Ulf Schrader is Professor of Economic Education and Sustain-

able Consumption at the Technical University of Berlin, Germany.

His main research interests are sustainable consumption, con-

sumer policy, sustainability marketing, and business ethics.

Thorsten Hennig-Thurau is Professor of Marketing and

Media Research at Bauhaus-University of Weimar and Research

Professor in Management at Cass Business School, City University

London. His main research interests are service management,

media management, and customer relationship management.


Cost Sharing under Uncertainty: An Algorithmic Approach toCooperative Interval-Valued Games Alf Kimms, Julia Drechsel



Cost Sharing under Uncertainty: AnAlgorithmic Approach to CooperativeInterval-Valued GamesAlf Kimms, Mercator School of Management, University of Duisburg-Essen, Germany, E-mail: [email protected]

Julia Drechsel, Mercator School of Management, University of Duisburg-Essen, Germany, E-mail: [email protected]

AbstractRecently, Branzei, Dimitrov, and Tijs (2003) introduced cooperative interval-valued games. Amongother insights, the notion of an interval core has been coined and proposed as a solution concept forinterval-valued games. In this paper we will present a general mathematical programming algorithmwhich can be applied to find an element in the interval core. As an example, we discuss lot sizingwith uncertain demand to provide an application for interval-valued games and to demonstrate howinterval core elements can be computed. Also, we reveal that pitfalls exist if interval core elements arecomputed in a straightforward manner by considering the interval borders separately.

Keywords: cooperative game theory, core, interval-valued games, mathematical programming,lot-sizing

Manuscript received June 22, 2009, accepted by Karl Inderfurth (Operations & InformationSystems) November 17, 2009.

1 A Primer on CooperativeInterval-Valued Games

In the classical literature a cooperative game isdefined by a pair (N , c) where N = 1, . . . ,n is anindex set of players and c : 2N → IR is a charac-teristic function which assigns to every coalitionS ∈ 2N a value c(S) (with c(∅) = 0). Let us assumethat c is a cost function (‘‘the lower the better’’),but in settings where c is a benefit function the fol-lowing material can be straightforwardly adapted.A solution of the cooperative game (with transfer-able utilities) is a distribution of the cost shares,i.e. a cost allocation πi ∈ IR for every player i ∈ N .One of the most prominent solution concepts wasintroduced by Gillies (1959) and is known as thecore. The concept of the core defines cost alloca-tions which are efficient and coalitionally rational(stable). Formally, the set of core allocations canbe specified as follows:

(1) C(N , c) = π ∈ IRn|∑i∈N

πi = c(N) and

∑i∈S

πi ≤ c(S) for all S ⊂ N ,S 6= ∅.

Branzei, Dimitrov, and Tijs (2003) have studiedbankruptcy situations and introduced cooperativeinterval-valued games in this context. Interval-valued games can be seen as a means to handleuncertain outcomes. A theory of interval-valuedgames was developed by Alparslan-Gök, Branzei,and Tijs (2008a, 2008b) as well as by Alparslan-Gök, Miquel, and Tijs (2009). See also Branzei,Tijs, and Alparslan-Gök (2008) for a survey. Acooperative interval-valued game is defined by apair (N , cI) where N is an index set of players asbefore, and cI : 2N → I(IR) is a characteristic (cost)function which assigns to every coalition S ∈ 2N aclosed interval cI(S) = [cI(S); cI(S)] (with cI(∅) =[0;0]) where I(IR) is the set of all closed intervals inIR. It is easy to see that classical cooperative gamesare a special case of cooperative interval-valuedgames where cI(S) = cI(S) for all coalitions S ⊆N . It should also be remarked that vector-valuedgames (seeFernandez,Hinojosa, andPuerto2002)are somewhat different to interval-valued games.To define a core variant that applies to interval-

206


valued games, the following notation turns out tobe helpful. The sum of two intervals I = [I; I] andJ = [J; J] is defined to be I + J = [I + J; I + J].The interval I is called weakly better than intervalJ (I J or J I), iff I ≤ J and I ≤ J.Alparslan-Gök, Branzei, and Tijs (2008a) definethe interval core of a cooperative interval-valuedgame as follows:(2) CI(N , cI)

= (I1, . . . , In) ∈ I(IR)n|∑i∈N

Ii = cI(N)

and∑i∈S

Ii cI(S) for all S ⊂ N ,S 6= ∅.

These authors, however, do not provide an algo-rithm to compute a core element of interval-valuedgames.Now we have introduced the theoretical back-ground that is necessary for understanding ourapproach. Our contribution is an algorithm whichcan be used to compute an interval core element.Besides this, we discuss pitfalls when implement-ing a more simple procedure based on an alreadyknown algorithm for core element computationin cases of independent determination of intervalborders. The rest of the paper is structured as fol-lows. In Section 2 we specify the straightforwardprocedure with separate interval border determi-nation that aims to find an element in the intervalcore (or detects that the interval core is empty).Section 3 introduces a lot-sizing problem with un-certain demand as an example for interval-valuedgames. Section 4 reveals that some odd situationsmay occur if the straightforward approach is ap-plied and left and right interval borders are usedindependently to compute interval core elements.Thus, a modification of the proposed algorithmwill be presented to avoid such problems. A com-putational study in Section 5 demonstrates thatthis modified algorithm is necessary in many casesand can indeedbe successfully applied. Concludingremarks in Section 6 finish the paper.

2 Computing Interval CoreElements

To compute an element in the interval core, wewill make use of a procedure by Drechsel andKimms (2007) which computes a core element.This procedure is employed as a subroutine. Tosee how this works we start with a reformulation

of interval-valued games and the interval core. Aninterval-valued game can be specified by a triple(N , cI , cI)where thenotationhas the samemeaningas above. Two sets of extreme cost allocations canthen be defined as

(3) CI(N , cI)

= π ∈ IRn|∑i∈N

πi = cI(N) and

∑i∈S

πi ≤ cI(S) for all S ⊂ N ,S 6= ∅

and

(4) CI(N , cI)


πi = cI(N) and

∑i∈S

πi ≤ cI(S) for all S ⊂ N ,S 6= ∅.

If we choose (I1, . . . , In) ∈ CI(N , c) and (I1, . . . ,

In) ∈ CI(N , c) then it follows immediately fromthe definitions of cI , Ii and CI(N , cI) that ([I1; I1],. . . , [In; In]) ∈ CI(N , cI) --- the reverse is true aswell. CI(N , cI) and CI(N , cI) define the cores of thecooperative games (N , cI) and (N , cI), respectively.This is along the lines of Remark 3.1 in the paperby Alparslan-Gök, Branzei, and Tijs (2008a), whostate that the time complexity of algorithms for thecomputation of elements of the interval core is thesame as the time complexity of related algorithmsfor computing elements of the core of a traditionalgame. This suggests that an element in the intervalcore can be easily computed if one can compute anelement in the core.The procedure of Drechsel and Kimms (2007) canbe applied to find core elements for the games(N , cI) and (N , cI) separately. To be self-contained,this procedure will be reviewed briefly. To ease thenotation, assume that a core element of a game(N , c) shall be computed. The definition of the corespecifies a constraint-satisfaction problem wherethe number of constraints is in the order of 2n.Hence, Drechsel and Kimms (2007) suggested torun a row-generation procedure.The master problem is a linear program of thefollowing form where S is a (small) subset ofcoalitions for which the core defining inequalityis explicitly taken into account.

207


Master problemMP(S):

(5) min v

s.t.

(6)∑i∈N

πi = c(N)

(7)∑i∈S

πi − v ≤ c(S) S ∈ S

(8) πi ∈ IR i ∈ N(9) v ≥ 0

The iterative row-generation procedure can be out-lined as follows:

1. Define a small initial set S, e.g. S = 1, . . . ,n.

2. Solve the linear programMP(S) optimally.

3. If v > 0 then stop. The game instance has anempty core.

4. Otherwise, find a coalition S′ 6∈ S (S′ 6= ∅)such that

∑i∈S′ πi > c(S′). Since S′ may not be

unique, we suggest looking for a coalition S′

that violates the core defining inequality most(i.e.

∑i∈S′ πi − c(S′) is maximized) --- subprob-

lem SP(π).

5. If no such coalition S′ can be found (SP(π)has a non-positive optimum objective functionvalue) then stop. The current values πi definea core allocation.

6. Otherwise, update S = S ∪ S′. Return to Step2.

The subproblem to be solved in Step 4 of thealgorithm is the only problem-specific part in thisapproach. In a subsequent section we will providethe details of it by means of a specific application.It should be noted that this procedure finds arather arbitrary element in the core (if the coreis not empty). Drechsel and Kimms (2007) havediscussed several modifications to the procedurewhich yield core elements that can be consideredasmore fair than an arbitrary core element. For thesake of simplicity, we will not repeat these modifi-cations here, but we would like to emphasize thatthese modifications can be straightforwardly inte-grated into the context of interval-valued games ifdesired.

3 An Application:Lot-Sizing with UncertainDemand

3.1 TheWagner-Whitin Problem

The starting point of our example is a well-knownlot-sizing problem which was introduced by Wag-ner andWhitin (1958) and which can be describedas follows: A single decision maker has to makeorder decisions for a single item. Given a planninghorizon of T time periods, a demand dt in periodt has to be met without backlogging and withoutshortages. In order to meet the demand in period tone may place an order in t or before. If the orderis placed before, the ordered items must be storedin inventory. If a fixed cost st is incurred wheneveran order is placed in period t and a unit holdingcost ht is charged for every item in stock at theend of period t, then we face the classical trade-offbetween saving fixed costs versus saving holdingcosts --- a lot-sizing problem occurs. In addition tothat a unit ordering cost pt may be incurred foreach item being ordered. The decision to be madeis qt, the quantity to be ordered in period t. De-pending on the order quantities and the demandwe have Invt units of the item in stock at the end ofa period t (Inv0 is the initial stock --- a given value).To formulate a mixed-integer program a furtherdecision variable xt may be used to indicate if anorder is placed in period t or not, and a parameterM which represents a large number may be usedfor technical reasons, as we see below:

(10) minT∑t=1

(stxt + htInvt + ptqt)

s.t.

(11) Invt = Invt−1 + qt − dt t = 1, . . . ,T

(12) qt ≤Mxt t = 1, . . . ,T

(13) qt, Invt ≥ 0 t = 1, . . . ,T

(14) xt ∈ 0,1 t = 1, . . . ,T

The objective (10) is to minimize the sum of fixedand quantity-dependent ordering costs and hold-ing costs. The inventory balance (11) states thatat the end of a period we have in stock what wasthere at the beginning of the period plus what wasordered in that period minus period demand. Ifthe order quantity is positive in period t then theindicator variable xt must be set to one as statedby (12). The domain of the decision variables is

208


specified in (13) and (14). Note that due to Invt ≥ 0all demand must be fulfilled right on time.

3.2 A Cooperative Lot-Sizing Problem

Cooperativeversionsof theWagner-Whitin-relatedproblem where several players (the set N) may acttogether have been studied by Chen and Zhang(2007), Drechsel and Kimms (2007), Guardiola,Meca, and Puerto (2006), and Van den Heuvel,Borm, and Hamers (2007). A model formulationclosely related to the single-player model shownabove can be given, simply by replacing the de-mand value dt with the total demand of all playersdt(N) =

∑i∈N dit where dit denotes the demand

of player i in period t. The characteristic functionvalue c(S) for S ⊆ N is defined to be the optimumobjective function value (10) using dt(S) as thedemand in (11).As shown by Drechsel and Kimms (2007) one canapply the row-generation procedure for finding acore element in the game (N , c) to handle instanceswith up to 150 players. The subproblem that has tobe solved in Step 4 of the procedure is the followingoptimization problem.

Subproblem SP(π):

(15) max

(−o +

∑i∈N

πizi

)s.t.

(16) Invt = Invt−1 + qt −∑i∈N

ditzi t = 1, . . . ,T

(17) qt ≤Mxt t = 1, . . . ,T

(18) o =T∑t=1

(stxt + htInvt + ptqt)

(19) qt, Invt ≥ 0 t = 1, . . . ,T

(20) xt ∈ 0,1 t = 1, . . . ,T

(21) zi ∈ 0,1 i ∈ N(22) o ≥ 0

Note that a coalition S′ to be considered in themaster problem is found if the optimum objectivefunction value of the subproblem is positive. S′ isdefined by the values of the zi variables (zi = 1 indi-cates i ∈ S′) and c(S′) = o. Note also that Drechseland Kimms (2007) have tested other subproblemformulations, and that selecting a coalition S′ bymeans of subproblem SP(π) turned out to requirefewer iterations of the row-generation procedure

and less total computation time than other selec-tion rules.It should be emphasized that the proposed row-generation approach to compute a core elementis very general and can be applied whenever thecharacteristic function is implicitly given, i.e. char-acteristic function values result from an optimiza-tion problem. Drechsel and Kimms (2008), forinstance, have shown that for much more complexlot-sizing situations with capacity constraints andtransshipments among the players basically thevery same approach is applicable. Only the sub-problem needs to be reformulated, because this isthe problem-specific part of the algorithm. How toformulate such subproblems for different applica-tions cannot be specified in general, but it requiresno more skills than formulating the underlyingoptimization problem as a mathematical program.As a side remark, it should be emphasized thatmany operations management planning problemshave to be solved repeatedly with a rolling hori-zon. Lot sizing is one example. Our example istherefore stylized by ignoring (uncertain demand)data beyond the period T and we admit that ourfocus is confined to interval-valued data. We alsodo not consider the possibility to condition futurelot-sizing decisions on realized past demands likein stochastic dynamic optimization procedures. Tothe best of our knowledge, such simplifying as-sumptions are also made by all authors who haveconsidered cost sharing based on multi-period op-timization problems. Hence, one should be awareof the fact that dynamic settings where uncertaintyis not revealed at once and the consequences oncost sharing have not been studied yet.

3.3 Cooperative Lot-Sizing withUncertain Demand

A source for uncertainty in a lot-sizing problemis the demand of a player. Let us assume that foreach player i the demand in a certain period isspecified by an interval [dit; dit] instead of a singlenumber dit. Consequently, the total demand dt(N)in a period t falls into the interval [dt(N); dt(N)] =[∑

i∈N dit;∑

i∈N dit]. From these interval borderswe can derive characteristic functions cI and cI

by inserting dt(S) and dt(S), respectively, into (11)for S ⊆ N . With this in mind, we can computean element in the interval core as described in aprevious section.

209


4 Phenomena of Interval Cores

4.1 ‘‘Pay Less for More’’

As some researchers have pointed out already (see,e.g., Young 1985), core allocations are not mono-tonic. This effect causes some problems in thecontext of interval cores as we will show by meansof an example. Consider an examplewithn = 3 andT = 6. Let the demand be dit = 10 and dit = 15,respectively, for all i and t. Suppose the followingcost data: holding cost coefficients ht = 5 for allt, setup cost coefficients st = 100 for all t, andproduction cost coefficients pt = 1 for all t. Thecharacteristic functions cI and cI are provided inTable 1.

Table 1: The Characteristic Functions ofthe Interval-Valued Game

S ∅ 1 2 3

cI(S) 0 510 510 510

cI(S) 0 615 615 615

S 1,2 1,3 2,3 1,2,3cI(S) 720 720 720 780

cI(S) 780 780 780 870

Apparently, we have cI(S) ≥ cI(S) in this exam-ple. This, by the way, is always true for lot-sizingproblems (with non-negative cost data), becauseadditional demand cannot decrease the (total)cost to fulfill the demand. Since the character-istic functions cI and cI define core cost allo-cations πi(cI) and πi(cI), respectively, we wouldexpect that πi(cI) ≥ πi(cI) is true for all playersi. In other words, we would expect that increas-ing the demand of a player does not lead to alower cost assignment, i.e. the principle of mono-tonicity holds. But this is not true in general. Forexample, π(cI) = (210,510,60) is a core cost al-location for the game (N , cI). On the other hand,π(cI) = (165,615,90) is a core cost allocation forthe game (N , cI). As we can see, a lower cost sharewould be assigned to player 1 (165 instead of 210)although his demand increased from 10 to 15 inevery period.Young (1985) has discussed the issue of mono-tonicity and showed that the Shapley value isthe unique symmetric allocation procedure thatis (strongly) monotonic. But using the Shapleyvalue as an interval border is not consistent withthe idea of the interval core, because the Shapley

value need not be in the core if the game is notconcave (see Shapley 1971).As a consequence of the phenomenon we mayobserve intervals Ii = [Ii; Ii] with Ii > Ii. This,by the way, is not valid for lot-sizing problemsonly, but to all kinds of applications, because thenon-monotonicity of core allocations is problemindependent.

4.2 Pitfalls

It seems to be plausible to interpret an elementfrom the interval core as a promise to the playerssuch that if (I1, . . . , In) ∈ CI(N , cI) is chosen inadvance, an ex post (i.e. after the uncertainty hasbeen resolved) core cost allocation with πi ∈ Iican be found, because the two extreme scenarioshave been considered to define the interval core.Hence, the problem of finding a core element expost seems to be of the following form:

(23) C|(I1,...,In)(N , c)


πi = c(N) and∑i∈S

πi ≤

c(S) for all S ⊂ N ,S 6= ∅ and πi ∈ Ii.

Note that this interpretation is meaningful iff theunderlying (ex post) characteristic function c ismonotone over the interval defined by cI and co-incides with cI and cI at the interval borders. Thisinterpretation, however, is wrong as we will showbe means of the example from Subsection 4.1.Recall that in the example the following intervalswere computed: I1 = [210;165], I2 = [510;615],and I3 = [60,90]. Assume that ex post the follow-ing demand data can be observed: d1t = 15 for allt, d2t = 10 for all t, and d3t = 15 for all t. As wecan see from Table 1, this leads to c(1) = 615,c(2) = 510, and c(3) = 615. The cost of thegrand coalition is c(1,2,3) = 840. The defini-tion of C|(I1,...,In)(N , c) requires π2 = 510. Hence,π1 + π3 = 840 − 510 = 330. But this is not possi-ble, because of π1 ≤ 210 and π3 ≤ 90. Hence, thedescribed approach leads to unreasonable results.

4.3 A Proper Definition ofInterval-Valued Cores

To avoid such problems with interval cores, weneed proper intervals, i.e. we need to compute in-terval borders πi(cI) and πi(cI) such that πi(cI) ≤

210


πi(cI) comes out. Such a result can be reachedbe computing the values πi(cI) and πi(cI) simul-taneously by solving a modified master problemMPI(S, S):

(24) min v + v

s.t.

(25)∑i∈N

πi(cI) = cI(N)

(26)∑i∈S

πi(cI) − v ≤ cI(S) S ∈ S

(27)∑i∈N

πi(cI) = cI(N)

(28)∑i∈S

πi(cI) − v ≤ cI(S) S ∈ S

(29) πi(cI) ≤ πi(c

I) i ∈ N(30) πi(c

I),πi(cI) ∈ IR i ∈ N

(31) v, v ≥ 0

This master problem can be solved by the iterativerow-generation algorithm as follows:

1. Define small initial sets S and S, e.g. S = S =1, . . . , n.

2. Solve the linear programMPI(S, S) optimally.

3. If v + v > 0 then stop. The game instance hasan empty interval core.

4. Otherwise:

(a) Find a coalition S′ 6∈ S (S′ 6= ∅) such that∑i∈S′ πi(cI) > cI(S

′). Since S′ may not beunique, we suggest looking for a coalitionS′ that violates the core defining inequalitymost (i.e.

∑i∈S′ πi(cI)−cI(S

′) ismaximized)

--- subproblem SP(π(cI)). Update S = S ∪S′, if such a coalition S′ was found.

(b) Find a coalition S′ 6∈ S (S′ 6= ∅) such that∑i∈S′ πi(cI) > cI(S′). Since S′ may not be

unique, we suggest looking for a coalitionS′ that violates the core defining inequalitymost (i.e.

∑i∈S′ πi(cI)−cI(S′) ismaximized)

--- subproblem ¯SP(π(cI)). Update S = S ∪S′, if such a coalition S′ was found.

5. If neither S′ nor S′ can be found then stop.The current values πi(cI) and πi(cI) define aninterval core element.

6. Return to Step 2.

Recall thenumerical example fromSection4.1. Thepresented algorithm yields the following allocationin the interval-valued core: π(cI) = (210,60,510)and π(cI) = (210,90,570).

5 Computational StudyTo test the proposed algorithm from Section 4.3we have implemented it with AMPL/CPLEX ver-sion 10.0.0 on an Intel Celeron PC with 2 GHz.We used random instances taken from Test Bed 1in Drechsel and Kimms (2007) to define the game(N , cI): random integers were drawn with uniformdistribution from certain intervals (dit ∈ [0,20],st ∈ [0,200], ht ∈ [0,10], and pt ∈ [0,15]),T = 6, and n ∈ 5,10,15,20,25,30. For eachnumber of players, a total of 15 random instanceswas constructed which gives a total of 6 × 15 = 90instances. These instances are provided as an elec-tronic companion to this paper as part of the onlineversion.1 The files are in ASCII format and can beusedby the commercial softwareCPLEX.Thegame(N , cI) was constructed by copying the parametervalues from the corresponding game (N , cI). Onlydemand values were chosen differently in the fol-lowing way. Suppose dit is the chosen demand inthe corresponding game (N , cI). The demand ditfor the game (N , cI) is a random integer drawnwith uniform distribution from [dit,20].The average number of iterations (as well as theminimum/maximum number of iterations over 15random instances) for the algorithm can be seenin Table 2. The table presents the results regardingnumber of iterations for the two subproblems SP

and ¯SP that are part of Step 4 of the algorithm.The numbers differ for most of the instances --- ofcourse, the bigger number also denotes the totalnumber of needed iterations.Note that in theworstcase the number of iterations could have been inthe order of 2n which practically never happensfromourexperience.Themaximumcomputationaltime (measured in CPU-seconds) to determine anelement in the interval-valued core was less than aminute for all instances n ≤ 20 and not more thanfive minutes for the larger instances.To compare the computational burden with thestraightforward approach from Section 2 --- the

1 See www.business-research.org.

211


Table 2: The Number of IterationsRequired to Find an Interval Core Element

n 5 10 15

SP4.87

(4/7)

20.47

(14/40)

146.27

(25/293)

¯SP4.73

(4/6)

21.27

(12/60)

146.27

(16/293)

n 20 25 30

SP406.73

(110/1036)

1044.60

(165/2364)

1043.53

(30/3946)

¯SP403.80

(110/1037)

1039.80

(165/2360)

1054.53

( 30/3941)

one which may lead to the problems discussed inSection 4.2 --- we also provide the average num-ber of iterations of the straightforward approachin Table 3. As one can see the straightforwardapproach terminates much faster on average (therun-time effort was less than one minute for eachinstance). Hence it might be worthwhile to runthe straightforward approach first and test if thecomputed interval borders indeed show the oddbehaviour discussed in Section 4.2. If this is notthe case, then we have found an interval core ele-ment. Otherwise, we need to run the algorithmfrom Section 4.3 which is necessary in most casesas we can see in Table 4. Table 4 shows the numberof instances in the test bed (a total of 15 instancesper value of n) where the algorithm from Section 2yields an improper interval for at least one player.

Table 3: The Average Number of IterationsRequired by the Straightforward Approachfrom Section 2

n 5 10 15

(N , cI) 4.0 10.5 15.3

(N , cI) 4.0 9.3 14.0

n 20 25 30

(N , cI) 61.8 29.5 50.0

(N , cI) 19.0 24.4 29.0

Table 4: The Number of Instances wherethe Algorithm from Section 2 YieldsImproper Intervals

n 5 10 15 20 25 30

14 13 12 4 11 11

6 ConclusionsIn recent years a new class of cooperative games,namely interval-valued games, has been inventedto handle situations with an uncertain outcome.The range of the outcome is specified by a closedinterval. A theory for interval-valued games hasbeen developed and solution concepts from classi-cal cooperative game theory have been adapted tothis new setting. The core which is one of the mostprominent solution concepts in classical coopera-tive game theory evolved to become the intervalcore.Remark 3.1 in a paper by Alparslan-Gök, Branzei,and Tijs (2008a) states that the time complexityof algorithms for computing elements of intervalcores equals the time complexity of correspondingalgorithms for computing core elements in a tradi-tional game. This statement suggests to computean element in the interval core by making use of aprocedure to find a core element.However, such analgorithm has not been provided by these authors.Our contribution is to provide a generalmathemat-ical programming approach which can be used tocompute an element in the interval core bymakinguse of a procedure to find core elements proposedby Drechsel and Kimms (2007) as a subroutine.The approach is general, because the character-istic function values which define the lower andupper bounds of the outcome interval may resultfrom solving general optimization problems. Wehave demonstrated that interval-valued games area consequence of uncertain parameter values ofthe optimization problem where the uncertaintyof these parameter values is specified by inter-vals. With the aid of a lot-sizing example we havedemonstrated the applicability of our approach bymeans of a small computational study.In addition to providing an algorithm we havealso contributed to the discussion of interval coresby pointing to a pitfall. A phenomenon known asnon-monotonicity exists, i.e. players may be betteroff in a cooperation, if their standalone situationis getting worse (in terms of objective functionvalues). In the lot-sizing example, for instance,we have demonstrated that a player with moredemand than before (increasing costs)may receivea lower cost share than before. A straightforwardinterpretation of an interval core element in thesense of cost limits that will be assigned to certainplayers is not possible if the interval core element is

212


not selected with care. To get out of this problem,we proposed a modified algorithm which yieldsan element from the interval core without suchdeficits.

7 AcknowledgementWe thank Rodica Branzei for her comments on anearly version of this paper.

References

Alparslan-Gök, Sirma Z., Rodica Branzei, and Stef H. Tijs(2008a): Cores and Stable Sets for Interval-Valued Games,Working Paper, SSRN.

Alparslan-Gök, Sirma Z., Rodica Branzei, and Stef H. Tijs(2008b): Convex Interval Games, Working Paper, SSRN.

Alparslan-Gök, SirmaZ., SylviaMiquel, and StefH. Tijs (2009):Cooperation under Interval Uncertainty, Mathematical Meth-ods of Operation Research, 69 (1): 99-109.

Branzei, Rodica, Dinko Dimitrov, and Stef H. Tijs (2003):Shapley-like Values for Interval Bankruptcy Games,EconomicsBulletin, 3 (9): 1-8.

Branzei, Rodica, Stef H. Tijs, and Sirma Z. Alparslan-Gök(2008): Cooperative Interval Games: A Survey,Working Paper,http://www.3.iam.metu.edu.tr/iam/images/8/80/survey.pdf(Access Date: 2009-12-03).

Chen, Xin and Jiawei Zhang (2007): Duality Approches toEconomic Lot-Sizing Games, Working Paper, SSRN.

Drechsel, Julia and Alf Kimms (2007): Computing Core Alloca-tions inCooperativeGameswith anApplication to ProcurementProblems, Working Paper, University of Duisburg-Essen.

Drechsel, Julia and Alf Kimms (2008): Solutions and Fair CostAllocations for Cooperative Lot Sizing with Transshipmentsand Scarce Capacities, Working Paper, University of Duisburg-Essen.

Fernandez, Francisco R. , Miguel A. Hinojosa, and Justo Puerto(2002), Core Solutions in Vector-Valued Games, Journal ofOptimization Theory and Applications, 112 (2): 331-360.

Gillies, Donald B. (1959): Solutions to General Non-Zero-SumGames, in: AlbertW. Tucker andR.DuncanLuce (eds.):Contri-butions to the Theory of Games 4, Princeton University Press:Princeton, 47-85.

Guardiola, Luis A. , Ana Meca, and Justo Puerto (2006): Co-ordination in Periodic Review Inventory Situations, WorkingPaper, Universidad Miguel Hernández de Elche.

Shapley, Lloyd S. (1971): Cores of Convex Games, InternationalJournal of Game Theory, 1 (1): 11-26.

Van den Heuvel, Wilco, Peter Borm, and Herbert Hamers(2007): Economic Lot-Sizing Games, European Journal ofOperational Research, 176 (2): 1117-1130.

Wagner, Harvey M. and Thomson M. Whitin (1958): DynamicVersion of the Economic Lot SizeModel,Management Science,5 (1): 89-96.

Young, H. Peyton (1985): Monotonic Solutions of CooperativeGames, International Journal of Game Theory, 14 (2): 65-72.

Biographies

Alf Kimms studied computer science and busi-ness administration at the University of Kiel. Hereceived his Ph.D. in business administration fromthe University of Kiel in 1996 with a thesis onmulti-level lot sizing and scheduling under thesupervision of Prof. Dr. Andreas Drexl, and heearned his postdoctoral lecturer qualification fromthe same university in 2001 with a thesis on fi-nancial objectives for managing projects. In 2001he became full professor at the Technical Univer-sity of Freiberg and held the chair of productionand logistics. In 2006 he was appointed as a fullprofessor at the University of Duisburg-Essen tohold the chair of logistics and traffic management.Professor Kimms has received several best-paperawards.

Julia Drechsel studied industrial engineering atthe Technical University of Freiberg. She receivedseveral prizes for her diploma thesis on task forcedeployment. From 2005 to 2009 she was a Ph.D.student under the supervision of Professor Kimmsworking at the University of Duisburg-Essen. Herresearch focus is on Supply Chain Management.Since 2009 she has been working for the companyBayer Technology Services GmbH.

213


The Role of Ambidexterity in Marketing Strategy Implementation:Resolving the Exploration-Exploitation Dilemma Christiane Prange, Bodo B. Schlegelmilch



215

1 Introduction Marketers are faced with a dilemma: they are ex-pected to consolidate their existing business while simultaneously finding new opportunities. They are torn between exploitation and exploration, or be-tween alignment and innovation. Thus, implement-ing the right strategy is often difficult. Business de-velopment may be eager to sell new products that have not yet completed the research process. R&D may develop new product ideas, but fail to commer-cialize them (Souder and Chakrabarti 1978; Griffin and Hauser 1996). Firms are market-driven while market-driving firms are more conducive to innova-tions (Kumar 1997; Jaworski, Kohli, and Sahay 2000). And as if these tensions were not enough, marketing managers are also challenged to act lo-cally but integrate globally, and to pursue differen-tiation but obverse low-cost strategies (Ghoshal and Bartlett 1994; Porter 1996; Tushman and O'Reilly 1996).

Despite the growing interest among researchers and the pressing need of practitioners, there is a dearth of research that investigates the implementation of conflicting marketing strategies. The handful of ex-isting studies to date has broadly focused on the marketing strategy formulation-implementation link (Bonoma 1984; Crespedes 1991; Walker and Rueckert 1987; White, Conand, and Echambadi 2003) with only tangential reference to theoretically derived implementation guidelines. Consequently, the literature suffers from several limitations. First, previous research has either focused on isolated fac-tors in support of strategy implementation or has highlighted rather complex frameworks (Li, Guohui, and Eppler 2008) with holistic studies using simple categorizations of various factors without theoretical integration (Noble and Mokwa 1999). Second, re-search has largely overlooked the various roles of managers and employees in the strategy implemen-tation process (Hart and Banbury 1994). This lack of an organization-wide perspective is noteworthy

The Role of Ambidexterity in Marketing Strategy Implementation: Resolving the Exploration-Exploitation Dilemma Christiane Prange, EM Lyon Business School, France, E-Mail: [email protected]

Bodo B. Schlegelmilch, WU Wien, Austria, and Leeds University Business School, UK, E-Mail: [email protected]

Abstract Formulating consistent marketing strategies is a difficult task, but successfully implementing them is even more challenging. This is even more pertinent as marketing strategies quite often incorporate inherent conflicts between major breakthroughs and consolidation. Consequently, marketers need to balance ex-ploratory and exploitative strategies. However, the literature lacks concrete insights for marketing man-agers as to how exploratory and exploitative strategies can be best combined. This paper addresses this is-sue by introducing a framework of multiple types of ambidexterity. Based on qualitative research, tools and procedures are identified to overcome marketing dilemmas and support strategy implementation by drawing on ambidextrous designs. Keywords: marketing dilemmas, marketing strategy implementation, exploration, exploitation, ambidex-terity Manuscript received November 19, 2008, accepted by Adamantios Diamantopoulos (Marketing), Novem-ber 11, 2009.


216

given the importance of functional interactions for implementation success (Chimhanzi and Morgan 2005). Finally, the majority of implementation stud-ies has acknowledged that the changing content of strategy holds few clear implications and has conse-quently emphasized the necessity for managers to continuously redefine tasks and communicate them throughout the organization (Shashittal and Wile-mon 1996). Notwithstanding the quest for ongoing adaptability of both strategy formulation and im-plementation processes, studies that focus on in-trinsic tradeoffs in marketing strategies are rare. These limitations present the starting point for our research and can be best illustrated with reference to the more content-oriented marketing literature where tradeoffs abound. For instance, customer sat-isfaction has become one of the unquestioned linch-pins to success in marketing (e.g. Gale 1992) and it has been suggested to combine service activities with increasing sales efforts. However, doing so si-multaneously is often problematic as customer ser-vice staff feels uncomfortable with the twin role of selling products while serving the customer (Akşin and Harker 1999; Evans, Arnold, and Grant 1999). While requests have been formulated to integrate internal structures, processes, goals and rewards to properly support the desired role behavior of cus-tomer contact personnel (Schneider and Bowen 1995), fine-grained insight on how this could be done is lacking. A related dilemma is the tension in new product de-velopment: Incremental product improvements typically arise from small changes where return are certain, risks are low, and where products exhibit little deviation from customers’ current experience. However, innovating incrementally may blind firms towards technology and product category leaps (O’Reilly and Tushman 2004). This may lead to what Leonard-Barton (1992) calls the “capability-rigidity paradox”. An excessive focus on gradual product improvement may imply technological ex-haustion in the market in which firms compete for developing new products (Lee and Ryu 2002). Eventually, firms need to incorporate both incre-mental and radical approaches to new product de-velopment into their strategy. A last example relates to conflicting requirements for chief marketing officers who are supposed to have creative flair as well as financial know-how (Quelch 2005), which is reflected in such different tasks as marketing controlling, and advertising. The

“Chief Marketing Officer’s Dilemma” (McEwen 2008) is further reinforced by striving for customer engagement, which depends on both communicat-ing and delivering the brand promise. While the first half of the input factors (communicating brand promise via advertising and marketing) is easily controlled, the second half depends on employee engagement (delivering upon brand promise via employees and operations). Combining both is likely to increase profitability (McEwen 2008). These examples underscore the existence of con-flicting marketing strategies which poses a dilemma for marketing managers and require a balance use of tools, procedures, resource endowments, struc-tures, and supportive contexts. While the marketing literature has provided us with a wealth of sugges-tions as to how single elements in the implementa-tion process can be optimized or which frameworks to use, the question of how tradeoffs between differ-ent strategies can be overcome in the implementa-tion process has been largely ignored. One exception is the seminal work by Kyriakopoulos and Moorman (2004) and its extension by Menuc and Auh (2008), who point to the role of market orientation in inte-grating ‘marketing exploitation strategies’ and ‘mar-keting exploration strategies’. The former refers to strategies that primarily involve improving and re-fining current skills and procedures associated with existing marketing strategies, including current segments, positioning, distribution, and other mar-keting mix strategies. Marketing exploration, on the other hand, “relates to strategies that primarily in-volve challenging prior approaches to interfacing with the market, such as a new segmentation, new positioning, new products, or new channels” (Kyria-kopoulos and Moorman 2004: 221). This research follows the exploration vs. exploita-tion paradigm and explicitly adopts an implementa-tion focus, which redefines marketing strategy im-plementation as the capability to translate intrinsi-cally conflicting demands into designs and toolsets that help to overcome marketing dilemmas. Draw-ing principally on the management literature, we focus on how marketers rely on structural and tem-poral solutions to solve implementation problems. We refer to the notion of ‘ambidexterity’ which originally relates to “the power of using two hands alike” (Oxford English Dictionary). The concept has first been applied to managerial contradictions by Duncan (1976) and has meanwhile entered various streams of research, like the strategic management


217

literature on alignment versus adaptability (Gibson and Birkinshaw 2004), the operations management literature on flexibility versus efficiency (Adler, Goldoftas, and Levine 1999) or the innovation man-agement literature on radical versus incremental, innovation (Tushman and O’Reilly 1994; Danneels 2002; O’Reilly and Tushman 2004). Strikingly, the marketing arena has largely ignored this promising stream of research. Against this backdrop, this paper has three objec-tives: First, it is among the first to apply the con-cepts of exploration, exploitation, and ambidexterity to the implementation of marketing strategies. Sec-ond, it analyzes different categories for implement-ing ambidexterity and suggests how various types may be best coordinated. In doing so, it provides novel insight as previous studies have only very rarely analyzed different types of ambidexterity in a single study. Third, it offers a multi-level and tem-poral framework that synthesizes previous research on ambidexterity and prepares the ground for some promising research avenues in marketing. The rest of the paper is structured as follows: In the next sec-tion, we provide a brief review of the literature on exploration and exploitation, which we will apply as a conceptual lens to redefine marketing dilemmas. Subsequently, we describe variants of ambidexterity which have been suggested to balance the dual processes of exploration and exploitation. This leads to our revised framework of ambidexterity, which we use to empirically investigate tools and proce-dures to support the implementation of seemingly conflicting marketing strategies. We present the re-search design, findings from four cases and discuss the results of our analysis. Finally, we identify some avenues for future research and managerial implica-tions.

2 The Role of Ambidexterity as a Framework for Solving Marke-ting Dilemmas

Most marketing dilemmas can be reframed by adopting the exploration versus exploitation lens (March 1991) that has attracted a large number of researchers in diverse fields of research (Birkinshaw and Gibson 2004; He and Wong 2004; Benner and Tushman 2003; Jansen, Van den Bosch, and Vol-berda 2005; Greve 2007). ‘Exploration’ refers to ex-perimentation with new alternatives, having returns that are uncertain, distant, and often negative. In

contrast, ‘exploitation’ describes the refinement and extension of existing competencies, technologies, and paradigms, exhibiting returns that are positive, proximate, and predictable. Keeping with the basic idea of March, most subsequent studies have fo-cused on the organization or the business unit as the respective level of analysis (March 1991; O’Reilly and Tushman 2004, 2007; Lavie and Rosenkopf 2006). However, terms have been used somewhat sloppily, invoking a variety of associated concepts to convey the nature of exploration and exploitation. For instance, exploitation has been associated with operational efficiency, control, and reliability (Por-ter 1996; Deming 1981; Juran and Gryna 1988). Similar confusion exists when exploration is differ-entially described as involving innovation, risk tak-ing, invention, new capability building, etc. (Benner and Tushman 2003). As interest grew further, ex-ploration and exploitation became widely popular phrases, capturing the fundamental tension in vir-tually any domain of organizational adaptation (McGrath 2001). A related discussion of organiza-tional learning processes also questions whether companies should engage in different types of inno-vation in order to maximize their outputs (He and Wong 2004; Jansen, Van den Bosch, and Volberda 2005). While a variety of innovation types is clearly seen as important, mixing them has often been con-sidered difficult – equally difficult as the combina-tion of seemingly contradictory marketing strate-gies. This has also been the customary tenet of strategists who argued that mixing low-cost and dif-ferentiation strategies is likely to result in lock in or “stuck-in-the-middle” positions (Porter 1980: 41) with “firms that are not particularly excellent at any thing” (Miller and Friesen 1986: 42). However, from a strategic perspective, the strict opposition has been relieved as both Porter (1996) and others think that it is possible to “play the spread” (Day 1989). Further input into the dynamic interplay of conflict-ing strategies comes from the punctuated equilib-rium model of change (Adner and Levinthal 2002) which describes different interacting modes of change and depicts organizations as evolving through long periods of stability (equilibrium peri-ods) punctuated by relatively short outbursts of fundamental change (revolutionary periods). Re-searchers have suggested that firms following this pattern are more successful (Miller and Friesen 1984) because they balance reactions to both inertial and disruptive forces. Indeed, temporal cycling be-


218

tween long periods of exploitation and short bursts of exploitation has been identified as a viable bal-ancing mechanism that may be both logical and practical (Brown and Eisenhardt 1997; Gupta, Smith, and Shalley 2006). While early studies within the exploration versus exploitation paradigm tell us that the two are rather exclusive, more recent research has emphasized the simultaneousness of exploitative and exploratory activities and their joint impact on performance (He and Wong 2004; Auh and Menguc 2005; Rothaer-mel and Deeds 2004). Indeed, initial suggestions have been made to structurally separate exploitation and exploration within different business units in order to increase organizational performance (Dun-can 1976). In addition, punctuated equilibrium the-ory convinces us that both types are inseparable and lead to performance enhancements if pursued al-ternately, i.e. via temporal separation (i.e. Gersick 1991). Paradoxes or dilemmas have long been discussed in the literature (Thompson 1967; Burns and Stalker 1961), However, it is only recently that the concept of the ‘ambidextrous organization’ (Lewis 2000; Tushman and O'Reilly 1996; Katila and Ahuja 2000; Birkinshaw and Gibson 2004; Raisch and Birkinshaw 2008) provided first insights into coex-istence and interdependencies of potentially con-flicting objectives. While the term ambidexterity has been used in different contexts, it has largely dealt with the twin concept of exploration versus exploita-tion in management-related studies (March 1991; Levinthal and March 1993; Danneels 2002; Benner and Tushman 2003; Jansen, Van den Bosch, and Volberda 2005). Here, we introduce the idea of ambidexterity as a conceptual framework for implementing inherently contradictory marketing strategies. In the literature, one of the common definitions of ambidexterity re-lates to the “pursuit of both exploration and exploi-tation via loosely coupled and differentiated sub-units or individuals, each of which specializes in ei-ther exploration or exploitation” (Gupta, Smith, and Shalley 2006: 693). For our purposes, this defini-tion is somewhat imprecise as ambidexterity can be understood as both, simultaneous or sequential bal-ance. Typically, the former is captured in traditional definitions of ambidexterity while the latter is sug-gested by punctuated equilibrium theory (Tusman and Romanelli 1994; Gersick 1991; Burgelman 2002) or temporal sequence approaches (e.g. Sig-

gelkow and Levinthal 2003). Contrary to other stud-ies, we explicitly use the notion of ‘punctuated am-bidexterity’ as a sequential approach to balancing contradictory marketing strategies. Thus, our un-derstanding is closer to He and Wong’s (2004: 483-484) research which defines ambidexterity in rela-tion to technological innovation: “We extend the ex-ploration versus exploitation construct to define …two generic dimensions (1) an explorative innova-tion dimension to denote […] activities aimed at en-tering new product-market domains, and (2) an ex-ploitative innovation dimension to denote […] ac-tivities aimed at improving existing market posi-tions.” Here the authors keep their definition open to include both temporal and sequential coordina-tion mechanisms. While different variants of ambidexterity have been discussed in the literature, we know relatively little about the systematic nature of implementing ambi-dexterity in an organization, and our ability to in-terpret the differences between various forms from a rigorous theoretical vantage point is limited. Therefore, we build on existing research in strategic management and organization theory to develop a novel framework based around the dimensions of coordination level and coordination logic by which different approaches to strategy implementation are highlighted. The first of these dimensions, namely the level of coordination, has a well-established tradition in the organization theory literature (Klein, Dansereau, and Hall 1994). Here, we refer to the levels of the individual and the organization and follow the quest by Simek (2009) to investigate multiple levels of ambidexterity. As for the organizational level, we consider ambidexterity a result of splitting the re-sponsibilities for different types of activities between business units. Thus, while decision-making re-mains within the organizational or corporate level, the implementation of ambidexterity occurs within business units with differentially allocated strategy variants. We further refer to the individual level, be-cause the individual employee impacts different strategies (Birkinshaw and Gibson 2004) and their implementation success. The second dimension, namely the distinction between simultaneity and sequence, contrasts the original ambidexterity con-cept with that of temporal separation, i.e. punctu-ated ambidexterity where organizations balance the pursuit of different marketing strategies over time. By integrating these dimensions, it is possible to


219

identify four generic types of ambidexterity which facilitate the implementation of contradictory mar-keting strategies (Figure 1). We discuss each of these variants in turn, and show how they relate back to prior research.

Figure 1: Variants of Ambidexterity

2.1 Structural Ambidexterity Marketing strategists have typically resolved the tension between exploration and exploitation by separating them in two different parts of the firm (e.g., marketing vs. sales; marketing vs. R&D; mar-keting vs. service). The first idea of structural sepa-ration is influenced by Duncan who argued that companies need to put in place ‘dual structures’ in order to manage conflicting demands (Duncan 1976). For instance, R&D and business development are given responsibility for new product develop-ment while the core business units focus on align-ment and exploitation. In this vein, McDonough and Leifer (1983) suggested that organizations use several structures simultaneously because the chal-lenges they face are so dramatically different that they cannot be managed within one organizational unit. Similarly, Christensen and Raynor (2003) proposed that established companies could pursue a disruptive innovation only in a separate business unit. Taking this further, Carson (2007) discusses the scope and required control mechanisms for out-sourcing new product development entirely – an ex-treme form of structural separation. Empirical research has further identified that struc-tural separation usually occurs at different levels of the firm, e.g., between headquarters and subsidiar-ies or global teams and centres of excellence. Alter-

natively, structural separation can be achieved by separating functions of a multi-product firm or by creating functions with a particular orientation. For instance, R&D is more oriented towards exploration while production units normally focus on exploita-tion. Related to the overall balance of strategies, the most common tenet has been to structurally sepa-rate business units: some with a focus on explor-ative innovation and others specializing in exploita-tive innovation (e.g. Ambos and Schlegelmilch 2005; Floyd and Lane 2000; Volberda 1998; O’Reilly and Tushman 2004; Benner and Tushman 2003). For instance, structural separation is under-taken by most large software companies which ex-ternalize specialist teams to develop revolutionary codes in distinct locations. With respect to implementation issues, structural separation is not without limits (Birkinshaw and Gibson 2004; Adler, Godolftas, and Levine 1999). Separation can lead to isolation of particular units as innovative ideas will not be transferred and inte-grated across the company. Where new product de-velopment is outsourced, idea transfer and integra-tion issues appear to be particularly pertinent. Fur-ther, due to the nature of its static approach of allo-cating contradictory tasks, units may loose their in-centives and turn inert over time. On the other hand, exploitation-oriented structures may lack creative outbursts, which may also result in inertia. Because of the inherent difficulties that arise from balancing exploration and exploitation, it has been suggested to adopt a longitudinal perspective where a unit’s focus changes over time.

2.2 Punctuated Ambidexterity Organizations may also disentangle different strate-gic marketing tasks by separating them over time. Major emphasis is placed on organizational units that focus on one type of strategy one day, and on a different set of innovations at another point in time (Brown and Eisenhardt 1997; Puranam, Singh, and Zollo 2006). Thus, organizations temporally sepa-rate between exploration and exploitation. For in-stance, the same decision unit may use a mechanis-tic structure for making routine decisions and then shift to an organic structure for making non-routine decisions. In the case of product development, temporal fluc-tuations may be best suited to cope with changes in technological jumps or breakthroughs which are likely to interpenetrate long periods of incremental


220

product development. In this situation, a whole unit adapts to the requirement of specific contingencies. Similarly, Siggelkow and Levinthal (2003) suggest that units switch over time by adopting more sepa-rated organizational design at one time and more integrated ones at another. These designs may dif-ferentially relate to information transfer and knowl-edge generation and, thus, have an impact on new product development rates. Occasionally, units or teams move from sales to ser-vice function or change from research-driven to commercial orientations. For instance, the RBC Fi-nancial Group in Canada is the largest bank as measured by assets and market capitalization and one of North America’s leading diversified financial services. In order to become more competitive, RBC realized that some of their units had to become more market-oriented. Management started a proc-ess whereby, over time, employees sensed that the culture was fundamentally changing. Their ideas were given due consideration, and they were being rewarded for market-oriented activities (Dobni 2006). One of the advantages of these shifts is that it prevented corporate units from turning inert. It also increased inter-departmental communication which has been identified as vital for successful strategy implementation (Forman and Argenti 2005). How-ever, there are also certain drawbacks, for instance, the ability to fluidly change organizational attributes and managerial approaches, with respect to strate-gic success criteria and incentive systems.

2.3 Contextual Ambidexterity The organizational literature also suggests that am-bidexterity emerges when leaders in a business unit develop a supportive context. The resulting type of contextual ambidexterity, which was formally intro-duced by Birkinshaw and Gibson (2004), is seen as the behavioral capacity to simultaneously demon-strate alignment and adaptability (Lubatkin, Simek, Lin, and Veiga 2006; Mom 2007; Carmeli and Halevi 2009; Guettel and Konlechner 2009). In-deed, contextual ambidexterity differs markedly from structural ambidexterity because it is “…a dual capacity … woven into the fabric of an organization on the individual level”. Instead of focusing on dual structures or on changing tasks for business units, organizations reconcile conflicting demands by building a set of processes, systems or contexts that enable and encourage individual employees to make their own judgement as to how to divide their time

between conflicting demands. For instance, market-ing managers decide whether they engage in new business development, service or sales activities, e.g. by involving customers with a need for ongoing in-cremental product development. Here, the contex-tual variant remains a set of stimuli that motivate marketing managers to act in a certain way. Previous research has acknowledged the role of managers in enabling and developing conditions for ambidexterity (Gibson and Birkinshaw 2004; Smith and Tushman 2005; Lubatkin, Simek, Lin, and Veiga 2006). Building predominantly on a leader-ship approach, studies have documented the impor-tance of contextual leadership that helps to balance requirements posed by contradictory activities and, consequently, sustains business unit performance (Osborn, Hunt, and Jauch 2002). Among scattered insights there is consensus on a few mechanisms that help to support contextual ambidexterity. For example, Tushman and O’Reilly (1997) argue that it is driven by internal processes that enable managers to handle large amounts of information and deci-sion alternatives to deal with conflicts. Carmeli and Haleyi (2009) suggest that managers should make active use of conflicts rather than avoiding them. As a consequence, behavioral complexity, i.e., the ca-pacity to adopt multiple leadership roles and change between them in selective contexts, is seen as pri-mordial (Lubatkin, Simek, Lin, and Veiga 2006). Most of the previous guidelines seem intuitively simple, but little has been said about their concrete implementation. As Mom (2007) remarks, there is clearly a lack of conceptually and empirically vali-dated understanding about exploration and explora-tion at the managerial level of analysis. This is sur-prising given the promising potential of related lit-eratures on organizational learning and top man-agement research. Furthermore, none of the exist-ing work has yet related to marketing tasks. Switch-ing between different contextual requirements seems indeed difficult as managers lack the ability of integrating exploitation into their mindset. Hence, they have to be substituted in order to pur-sue consolidation or exploitation strategies. This is captured within the last cell of peripatric ambidex-terity.

2.4 Peripatric Ambidexterity While the previous types of separation are well an-chored in the existing literature on ambidexterity, no explicit reference has yet been made to coordi-


221

nating marketing strategies by exchanging the top management or founding team. For this variant, we initiate the notion of ‘peripatric ambidexterity’, a term borrowed from genetics. This specifies the formation of a new species through during evolution and is often tied to the idea of a founder who devel-ops a new population within an isolated niche (Mayr 1942). In the business world we can find some analogies which emphasize the substitution of one ‘species’ by another. Take the example of Nissan and Renault, now headed by Carlos Ghoshn. The man-ager has instilled a vision of innovative and explora-tory leadership at the two companies that his prede-cessors failed to achieve. However, people presently ask whether he might be better off in the saviour rather than the consolidator and incremental adapter role. Thus, management may again have to change or evolve further. Similarly, when Goodyear was confronted by Michelin’s introduction of the ra-dial tire, Goodyear’s senior management team first focused on the existing product and avoided radical challenge. When they finally introduced the radial tire, Goodyear completely shifted from bias-ply tires to radicals. This strategic shift at Goodyear led to the creation of a new senior management team (Virany, Tushman, and Romanelli 1992; Smith and Tush-man 2005). The literature on leadership covers this phenome-non with reference to the ‘human trait of ambidex-terity’ (Lubatkin, Simek, Lin, and Veiga 2006). Fur-ther insight comes from research on founding teams and their prior affiliations which are presumed to shape new firm behaviors (Beckman 2006). Execu-tive behavior is supposed to be driven by past in-sights and experiences. Thus, previous career ex-periences shape the range of action that is available to formulate and implement strategies, i.e., engage in either explorative or exploitative behavior (Boeker 1997; Kraatz and Moore 2002). While a va-riety of perspectives stimulates exploration, a singu-lar company affiliation is likely to improve existing routines and consolidate existing capabilities. The size of the firm may play another important role for choosing peripatric ambidexterity. While large firms with a variety of business units are likely to pursue both exploration and exploitation, an entre-preneurial firm exists as a single business unit. Thus, the choice of a homogeneous team defines the corporate orientation to strategy variants. Contrary to punctuated ambidexterity, the peripatric version posits that it is difficult for individuals or teams to

change behaviors. Once they surpass a critical size, only a few managers from the original founding team are normally able to manage the shift. As sug-gested in situational leadership theories, it is much easier to find the manager and identify situational fit (Fiedler 1967). Thus, a firm needs to exchange its top management if it intends to implement a drastic change in corporate strategy. In summary, the description of the four variants of ambidexterity has shown that each of them is a suit-able solution for not only balancing exploration and exploitation within an organization, but also for solving concrete marketing dilemmas by imple-menting ambidextrous designs. Table 1 summarizes the major insights from the literature, highlights se-lected marketing dilemmas and implementation categories, and presents the few existing insights on the practice of ambidexterity. As these insights are largely anecdotal, our study is largely exploratory (Churchill and Iacobucci 2005; Srnka and Koeszegi 2007) and investigates implementation categories in more detail.

3 Research Design The findings presented below are part of a larger re-search project on ambidextrous innovation in global firms. In the first (and ongoing) stage of the project, we have conducted 50 interviews in Germany, Aus-tria, France, The Netherlands, Switzerland, and the UK. Industry sectors included banking, engineering, automotive, information technology, consulting, fast moving consumer goods, and biochemistry. In the second stage, we purposefully re-contacted four firms (Patton 1990), Apple, Deutsche Bank, Deloitte, and Celltech (UCB), and conducted in-depth interviews to focus more specifically on how diverse marketing strategies can be best imple-mented by using ambidextrous designs. More de-tailed information on the research design can be found in the appendix.

4 Presentation of Findings We start the presentation of findings by reporting the results of our content-analysis per firm (Miles and Huberman 1984). The structure of each case presentation begins with a short overview of the firm and major conflicts or dilemmas. We discuss relevant implementation categories of ambidextrous designs in detail and illustrate how they support the combination of controversial marketing strategies.


222

Table 1: Major Approaches to Ambidexterity, Selected Marketing Dilemmas, and Implementation Criteria

Structural

Ambidexterity

Punctuated Ambi-dexterity

Contextual Ambidex-terity

Peripatric Ambi-dexterity

Major Studies Duncan (1976); O’Reilly and Tushman (2004); Tushman and O’Reilly (1997)

Jansen, Van den Bosch, and Volberda (2009)

Punctuated equilib-rium (Gersick 1991; Tusman and Ro-manelli 1994; Burgel-man 2002)

Temporal separation (Siggelkow and Levin-thal 2003)

Gibson and Birkinshaw (2004); Smith and Tushman (2005); Lubatkin, Simek, Lin, and Veiga (2006); Mom (2007); Carmeli and Halevi (2009)

Genetics (Mayr, 1942) Leadership research

Burns (1978), Bass (1985)

Definition of Ambidex-terity

”Ambidexterity refers to the synchronous pursuit of both exploration and exploitation via loosely coupled and differenti-ated subunits or indi-viduals, each of which specializes in either ex-ploration or exploita-tion.” (Gupta, Smith, and Shalley 2006: 698)

“Temporal cycling be-tween long periods of exploitation and short bursts of exploration” Gupta, Smith, and Shalley 2006: 693)

“Contextual ambidexte-rity is the behavioral ca-pacity to simultaneously demonstrate alignment and adaptability across an entire business unit” (Gibson and Birkinshaw 2004: 209)

“Change towards a new evolutionary period of the firm which is inextricably linked to a CEO or Top Management Team” (inspired by Mayr, 1942)

Description Dual structures, differ-ent units focus on dif-ferent types of activities or challenges

Single units, shifting focus over time, i.e. concentrating sequen-tially on different types of marketing

Individuals in business units, who decide on which type of activity they want to focus

Top Management Team / CEO focuses on one particular type of strategy at a given time

Contribution of Ambi-dexterity Framework to Resolving Marketing Dilemma

Combination of brands and products which may otherwise canni-balize each other; tar-geting of incompatible customer segments (e.g. Deutsche Bank vs. Maxblue Investment Platform; SAP Soft-ware Development )

Disruptive technology changes versus incre-mental product inno-vation

(e.g. Siemens VDO has, at a point in time, ac-counted for 90% of all innovations)

Short-term balance of existing and new mar-keting trajectories

(e.g. Google with indi-vidual responsibility for radical product innova-tion and adjustment)

Sales-driven vs. con-solidation driven marketing; emotion-nal versus informa-tive marketing (e.g. consolidation of Ger-man Metallgesell-schaft by Kajo Neun-kirchen or Renault/ Nissan with Carlos Ghoshn)

Implementation of Am-bidextrous Designs

Strong autonomy; cross-unit relationships (Raisch 2008); integra-tion through teams and organizational mecha-nisms (Jansen, Tempe-laar, Van den Bosch, and Volberda (2009),

Focus on specialized skills closely tied to the requirements of one activity or period. Change management

Behavioral complexity (Hambrick and Mason 1994)

Organizational context (Ghoshal and Bartlett, 1994), Culture (Ne-manich and Vera 2009)

No explicit studies as construct has not yet been used in the lit-erature

Selected Implementa-tion Categories from the (Marketing) Strategy Literature

Structural variables: company’s marketing function, control systems, policy directives (Bonoma and Crittenden 1988); formal firm structure and networks of informal communications (Franckwick, Ward, Hutt, and Reingen 1994);

Behavioral variables: skills, resource allocation, ability to develop informal structures (Bonoma and Crittenden 1988);

Interpersonal processes: strategic consensus and autonomy, leadership and implementation styles, communication (Noble and Mokwa 1999), shared understanding (Woolridge and Floyd 1989);

Individual-level processes: cognition, organizational roles, commitment (Noble and Mokwa 1999);

Managerial levers: goals, organizational structure, leadership, communications, incentives (Noble 1999)


223

We continue with a discussion of commonalities across cases and compare findings with relevant in-sights from the literature.

4.1 Q110 - Deutsche Bank der Zukunft: Structural Ambidexterity

Deutsche Bank, with headquarters in Frankfurt/-Germany, is one of the leading financial institutions with some 80,000 employees in more than 72 coun-tries. With a large presence in Europe, the Americas, Asia Pacific and the emerging markets, the bank of-fers financial products and services for corporate and institutional clients along with private and business clients. In 2005, having faced hugely nega-tive customer satisfaction and retention rates, espe-cially with outsourcing its online subsidiary, Deutsche Bank24, the company decided to break new grounds. Driven by the Private Client and Asset Management division (PCAM) and its subdivision Private & Business Clients (PBC), which provides private individuals and small to medium-sized businesses with a full range of traditional banking products, the company decided to launch a new in-novative subsidiary in the city centre of Berlin, Germany. The creation of “Q110 - Deutsche Bank of the Future”, as it is referred to, was largely market-ing-driven and (originally) had the sole purpose of bringing customer satisfaction rates to new heights. Q110 is tangibly different from corporate headquar-ters and other subsidiaries as it draws inspiration from unusual ideas and strikes out new paths – of-ten involving innovative technology. Right from its inception, the subsidiary was created as a translu-cent building, integrating shops, service offerings, a lounge for relaxation and some 1260 square meters for interaction. No counters are allowed to interfere with the boundary-free interior design and the open spirit of the unit’s team. As one interviewee (B3) put it: “Each time we create an encounter, we do so ex-actly in the way the customer prefers it….this could be a coffee in the lounge, a product show on one of the oversized TV screens, a guided tour ….or, if the customer prefers, a quiet room. And even more, the customer does not necessarily come here to de-mand a banking service, but rather because he likes the atmosphere or looks for inspiration and design products.” Thus, the centre is characterized by customer orientation, proximity and approach-ability, which also includes opening hours beyond the expected.

When launching structural separation, one of the major challenges of the new subsidiary was to intro-duce new customer and service experiences and market them differently. However, this new market-ing approach needed to be compatible with a tradi-tionally conservative approach to banking and fi-nancial services. Ambitiously formulated (B2), “The objective was to win all private customers for our subsidiary.” Thus, within the overall conglomerate of Deutsche Bank, it was clearly decided that Q110 was to act beyond traditional structures supporting a context of creativity. While the structural separa-tion of the unit was one issue, from a corporate per-spective, it was even more important to integrate different marketing strategies and underlying value systems as both, Q110 and more exploitative sub-sidiaries were to coexist under the umbrella of Deutsche Bank AG. The activities launched by Deutsche Bank to implement structural ambidexter-ity can be described in terms of the nine implemen-tation categories (see Table A.2 in the Appendix). Goal-setting is an important aspect of a firm’s im-plementation process and clear objectives should be communicated (Noble 1999). At the stage of imple-menting Q110, the formulation of strategic objec-tives was in line with the exploratory function of the unit, e.g., its overachieving goal of customer devel-opment and whole-hearted dedication to customer satisfaction. The implementation literature also em-phasizes that without knowledge of the strategic vi-sion, functions would not be able to implement a marketing strategy within a broader context (Noble 1999). Initially, there were no immediate financial constraints, which would have been imposed on a more exploitative unit. In this sense, achieving am-bidexterity created paradoxical situations because the short-term efficiency and control focus of head-quarter-dominated subsidiaries was at odds with the long-term experiential focus of Q110 (see also Floyd and Lane 2000). In terms of structure, strategy implementation is often difficult when units are unequally affected or represented, e.g., when one unit is considered more important than the other. However, separating Q110 as a different unit from corporate headquarters was not seen as a major issue (B5: “We all work for Deutsche Bank, we do not talk about Q110 to our customers.”) While structural separation was driven by the top management, with people working in clearly distinct locations, employees’ mindsets were not completely detached from other units. In part


224

this is due to the overall spirit and vision Deutsche Bank integrates in the different units. Further, an ongoing communication process with colleagues of other units reflects the typical work flow when as-signments and clients are handed over. Formal meetings are complemented by informal communi-cation, especially when the objective is serving the customer better (B6:”When there is a customer, and even if we do not have the time, we talk to our colleagues and we try to constantly be in touch to see who has some availabilities… we normally find a solution”). Incentives systems are important tools in imple-menting strategy and include both monetary and non-monetary rewards (Walker, Churchill, and Ford 1977). Within Q110 no special financial incen-tives were put in place (e.g. B4: “We do not receive any bonus payments simply because we work for Q110…even though the work is quite different and we have long working hours, including even Sat-urdays.”) However, what was clearly mentioned were immaterial incentives arising both from the team atmosphere and, most prominently, from fu-ture career perspectives (e.g. B3: “…many people, after they left Q110 have continued their manage-ment career somewhere else within Deutsche Bank. Having worked for Q110 is apparently seen as a preparation for major career jumps”). Apart from being an individual incentive, the promotion of peo-ple into other units serves a means of integration and transferring exploratory knowledge to exploita-tive units. The organizational context, e.g. the systems and processes that define the context for ambidexterity is strongly related to individual empowerment through leadership. This reflects the criteria of dis-cipline and stretch as illustrated by Ghoshal and Bartlett (1994), who propose that employees volun-tarily strive to meet expectations. But more impor-tantly, organizational stretch induces them to strive for more, rather than less ambitious goals. While employees work in a structurally separated context, it is these criteria that facilitate contextually ambi-dextrous behaviors within their explorative unit. It seems as if this has been an objective from the be-ginning as recruiting policies did not only focus on exploration-oriented people but on diversity. For instance, a formal banking education was not part of the predominant catalogue of criteria. (B3: “I was first a carpenter, then I had a traineeship with the bank, did several things in between and now I am

here. What they require is more a certain attitude and willingness to learn than a formal education. However, this has changed a bit in the meantime and I think everybody now builds on an appren-ticeship with Deutsche Bank”). The diversity in em-ployee background and culture was seen as one re-quirement to increase behavioral complexity. This is vital in a context that incorporates all sorts of service offerings beyond the typical banking products. The case of Deutsche Bank was initially selected to investigate structural ambidexterity, which occurs on several levels of the firm. For instance, it was the driving element to distinguish both between corpo-rate divisions and between sub-divisions like Private and Business Clients (PBC) and Asset and Wealth Management (AWM). In several geographic re-gions, AWM is placed in attractive locations and buildings clearly distant from the PBC functions. Spatial/geographical separation also presented the founding principle for Q110 in Berlin, which was dislocated from the corporate functions in Frank-furt/Main. However, like a ‘nested design’, Q110 does not only serve as an example of structural separation, but is also built on contextual ambidex-terity and, to a minor degree, incorporates elements of punctuated ambidexterity (see further explana-tion in the cross-case discussion).

4.2 Deloitte Consulting: Contextual Ambidexterity

In 1995, the partners of Deloitte Touche Tohmatsu voted to create Deloitte & Touche Consulting Group (later Deloitte Consulting) which, after creation, remained part of the overall Deloitte group. By inte-grating skills from across the firm, Deloitte answers clients' demands for single-source solutions. Glob-ally, Deloitte is present in over 130 countries and employs more than 90,000 people. Consulting ser-vices are focused on three main areas: Strategy & Operations, Technology Advisory Services, and Human Capital Advisory Services. In terms of reve-nues, the company achieved $6.3 billion in 2008. The need for ambidextrous designs was driven by several developments within the consulting indus-try, most notably customer demand, information technology, changing regulation, and increased competition. These trends confronted managers with the challenge to increasingly focus on both ex-ploration and exploitation activities. Within Deloitte Consulting, this challenge is most prevalent on the partner and manager level (D1: “The partner role is


225

naturally one of developing the business, but you also have to capitalize on your existing abilities and current projects, and this is precisely the difficulty – doing two things that do not quite go well to-gether’”). Indeed, it is clearly expected that partners conduct both exploitative (e.g. striving for cost-efficiency) and explorative activities (anticipating industry changes) to deal with current and future requirements of the firm. As front-line experts, they are so close to customers and their specialized ex-perience makes them ambassadors of their com-pany’s services. At the same time, they need to ex-tend previously successful projects and procedures with seemingly minor modifications to fit the new clients. In the firm, roughly 30% of the work time is dedicated to new developments and research, while the dominant focus remains on working with the client. However, managers and partners have lee-way in deciding upon their current focus. Ambidex-trous challenges also occur in the capability-based marketing approach of the firm. In building trust with clients and potential employees, knowledge and competence are seen as major drivers. This im-plies that both existing competencies need to be nurtured, while new knowledge is required to signal cutting-edge technical expertise. In order to implement these challenges, a variety of implementation instruments was used. On a strate-gic level, the overall goals of the firm do not change while it continuously reorients itself towards market demands. Growth is certainly an issue and the firm tries to proportionally manage growth while expect-ing profitability rates to rise accordingly. This im-plies a multi-market approach (D4: “It also depends on the location, some are exciting in terms of firms coming into the country, others are more estab-lished; you definitely do not want to be limited to local markets especially if you look at markets that are definitely going to grow […] but this requires flexible adaptation”). When this type of flexible adaptation has been suc-cessfully achieved, “every individual in a unit can deliver to existing clients … but at the same time every individual is on the lookout for changes in the task environment, and acts accordingly. This is po-tentially a more sustainable business model … be-cause it facilitates the adaptation of an entire busi-ness unit, not just the separate units or functions responsible for new business development” (Gibson and Birkinshaw 2004: 211). Within Deloitte, this is reflected in their ‘multiple market approach’, i.e.,

offering global services adjusted to local contexts. This requires that managers individually adjust be-haviors to national clients while representing a global firm. The succinct balance arises from con-tinuous adaptation and re-adaptation. In doing so, continuous learning is one of the major competences of the firm which does not only affect the managerial level but transcends the organiza-tion. Thus, all employees are actively encouraged to develop, improve, and refine in-depth knowledge pertaining to a certain market segment, country, culture, product, service, or internal process which is anchored in the organizational context and sup-ported by suitable structures. The organizational context can be described according to the four ele-ments of discipline, stretch, support, and trust (Bartlett and Ghoshal 1994). Within Deloitte, disci-pline is already reflected in the recruiting process where the firm is being committed to being trans-parent about the progress toward global goals, both for employees and the firm. Member firms, partners and employees are all united in the effort to collec-tively achieve these goals and help turn a shared vi-sion into results. The development of the vision in-corporates ‘stretch’ so as to inspire people with the abilities to organize themselves and move upwards. With regard to more exploratory activities, the value system emphasized diversity from which a broader variety of new inputs is believed to emerge. As a ba-sic prerequisite, employees are recruited according to criteria such as openness, ability to learn and re-lationship skills with clients. Internally, these skills are constantly challenged and stretched to facilitate optimum careers (D4: “I always try to see the po-tential of people, I listen intensively, challenge them… help them, and discuss options with them”). In addition, the career path is open to focus more on technical expertise or managerial skills where peo-ple can change the percentages and requirements of explorative versus exploitative behavior. These moves are actively supported by the management which lends access to resources and interchange. Finally, the element of trust plays a particular im-portant role not only within the firm, but also in re-lation to its clients. Trust is seen as the selection cri-teria that facilitates relationship building and long-term developments. This is also reflected in the firm’s approach to culture which takes on open and supportive stance and focuses on the idea that de-velopment is a voluntary act (D1:”People want to learn. If we want to change something we need to


226

explain it to them so that they see the benefits and can decide themselves which way to follow”). Cul-turally-driven knowledge development initiatives serve this purpose of interpersonal support and ex-change as they are normally integrated into daily operations and customer needs. In summary, the example of contextual ambidexter-ity shows that soft implementation factors dominate the balance of marketing strategies. This is in line with an understanding of knowledge management and capability development as voluntary actions transcending different levels of the firm.

4.3 Celltech: Punctuated Ambidexterity The origin of Celltech, one of the largest biopharma-ceutical companies in Europe, dates back to the year 1980, when investment funds were obtained to build up a strong foundation around diagnostics, nutritional and contract business followed by thera-peutics in the late eighties. Celltech can be viewed as having four basic historical periods, which link to the balance of exploration and exploitation. First, the two different strands of biologics and therapeu-tics were meant to subsidize each other. Between 1990 and 1992, the future was seen in the develop-ment of innovative drugs, thus in expanding ex-ploratory innovation of the therapeutics unit. In the third period between 1992 and 1996, resources were equally distributed to both units, while at the same time fostering a strategy of external collaboration. In this period, business units reaped off the benefits from exploiting their existing knowledge via cash milestone payments from their collaborators with-out selling a full interest in the downstream prop-erty rights. Finally, in 1996 the biologics division was sold. From now on, the unique focus of the therapeutics unit changed from exploitation to ex-ploration and the success of this strategy was evi-denced in a 25% increase in share prices in 1998. However, there were difficult challenges associated with the unit’s shift between exploitative and ex-plorative strategies. Viewed across extended periods of time, balance was achieved by maintaining three explorative (discovery of new drugs, phase one clini-cal trials, development of collaborative capability) and two exploitative activities (phase two and three clinical trials, management of prestige alliances, which facilitate access to world-class capabilities). The resulting question is how Celltech managed to induce and maintain this balance.

Initially, management support in developing a shared culture and language across the firm was one of the major facilitators (McNamara and Baden-Fuller 1999; Dodgson 1991). Emphasizing the need for exploration originated from an ongoing crisis at the end of the 1980s when the biology unit had been the major source of revenues. At that time, it was largely based on the exploitation of academic knowledge. On the therapeutics side, exploitation did not achieve sustainable revenues, and change would have to overcome the inert culture and an over-reliance on existing knowledge. External re-quirements from stakeholders demanded new di-rections and new capabilities had to be developed in order to focus on explorative product development rather than on technical excellence. One of the ma-jor associated challenges of putting these opposing strategies into place resulted in ‘unlearning’ old be-haviors. This was seen as difficult because efforts and risks in switching from one capability to an-other can be substantial while necessary to substan-tiate marketing activities. Celltech managed to do this by applying several managerial techniques which we present with reference to our category scheme: In terms of corporate goals, the changes within Celltech were driven by external market contingen-cies. Celltech’s corporate strategy has been adapted over its 11 years of existence and the firm’s man-agement has defined respective ‘tipping points’ which characterized the shifts from more explor-ative to exploitative periods and vice versa (Dodg-son 1991). For instance, the choice of the new thera-peutics focus was driven by a new leadership efforts (Clit: ”It came from the top. Of course, there were managers further down but the way we were go-ing to organize ourselves came from David Blox-ham (Director of Research)”; McNamara and Ba-den-Fuller 1999: 7). Essentially, the company’s strategy has changed from being a research com-pany to a manufacturing company to a fully inte-grated biopharmaceutical company developing, manufacturing and marketing drugs. Implementing these changes in the skill base, associated changes in strategy and learning processes needed to be dif-fused throughout the units. The internal learning process was largely influenced by soft implementation criteria, such as culture, communications, and leadership styles: When the unit changed from a discipline and technology-based capability to explorative therapeutics, chem-


227

ists were thrust together with biologist leading to differences in understanding. Disciplines that had previously worked in isolation now had to converse and train each other in the basics of their discipline. (C1: ”It was quite amazing. Formerly, you were a specialist in your field... you still continued to be one but the new task was move out of the box and increase interdepartmental knowledge sharing”). An understanding of the language and mindset of each other facilitated a deeper understanding. Trig-gers for innovative solutions were set off through this process of developing a shared understanding at the level of bench scientists. The firm’s organiza-tional culture or ’social fabric’ played a major role in determining the speed of learning a new explorative approach to innovation (Brown and Dugoid 1991). Indeed, particular attention was paid to managing exploration for new knowledge. This involved in-formal interaction mechanisms as staff was located in close proximity and the layout of the building was specifically designed to facilitate interactions. Sup-ported by more formal mechanisms, like quarterly reviews, the team was forced to constantly interact with the senior management in order to formulate realistic forecasts and have new ideas bubbling up. In terms of recruitment, changes were supported by broadening the existing skill-base (Clit: ”The new people faced the informal and social club of the old.. it took time to integrate them but the process of learning to work together created a new knowl-edge base and capability within the firm”; McNa-mara and Baden-Fuller 1999: 11). While employees previously used to work in functional teams, they were now required to launch projects by tapping dif-fering functional expertise within one group. Thus, diversity of knowledge and the combination of ex-pertise was deliberately used as a driver of knowl-edge development. Staff from different disciplines was hired to support the formation of new team structures and infuse the unit with new resources. The required roles and skills reflected the predomi-nant mindset of the unit at specific points in time, while a major focus was placed on extensive training support to increase expert knowledge. This was tied to individualized incentive systems which rewarded enthusiasm and commitment, whereby, the com-pany had considerable flexibility in its employment system to allow for this. While expert knowledge remained the central recruiting criterion, it was immediately complemented by extensive training. To reward expert knowledge, the role of Principal

Scientist was created for those people wishing to concentrate on developing scientific expertise, rather than undertaking management responsibili-ties (Dodgson 2001: 146). Taken collectively, the move from exploitation to exploration was considered risky, but the unit’s suc-cess helped the organization to survive and grow. One of the major lessons learned was the fact that the company did not only require a change in scien-tific, but also in managerial capabilities in order to succeed. These relied predominantly on knowledge-sharing skills. Further, systems were put in place to ensure the long-term balance between exploration and exploitation by inserting discovery projects into the life-cycles of knowledge exploitation. This helped to prevent exploration being driven out by exploitation which has often been mentioned as one of the major dangers (McNamara and Baden-Fuller 1999). In terms of the organizational context, sev-eral monitoring procedures were implemented to guarantee the ongoing success of the new strategic orientation. In 2004, Celltech was acquired by UCB while Celltech’s innovation stance within the larger group remained the same.

4.4 Apple Computer and Peripatric Ambidexterity

In 1976, Apple was founded by Steve Jobs and Steve Wosniak. Within several months, they had pro-duced 200 computers. Jobs made it Apple’s mission to bring easy-to-use computers to the market. In 1978, the company launched the Apple II that sparked a computing revolution that drove the PC industry to $1 billion in annual sales in less than three years. Apple quickly became the industry leader and, in 1980, launched a successful IPO. The competitive position changed in 1980, when IBM entered the PC market and Apple’s market share continued to drop. In 1985, Apple removed Jobs from his operational role and handed over to John Sculley, a former Pepsi executive. Under Sculley, Apple worked to drive costs down and sustain prof-itability. However, in a lacklustre period during 1989-1997 Apple was nearly written off but it had an impressive and dynamic comeback after Jobs re-joined the company in 1998. The restructuring ef-forts continued. Between 2003 and 2008, sales multiplied to $24 billion and it topped Fortune´s 500 companies in terms of total return to share-holders in 2003-2008 with a 94% return as well as


228

during the period of 1998-2008 with a 51% return (Morris 2008: 68). One thing that makes the firm special is Steve Jobs, the co-founder and CEO of Apple, who has always been a persuasive and charismatic evangelist for the company. (A4: “No, by no means is Steve Jobs a normal boss. He is in the firm, the spirit is alive, and he is always present”). While Jobs has power-fully put forward the growth of the firm, an indus-try-wide sales slump towards the end of 1984 caused deterioration in Jobs' working relationship with then CEO John Sculley. By the end of May 1985, following an internal power struggle and an announcement of significant layoffs, Jobs was ousted before making a triumphant return in 1997. But why has the rise and fall of Apple been so closely linked to the innovation-driven leadership style of Steve Jobs? And why did the company temporarily have to get rid of him in order to consolidate operations? One of the major explanations builds on his personality. Jobs has always been driven by the energy to change the world with exciting new products. Fundamen-tally, the design and development of the Macintosh in 1984, Apple’s flagship product, is an emblem of Job´s entrepreneurial spirit that characterized Apple in its garage-shop days. Driven by explorative innovation strategies, Jobs unleashed the creativity of highly individualistic and talented soft- and hardware designers. He infused the development team with a renegade spirit, often shouting the battle cry, “It’s better to be a pirate than join the Navy” (Sculley and Byrne 1987: 147). Indeed, the pirate metaphor was more real than fiction as Jobs, as well as his frenetic work schedule of 48-hours of straight programming and breaks with pineapple-topped pizza generated a sense of shared respon-sibility and togetherness. However, while the new product generated publicity, company morale was low. Apple had become a company that was too preoccupied with management and groups rather than with inspiring new products. Jobs’ personal success gave way to a professional nemesis as his leadership and inno-vation style did not fit any longer to the company’s immediate requirements. Executing a shift in stra-tegy from exploration to exploitation was seen as a challenge for top executives. While competitive demands from the external environment left no manœuvering space, the firm’s internal social struc-tures, roles, and norms were still influenced by an

explorative forward-looking management style. Human agency, in terms of a different personality, was seen as the primordial means to execute a change (consolidation) in the firm’s strategy. How-ever, once Jobs left the company, Apple became so marginal in the computer industry and losing so much money that analysts debated whether it would implode or be sold. Jobs, who returned to Apple in 1997 after years of exile, was still attracted to devices that define new categories, rather than compete in large, pre-existing industries. His comeback was seen as Apple’s desperate attempt to survive one of its worst phases and the company again needed a charismatic leader who would revive its fortunes (A4: “The years in which he was not there were ok, we still had a huge hype but he was missing. When he came back in 1998, we again had all the atten-tion“). And since Jobs himself has proved unrivalled in the art of managing disruption, he was again able to take Apple to new heights. In doing so, he built on his ability to get his employees so motivated that they maniacally achieved goals or created technology far beyond what they thought was possible. The story of how Jobs built the Apple empire, how and why he was thrown out of it and the reasons behind his return to Apple is essentially the story of an entrepreneur and a firm’s peripatric ambid-exterity – a prominent corporate marketing strategy tied to a single person. It shows that a leader’s personality influences a company and indeed a fundamental change can only be achieved once the leader is dismissed – or resurrected. (A4: “There will be a big bang when he is no longer there…what is obvious, the brand is tied to him, it is crazy how one person drives this firm”). In terms of implementing the current strategy, several elements were observed. Fundamentally, the firm is driven by the charisma and inspiration of one person, even though the top management has a proper function in the actual operations. But the role of a (this) CEO is one of the fundamental driv-ers for firm strategy. Building on his charisma, em-ployees are inspired to work for the firm rather than by being attracted by impressive financial bonus systems. The company has a very hierarchic struc-ture (A4: “You may check back with London (the European headquarters) but then, quite quickly, you are in the US”) which also implies that deci-sions are often checked and double-checked rather than spontaneously implemented by competent


229

managers. Various control systems are in place and often employees rather try to play it safe. In terms of communications, the company has an explicit non-disclosure policy and even internally, cross-functional communication is often limited, some-times by design, sometimes by default. In critical areas, knowledge transfer is inhibited by different locations (structural separation) in others, simply by a lack of engagement into inter-departmental com-munication (A5: You won’t believe it but there are many areas where access is restricted and you have no idea what people are doing, you have no idea what the company is heading at”). It seems like these management principles match certain employee characteristics. People working for the firm have been describes as all “being somewhat crazy”, A4) and the firm’s recruiting policy might already include a self-selection process as to which people apply for a job. Once in the business, people either conform to the vision of the firm (that of Steve Job) or leave. For those who have been doing the job for a while, it is still astonishing how much they feel the daily presence of the firm leader (A5: While it looks good, it is not the touch and feel to make everybody happy ... it is rather tough be-cause you feel obliged to function, and I say ‘func-tion’ because sometimes you just have to do what you are being told. But somehow, it is strange, you do everything for a glimpse of attention”). The du-ality of micro-management and charismatic inspira-tion provides some of the driving forces for the im-plementation of explorative activities. While this strategy has proven successful over the past, the firm may require exploitative moves. However, with such a strong person-dominated culture and strat-egy, the decisive question is whether the firm is able to change. One of the major concerns is that a new CEO might change the overall orientation of the firm and, thereby, destroy what makes it special. In summary, the four cases provide insights into how marketers could coordinate different strategies by implementing ambidextrous designs. All cases started from different contexts and highlighted dif-ferent implementation criteria. While each firm applied specific implementation tools and proce-dures, there are several commonalities across cases, which we discuss with reference to the relevant literature.

5 Discussion For every firm it is essential to achieve both growth and profitability which often hinges upon the suc-cessful implementation of marketing strategies. This includes the combination of exploratory and ex-ploitative activities. A focus on one of these tasks may be easy but combining and balancing them poses a major challenge. Addressing the question of how this dilemma could be solved, this study con-tributes to the emerging literature of ambidextrous designs and investigates them through the lens of a marketing perspective. The discussion of four dif-ferent types of ambidexterity has illustrated an in-creasing number of studies that deals with struc-tural, contextual, and punctuated ambidexterity. In addition, we have introduced a fourth type of peri-patric ambidexterity that has so far not been dealt with in the literature. While there has recently been an increasing interest into the issue of ambidexter-ity, insights into how ambidextrous designs can be implemented in order to solve marketing dilemmas are largely lacking. In addition to specific findings from each of our case firms, several commonalities have emerged that extend the previous literature on ambidexterity. In the case of structural ambidexterity, the decision to coordinate different activities across organiza-tional units has been identified as an important step (O’Reilly and Tushman 2005). For Deutsche Bank this was an essential move forward to develop a new business approach outside the confines of corporate hierarchies. This solution has also been suggested when executives apply paradoxical cognitions (Smith and Tushman 2005) to frame exploration vs. exploration tensions, e.g. traditional values vs. ex-citement in banking, and consequently actions in the firm. However, separating units is only one side of the coin and successfully managing an explorative or exploitative unit does not provide insight into how value generated in this unit can be transferred back to other units or even corporate headquarters. While one of the advantages of structural ambidex-terity is that it creates “pragmatic boundaries” (Car-lile 2004) which safeguards conflicting activities, our findings suggest that employees working in separate units realize the potential paradox of ex-ploration and exploitation but do not necessarily see them as constraints. For instance, Q110 employees capitalize on their specific status, but integrate and reintegrate values and experiences of the different


230

units by regular exchanges and potential career moves. Thus, while structural ambidexterity was used to kick-start exploratory innovation in bank-ing, contextually ambidextrous individuals were seen as a necessary condition to maintain the spe-cific status of the subsidiary. Against conventional insight which focuses on the top management as integrators (O’Reilly and Tush-man 2004; Smith and Tushman 2005; Lubatkin, Simek, Lin, and Veiga 2006), Deutsche Bank ap-plies a ‘nested design’. On the (inter)personal level, it relies on informal knowledge flows and the re-cruitment of contextually ambidextrous people. While selected mechanisms like senior team inte-gration or cross-functional interfaces have been dis-cussed, the more informal integration mechanisms should not be underestimated (Jansen, Tempelaar, Van den Bosch, and Volberda 2009). In addition to formal and often pre-established integration mechanisms, informal procedures refer to emergent social properties that influence the way employees exchange knowledge across boundaries (Tsai 2002). In the case of Deutsche Bank, both formal and in-formal communication channels complement each other, information barriers between units are low, and knowledge exchange is driven by individuals. In comparing structural and contextual ambidexter-ity, the former allocates conflicting tasks to different units, but misses the value creating component if subsequent integration is not sufficiently ensured. Hence, the interplay between structural and contex-tual ambidexterity provided a basic means to sup-port knowledge transfer from exploratory to ex-ploitative units. People with the ability to behave differently in the same context (i.e. contextually ambidextrous managers) or those who are able to switch contexts (which may drive the shift of whole teams towards explorative or exploitative functions, i.e. punctuated ambidexterity) can be seen as vital agents of balancing conflicting marketing strategies. For these capabilities and their efficient allocation across the boundaries of structurally separated units, the organization needs to provide career in-centives to support contextually sensitive managers. This leads to continuously creating new combina-tions of exploratory and exploitative marketing strategies at other levels of the firm (Sirmon, Hitt, and Ireland 2007). This is what both Deutsche Bank and also Celltech did. In promoting former Q110 employees into higher management positions they transferred a creative customer and marketing focus

into more exploitation-oriented units. In the case of Celltech, diluting exploitative constraints and pro-gressing to exploration was also based on career op-tions and diversified recruiting. In the case of Apple, more formal communication structures dominate. What is communicated largely ties to the vision which drives the overall marketing approach, and has spill-over effects to structures and leadership principles. Most implementation pa-rameters are tied to the dominant orientation and implementation success becomes a matter of both communication quality and quantity. For instance, Apple deliberately minimizes communication flows between hardware and software developers (struc-tural separation) and strategic communication channels are dominated from the top. This is equally manifested in the external non-disclose pol-icy as well as limited interdepartmental communi-cation. While it is important that visions are com-municated to rationalize both exploratory and ex-ploitative innovation (Jansen, Van den Bosch, and Volberda 2006), in the peripatric version of ambi-dexterity, one type of strategy at a specific time is emphasized. In order to implement explorative marketing strategies the dominant vision is also re-flected in recruiting practices. These focus on intrin-sically motivated people who share the approach and are ‘fanatics’ by themselves. While other ambi-dextrous designs may fail in this constellation, here the dominance of a top innovator and explorer pro-vides the glue that welds everything together. This is in stark contrast to contextual ambidexterity (Deloitte Consulting), which integrates an openness of mind into value systems and promotes flexibility in deriving at customer-centric work with high satis-faction rates. While value systems are inspiring, the continuity of the business requires that strategy is flexibly adapted (even though this is unlikely to oc-cur in the form of dramatic jumps). In turn, it re-quires managerial mindsets that are equally open to conflicting options of exploration and exploitation, especially when acting in an international context. Recent insights in this typical international market-ing dilemma have been discussed with reference to the notion of strategic ambidexterity. This captures the trade-off between a multiple pro-growth vs. a focused pro-profit strategy (Han 2007) where the former favours multinational marketing and the lat-ter standardized market operations. In comparing contextual and punctuated ambidex-terity from a knowledge development and recruiting


231

perspective, several insights emerge. Both can be explained with reference to the dynamic capability view (Teece, Pisano, and Shuen 1997) which has re-cently been extended to include marketing dynamic capabilities (MDCs), i.e. “the responsiveness and efficiency of cross-functional business processes for creating and delivering customer value in response to market changes” (Fang and Zou 2009: 3). MDCs are required because the ability to create and deliver superior customer value through efficient and fast-responding marketing processes has become one of the major sources of competitive advantage to firms (Day 1994). It is the customer focus which makes MDCs particularly relevant for this research as it distinguishes MDCs from other dynamic capabili-ties. In the case of Deloitte, MDCs are specifically related to customer relationship building via trust and capability development which focuses on multi-context and multi-client adaptability. This is in line with what Bartlett and Ghoshal’s (1989) focus on integrating recruitment and selection, training and career path management as ways of stimulating a company to become globally integrated and locally responsive at the same time. In contrast, Celltech’s capability development is rather product-based (drug development) and process-based (learning to collaborate) where expert competences need to be translated into core technologies and cutting edge offerings. The differences between variations of ambidexterity is also reflected in the recruiting approaches, which focus either on generalist and client-compatible atti-tudes (contextual ambidexterity) or in-depth spe-cialist knowledge (punctuated ambidexterity) or on both to be allocated to different units (structural ambidexterity). A specialist focus is required when firms transgress through certain stages, and special-ist knowledge provides psychological safety (Ne-manich and Vera 2009). However, both the Celltech case and the literature suggest that strategy imple-mentation is improved through flexibly inserting new and disconnected knowledge into the firm which challenge the existing resource base (Guettel and Konlechner 2009). This is especially important when firms experience core technology changes which require the speedy acquisition of new knowl-edge without rendering previous expert knowledge superfluous. Here, complementing specialists’ knowledge by ongoing learning routines which ad-just the corporate knowledge base has proved use-ful. In comparison, peripatric ambidexterity is likely

to rely on specialist knowledge to promote and im-plement one type of strategy without engaging em-ployees into too many cognitive paradoxes (Smith and Tushman 2005). This is not to say that learning does not play a role in a peripatric orientation, but companies need the brightest technological people and often leave soft-skills development to individual initiatives. In summary, both within and across-case discus-sions have highlighted different implementation in-struments which can be applied to solve marketing dilemmas. These insights may be used for formulat-ing further challenges for research.

6 Implications for Further Re-search and Marketing Practice

Implications for further research can be derived from both the limitations and the contributions of this study. The limitations of this research can be divided into three categories. First, in exploratory research small numbers of cases are acceptable, but investigating additional and more varied cases is likely to yield promising insights. Second, with the choice of Apple and Celltech, we presented two ex-amples focused on (the move towards) explorative strategy while insights into the opposite direction would also be useful. Third, in investigating imple-mentation categories, we relied on a mix of existing categories from the marketing strategy and ambi-dexterity literature. However, both streams of re-search have not yet reached a stage of consolidated insights which is evidenced by the heterogeneity of approaches and elements (Li, Guohui, and Eppler 2008). One of the immediate consequences in this research relates to initially low inter-rater reliability percentages due to overlapping categories. Thus, further research may start from redefining catego-ries based on empirical insights. Our results demonstrate that knowledge of ambi-dextrous designs, associated tools and procedures is a prominent requirement when companies want to solve their marketing dilemmas. Each of the four ambidextrous designs facilitates the solution of dif-ferent marketing dilemmas and requires a different, partly overlapping, set of implementation factors. These relate both to the personal and the organiza-tional level and warrant further attention (Table 2). Further, studies should not only investigate the im-plementation of one of the four types of ambidexter-ity in isolation but focus on their dynamic interplay both within organizations and over time.


232

Table 2: Summary and Research Implications

Marketing Driver of Ambi-dexterity

Implementation Personal Level

Implementation Organizational Level

Structural ambidexterity New business approach out-side stagnating core busi-ness

Conflicting values of mar-keting message (e.g. excite-ment vs. tradition)

Integration through infor-mal knowledge flows

Recruiting of contextually ambidextrous people

Transversal careers across units

Top management provides vision and goals but grants autonomy in procedures

Peripatric ambidexterity Personality-driven product and marketing approach

Desired spill-over effects (person – organization - brand)

Recruiting focus explora-tion-oriented people

Intrinsic motivation drivers

Clear guidelines on work procedures

Limited cross-functional interfaces (in critical areas)

Contextual ambidexterity Market approach

Marketing dynamic capa-bilities

High motivated to learn and quickly adjust to new con-tingencies

Trust as essential personal and interpersonal compe-tence

Career development with leeway for technical and managerial careers

Managerial support throughout business proc-esses, task assignments, ca-reer perspective

Punctuated ambidexterity

Core technology changes (product substitutions)

Core competence-based marketing

Expert knowledge

Willingness to engage in learning and knowledge sharing in addition to ex-pert status

Effective linkages between teams through shared cul-ture

Multi-functional project or-ganization

In addition, attention should also be paid to situa-tions when firms adopt temporal models of ambi-dexterity as they are required to optimize the fit be-tween internal design parameters and constantly changing external contingencies. Research may be further inspired by configuration theories (Miller and Friesen 1984) which discuss the interplay be-tween these two types of parameters. It would also be particularly important to analyze whether the use of different types of ambidexterity is conducive to explaining underlying differences in terms of per-formance. For marketing practitioners, an understanding of the four types of ambidexterity and the associated implementation tools provides a basis for finding systematic solutions to several common marketing dilemmas. This could imply the tension between marketing established or conventional products and emerging new products which may be resolved by structural ambidexterity. Conflicts between a zeal-ous business development and a sluggish R&D may point towards the need for a better synchronisation through punctuated ambidexterity. In contrast, if senior management finds its company too research driven but lacking in commercialisation, a new mar-keting director may offer a solution, i.e., peripatric

ambidexterity might be called for. Finally, senior marketing managers will have to give their brand managers sufficient latitude to strike the fine bal-ance between the introduction of new products and the extension of existing products. Thus, creating the right environment for contextual ambidexterity would be the order of the day. However, as this study has illustrated, rather than focusing on isolated marketing dilemmas, managers will most likely require a more integrated approach. Thus, meeting the multi-layered challenges of suc-cessful marketing, they are likely to implement dif-ferent types of ambidexterity in the firm. Fulfilling singular requirements may be relatively easy and previous research has extensively dealt with the du-alism of exploration and exploitation strategies, but as Raisch (2008: 484) urges us “further insights are needed about the specific contexts [of ambidextrous designs], which factors contribute to their successful implementation, and what outcomes are likely to result from their use.” This implies that understand-ing the compatibility of implementation instru-ments becomes a vital part in pursuing balanced marketing strategies.


233

7 Conclusions This study is among the first that adopts a market-ing lens to investigating ambidexterity. We contrib-ute to both previous research and managerial prac-tice as to how firms in general - and marketing management in particular - may manage and organ-ize the dual pursuit of exploratory and exploitative strategies. Even though a few existing studies have investigated selected types of ambidexterity, we are currently unaware of any research that includes the notion of peripatric ambidexterity and further dis-cusses the interplay of four different balancing mechanisms. Our findings illustrate that firms use these mechanisms by drawing on a variety of im-plementation elements. In changing or extending their marketing orientation, managers need to be well aware of the complementary and potentially conflicting use of implementation designs.

Appendix - Details of Methodological Approach Each of the four firms, Celltech, Apple, Deloitte, and Deutsche Bank, was approached to provide addi-tional data on the specific ambidexterity type it scored most prominently in the first round, i.e., Deutsche Bank for structural ambidexterity, Deloitte for contextual ambidexterity, Apple for pe-ripatric ambidexterity, and Celltech for punctuated ambidexterity. However, during interviews, new evidence for different types of ambidexterity was found, which gave rise to further analysis. When (re)contacting these firms and respective in-terview partners, we encountered a differential will-ingness to answer further detailed questions. While interviews with Deutsche Bank and Deloitte were continued with semi-structured questionnaires, both Celltech and Apple required a change in the interview approach. Celltech is a company, which was acquired by UCB in 2004 and no longer exists in its original form. In the first round, interviews in-volved broader aspects of innovation and strategy; but the focus of this second round-investigation re-lates to very specific developments in the past, and suitable interview partners that experienced Cell-tech’s strategy changes were difficult to find. How-ever, we decided to maintain the firm as it is one of the typical classic cases of punctuated ambidexterity (equilibrium), which is well documented in the case literature (e.g. McNamara and Baden-Fuller 1999; McNamara, Baden-Fuller, and Howell 1999). Con-

sequently, our analysis of Celltech is literature-based and complemented by a few selected inter-view insights. In the case of Apple we encountered further challenges, which we did not fully realize in the first round. Due to a corporate non-disclosure policy it was difficult to obtain more fine-tuned in-formation. We, therefore, included external partners and former employees. In addition, we used projec-tive techniques in the interview process. Projective or third person techniques are used when the re-quired information cannot be accurately obtained by direct methods (or directly concerned interview partners). They are also highly recommended in ex-ploratory research to provide significant insights (Webb 1992). As qualitative analysis and interpreta-tion of projective techniques are no different from the procedures for qualitative research in general, we continue the analysis phase in the same way as for direct interviewing. In sum, in the second round we conducted 20 inter-views, by applying three complementary methods: traditional interviews based on semi-structured questionnaires, literature-based case analysis, and interviews based on projective techniques. All inter-views lasted, on average 90 minutes. When contact-ing interview partners, all participants were in-formed that the study was primarily about balanc-ing difficult marketing-related tasks and conflicts, and how to communicate and market these chal-lenges both internally and externally. The term “ambidexterity” was not mentioned initially as pre-vious experience indicates that the majority of man-agers has not yet encountered the notion of ambi-dexterity even though most managers use what the literature has identified as ‘ambidextrous coordina-tion designs’. This discrepancy is a pertinent issue as managers often use different words than research-ers and a shared meaning is required before advanc-ing with the issue (Astley and Zammuto 1992). In addition to interview data, available material both from the public press, company reports and existing previous research was included for data tri-angulation (Jick 1979). Resulting insights illustrate our theoretical perspective and describe the chal-lenges associated with implementing each type of ambidexterity. In order to prepare data for analysis, interviews were audio-taped, transcribed and, when necessary, translated into English. Full transcripts of interviews were prepared and, subsequently, abridged to capture the data directly relevant to our research topic. We assigned code numbers to each


234

interview per firm (A1, A2, etc. for Apple, B1, B2 etc. for Deutsche Bank, C1, C2, etc. for Celltech, and D1, D2, etc. for Deloitte), which later facilitated the identification of interview quotes and multiple men-tions. In the categorization process, nine major categories emerged. Assigning categories was based on a deductive-inductive procedure which combines the advantages of reliability (with criteria derived from theory) and those of validity (which suggest an inductive development that captures the essence of the phenomenon) (Druckman and Hopmann 2002). We started with categories from the imple-mentation literature and complemented them with

those mentioned for ambidextrous strategies. We decided to define main categories (9) and main-tained sub-categories (24) whenever this seemed useful to support subsequent coding (Srnka and Koeszegi 2007). In the coding process, we assigned interview statements to categories. Statements or “sense units” (Bubert, Gadner, and Richards 2004) have been identified as the best basis for coding and analysis. We independently categorized the 127 statements, which we identified in the interviews, into the nine categories illustrated in the strategy and ambidexterity implementation literature (see Table A.1 for an example of the coding process).

Table A.1: Coding Example

Definition – Main Category Coding: Firm/Interview No.-Category-Subcategory: Sense unit with key words in italics

Goals: current and future objectives (financial and strategic) of the firm, both corporate and business unit level; vision of the firm, vision of CEO, firm development

A3-G(oals)-St(rategic Vision): Sense unit: “[…] and when you are asked to judge on the future; what you can always say is that there will be something new, there is constant innovation,… the firm is constantly re-inventing itself and this is how it defines its strategy.”

Organizational Context: formal guidelines, controls, monitor-ing, daily work context that affects motivation, organization of work environment

B3-O(rganizational)Context-Au(tonomy): Sense unit: “You know, there are constraining and supporting factors but we work here be-cause we like the context; we organize our work largely by ourselves, .. Sense unit: we kind of create our own environment so that eve-rybody can work according to the best of his abilities and for the maximum benefit of Q110…”

As one of the criteria for “good” science is based inter-subjectivity through multiple person in-volvement, we independently engaged two people in the process of allocating statements to the nine major categories identified in the literature. The researchers initially achieved an agreement of 65% which is not very high. Major differences oc-curred in the categories leadership style vs. com-munications and behavior complexity vs. recrui-

ting. Referring back to the strategy implementa-tion literature did, indeed, provide evidence that these categories have often been diluted (Li, Guo-hui, and Eppler 2008). After in-depth discussion, we increased agreement and were able to allocate all statements to nine categories. The resulting categories and subcategories are summarized in Table A.2 which also depicts the number of men-tions per category and some illustrative quotes.

Table A.2: Implementation Categories, Subcategories, and Interview Examples

Implementation Categories (num-ber of mentions)

Litera-ture Re-ferences

Subcategories Illustrations

Goals (11x)

Noble (1999)

Corporate goals

Unit goals Strategic

vision

A4: ”There is always the run after the next hype, you feel the drive and you know what, when Steve was away, it was different“ B2: “The concept [Q110] was established with a long-term perspective in mind, but of course it is subject to change. This is pretty clear when we want to fill the notion ‘bank of the future’ with life…what is important is new client generation and we have roughly 50% higher rates than other subsidiaries; our client focus in clearly king“ Clit: ”It came from the top. Of course, there were managers further down but the way we were going to organize ourselves came from David Bloxham (Director of Research)”; (McNamara et al. 1999: 7)


235

Table A.2 continued: Implementation Categories, Subcategories, and Interview Examples




D2: “We always have very clear and challenging objective and we keep them high… but , actually our strategy is not changing dramatically but the envi-ronment is and we need to update our knowledge”

Structures (12x)

Govinda-rajan (1988) Drazin and Howard (1984)

Organization structures

Team struc-ture

Integration

A1: “We have several clearly defined structural devices, e.g. international sales and all the financial functions are coordinated in Cork… but even here, there are regional structures, even though with few hierarchies… and very quickly you are in the US” B8: “We have two teams which deal with customers in a different way. First, there is the Forum Team, which is the first contact for everybody. These people are moving freely in the area. Then, we have the Client Team, which is responsible for fixed meetings, and for coordination subsequent meet-ings.” C2: “It was very fluid…you were in contact with almost everybody…in the old times, you went to work and you had your three colleagues, that was it..” D1: “Of course we have clear structures, but it depends on the level, and it is more that you facilitate exchange than imposing constraints.”

Leadership Style (14x)

Gupta and Govinda-rajan (1984)

Support Motivation Empathy

A4: “He clearly is the great motivator…when you listen to his public speeches, and even more, when you meet him internally, the vision is alive, it lives, and he is by no means a normal boss” A4: “It is crazy how one person drives this firm…you are always expecting him to introduce the next hype and you discuss it in your team with your colleagues and you also think about new ideas” B7: “You know, we all work here because we wanted to; if we work long hours, this does not really matter because we are motivated, we believe we are doing an exciting job here, and what’s most important…we like it” Clit: “New management entered the firm but was cautious at first, galvaniz-ing the commitment of a key group of scientists and administrators prior to announce a change in strategy” (McNamara and Baden-Fuller, 1999: 304) D4: “I always try to see the potential of people, I listen intensively, challenge them… help them, and discuss options with them”

Communications (21x)

Westley (1990) Rapert, Velliquet-te, and Garretson (2002)

Informal communica-tion

Formal communica-tion

A2: “You get to know what is important for your work but sometimes I would like to be more involved. Also, you hear different things from different people as if they do not communicate with each other” B8: „It is important that you do not only focus on your own things but also show responsibility towards each other and act to support the customer best. This requires that you, kind of anticipate what others will do but also that you constantly communicate“ C1:”There were so many diverse opinions, people looked at the world differ-ently…we were constantly in the process of negotiation, you know this was everywhere, we did not rely on any official communications channels…but of course, there was also information coming from the boss” D4: ”Much is based on trust. I try to communicate clearly and build a rela-tionship. It is like with external clients. Once they work with you they know they can trust us and we help them. It is sometimes that these relationships are more important than the knowledge because others also have this knowledge”

Incentives (9x)

Walker, Churchill, and Ford (1977)

Financial Immaterial

A1: “Formal financial incentives are very rare…this was different earlier, but today, you have to launch a big bang. But you work in this company and quite often you feel this is an incentive…” B3: ”There is no differences in financial rewards, this is pretty much the same across all subsidiaries.. B3: “…many people, after they left Q110 have continued their management career somewhere else within Deutsche Bank. Having worked for Q110 is apparently seen as a preparation for major career jumps”


236





C2: ”We had very formal appraisal systems and there was an additional in-centive system with quarterly bonus awards for exceptional contributions.” D2: “You can imagine that incentives are attractive and we expect a lot. In the long run, it will be your overall performance that brings you to partner level and, apart from everything else, this is well rewarded”

Culture (16x)

Nemanich and Vera (2009)

Team spirit Openness

A5: “You find many people here late in the night who discuss and experi-ment with colleagues…it is there that you develop the best ideas” B1: ”We have a strong team spirit and we help each other even if this is sometimes difficult … especially if you look at the large area here. But it is clear that we try to permanently be in touch, this also relates to communica-tions between teams and between bankers and shop employees and – you know, even though the latter are not bankers, they are employed by DB, we have no typical shop in the shop principle here” Clit: “We created a culture of flexibility. For instance, one principle is the sanctioning of scientists spending up to 10 percent of their time on individ-ual projects…We want them the use their knowledge in a direction uncon-strained by project requirements” (McNamara, Baden-Fuller, and Howell 1999: 12) D3: “We have a very open culture, where we are critical and challenge each other. We are interested in what people and their potential to contribute, and this is what we permanently encourage”

Organizational Context (11x)

Bartlett and Gho-shal (1994); Jaworski and McIn-nis (1989)

Autonomy Guidelines /

Control Managerial

Support

A1: “It’s simply that you can do what you like most. People hang around late hours and discuss the newest gadgets, others go early or work from home… you can really influence your own work environment” A1: “There are very clear and strict guidelines and you normally check with your boss and make sure he agrees” B2: “Of course, there were guidelines but they used to be much stronger ... Today, there are many things we can largely decide on our own and quite often only briefly check back with the HQ. In fact, is has happened that ideas developed from the HQ did not work at all in our subsidiary” B4: “People who work here have a good feeling for what works and what it is that customers like” Clit: ”You manage the projects by objectives and milestones.. You have reg-ular quarterly reviews, after all, we are a small company so you can monitor things reasonably closely”. (McNamara, Baden-Fuller, and Howell 1999: 10). D3: “We have, and need to have financial objectives, which means you should really spend your time on billable hours with the client. But the natu-ral context allows for time to do research, and you are not far away from your clients when you develop something because we listen to clients and involve them in research”

Behavioral Com-plexity (17x)

Sproull and Hof-meister (1986)

Role Behav-ior

Flexibility Commitment

A5: “While people seem to have huge leeway in what they do, I have the feeling that there are serious constraints in how flexibly they can really act” B3” We need to be flexible here, we incorporate different roles, from event manager to conservative banker. We need to adjust to our clients because client satisfaction is key” B5 “every employee is not only banker but always something like a customer relationship manager, a PR specialist, in short someone who thinks out of the box. You need to sell not only banking service but also kitchen accesso-ries and concert tickets and you need to live this!!” C1: ”It was quite amazing. Formerly, you were a specialist in your field.. you still continued to be one but the new task was to move out of the box and increase interdepartmental knowledge sharing” D1: “People travel from London to Spain to Portugal, all within one week. They communicate fluently in these languages and shift between cultures. Of course, you also have this in client relations. For some consultants, it is cer-tainly less challenging to spend more time on office work”


237





Recruiting (16x)

Pearce and Robinson (2005)

Personality Education Learning

Skills

A4: “It’s not a normal firm, they are all fanatics. Even our bankers are not normal bankers. You have to be a special type of person here and you need to like it. Perhaps, this is already a self-selection criterion in the recruitment process…” B4: “Criteria for recruiting are fairly simple: openness, a basic interest in the job, you use of body language. Probably, there is a self-selection process, people come here because they want to work here, later on they receive trainings and coaching.” Clit:”…simultaneously, 35 medical chemists were brought into the firm. This new blood not only provided key skills needed to implement the new strategy, but also a group of people who could stimulate and challenge” (Dodgson, 2001: 144) D4: “Individual objectives and career perspectives can change and then we discuss new options.. and there often is an option to change careers in the firm, like from managerial aspirations to technical expertise…so we are basi-cally looking for open-minded people who can communicate well and who also have the personality to stand in front of the client”

References

Adner, Ron and Daniel A. Levinthal (2002): The Emergence of Emerging Technology, California Management Review, 45 (1): 50-66.

Adler, Paul, Barbara Godolftas, and David Levine (1999): Flexibility versus Efficiency? A Case Study of Model Change-overs in the Toyota Production System, Strategic Management Journal, 24: 1011-1025.

Akşin, O. Zeynep and Patrick T. Harker (1999): To Sell or Not to Sell: Determining the Trade-offs Between Service and Sales in Retail Banking Phone Centers, Journal of Service Research, 2: 19-33.

Ambos, Bjoern and Bodo B. Schlegelmilch (2005): In Search of Global Advantage, European Business Forum, 21: 23-24.

Astley, W. Graham and Raymond F. Zammuto (1992): Organi-zation Science, Managers, and Language Games, Organization Science, 3: 443-460.

Auh, Seigyoung and Bulent Menguc (2005): Balancing Ex-ploration and Exploitation: The Moderating Role of Com-petitive Intensity, Journal of Business Research, 58: 1652-1661.

Bass, Bernard M. (1985): Leadership and Performance Beyond Expectation, Free Press: New York.

Beckman, Christine M. (2006): The Influence of Founding Team Company Affiliations on Firm Behavior, Academy of Man-agement Journal, 49: 741-758.

Birkinshaw, Julian and Cristina Gibson (2004): Building Am-bidexterity into an Organization, Sloan Management Review, 45 (4): 47-55.

Benner, Mary J. and Michael L. Tushman (2003): Exploita-tion, Exploration, and Process Management: The Produc-tivity Dilemma Revisited, Academy of Management Review, 28: 238-256.

Boeker, Warren (1997): Executive Migration and Strategic Change: The Effect of Top Manager Movement on Product-Market Entry, Administrative Science Quarterly, 42: 213-236.

Bonoma, Thomas V. (1984): Making Your Marketing Strat-egy Work, Harvard Business Review, 62 (2): 69-76.

Bonoma, Thomas V. and Victoria L. Crittendon (1988): Managing Marketing Implementation, Sloan Management Review, 29 (4): 7-14.

Brown, John S. and Paul Dugoid (1991): Organizational Learning and Communities of Practice: Towards a Unified View of Working, Learning, and Innovation, Organization Science, 2: 40-57.

Brown, Shona L. and Kathleen M. Eisenhardt, (1997): The Art of Continuous Change: Linking Complexity Theory and Time-Paced Evolution in Relentlessly Shifting Organizations, Admin-istrative Science Quarterly, 42: 1-34.

Bubert, Renate, Johannes Gadner, and Lyn Richards (2004): Applying Qualitative Methods to Marketing Management Re-search, Palgrave McMillan: Houndsmill et al.

Burgelman, Richard A. (2002): Strategy as a Vector and the Inertia of Co-evolutionary Lock-in, Administrative Science Quar-terly: 47: 325-358.

Burns, James (1978): Leadership, Harper &Row: New York.

Burns, Tom and George M. Stalker (1961): The Management of Innovation, Tavistock: London.

Carlile, Paul R. (2004): Transferring, Translating and Trans-forming: An Integrated Framework for Managing Knowledge Across Boundaries, Organization Science, 15: 555-568.

Carmeli, Abraham and Meyrav Y. Halevi (2009): How Top Management Teams Behavioral Integration and Behavioral Complexity enable Organizational Ambidexterity. The Moder-ating Role of Contextual Ambidexteritxy, The Leadership Quar-terly, 20: 207-218.

Carson, Stephen (2007): When to Give Up Control of Out-sourced New Product Development, Journal of Marketing, 71 (1): 49-66.

Chimhanzi, Jaqueline and Robert E. Morgan (2005): Explana-tions from the Marketing/Human Resource Dyad for Market-


238

ing Strategy Implementation Effectiveness in Service Firms, Journal of Business Research, 58: 87-796.

Cespedes, Frank (1991): Organizing and Implementing the Market-ing Effort, Addison-Wesley: Reading, Mass.

Christensen, Clayton M. and Michael Raynor (2003): The Inno-vator's Solution, Harvard Business School Press: Boston.

Churchill, Gilbert A. and Dawn Iacobcci (2005): Marketing Research: Methodological Foundations, 9th Edition, Thompson South-Western: Mason, Ohio.

Danneels, Erwin (2002): The Dynamics of Product Innova-tion and Firm Competencies, Strategic Management Journal, 23: 1095-116.

Day, George S. (1994): The Capabilities of Market-Driven Organizations, Journal of Marketing, 58 (2): 37-52.

Day, George S. (1989): Deciding How to Compete, Planning Review, 17 (5): 18-23.

Deming, W.E. (1981): Management of Statistical Techniques for Quality and Productivity, New York University Graduate School of Business: New York.

Dobni, C. Brooke (2006): Developing an Innovation Orienta-tion in Financial Services Organizations, Journal of Financial Services Marketing, 11: 166-179.

Dodgson, Mark (1991): Technology Learning, Technology Strategy and Competitive Pressures, British Journal of Man-agement, 2: 133-149.

Drazin, Robert and Peter Howard (1984): Strategy Implemen-tation: A Technique for Organizational Design, Columbia Jour-nal of World Business, 19: 40-46.

Druckman, Daniel and Terrance Hopmann (2002): Content Analysis, in: Kremenyuk, V.A. (ed.), International Negotiation: Analysis, Approaches, Issues, Jossey-Bass: San Francisco, 288-314.

Duncan, Robert B. (1976): The Ambidextrous Organization: Designing Dual Structures for Innovation, in: Pondy, Lois R. and Dennis P. Slevin (eds.): The Management of Organization Design: Volume, Strategies and Implementation, Elsevier: New York, 167-188.

Evans, Kenneth R., Todd J. Arnold, and John A. Grant (1999), Combining Service and Sales at the Point of Customer Contact: A Retail Banking Example, Journal of Service Research, 2 (1): 34-49.

Fang, Eric and Shaoming Zou (2009): Antecedents and Conse-quences of Marketing Dynamic Capabilities in International Joint Ventures, Journal of International Business Studies, 40: 742-761.

Fiedler, Fred (1967): A Theory of Leadership Effectiveness, McGraw-Hill: New York.

Floyd, Steven and Peter W. Lane (2000): Strategizing through-out the Organization: Managing Role Conflict in Strategic Re-newal, Academy of Management Review, 25: 154-177.

Forman, Janis and Paul A. Argenti (2005), How Corporate Communication Influences Strategy Implementation, Reputa-tion and the Corporate Brand: An Exploratory Qualitative Study, Corporate Reputation Review, 8: 245-264.

Franckwick, Gary, James C. Ward, Michael D. Hutt, and Peter Reingen (1994): Evolving Patterns of Organizational Belief in

the Formation of Strategy, Journal of Marketing, 58 (2): 96-110.

Gale, Bradley (1992): Monitoring Customer Satisfaction and Market-perceived Quality, American Marketing Association Worth Repeating Series, Number 022CSO1.

Gersick, Connie (1991): Revolutionary Change Theories: A Multilevel Exploration of the Punctuated Equilibrium Paradigm, Academy of Management Review, 16: 10-36.

Ghoshal Sumantra and Christopher A. Bartlett (1994): Linking Organizational Context and Managerial Action: The Dimen-sions of Quality of Management, Strategic Management Journal, 15: 91-112.

Gibson, Cristina and Julian Birkinshaw (2004): The Antece-dents, Consequences, and Mediating Role of Organizational Ambidexterity, Academy of Management Journal, 47: 209-226.

Govindarajan, Vijay. (1988): A Contingency Approach to Strat-egy Implementation at the Business Unit Level: Integrating Administrative Mechanisms with Strategy, Academy of Man-agement Journal, 27: 25-41.

Greve, Heinrich R. (2007): Exploration and Exploitation in Product Innovation, Industrial and Corporate Change, 16: 945-975.

Griffin, Abbie and John H. Hauser (1996): Integrating R&D and Marketing: A Review and Analysis of the Literature, Jour-nal of Product Innovation Management, 13: 191-215.

Güttel, Wolfgang H. and Stefan W. Konlechner (2009): Con-tinuously Hanging by a Thread: Managing Contextually Ambi-dextrous Organizations, Schmalenbach Business Review, 61: 149-171.

Gupta, Anil K., Ken G. Smith, and Christina E. Shalley (2006): The Interplay Between Exploration and Exploitation, Academy of Management Journal, 49: 693-706.

Han, Mary (2007): Achieving Superior Internationalization Through Strategic Ambidexterity, Journal of Enterprising Culture, 15: 43-77.

Hart, Stuart and Catherine Banbury (1994): How Strategy-Making Processes Can Make a Difference, Strategic Management Journal, 15: 251-268.

He Zi-Lin and Poh-Kam Wong (2004): Exploration versus Ex-ploitation: An Empirical Test of the Ambidexterity Hypothesis, Organization Science, 15: 481-494.

Jansen, Justin J. P., Frans A. J. Van den Bosch, and Henk W. Volberda (2005): Exploratory Innovation, Exploitative Innovation, and Ambidexterity: The Impact of Environ-mental and Organizational Antecedents, Schmalenbach Busi-ness Review, 57: 351-363.

Jansen, Justin J. P., Michiel P. Tempelaar, Frans A. J. Van den Bosch, and Henk W. Volberda (2009): Structural Dif-ferentiation and Ambidexterity: The Mediating Role of In-tegration Mechanisms, Organization Science, 20: 797-811.

Jaworski, Bernard and Deborah J. McInnis (1989): Marketing Jobs and Management Control: Towards a Framework, Journal of Marketing Research, 26: 406-19.

Jaworski, Bernard., Ajay K. Kohli, and Arvind Sahay (2000): Market-driven versus driving-markets, Journal of the Academy of Marketing Science, 28: 45-54.


239

Jick, Todd D. (1979): Mixing Qualitative and Quantitative Methods, Triangulation in Action, Administrative Science Quar-terly, 24: 602-611.

Juran Joseph M. and Frank M. Gryna (1988): Juran’s Quality Control Handbook, McGraw Hill, Inc.: New York.

Katila Riita and Gautam Ahuja (2000): Something Old, Some-thing New: A Longitudinal Study of Search Behavior and New Product Introductions, Academy of Management Journal, 45: 1183-1194.

Klein, Katherine J., Fred Dansereau, and Rosalie I. Hall (1994): Levels Issues in Theory Development, Data Collection, and Analysis, Academy of Management Review, 19: 195-229.

Kraatz Matthew S. and James H. Moore (2002): Executive Mi-gration and Institutional Change, Academy of Management Jour-nal, 45: 120-143.

Kumar, Nirmalya (1997): The Revolution in Retailing: from Market Driven to Market Driving, Long Range Planning, 30: 830-836.

Kyriakopoulos, Kyriakos and Christine Moorman, C. (2004): Tradeoffs in Marketing Exploitation and Exploration Strate-gies: The Overlooked Role of Market Orientation, International Journal of Research in Marketing, 21: 219-240.

Lavie Dovev and Lori Rosenkopf (2006): Balancing Explora-tion and Exploitation in Alliance Formation, Academy of Man-agement Journal, 49: 791-818.

Lee, Jeho and Young U. Ryu (2002): Exploration, Exploitation, and Adaptive Rationality: The Neo-Schumperian Perspective, Simulation Modeling Practice and Theory, 10 (5-7): 297-320.

Leonard-Barton, Dorothy (1992): Core Capabilities and Core Rigidities: A Paradox in Managing New Product Development, Strategic Management Journal, 13 (S1): 111-125.

Lewis, Marianne W. (2000): Exploring Paradox: Toward a More Comprehensive Guide, Academy of Management Review, 25: 760-776.

Levinthal Daniel A. and James G. March (1993): The Myopia of Learning, Strategic Management Journal, 14: 95-112.

Li , Yang, Sun Guohui, and Martin J. Eppler (2008): Making Strategy Work: A Literature Review on the Factors Influencing Strat-egy Implementation, ICA Lugano, Working Paper 2/2008: 1-46.

Lubatkin, Michael H., Zeki Simek, Yan Lin, and John F. Veiga (2006): Ambidexterity and Performance in Small-to Medium- Sized Firms: The Pivotal Role of Top Manage-ment Team Behavioral Integration, Journal of Management, 32: 646-672.

March, James G. (1991): Exploration and Exploitation in Organizational Learning, Organization Science, 2: 71-87.

Mayr, Ernst (1942): Systematics and the Origin of Species, Co-lumbia University Press: New York.

McDonough Edward F. and Richard Leifer (1983): Using Simultaneous Structures to Cope with Uncertainty, Academy of Management Journal, 26: 727-735.

McEwen, William (2008): The Chief Marketing Officer’s Dilemma, Gallup Management Journal, March 13, 2008, http://gmj.gallup.com/content/104704/Chief-Marketing-Officers-Dilemma.aspx (Access Date: 5 April, 2009)

McGrath, Rita G. (2001): Exploratory Learning, Innovative Capacity and Managerial Oversight, Academy of Manage-ment Journal, 44: 118-131.

McNamara, Peter and Charles Baden-Fuller, C. (1999): Les-sons from the Celltech Case: Balancing Knowledge Explora-tion and Exploitation in Organizational Renewal, British Journal of Management, 10: 291-307.

McNamara, Peter, Charles Baden-Fuller, and James Howell (1999): The Rejuvenation of Celltech in the1990s: Laying the Foundations of a FTSE 100 Company, ECCH Case Study No. 300-072-1.

Menguc, Bulent and Seigyoung Auh (2008): The Asymmet-ric Moderating Role of Market Orientation on the Ambidex-terity – Firm Performance Relationship for Prospectors and Defenders, Industrial Marketing Management, 37: 455-470.

Miles, Matthew, B. and A. Michael Huberman (1984), Quali-tative Data Analysis, 2nd ed., Sage Publications: London.

Miller, Danny and Peter H. Friesen (1984): Organizations: A Quantum View, Prentice-Hall, Englewood Cliffs, NJ.

Miller, Danny and Peter H. Friesen (1986): Porter’s Generic Strategies and Performance, Organization Studies, 7: 37-56.

Mom, Tom J.M. (2007): Managers’ Exploration and Exploitation Activities: The Influence of Organizational Factors and Knowledge Inflows, Dissertation, Erasmus University Rotterdam.

Morris, Betsy (2008): What Makes Apple Golden, Fortune, March 17, 2008, 157 (5): 68-71.

Nemanich, Loise A. and Vera Duysa (2009): Transforma-tional Leadership and Ambidexterity in the Context of an Acquisition, The Leadership Quarterly, 20: 19-33.

Noble, Charles H. (1999): Building the Strategy Implemen-tation Network, Business Horizons, 42 (6): 19-28.

Noble, Charles H. and Michael P. Mokwa (1999): Imple-menting Marketing Strategies: Developing and Testing a Managerial Strategy, Journal of Marketing, 63 (4): 57-73.

O’Reilly Charles A. and Michael L. Tushman (2004): The Ambidextrous Organization, Harvard Business Review, 82 (4): 74-81.

O’Reilly Charles A. and Michael L. Tushman (2007): Ambi-dexterity as a Dynamic Capability: Resolving the Innova-tor’s Dilemma, Research Paper No. 1963, Stanford Gradu-ate School of Business.

Osborn, Richard N., James G. Hunt, and Lawrence R. Jauch (2002): Toward a contextual theory of leadership, The Leader-ship Quarterly, 13: 707-837.

Peirce, John and Richard Robinson (2005): Formulation, Im-plementation, and Control of Competitive Strategy, New York: McGraw Hill:

Porter, Michael E. (1980): Competitive Strategy, The Free Press: New York.

Porter, Michal E. (1996): What is Strategy?, Harvard Business Review, 74 (6): 61-78.

Puranam, Pranish, Harbir Singh, and Maurizio Zollo (2006): Organizing for Innovation: Managing the Coordi-nation-Autonomy Dilemma in Technology Acquisitions, Academy of Management Journal, 49: 263-280.


240

Quelch, John A. (2005): Ambidextrous Marketing, The Wall Street Journal, October 11, B2.

Raisch, Sebastian (2008): Balanced Structures: Designing Or-ganizations for Profitable Growth, Long Range Planning, 41: 483-508.

Raisch, Sebastian and Julian Birkinshaw (2008): Organiza-tional Ambidexterity: Antecedents, Outcomes, and Moderators, Journal of Management, 34: 375-409.

Rapert, Molly I., Anne Velliquette, and Judith A. Garretson (2002): The Strategic Implementation Process Evoking Strate-gic Consensus through Communication, Journal of Business Re-search, 55: 301-310.

Roethaermel Frank T. and David L. Deeds (2004): Exploration and Exploitation Alliances in Biotechnology: A System of New Product Development, Strategic Management Journal, 25: 201-221.

Shashittal, Hemant C. and David Whilemon (1996): Marketing Implementation in Small and Midsized Industrial Firms, Indus-trial Marketing Management, 25: 67-78.

Schneider, Benjamin and David E. Bowen (1995): Winning the Service Game, Harvard University Press: Boston.

Sculley John and David Byrne (1987): Odyssey: Pepsi to Apple, A Journey of Adventure. Ideas and the Future, Harper Collins: New York, 147.

Siggelkow Nicolai and Daniel A. Levinthal (2003): Temporarily Divide to Conquer: Centralized, Decentralized, and Reinte-grated Organizational Approaches to Exploration and Adapta-tion, Organization Science, 14: 650-669.

Simek, Zeki (2009): Organizational Ambidexterity: Towards a Multilevel Understanding, Journal of Management Studies, 46: 597-624.

Sirmon, David G., Michael Hitt, and R. Duane Ireland (2007): Managing firm-resources in dynamic environments to create value: Looking inside the black box, Academy of Management Review, 32: 273-292.

Smith, Wendy and Michael L. Tushman (2005): Managing Strategic Contradictions: A Top Management Model for Man-aging Innovation Streams, Organization Science, 16: 522-536.

Souder, William E. and Alok C. Chakrabarti, (1978): The R&D/Marketing Interface: Results from an Empirical Study on Innovation Projects, IEEE Transaction on Engineering, 25 (4): 88-93.

Sproull, Lee S. and Kay R. Hofmeister (1986): Thinking about Implementation, Journal of Management, 12: 43-60.

Srnka, Katharina J. and Sabine T. Koeszegi (2007): From Words to Numbers: How to Transform Qualitative Data into Meaningful Quantitative Results, Schmalenbach Business Review, 59: 29-57.

Teece, David, Gary Pisano, and Amy Shuen (1997): Dynamic Capabilities and Strategic Management, Strategic Management Journal, 18: 509-533.

Thompson James D. (1967): Organizations in Action: Social Sci-ences Bases of Administrative Theory, McGraw Hill: New York.

Tsai, Wenpin (2002): Social structure of ‘coopetition’ within a multiunit organization: Coordination, competition, and intra-

organizational knowledge sharing, Organization Science, 13: 179-190.

Tushman, Michael L. and Elaine Romanelli (1994): Organiza-tional Transformation as Punctuated Equilibrium: An Empiri-cal Test, Academy of Management Journal, 37: 1141-1166.

Tushman Michael L. and Charles A. O'Reilly (1996): Ambidex-trous Organizations: Managing Evolutionary and Revolution-ary Change, California Management Review, 38 (4): 8-30.

Tushman Michael L. and Charles A. O'Reilly (1997): Winning Through Innovation: A Practical Guide to Leading Organizational Change and Renewal, Harvard University Press: Boston.

Virany, Beverly, Michael Tushman, and Elaine Romanelli (1992): Executive Succession and Organizational Outcomes in Turbulent Environments: An Organizational Learning Ap-proach, Organization Science, 3: 72-92.

Volberda, Henk W. (1998): Building the Flexible Firm, Oxford University Press: Oxford.

Walker, Orville C. and Robert Rueckert (1987): Marketing’s Role in the Implementation of Business Strategy: A Critical Review and Conceptual Framework, Journal of Marketing, 51 (3): 15-33.

Walker, Orville C., Gilbert A. Churchill, and Neil M. Ford (1977): Motivation and Performance in Industrial Selling: Pre-sent Knowledge and Needed Research, Journal of Marketing Re-search, 14: 156-168.

Webb, James (1992): Understanding and Designing Marketing Research, Academic Press: London.

Westley, Frances (1990): Middle Managers and Strategy: Mi-cro-dynamics and Inclusion, Strategic Management Journal, 11: 337-51.

White, J. Chris, Jeffrey S. Conant, and Raj Echambadi (2003): Marketing Strategy Development Styles, Implementation Ca-pability, and Firm Performance: Investigating the Curvilinear Impact on Multiple Strategy-Making Styles, Marketing Letters, 14: 111-124.

Woolridge, Bill and Steven W. Floyd (1989): Research Notes and Communications: Strategic Process Effects on Consensus, Strategic Management Journal, 10: 295-302.

Biographies

Christiane Prange, Dipl.-Kffr., Ph.D., is an Associate Profes-sor of International Marketing and Strategy at EM Lyon Busi-ness School in France. After several years in industry and man-agement consulting she returned to academia and obtained her Ph.D. from Geneva University. Switzerland. She held both teaching and research positions with the Open University and the University of Liverpool in the UK, the Vienna University of Economics and Business Administration in Austria as well as visiting assignments in several Asian and European countries.

Bodo B. Schlegelmilch, M.Sc., Ph.D., D.Litt., is Dean of the WU Executive Academy and Professor of International Market-ing and Management at the Vienna University of Economics and Business Administration (WU) and at Leeds University Business School. He is also a Fellow of the Chartered Institute of Marketing and Adjunct Professor of International Business Studies at the University of Minnesota, Carlson School of Man-agement and at Kingston University, London, UK.

V o l u m e 2 | I s sue 2 | December 2009 | w w w . b u s i n e s s - r e s e a r c h . o r g | I SSN 1866-8658

volume2 issue2 december 2009 final

Documents