Practice-Based Learning as a Dynamic

Capability: A Longitudinal Study of Public

Hospitals in England

Prof. Dr. Antonio Vera German Police University, Department of Organization and HR Management

Zum Roten Berge 18-24, 48165 Muenster, Germany

Email: antonio.vera@dhpol.de

Jun.-Prof. Dr. Torsten Oliver Salge Ruhr-University Bochum, Faculty of Management and Economics

Universitaetsstrasse 150, 44780 Bochum, Germany

Email: oliver.salge@rub.de

11th Public Management Research Association Conference

The Maxwell School of Citizenship and Public Affairs

Syracuse University

June 2-4, 2011

Please do not quote or circulate.

2

ABSTRACT

This paper examines the antecedents, consequences and moderators of practice-based learning

capabilities, understood as organizations’ ability to routinely engage in learning by doing,

using and interacting. As such, this mode of learning forms a natural antithesis to science-

based learning, which is fueled by novel scientific and technological insights. Conceptualized

as an incremental dynamic capability, practice-based learning is expected to be a vital driver

of gradual organizational adaptation processes. As dynamic capabilities consist of bundles of

relatively stable routines, we propose that an organization’s level of practice-based learning

capabilities will be highly persistent over time. We also argue that building and exercising

practice-based learning capabilities is resource-intensive and will as such tend to rely on the

availability of sufficient slack resources. Last, we suggest that practice-based learning will be

positively related to organizational performance, especially when the underlying business

model is labor- rather than capital-intensive. To test our theoretical ideas, we draw on unique

panel data from all public non-specialist hospitals in England and find broad support for our

hypotheses. (171 words)

Keywords: Organizational Learning; Dynamic Capabilities; Public Management; Hospitals.

3

INTRODUCTION

“[…] I see innovation in two ways. The first is very much related to biomedical research as

you can’t find a new gene for instance unless you do research - one really doesn’t exist

without the other. The second type of innovation is to do with more practical things which

come out of patient care. It is to a lesser extent related to research because it’s by doing […]

by caring for patients, by seeing the difficulties they encounter that you come up with new

ideas to help them […] So, I really see a split there.”

(Interview with the R&D director of an Acute Trust in the English National Health Service,

August 2007)

This statement illustrates the multifaceted nature of innovation and learning in public health

care and beyond. More specifically, it highlights the practical and conceptual meaningfulness

of Jensen et al.’s (2007) distinction between science- and practice-based learning.1 Science-

based learning is fueled by advances in science, technology and innovation and contributes to

the production of often codifiable global knowledge. Practice-based learning in turn tends to

add to tacit local knowledge and thrives on doing, using and interacting as essential

constitutive elements of daily work. Science- and practice-based learning can be thought of as

two ideal-type modes located at opposing ends of the spectrum of organizational learning

processes. Brown and Duguid (1991, p. 53) seem to concur suggesting that “innovating and

learning in daily activity lie at one end of a continuum of innovating practices that stretches to

radical innovation cultivated in research laboratories at the far end.”

Although extant research demonstrates the practical and theoretical relevance of both modes

of learning (e.g. von Hippel, 1976; Rosenberg, 1982; Pavitt, 1984), quantitative research has

                                                            1 Jensen et al. (2007) use alternative labels to denote both learning modes. As for practice-based learning, they

employ the term “Doing, Using and Interacting Mode” (DUI), while they speak of science-based innovation as the “Science, Technology and Innovation (STI) Mode”. In the interest of simplicity, we opted for a more concise set of labels.

4

focused almost exclusively on science-based learning (Jensen et al., 2007). That said, an

important body of conceptual or qualitative research has provided rich insights into more

mundane, informal and experiential forms of practice-based learning. Among others, such

studies have explored situated learning (e.g. Brown and Duguid, 1991; Lave and Wenger

1991), dispersed learning (e.g. Sole and Edmondson, 2002), incremental learning (e.g.

Edmondson, 2002), learning from errors (e.g. Provera, Montefusco and Canato, 2010), or

learning through experimenting (Pablo et al., 2007). Often informed by a practice perspective

(Brown and Duguid, 2001), this literature has contributed much to understanding the nature of

learning in work settings. In particular, it revealed that working, learning and innovating are

closely interrelated and at times even inseparable (Brown and Duguid, 1991; Orr, 1996).

Learning thus often occurs during actual practice, defined as the “coordinated activities of

individuals and groups in doing their ‘real work’ as it is informed by a particular

organizational or group context” (Cook and Brown, 1999, p. 386-387). As such, practice-

based learning shapes - and is shaped by - the particular social context it is situated in (Lave

and Wenger, 1991).

We believe that quantitative studies can complement this body of literature meaningfully by

examining possible antecedents, consequences and moderators of practice-based learning. In

doing so, quantitative research can test not least the underlying assumption of many practice-

based studies that learning as an integral part of daily work is indeed beneficial (Brown and

Duguid, 1991). This paper seeks to contribute to this endeavor both conceptually and

empirically. Drawing on the Dynamic Capabilities View in general (e.g. Teece, Pisano and

Shuen, 1997; Zahra, Sapienza and Davidsson, 2006) and the notion of incremental dynamic

capabilities in particular (Ambrosini, Bowman and Collier, 2009), we conceptualize practice-

based learning as a set of routines that enable organizations to gradually reconfigure their

resource base in response to changing requirements. Organizations’ practice-based learning

capabilities are expected to be not only highly persistent over time, but also reliant on the

5

availability of slack resources. Moreover, we suggest that the small steps triggered by

practice-based learning lead to incremental improvements enhancing organizational

performance. This should hold in particular when the underlying business model is labor-

rather than capital-intensive.

We examine these ideas in the context of the public sector, which dedicates growing attention

to innovation and learning (Hartley et al., 2008). Ferlie, Hartley and Martin (2003) explicitly

call for quantitative research to complement existing qualitative studies and highlight the

public sector as a particularly promising setting for conducting robust panel data analyses.

Similarly, Easterby-Smith, Lyles and Peteraf (2009) highlight the need for longitudinal

dynamic capabilities research in under-explored contexts such as the public sector.

Responding to these calls, we present findings from dynamic panel data analyses based on a

dataset covering all 153 public non-specialist hospitals in England over the three-year period

from April 2004 to March 2007.

Overall, this study is dedicated to extending our understanding of the complex phenomenon of

practice-based learning. To do so, we employ the dynamic capability view as our overarching

theory. The latter provides the basis for decomposing practice-based learning into its

constitutive routines and for developing four distinct hypotheses, which are tested by drawing

on panel data from the particular setting of public, non-specialist hospitals. Our study thus

contributes to research on organizational learning (e.g. Zietsma et al., 2002) and dynamic

capabilities (e.g. Easterby-Smith, Lyles and Peteraf, 2009). At the same time, the empirical

setting this study is located in allows us to add to an important line of research on

organizational change and learning in the public sector in general (e.g. Cinite, Duxbury and

Higgins, 2009; Hartley, Benington and Binns, 1997) and in public health care in particular

(e.g. Ashburner, Ferlie and FitzGerald, 1996; Desai, 2010; Desombre et al., 2006; McNulty,

2003; Oborn and Dawson, 2010).

6

CONCEPTUAL BACKGROUND

Modes of Organizational Learning

In this study, we draw on recent conceptual developments by Jensen et al. (2007), who

described, operationalized, and empirically investigated an ideal type distinction between

practice-based and science-based learning, or as they labeled it, between a ‘doing, using, and

interaction’ (DUI) and a ‘science, technology, and innovation’ (STI) mode of learning.

Science-based learning relies primarily on explicit, global know-what and know-why. It

explores fundamentally new knowledge that might challenge the firm’s basic assumptions.

Science-based learning is driven primarily by the generation or adaptation of new scientific

knowledge, which is usually codified to assume a more global meaning beyond the specific

context of its emergence. Given the obvious complexity, science-based learning projects are

frequently concentrated on an ‘elite’ group of highly qualified scientists and engineers, and

tend to be conducted as dedicated R&D projects that are somewhat removed from everyday

operations. Science-based learning is thus strongly skewed towards technology-intensive,

radical learning.

Practice-based learning, in contrast, relies primarily on implicit, local know-how and know-

who. It is typically driven by rather mundane challenges and is motivated by a wide range of

pragmatic problems encountered during regular work activities. Practice-based learning

gradually expands and exploits the firm’s current knowledge base; it typically does not

fundamentally depart from, but improves upon current knowledge. It largely thrives on

learning by doing and learning by using. Internal ownership is hence ideally not limited to a

specific hierarchical level or functional specialization but highly distributed across the entire

organization. Moreover, practice-based learning typically occurs informally as part of regular

work activities. The same individual can hence naturally integrate the role of a worker, learner

7

and innovator (Brown and Duguid, 1991), rendering the common separation between learning

and work practice less appropriate.

Table 1 juxtaposes these two stylized learning modes. It is important, though, to emphasize

that this is an ideal-typical distinction for analytical purposes only. Science- and practice-

based learning as depicted here can be thought of as constituting the two extreme points of a

continuum (Brown and Duguid, 1991). The demarcation between these learning modes is

hence rather fuzzy, such that actual learning activities will more often than not include

elements of both learning modes.

_______________________________

INSERT TABLE 1 ABOUT HERE _______________________________

Practice-Based Learning as a Dynamic Capability

The Dynamic Capabilities View understands less organizational resources per se than their

effective (re)configuration as a key source of sustained superior performance (Zollo and

Winter, 2002). From a dynamic capabilities perspective, the competitive advantage of

organizations results from their superior ability to adapt, adjust or reconfigure their resource

base and operating routines in response to changing requirements (Eisenhardt and Martin,

2000; Teece, Pisano and Shuen, 1997). We follow Zollo and Winter (2002, p. 340) in defining

dynamic capabilities as “a learned and stable pattern of collective activity through which the

organization systematically generates and modifies its operating routines in pursuit of

improved effectiveness.” Ambrosini, Bowman and Collier (2009) distinguish between three

levels of dynamic capabilities, which they label incremental, renewing and regenerating

dynamic capabilities. Building on this theoretical work, we conceptualize practice-based

learning as an incremental dynamic capability given its focus on continuously improving

rather than radically altering the organizational resource base.

8

An understanding of practice-based learning as an incremental dynamic capability derives

much of its appeal from a focus on routines. Capabilities, regardless of whether they are

dynamic or substantive, are conceived of as bundles of organizational routines (Helfat and

Peteraf, 2003; Macher and Mowery, 2009). According to Edmondson, Bohmer and Pisano

(2001, p. 686) routines are “repeated patterns of behavior bound by rules and customs”.

Although routines have long been thought of as a possible cause of organizational inertia

(Hannan and Freeman, 1984), there is now a growing recognition of the transformative

potential of routines (Feldman, 2000; Howard-Grenville, 2005; Amburgey, Kelly and Barnett,

1993). It is therefore not surprising that Levitt and March (1988) conceptualize organizational

learning as being composed of distinct routines. Such learning routines can hence be the

building blocks for dynamic learning capabilities (Dyer and Nobeoka, 2000). Consequently, it

is by decomposing practice-based learning capabilities into its constitutive routines that we

seek to explicate the complex nature of the concept.

Informed by previous research on dynamic capabilities and organizational learning (e.g.

Brown and Duguid, 1991; Chiva and Alegre, 2009; Crossan, Lane and White, 1999; Day and

Schoemaker, 2004; Teece, 2007; Zollo and Winter, 2002), we identify three constitutive

routines of practice-based learning capabilities, which we label ‘detecting problems’,

‘suggesting ideas’, and ‘participating in change’.2

Detecting problems

The first routine consists in observing, recognizing and reporting problems and errors that

occur as part of daily work activities. Such problems encountered in actual practice can

constitute important “moments of breakdown” (Gherardi and Nicolini, 2001, p. 49) that

provoke reflection and thereby trigger learning processes (Cook and Brown, 1999). This

                                                            2 As suggested by the information processing view, the sequence of these three routines is not strictly linear

and unidirectional. Rather, there will be several feedback loops, through which outcomes at one stage shape subsequent iterations of information processing activities at earlier stages (Day and Schoemaker, 2004).

9

perception is shared by behavioral research. Cyert and March (1963) for instance suggest that

perceived performance problems constitute a vital trigger of organizational learning and

adaptation - a proposition that has received strong empirical support in the private (Greve,

2003) and the public sector (Salge, 2011). Similarly, Borins (2000) analyzed more than 400

innovation projects in the public sector and found that most were indeed driven by internal

problems or crises. The fundamental importance of errors, near misses or incidents as key

occasions for organizational learning is also highlighted by the literature on learning from

failure (Baum and Dahlin, 2007; Haunschild and Sullivan, 2002; Perrow, 1984; Weick and

Sutcliffe, 2007; Zhao, 2010). Indeed, this literature even suggests that organizations learn

more effectively from prior failure than they do from previous success (Madsen and Desai,

2010; Sitkin, 1992).

Suggesting Ideas

Problem detection by itself is obviously not sufficient for any meaningful learning to occur.

Rather, ideas to alleviate the performance problem at hand need to be proposed. Suggesting

such ideas is therefore the second constitutive routine of practice-based learning. As von

Hippel (1999) observes, such ideas are often best developed by those intimately familiar with

the problem rather than by “specialists” detached from the specific context. When confronted

with a problem or puzzle in their own community of practice, members are hence best

positioned to engage in so-called “productive inquiry” designed to find a solution to the

problem at hand (Cook and Brown, 1999). This process involves developing ideas for

possible solutions, testing them by means of small experiments and discussing the outcomes

(Brown and Duguid, 1991). Productive inquiry is hence a highly discursive activity, as part of

which members of a community of practice jointly seek to identify the most appropriate

solution (Gherardi and Nicolini, 2002; Wenger, 1998). In this process, ideas might be

generated that break with existing work practices and suggest something new (Orr, 1996).

10

Developing and suggesting such ideas is at the very heart of practice-based learning and

represents the creative variation generating mechanism that constitutes the vital engine of all

evolutionary processes (Nelson and Winter, 1982). Given its importance, organizations need

to encourage the suggestion of new ideas (Brown and Duguid, 1991). This can be achieved by

creating a climate of psychological safety consisting in “a shared belief that the team will not

embarrass, reject, or punish someone for speaking up” (Edmondson, 1999, p. 354). If this is

coupled by active staff encouragement (Edmondson, 2003; Nembhard and Edmondson, 2006)

and a positive attitude towards risk and experimentation (Bozeman and Kingsley, 1998),

chances are that employees feel motivated to suggest new and potentially challenging ideas

(Zollo and Winter, 2002).

Participating in change

For practice-based learning to occur, new ideas need to become part of actual work activities.

If learning is understood as “competent participation in practice” (Wenger, 1998, p. 137), it

becomes apparent that members of a community of practice need to be involved in deciding

upon changes that affect their work area. Only then can they feel and act as legitimate

members of their respective community (Gherardi and Nicolini, 2002; Lave and Wenger,

1991). The third routine thus consists in collectively implementing those novel ideas that are

perceived as being particularly effective in addressing specific problems (Chiva and Alegre,

2009). As research from social psychology shows (Baer and Frese, 2003), an organizational

climate that advocates staff initiative and participatory decision-making is likely to positively

affect the organizational ability to learn and to reap the benefits of innovative ideas. Dialogue

and joint action are crucial to the development of shared understanding and to the

institutionalization of learning, that is, to embedding learning into organizational systems,

structures, procedures and strategy (Crossan, Lane and White, 1999). Involving staff from

across the organization is particularly important for practice-based learning, which relies

11

heavily on front-line employees’ learning by doing and learning by using (Brown and Duguid,

1991; Orr, 1996; Rosenberg, 1982).

All three routines depicted above are expected to play a vital role for practice-based learning

across a range of settings. This however applies only when each routine is examined at a

relatively high level of abstraction, which is the case in our quantitative study. That said,

learning is always “enacted within the boundaries of a domain of knowing and doing”

(Gherardi, 2001, p. 131). When examined at a more concrete level, the specific nature of each

learning routine is hence likely to vary depending on the particular context it is situated in

(Lave and Wenger, 1991; Gherardi and Nicolini, 2001).

HYPOTHESES

Temporal Persistence

The assumption of temporal persistence is implicit in the conceptualization of practice-based

learning as a dynamic capability consisting of a bundle of organizational routines defined as

“a learned and stable pattern of collective activity” (Zollo and Winter, 2002, p. 340, emphasis

added). We therefore expect considerable path-dependence in the extent to which organi-

zations demonstrate practice-based learning capabilities (Sydow, Schreyögg and Koch, 2009).

Several arguments can be advanced in support of this proposition.

First, positive market feedback is expected to drive a virtuous cycle between learning

capabilities, learning success and organizational performance (Malerba, Orsenigo and Peretto,

1997; Nelson and Winter, 1982). Such ‘success-breeds-success’ models indicate that

successful engagement in learning leads to greater financial and non-financial returns, which

are invested in resources that facilitate future engagement in learning activities. Organizations

with effective practice-based learning routines will thus be inclined to further invest in

strengthening them.

12

Second, temporal persistence in practice-based learning capabilities may also be the result of

the cumulative nature of organizational learning. What an organization learns depends on

what it already knows, on the length of its history, and on the development stage of its

organizational routines (Cohen and Levinthal, 1990; Zahra and George, 2002). More

specifically, engagement in practice-based learning is likely to improve not only specific

operating routines but also the practice-based learning routines themselves through ‘learning-

by-doing’ and ‘learning-to-learn’ effects (Geroski, Van Reenen and Walters, 1997; Malerba,

Orsenigo and Peretto, 1997).

Third, organizations dedicate considerable attention and resources to the institutionalization of

dynamic capabilities (Teece, Pisano and Shuen, 1997). As for practice-based learning

capabilities, such investment might consist in establishing formal idea suggestion systems,

quality improvement circles or specific incentive schemes (Jensen et al., 2007). Such support

infrastructure is likely to contribute to greater temporal stability in the level of practice-based

learning capabilities. We therefore expect the latter to be characterized by considerable path-

dependence. Hence:

Hypothesis 1: Practice-based learning capabilities will demonstrate high temporal

persistence, such that past levels of practice-based learning capabilities will

be closely related to current levels.

Slack Resources

The development, maintenance and exercise of dynamic capabilities are highly resource-

intensive activities (Zahra, Sapienza and Davidsson, 2006). This is also likely to hold for

practice-based learning capabilities. Organizations will have to invest in developing an

appropriate organizational infrastructure and a supportive organizational climate, if they are to

build and maintain effective practice-based learning capabilities. Similarly, human and

13

financial resources will be required for problem detection, idea suggestion and change

implementation. As resources are scarce, those dedicated to building, maintaining and

exercising practice-based learning capabilities might be lacking elsewhere within the

organization. The extent to which organizations can invest in their practice-based learning

capabilities is thus likely to depend on their level of uncommitted excess resources (Voss,

Sirdeshmukh and Voss, 2008). Operational slack - for instance in form of uncommitted staff

time - appears particularly important for practice-based learning in that it provides employees

with the autonomy and flexibility required to engage in problem detection, idea suggestion

and change implementation (Singh, 1986). Consistent with these theoretical arguments, recent

empirical studies have revealed a positive link between slack and innovation in a number of

private and public sector settings (Chen, 2008; Danneels, 2008; Greve, 2003; Kim, Kim and

Lee, 2008; Salge, 2011).

At the same time, researchers have argued that excessive levels of slack can be detrimental for

innovation and learning (Jensen, 1986; Jensen, 1993; Nohria and Gulati, 1996). In particular,

excessive levels of slack are assumed to result in loose managerial controls and ineffective

use of organizational resources for initiatives that are more closely aligned with employees’

own preferences than with organizational goals. Perhaps more importantly, however,

excessive slack can provide a false sense of security that contributes to organizational inertia

at the expense of continuous adaptation (Hannan and Freeman, 1984). In such a case,

decision-makers are likely to favor stability, hence dedicating less attention and resources to

building, maintaining and exercising practice-based learning capabilities.

Considering these arguments, it appears likely that the marginal effect of slack on learning

will decline as operational slack increases. As a consequence, operational slack will be

positively associated with practice-based learning up until the point where slack becomes

excessive. Once this turning point has been reached, however, excessive slack is expected to

14

drive inertia rather than adaptation leading to a negative marginal effect of slack on learning.

Overall, we hence assume a curvilinear, inverse U-shaped relationship between operational

slack and the level of practice-based learning capabilities. The latter will hence be optimized

at moderate levels of slack, while both too much and too little slack is likely to be suboptimal.

Although this proposition has not yet been examined with regards to practice-based learning

capabilities, Nohria and Gulati (1996) found empirical support for a curvilinear link between

slack and innovation. We therefore suggest:

Hypothesis 2: There will be a curvilinear (inverse U-shaped) relationship between opera-

tional slack and the level of practice-based learning capabilities.

Performance Consequences

Dynamic capabilities are widely assumed to positively affect organizational performance (e.g.

Helfat and Peteraf, 2003; Teece, Pisano and Shuen, 1997; Winter, 2003). In particular,

dynamic capabilities are conceived of as being the engine of organizational adaptation to

changing external requirements. More specifically, dynamic capabilities allow organizations

to renew their resource base and operating routines (Eisenhardt and Martin, 2000). This is of

great strategic value in times of competitive and fast moving market and technology environ-

ments (Teece, 2007). Consequently, organizations with strong dynamic capabilities are

expected to exhibit superior performance (Zahra, Sapienza and Davidsson, 2006; Zott, 2003).

Following these arguments, we suggest that practice-based learning capabilities are likely to

enhance organizational performance. Detecting problems, suggesting ideas and implementing

change clearly constitute key routines to modify or reconfigure existing operating processes,

thereby adapting to changing environmental requirements. Over time, practice-based learning

capabilities are hence expected to contribute to extending and updating the organizational

knowledge base. They therefore enable the organization to change the way it solves its

15

problems in pursuit of improved effectiveness. Although such change might come in small

steps, we expect their sum to matter greatly. Consistent with extant learning (e.g. Pisano,

Bohmer and Edmondson, 2001) and innovation research (e.g. Damanpour, Walker and

Avellaneda, 2009; Jansen, Van Den Bosch and Volberda, 2006; Han, Kim and Srivastava,

1998), we therefore propose that organizational performance will tend to improve with higher

levels of practice-based learning capabilities. Thus:

Hypothesis 3: There will be a positive relationship between the level of practice-based

learning capabilities and organizational performance.

Moderating Role of the Business Model

Despite its hypothesized overall performance-enhancing effect, we do not expect practice-

based learning capabilities to be of universal value. Rather, we assume that organizations are

likely to differ in the extent to which they will benefit from practice-based learning. We thus

advocate a contingency view of practice-based learning and examine the extent to which the

pay-off from practice-based learning is contingent on the type of business model adopted by

the focal organization - a crucial factor that has hitherto received little attention in organi-

zational learning research (Chesbrough and Rosenbloom, 2002).

A business model is of vital strategic importance in that it constitutes an abstract repre-

sentation of the economic “money earning” logic of an organization (Froud et al., 2009;

Magretta, 2002; Zott and Amit, 2010). As such, business models specify the customer value

proposition as well as the mechanisms for value creation and value capture (Chesbrough,

2010; Teece, 2010). Since business models constitute complex systems, they can vary along

various dimensions including the targeted market segment or the value chain elements

covered (Chesbrough and Rosenbloom, 2002). Inter-organizational variation along these

dimensions is likely to be reflected in distinct resource configurations. As the latter are at the

heart of the dynamic capability view, we propose a resource-oriented distinction.

16

Although resources come in various forms and shapes (Barney, 1991), we focus on the

relative importance of labor and capital as two elementary, mutually substitutable resource

types. Following this approach, business models can be positioned on a continuum with

highly labor- and highly capital-intensive business models as its two stylized endpoints. Both

ideal-type business models are likely to have very different learning requirements.

Organizations with highly capital-intensive business models tend to rely primarily on

sophisticated state-of-the-art technologies and facilities. To adapt and reconfigure such a

resource base, science- rather than practice-based learning capabilities appear to be

paramount. Organizations with high relative capital intensity are hence expected to benefit

more from rigorous R&D activities conducted by highly qualified experts than from more

mundane and pragmatic problem detection, idea suggestion and change implementation

routines fueled by front-line employees. Conversely, organizations with a highly labor-

intensive business model rely heavily on motivated and skilled personnel and its ability to

satisfy customer needs. As operating routines tend to be less sophisticated technologically, its

continuous reconfiguration is not the sole responsibility of a small scientific elite, but the

collective mission of all employees. In doing so, the latter detect problems, suggest ideas and

implement change in a joint endeavor to adjust and further improve operating routines.

Given these distinct learning requirements, the type of business model is expected to

moderate the pay-off from learning. As for practice-based learning, we suggest that the

performance effect is likely to increase with the level of relative labor intensity. Hence:

Hypothesis 4: The level of relative labor intensity will positively moderate the relationship

between practice-based learning capabilities and organizational perfor-

mance, such that the effect is higher in organizations with a labor-intensive

than with a capital-intensive business model.

Figure I summarizes our hypotheses and illustrates our theoretical model.

17

_______________________________

INSERT FIGURE I ABOUT HERE _______________________________

METHODOLOGY

Setting and Data

We examine our hypotheses in the context of public hospital services in England. As health

care is a highly knowledge intensive service facing the continuous challenge of optimizing

clinical outcomes with increasingly limited resources, innovation and learning are clearly

paramount (Djellal and Gallouj, 2005; Metcalfe, James and Mina, 2005; Salge and Vera,

2009). In choosing this setting, we also contribute to broadening the empirical evidence base

on organizational learning and dynamic capabilities – two research areas, where large-scale

empirical studies located in the public sector remain surprisingly absent despite their apparent

relevance (e.g. Easterby-Smith, Lyles and Peteraf, 2009; Harvey et al., 2010).3 This is all the

more important, as public sector organizations tend to differ significantly from their private

sector counterparts not least with regards to their stronger exposure to political influence,

competitive forces, public scrutiny and internal bureaucracy (Bozeman and Bretschneider,

1994; Boyne, 2002). All of these are likely to affect the highly context-specific phenomenon

of organizational learning (e.g. Miner and Mezias, 1996; Bapuji and Crossan, 2004).

Public hospitals in England operate as so-called Acute Trusts under the umbrella of the

National Health Service (NHS). Each Acute Trust is an independent legal entity with its own

board and annual financial statements. Our dataset includes all 153 non-specialist Acute

Trusts in England. Each is observed over the three-year period from April 2004 to March

                                                            3 See Moynihan and Landuyt (2009) for a rare exception.

18

2007 for a total of 459 observations. It is precisely the absence of such panel datasets that has

long constituted a major obstacle to empirical learning research in the public sector (Ferlie,

Hartley and Martin, 2003). Similarly, such a dataset is needed to respond to recent calls for

more longitudinal research on dynamic capabilities (Easterby-Smith, Lyles and Peteraf,

2009). The database underlying this study was populated by integrating archival data provided

by numerous actors in the English health system, which include the Department of Health, the

NHS, Dr Foster, the Care Quality Commission and Monitor. Whenever archival records were

incomplete, we collected the missing information directly from the individual Acute Trusts

drawing on published material such as annual reports or responses to our own information

requests. This process resulted in a complete and balanced panel dataset that was not affected

by any missing data problems.

Measures

Practice-based Learning Capabilities

The measurement of dynamic capabilities in general and practice-based learning capabilities

in particular is inherently difficult, as they materialize only when exercised (Easterby-Smith,

Lyles and Peteraf, 2009). Jensen et al. (2007) for instance relied on a survey to solicit top

management’s view on the presence of certain organizational practices thought to enhance

practice-based learning. These included interdisciplinary work groups, quality circles and

suggestion systems among others. Although clearly valid, this approach captures only mana-

gerial perceptions, which are likely to diverge from organizational reality.

In the present study, we therefore follow recent calls for multi-informant designs (e.g. Doty

and Glick, 1998; Van Bruggen, Lilien and Kacker, 2002; Walker and Enticott, 2004) and rely

on the assessments of a large number of employees in each organization distributed across

hierarchical levels and functional areas. Such unique multi-informant data on practice-based

19

learning was collected by means of three waves of an annual staff survey carried out by the

NHS. This survey covers all NHS Trusts, each of which draws a random quota sample among

all of its staff members. Response rates across all Trusts amount to 56.58 percent in 2004,

54.54 percent in 2005 and 51.55 percent in 2006. As for the practice-based learning items, for

the three-year period from April 2004 to March 2007, an average of 409.21 individual staff

responses are available per organization and year (sd=69.94; min=131; max=563).

Consistent with our theoretical conception and recent empirical studies in the dynamic

capabilities literature (e.g. Macher and Mowery, 2009), we adopt a routine-based

measurement approach. In particular, we measure practice-based learning capabilities by

means of three items with five-point Likert-type scales ranging from “strongly disagree” to

“strongly agree”. These are “My Trust encourages us to report errors, near misses or inci-

dents” for routine 1 (‘detecting problems’), “Managers encourage staff to suggest new ideas

for improving services” for routine 2 (‘suggesting ideas’), and “I am involved in deciding on

the changes introduced that affect my work area / team / department” for routine 3 (‘partici-

pating in change’). We then aggregated these individual-level perceptions of learning routines

to the hospital level. To justify such aggregation, we needed to demonstrate the reliability of

learning routine means across units. For this purpose, we calculated intraclass correlation

(ICC) statistics and found an ICC(2) of 0.92, well above the 0.70 benchmark required to

justify aggregation (Klein and Kozlowski, 2000). Moreover, the overall measure of practice-

based learning capabilities demonstrated an adequate reliability coefficient alpha of 0.76.

Organizational Performance

When examined from a service user perspective (Ferlie, Hartley and Martin, 2003; Vigoda-

Gadot et al., 2008), the quality of public services is expected to be the central performance

metric. Given hospitals’ primary mission to cure ill or injured patients, quality, at the most

basic level, consists in guaranteeing the survival of as many admitted patients as possible, that

20

is, low mortality rates. The latter, however, are obviously affected by the specific

characteristics of a hospital’s patient mix and hence need to be risk-adjusted to take inter-

hospital differences in patient mix into account. For this purpose, the so-called Hospital

Standardized Mortality Ratio (HSMR) has been developed and is now widely established as a

standard measure of hospital quality (e.g. Jarman et al., 1999). The HSMR corresponds to the

ratio of actual deaths to expected deaths. Risk-adjustment is achieved by taking into account

standardization factors such as gender, age, primary diagnosis and co-morbidities. The HSMR

is centered to 100 corresponding to the expected death rate with values greater (smaller) than

100 indicating a higher (lower) death rate than normally expected. Since a higher HSMR

represents lower performance, we calculated organizational performance as 200 - HSMR.4

Operational Slack

Operational slack consists in unused or underutilized physical resources, e.g. excess

production capacities, unoccupied facilities or excess staffing levels. In the hospital sector, a

key physical resource is the number of beds, which is also by far the most common indicator

for hospital size (e.g. Kimberly and Evanisko, 1981; Goes and Park, 1997). Therefore, in the

present study we measure operational slack by the share of unoccupied hospital beds.

Relative Labor Intensity of Business Model

We argued that relative labor intensity, i.e. the ratio between labor to capital investments, is a

key factor to distinguish different types of business models. Consistent with previous research

(e.g. Dewenter and Malatesta, 2001), we measure relative labor intensity as the ratio between

the number of full-time equivalents employed by the Trust and the value of fixed assets as

specified in the hospital’s balance sheet.

                                                            4 While the HSMR reflects in particular a hospital’s success at addressing the most severe cases where survival

is at stake, it might also serve as a suitable proxy for the quality of care in less severe cases. For instance, a low HSMR is likely to reflect a superior ability to avoid medical errors and hospital acquired infections, both of which can be a cause of death among less severely ill patients.

21

Control Variables

We also controlled for a number of factors that are regularly associated with differences in

hospital performance (Chadwick, Hunter and Walston, 2004; Fottler, 1987; McCracken,

McIlwain and Fottler, 2001). First, we measured the size of each hospital by the total number

of hospital beds available for use - the dominant size measure in the health care management

literature and commonly assumed to be an important predictor of care quality (e.g. Wouters,

Jansen-Landheer and van de Velde, 2010). Second, by means of a dummy variable we

captured whether an organization was granted foundation trust status leading to greater

managerial autonomy. Third, we included a dummy measure of the hospital’s university

affiliation. Fourth, given the increasingly competitive nature of the public hospital landscape

in England (Martin and Smith, 2005), we also control for environmental competitiveness

measured as the share of unoccupied beds within the local market, assuming that higher

excess bed capacity within a market leads to intensified competition for activity volumes

among local hospital service providers. Fifth, the environmental rurality of the hospital’s

location measured by a rurality index on a scale from 0 (metropolitan) to 100 (rural) was used

as a control variable. Last, we control for year effects by means of time dummies.

Analyses

Our dataset includes a cross-sectional and a temporal dimension and, consequently, calls for

suitable panel data techniques such as random or fixed effects estimators (Halaby, 2004). To

test hypothesis 1 pertaining to the temporal persistence of practice-based learning capabilities,

however, we needed to include a lagged dependent variable. Such an autoregressive model

cannot be estimated reliably using standard panel data techniques given the risk of potential

endogeneity biases due to the inclusion of a lagged dependent variable (Cefis and Orsenigo,

2001). We therefore rely on a one-step system General Method of Moments (GMM) estimator

explicitly developed for estimating autoregressive models based on panel data with many

22

cross-sectional, yet few temporal observations (Arellano, 2003). GMM models directly

address the endogeneity bias that can arise particularly in so-called “small T, large N” panels

with lagged dependent variables, which are necessarily correlated with the error term. GMM

models do so by using lagged values of the regressors as instruments, instead of drawing on

instrumental variables external to the structural equation. Furthermore, GMM estimation

accounts for time-invariant, unobserved organization-specific heterogeneity, which is elimi-

nated by transforming all regressors, typically by first-differencing, and thereby minimizes the

risk of spurious persistence estimates. Given these advantages, we employ the Arellano-Bond

System GMM estimator with heteroscedasticity and autocorrelation consistent standard errors

(Wawro, 2002; Uotila et al., 2009).5 Consistent with good practice in strategic action-

performance studies, we lag all independent variables by one year to establish the temporal

sequencing that is required when seeking to identify cause and effect relationships.

RESULTS

Table 2 presents descriptive statistics and pairwise correlations. Descriptive statistics indicate

considerable inter-organizational variation in all variables. Given the purpose of this study, it

is particularly important to note the considerable variation in organizational performance and

the low correlations between our key variables organizational performance, practice-based

learning, relative labor intensity and operational slack.

_______________________________

INSERT TABLE 2 ABOUT HERE _______________________________

Table 3 presents GMM panel data regression results with respect to the antecedents of

practice-based learning capabilities. Model 1 contains only controls. Main effects are then

introduced in model 2.                                                             5 In our GMM models, we follow previous research and treat time effects and control variables as exogenous,

main and interaction effects variables as endogenous and all other independent variables as predetermined.

23

_______________________________

INSERT TABLE 3 ABOUT HERE _______________________________

As for the control variables, coefficient estimates are negative and statistically significant for

‘organizational size’ and ‘environmental rurality’, but positive for ‘foundation trust status’.

These findings are consistent across models 1 and 2, suggesting that small foundation trusts in

metropolitan areas have particularly high levels of practice-based learning capabilities.

Hypothesis 1 suggests a positive relationship between past and current levels of practice-

based learning capabilities. More specifically, the coefficient estimate for the lagged practice-

based learning variable in model 2 - the persistence parameter - indicates the degree of

temporal persistence in practice-based learning. This parameter is expected to fall within the

range from zero to one (Roberts and Dowling, 2002). Values approaching one indicate high

persistence, while those approaching zero indicate low persistence (O'Toole and Meier, 1999).

The persistence parameter in model 2 is 0.44 and highly significant, pointing to moderate

temporal persistence in practice-based learning capabilities. Hypothesis 1 is thus supported.

According to hypothesis 2, we expect an inverse U-shaped relationship between operational

slack and the level of practice-based learning capabilities. To support such a curvilinear link,

the coefficient of the linear term for operational slack in model 2 needs to be positive, while

the coefficient of the squared term needs to be negative and statistically significant. Both

coefficients demonstrate the expected algebraic signs. However, the effects are very small in

magnitude and fail to achieve statistical significance. Hypothesis 2 is thus not supported.6

Table 4 presents the results from GMM analyses explaining hospital performance. Model 3

contains only controls. Main effects are added in model 4 and interaction effects in model 5.

                                                            6 The lack of statistical significance might be a consequence of our indirect measurement approach of

operational slack. Although it appears plausible to assume that lower bed occupancy will be associated with higher uncommitted staff time, it would have been better to measure the latter directly. As an alternative explanation, practice-based learning might be less dependent on uncommitted resources in that it can potentially make use of the resources regularly available as part of daily work activities.

24

_______________________________

INSERT TABLE 4 ABOUT HERE _______________________________

As for the control variables, the strong and highly significant coefficients of ‘lagged

performance’ in models 3 to 5 support a conception of public service organizations as

autoregressive, inertial systems, whose past performance is often one of the best predictors of

current performance (O’Toole and Meier, 1999; Meier and O'Toole, 2003). We also identify

significant, positive effects of ‘university affiliation’ and ‘environmental rurality’, which are

consistent across models 3 to 5 indicating particularly high patient survival rates for university

hospitals in rural areas. Although hospital size is frequently considered an important predictor

of care quality, we do not find larger hospitals to demonstrate higher patient survival rates.

Hypothesis 3 suggests a positive relationship between the level of practice-based learning

capabilities and organizational performance. In line with our theoretical expectation, the

coefficient estimate for practice-based learning in model 4 is positive and statistically

significant. The effect size is substantial and practically meaningful in that it suggests that a

one point increase in practice-based learning capabilities, measured on a 5 point scale, is

associated with a 14.07 point increase in risk-adjusted patient survival, centered on 100. Even

a one standard deviation increase in practice-based learning capabilities, i.e. an increase by

0.1 points, will be associated with a 1.41 point increase in risk-adjusted patient survival.

Hypothesis 3 is thus supported.

Following hypothesis 4, we expect the positive relationship between practice-based learning

capabilities and performance to be positively moderated by the relative labor intensity of

hospitals’ business models. We thus expect the performance-enhancing effect of practice-

based learning capabilities to become stronger as hospitals’ ratio of labor to capital

investments increases. The interaction term in model 5 is positive and significant, suggesting

that hospitals with greater relative reliance on human resources tend to benefit significantly

25

more from practice-based learning capabilities than their more capital-intensive counterparts.

Hypothesis 4 is thus supported.

DISCUSSION

In our study, we examined the phenomenon of practice-based learning, employing the

dynamic capability view as our overarching theory and drawing on panel data from the

specific setting of public hospital services. As a result, our findings are likely to have

meaningful implications for research on (1) organizational learning, (2) dynamic capabilities

and (3) public and health care management as well as for (4) practice and policy.

Implications for Research on Organizational Learning

First and most importantly, our study adds to the literature on organizational learning. In

particular, we provide one of the first quantitative and longitudinal analyses of incremental,

practice-based learning. In doing so, we complement previous qualitative studies taking a

practice perspective on learning (e.g. Brown and Duguid, 1991; Lave and Wenger 1991; Sole

and Edmondson, 2002). Such research has been highly useful for providing rich insights into

the nature of learning as an integral part of regular work practices. However, it is less well

suited for identifying systematic relationships pertaining to the antecedents, consequences and

moderators of practice-based learning. Brown and Duguid (1991, p. 55) for instance highlight

that “it has been unstated assumption that a unified understanding of working, learning and

innovation is potentially highly beneficial.” Our study provides robust empirical support of

this assumption, finding that practice-based learning capabilities indeed enhance

organizational performance in a statistically significant way. This finding complements

Jensen et al.’s (2007) work that uncovered innovation-enhancing effects of practice-based

learning. There is hence now robust evidence suggesting the practice-based learning is

beneficial not only for innovation but also for learning.

26

Add these findings to the fact that science-based learning is equally a key contributor to

innovation and performance (Jensen et al., 2007, Salge and Vera, 2009). It then becomes

apparent that the entire spectrum of organizational learning activities is highly valuable. Our

findings also suggest that the payoff from practice-based learning is likely to depend on the

respective business model and the resource configuration adopted. We therefore contribute to

the contingency perspective on learning that has already identified a number of essential

boundary conditions for effective organizational learning (e.g. Bapuji and Crossan, 2004;

Miner and Mezias, 1996; Zahra, Sapienza and Davidsson, 2006, Nembhard and Edmondson,

2006). As quantitative research continues to demonstrate a pronounced bias towards the

science-based end of the continuum (Jensen et al., 2007), further research is required to

explore the practice-based end of the spectrum in greater detail. Such research can potentially

benefit from our study that started to examine possible boundary conditions and decomposed

the concept of practice-based learning into its constitutive routines thereby making it

accessible to systematic measurement.

Our study focused on organizations’ aggregate practice-based learning capabilities and hence

examined the three constitutive routines as a bundle rather than individually. That said, our

findings also provide indirect support for the performance-enhancing effects of problem

detection, idea suggestion and change implementation routines. This holds for instance with

regards to problem detection and reporting as an essential precondition for practice-based

learning. In particular, our findings are consistent with Gherardi and Nicolini’s (2001) ideas

of problems as “moments of breakdown” that trigger reflection and learning. This reinforces

the vital importance of error reporting (Zhao and Olivera, 2006) and of creating a climate of

psychological safety that reduces the interpersonal risks potentially associated with detecting

and reporting errors (Nembhard and Edmondson, 2006; Tucker and Edmondson, 2003;

Edmondson 2003; Edmondson 1999).

27

Implications for Research on Dynamic Capabilities

Second, our study contributes to the literature on dynamic capabilities that has long been

limited by the lack of longitudinal studies designed to uncover temporal dynamics and causal

relationships (Easterby-Smith, Lyles and Peteraf, 2009). This holds not least for incremental

dynamic capabilities meant to gradually modify the organizational resource base. As one of

few large-scale longitudinal studies explicitly testing central predictions of the Dynamic

Capabilities View, our research contributes to filling this research gap.

For this purpose, we respond to frequent calls for greater conceptual clarity in dynamic

capabilities research (Zahra, Sapienza and Davidsson, 2006). We thus focus on one specific

incremental dynamic capability, for which we provide a concrete definition, a

conceptualization based on three constitutive routines as well as a measurement model. This

approach makes the abstract concept of dynamic capabilities amenable to hypothesis testing.

In particular, our finding that practice-based learning capabilities are persistent over time, i.e.

path-dependent, is consistent with a theorization of dynamic capabilities as bundles of

routines, defined as “learned and stable pattern[s] of collective activity” (Zollo and Winter,

2002, p. 340, emphasis added). Moreover, our results support the central prediction that

dynamic capabilities enhance operating routines and organizational performance (Zott, 2003)

even in a public sector setting that is partly shielded from competitive market pressures

(Easterby-Smith, Lyles and Peteraf, 2009). Perhaps more interestingly from a conceptual

point of view, we demonstrate that the value of dynamic capabilities depends not only on

environmental factors (Zahra, Sapienza and Davidsson, 2006), but also on the type of

business model adopted by the focal organization - a crucial factor that has hitherto received

little attention by dynamic capability researchers. Future contingency studies might thus wish

to explore internal in addition to external contingency factors potentially moderating the

relationship between dynamic capabilities and performance.

28

Implications for Research on Public and Health Care Management

Third, relevant implications for public and health care management research can be derived

from our study. Here, our study contributes not least to the important literature stream that

seeks to explain performance differences among public organizations in general (e.g. Boyne

et al., 2005; Meier et al., 2007; Hartley et al., 2008) and hospitals in particular (e.g. Chadwick,

Hunter and Walston, 2004; Fottler, 1987). Previous research in this area has revealed the

importance of organizational structure, culture and strategy as determinants of performance.

Recent qualitative studies, however, suggest that specific dynamic capabilities such as the

implementation of innovations (Piening, 2011) or learning through experimenting (Pablo et

al., 2007) might play a vital role in explaining performance differences among public

organizations. Our longitudinal study provides robust evidence in support of this proposition.

In particular, our findings highlight that hospitals with more pronounced practice-based

learning capabilities demonstrate notably higher risk-adjusted patient survival rates than those

with less developed practice-based learning capabilities. This finding underlines the

considerable value of the dynamic capability and organizational learning concepts for public

and health care management research, but also points to the vital importance of practice-based

learning routines. Especially in health care, the detection and reporting of problems that might

take the form of errors, near misses or incidents appears particularly important, as errors will

often have severe consequences (Dean et al., 2002). Our research indicates that strong

routines for detecting, reporting and learning from such errors have considerable potential for

improving the quality of care in hospital settings (Edmondson, 1996). That said, all three

practice-based learning routines merit further, more detailed longitudinal analysis.

29

Implications for Practice and Policy

Last, some implications for practice and policy emerge from this study. Perhaps most

importantly, our findings point managers and policy makers to the notable importance of

dynamic capabilities in general and practice-based learning in particular. Clearly, the close

association between practice-based learning and patient survival suggests that encouraging

incremental learning occurring as an integral part of daily work activities is likely to

contribute to tangible performance improvements. This reinforces findings from qualitative

studies on innovation suggesting that “hidden innovation often represents the innovation that

matters - the innovation that most directly contributes to the real practice and performance of

a sector” (NESTA, 2007, p.4). Interviews conducted with over seventy practitioners in the

English NHS indicate, however, that practice-based learning routines receive surprisingly

little managerial and political attention. In contrast, more science-based initiatives enjoy

substantial support from top management, R&D-specialists and external funding bodies.

Similar observations have been made in other settings such as the Danish private sector

(Jensen et al., 2007). There hence appears to be a general predisposition towards high-tech,

science-based forms of learning. It is against this backdrop that this paper highlights that

practice-based learning routines merit greater support by both managers and policy makers.

By decomposing practice-based learning into three of its constitutive routines, our study

guides managers in this endeavor. For these routines to flourish, decision-makers need to

encourage the reporting and open discussion of errors, incidents and other problems (Provera,

Montefusco and Canato, 2010), value and act upon the ideas submitted by all staff members

and involve the latter in key decisions that affect their work (Tucker and Edmondson, 2003;

Edmondson 2003; Nembhard and Edmondson, 2006).

30

Limitations and Future Research

A number of important limitations remain for future research to be addressed.

First, our study draws on existing survey data to enable longitudinal analyses. As such, it was

limited with regards to the number of items available to measure the practice-based learning

construct. Therefore, we are only able to examine each routine in a rather abstract way that

hides a wealth of detail and hence does not reflect adequately their inherent complexity.

Future research could address this important limitation of our study by developing a broader

set of items capturing the various facets of each learning routine.

Second, we rely on subjective staff perceptions of the three practice-based learning routines.

Although our multi-informant approach benefits from an average of 409 staff responses per

organization and year, we do not measure actual problem reporting, idea suggestion or change

implementation behaviors. Future research might thus wish to collect objective rather than

subjective data to capture the constitutive routines of practice-based learning.

Last, this study relies on data from public hospital services in England. Given the specificity

of this setting, the question of generalizability of our findings merits particular attention.

Clearly, our empirical results hold primarily for the specific context of this study. Although

we believe that similar relationships may be revealed in alternative health care and public

sector settings, this assumption should not remain untested. Indeed, we hope that this research

will trigger further research into the nature and value of practice-based learning in the public

sector and beyond.

31

REFERENCES

Ambrosini, V., C. Bowman and N. Collier (2009). ‘Dynamic capabilities: An exploration of how firms renew their resource base’, British Journal of Management, 20(s1), pp. S9–S24.

Amburgey, T.L., D. Kelly and W.P. Barnett (1993). ‘Resetting the clock: The dynamics of organizational change and failure’, Administrative Science Quarterly, 38(1), pp. 51-73.

Arellano, M. (2003). Panel Data Econometrics. Oxford: Oxford University Press.

Ashburner, L., E. Ferlie and L. FitzGerald (1996). ‘Organizational transformation and top-down change: The case of the NHS’, British Journal of Management, 7(1), pp. 1-16.

Baer, M. and M. Frese (2003). ‘Innovation is not enough: climates for initiative and psychological safety, process innovations, and firm performance’, Journal of Organizational Behavior, 24(1), pp. 45-68.

Bapuji, H. and M. Crossan (2004). ‘From Questions to Answers: Reviewing Organizational Learning Research’, Management Learning, 35(4), pp. 397-417.

Barnett, W.P. and J. Freeman (2001). ‘Too much of a good thing? Product proliferation and organizational failure’, Organization Science, 12(5), pp. 539-558.

Barney, J.B. (1991). ‘Firm resources and sustained competitive advantage’, Journal of Management, 17(1), pp. 99-120.

Baum, J.A.C. and K.B. Dahlin (2007). ‘Aspiration performance and railroads' patterns of learning from train wrecks and crashes’, Organization Science, 18(3), pp. 368-385.

Borins, S. (2000). ‘Loose Cannons and Rule Breakers, or Enterprising Leaders? Some Evidence About Innovative Public Managers’, Public Administration Review, 60(6), pp. 498-507.

Borins, S. (1998). Innovating With Integrity: How Local Heroes Are Transforming American Government. Washington D.C.: Georgetown University Press.

Boyne, G.A. (2002). ‘Public and private management: What's the difference?’ Journal of Management Studies, 39(1), pp. 97-122.

Boyne, G.A., K.J. Meier, L.J. O’Toole and R.M. Walker (2005). ‘Where next? Research directions on performance in public organizations’, Journal of Public Administration Research and Theory, 15(4), pp. 633-639.

Bozeman, B. and S. Bretschneider (1994). ‘The "publicness puzzle" in organization theory: A test of alternative explanations of differences between public and private organizations’, Journal of Public Administration Research and Theory, 4(2), pp. 197-224.

Bozeman, B. and G. Kingsley (1998). ‘Risk culture in public and private organizations’, Public Administration Review, 58(2), pp. 109-118.

Brown, J.S. and P. Duguid (2001). ‘Knowledge and organization: A social-practice Perspective’, Organization Science, 12(2), pp. 198-213.

Brown, J.S. and P. Duguid (1991). ‘Organizational learning and communities-of-practice: Toward a unified view of working, learning, and innovation.’, Organization Science, 2(1), pp. 40-57.

Cefis, E. and L. Orsenigo (2001). ‘The persistence of innovative activities: A cross-countries and cross-sectors comparative analysis’, Research Policy, 30(7), pp. 1139-1158.

Chadwick, C., L.W. Hunter and S.L. Walston (2004). ‘Effects of downsizing practices on the performance of hospitals’, Strategic Management Journal, 25(5), pp. 405-427.

32

Chen, W.-R. (2008). ‘Determinants of Firms' Backward-and Forward-Looking R&D Search Behavior’, Organization Science, 19(4), pp. 609-622.

Chesbrough, H. (2010). ‘Business model innovation: opportunities and barriers’, Long Range Planning, 43(2), pp. 354-363.

Chesbrough, H. and R.S. Rosenbloom (2002). ‘The role of the business model in capturing value from innovation: evidence from Xerox Corporation's technology spin-off companies’, Industrial and Corporate Change, 11(3), pp. 529-555.

Chiva, R. and J. Alegre (2009). ‘Organizational learning capability and job satisfaction: An empirical assessment in the ceramic tile industry’, British Journal of Management, 20(3), pp. 323-340.

Cinite, I., L.E. Duxbury and C. Higgins (2009). ‘Measurement of perceived organizational readiness for change in the public sector’, British Journal of Management, 20(2), pp. 265-277.

Cohen, W.M. and D.A. Levinthal (1990). ‘Absorptive capacity: A new perspective on learning and innovation’, Administrative Science Quarterly, 35(1), pp. 128-152.

Cook, S.D.N. and J.S. Brown (1999). ‘Bridging epistemologies: The generative dance between organizational knowledge and organizational knowing’, Organization Science, 10(4), pp. 381-400.

Crossan, M.M., H.W. Lane and R.E. White (1999). ‘An organizational learning framework: from intuition to institution’, Academy of Management Review, 24, pp. 522-37.

Cyert, R.M. and J.G. March (1963). A Behavioral Theory of the Firm. Englewood Cliffs, NJ: Prentice Hall.

Damanpour, F., R.M. Walker and C.N. Avellaneda (2009). ‘Combinative Effects of Innovation Types and Organizational Performance: A Longitudinal Study of Service Organizations’, Journal of Management Studies, 46(4), pp. 650-675.

Danneels, E. (2008). ‘Organizational antecedents of second-order competences’, Strategic Management Journal, 29, pp. 519-543.

Day, G.S. and P.J.H. Schoemaker (2004). ‘Driving Through the Fog: Managing at the Edge’, Long Range Planning, 37(2), pp. 127-142.

Dean, B., M. Schachter, C. Vincent and N. Barber (2002). ‘Causes of prescribing errors in hospital inpatients: a prospective study’. The Lancet, 359(9315), pp. 1373–1378.

Desai, V.M., (2010). ‘Ignorance isn't bliss: Complaint experience and organizational learning in the California nursing home industry, 1997-2004’, British Journal of Management, 21(4), pp. 829-842.

Desombre, T., C. Kelliher, F. Macfarlane and M. Ozbilgin (2006). ‘Re-organizing work roles in health care: Evidence from the implementation of functional flexibility’, British Journal of Management, 17(2), pp. 139-151.

Dewenter, K.L. and P.H. Malatesta (2001). ‘State-Owned and Privately Owned Firms: An Empirical Analysis of Profitability, Leverage, and Labor Intensity’, American Economic Review, 91(1), pp. 320-334.

Djellal, F. and F. Gallouj (2005). ‘Mapping innovation dynamics in hospitals’, Research Policy, 34(6), pp. 817-835.

Doty, D.H. and W.H. Glick (1998). ‘Common Methods Bias: Does Common Methods Variance Really Bias Results’, Organizational Research Methods, 1, pp. 374-406.

Dyer, J.H. and K. Nobeoka (2000). ‘Creating and managing a high-performance knowledge-sharing network: The Toyota case’, Strategic Management Journal, 21(3), pp. 345-367.

Easterby-Smith, M., M.A. Lyles and M.A. Peteraf, 2009. ‘Dynamic capabilities: Current

33

debates and future directions’. British Journal of Management, 20(s1), pp. S1–S8.

Edmondson, A.C. (1996). ‘Learning from mistakes is easier said than done: Group and organizational influences on the detection and correction of human error’. Journal of Applied Behavioral Science, 32(1), pp. 5–28.

Edmondson, A.C. (1999). ‘Psychological safety and learning behavior in work teams’, Administrative Science Quarterly, 44(2), pp. 350–383.

Edmondson, A.C. (2002). ‘The local and variegated nature of learning in organizations: A group-level perspective’, Organization Science, 13(2), pp. 128-146.

Edmondson, A.C. (2003). ‘Speaking Up in the Operating Room: How Team Leaders Promote Learning in Interdisciplinary Action Teams’, Journal of Management Studies, 40(6), pp. 1419-1452.

Edmondson, A.C., R. Bohmer and G.P. Pisano (2001). ‘Disrupted Routines: Team Learning and New Technology Adaptation’, Administrative Science Quarterly, 46(12), 685-716.

Eisenhardt, K.M. and J.A. Martin (2000). ‘Dynamic capabilities: what are they?’ Strategic Management Journal, 21(10-11), pp. 1105-1121.

Feldman, M.S. (2000). ‘Organizational Routines as a Source of Continuous Change’, Organization Science, 11(6), pp. 611-629.

Ferlie, E., J. Hartley and S. Martin (2003). ‘Changing public service organizations: Current perspectives and future prospects’, British Journal of Management, 14, pp. S1-S14.

Fottler, M.D. (1987). ‘Health care organizational performance: Present and future research’, Journal of Management, 13(2), pp. 367-391.

Froud, J., S. Johal, A. Leaver, R. Phillips and K. Williams (2009). ‘Stressed by choice: A business model analysis of the BBC’, British Journal of Management, 20(2), pp. 252-264.

Geroski, P.A., J. Van Reenen and C.F. Walters (1997). ‘How persistently do firms innovate?’ Research Policy, 26(1), pp. 33-48.

Gherardi, S. (2001). ‘From Organizational Learning to Practice-Based Knowing’, Human Relations, 54(1), pp. 131-139.

Gherardi, S. and D. Nicolini (2002). ‘Learning in a Constellation of Interconnected Practices: Canon or Dissonance?’ Journal of Management Studies, 39(4), pp. 419-436.

Gherardi, S. and D. Nicolini (2001). ‘The Sociological Foundation of Organizational Learning’, in Dierkes M., A. Berthoin Antal, J. Child. and I. Nonaka (eds.), Handbook of Organizational Learning. Oxford: Oxford University Press, pp. 35-60.

Goes, J.B. and S.H. Park (1997). ‘Interorganizational links and innovation: The case of hospital services’, Academy of Management Journal, 40(3), pp. 673-696.

Greve, H. (2003). ‘A behavioral theory of R&D expenditures and innovations: Evidence from shipbuilding’, Academy of Management Journal, 46(6), pp. 685-702.

Halaby, C.N. (2004). ‘Panel models in sociological research: Theory into practice’, Annual Review of Sociology, 30(1), pp. 507-544.

Han, J.K., N. Kim and R.K. Srivastava (1998). ‘Market orientation and organizational performance: Is innovation a missing link?’ Journal of Marketing, 62(4), pp. 30-45.

Hannan, M. and J. Freeman (1984). ‘Structural inertia and organizational change’, American Sociological Review, 49(2), pp. 149-164.

Hartley, J. (2005). ‘Innovation in governance and public services: Past and present’, Public Money & Management, 25(1), pp. 27-34.

Hartley, J., J. Benington and P. Binns (1997). ‘Researching the roles of internal-change agents in the management of organizational change’, British Journal of Management, 8(1), pp. 61–73.

34

Hartley, J., C. Donaldson, C. Skelcher and M. Wallace (2008). Managing to Improve Public Services. Cambridge: Cambridge University Press.

Harvey, G., C. Skelcher, E. Spencer, P. Jas and K. Walshe (2010). ‘Absorptive Capacity in a Non-Market Environment’, Public Management Review, 12(1), pp. 77-97.

Haunschild, P.R. and B.N. Sullivan (2002). ‘Learning from Complexity: Effects of prior accidents and incidents on airlines' learning’, Administrative Science Quarterly, 47, pp. 609-643.

Helfat, C.E. and M.A. Peteraf (2003). ‘The dynamic resource-based view: capability lifecycles’, Strategic Management Journal, 24(10), pp. 997-1010.

Howard-Grenville, J.A. (2005). ‘The Persistence of Flexible Organizational Routines: The Role of Agency and Organizational Context’, Organization Science, 16(6), pp. 618-636.

Jansen, J., F. Van Den Bosch and H. Volberda (2006). ‘Exploratory innovation, exploitative innovation, and performance: Effects of organizational antecedents and environmental moderators’, Management Science, 52(11), pp. 1661-1674.

Jarman, B., S. Gault, B. Alves, A. Hider, S. Dolan, A. Cook, B. Hurwitz and L. Iezzoni (1999). ‘Explaining differences in English hospital death rates using routinely collected data’, British Medical Journal, 318, pp. 1515-1520.

Jensen, M.G. (1986). ‘Agency costs of free cash flow, corporate finance, and takeovers’, American Economic Review, 76, pp. 323-329.

Jensen, M.G. (1993). ‘The modern industrial revolution, exit, and the failure of internal control systems’, Journal of Finance, 48, pp. 831-880.

Jensen, M.B., B. Johnson, E. Lorenz and B.A. Lundvall (2007). ‘Forms of knowledge and modes of innovation’, Research Policy, 36(5), pp. 680-693.

Kim, H., H. Kim and P.M. Lee (2008). ‘Ownership structure and the relationship between financial slack and R&D investments’, Organization Science, 19(3), pp. 404-418.

Kimberly, K.R. and M.J. Evanisco (1981). ‘Organizational innovation: The influence of individual, organizational, and contextual factors on hospital adoption of technological and administrative innovations’, Academy of Management Journal, 24(4), pp. 689-713.

Klein, K.J. and S.W.J. Kozlowski. (2000). ‘From micro to meso: Critical steps in conceptualizing and conducting multilevel research’, Organizational Research Methods, 3(3), pp. 211 -236.

Lave, J. and E. Wenger (1991). Situated Learning: Legitimate Peripheral Participation. Cambridge: Cambridge University Press.

Levitt, B. and J.G. March (1988). ‘Organisational learning’, Annual Review of Sociology, 14(3), pp. 319-340.

Macher, J.T. and D.C. Mowery (2009). ‘Measuring dynamic capabilities: Practices and performance in semiconductor manufacturing’, British Journal of Management, 20(s1), pp. S41–S62.

Madsen, P.M and V. Desai (2010). ‘Failing to learn? The effects of failure and success on organizational learning in the global orbital launch vehicle industry’, Academy of Management Journal, 53(3), pp. 451-476.

Magretta, J. (2002). ‘Why Business Models Matter’, Harvard Business Review, 80(5), pp. 86-92.

Malerba, F., L. Orsenigo and P. Peretto (1997). ‘Persistence of innovative activities, sectoral patterns of innovation and international technological specialization’, International Journal of Industrial Organization, 15(6), pp. 801-826.

35

Martin, S. and P.C. Smith (2005). ‘Multiple public service performance indicators: Toward an integrated statistical approach’, Journal of Public Administration Research and Theory, 15(4), pp. 599-613.

McCracken, M.J., T.F. McIlwain and M.D. Fottler (2001). ‘Measuring organizational performance in the hospital industry: An exploratory comparison of objective and subjective methods’, Health Services Management Research, 14(4), pp. 211-219.

McNulty, T. (2003). ‘Redesigning public services: Challenges of practice for policy’, British Journal of Management, 14(s1), pp. S31-S45.

Meier, K.J. and L.J. O'Toole (2003). ‘Public management and educational performance: The impact of managerial networking’, Public Administration Review, 63(6), pp. 689-699.

Meier, K.J., L.J. O’Toole, G.A. Boyne and R.M. Walker (2007). ‘Strategic management and the performance of public organizations: Testing venerable ideas against recent theories’, Journal of Public Administration Research and Theory, 17(3), pp. 357-377.

Metcalfe, J.S., A. James and A. Mina (2005). ‘Emergent innovation systems and the delivery of clinical services: The case of intra-ocular lenses’, Research Policy, 34(9), pp. 1283-1304.

Miner, A.S. and S.J. Mezias (1996). ‘Ugly duckling no more: Pasts and futures of organizational learning research’, Organization Science, 7(1), pp. 88-99.

Moynihan, D.P. and N. Landuyt (2009). ‘How do public organizations learn? Bridging cultural and structural perspectives’, Public Administration Review 69, pp. 1097-1105.

Nelson, R.R. and S.G. Winter (1982). An Evolutionary Theory of Economic Change. Boston: Harvard University Press.

Nembhard, I.M. and A.C. Edmondson (2006). ‘Making it safe: The effects of inclusiveness and professional status on psychological safety and improvement efforts in health care teams’, Journal of Organizational Behaviour, 27, pp. 941-966.

NESTA (2007). Hidden Innovation: How Innovation Happens in Six ‘Low Innovation’ Sectors. London: National Endowment for Science Technology and the Arts.

Nohria, N. and R. Gulati (1996). ‘Is slack good or bad for innovation?’ Academy of Management Journal, 39(5), pp. 1245-1264.

Oborn, E. and S. Dawson (2010). ‘Learning across communities of practice: An examination of multidisciplinary work’, British Journal of Management, 21(4), pp. 843-858.

Orr, J.E. (1996). Talking about Machines: An Ethnography of a Modern Job, Ithaca (NY): Cornell University Press.

O'Toole, L.J. and K.J. Meier (1999). ‘Modeling the impact of public management: Implications of structural context’, Journal of Public Administration Research and Theory 9(4), pp. 505-526.

Pablo, A.L., T. Reay, J.R. Dewald and A.L. Casebeer (2007). ‘Identifying, enabling and managing dynamic capabilities in the public sector’, Journal of Management Studies, 44(5), pp. 687-708.

Pavitt, K. (1984). ‘Sectoral patterns of technical change: towards a taxonomy and a theory’, Research Policy, 13, pp. 343-373.

Perrow, C. (1984). Normal Accidents: Living with High-Risk Technologies. New York: Basic Books.

Peters, B. (2009). ‘Persistence of innovation: Stylised facts and panel data evidence’, The Journal of Technology Transfer, 34(2), pp. 226-243.

Piening, E. (2011). ‘Insights into the process dynamics of innovation implementation’. Public Management Review, 13(1), pp. 127-157.

36

Pisano, G.P., R.M.J. Bohmer and A.C. Edmondson (2001). ‘Organizational differences in rates of learning: Evidence from the adoption of minimally invasive cardiac surgery’, Management Science, 47(6), pp. 752-768.

Provera, B., A. Montefusco and A. Canato (2010). ‘A 'no blame' approach to organizational learning’, British Journal of Management, 21(4), pp. 1057-1074.

Roberts, P.W. and G.R. Dowling (2002). ‘Corporate reputation and sustained superior financial performance’, Strategic Management Journal 23(12), pp. 1077-1093.

Roper, S. and N. Hewitt-Dundas (2008). ‘Innovation persistence: Survey and case-study evidence’, Research Policy, 37, pp. 149-162.

Rosenberg, N. (1982). Inside the Black Box: Technology and Economics. Cambridge: Cambridge University Press.

Salge, T.O. (2011). ‘A behavioral model of innovative search: Evidence from public hospital services’, Journal of Public Administration Research and Theory, 21(1), pp. 181-210.

Salge, T.O. and A. Vera (2009). ‘Hospital innovativeness and organizational performance: Evidence from English public acute care’, Health Care Management Review, 34(1), pp. 54-67.

Singh, J.V. (1986). ‘Performance, slack, and risk taking in organizational decision making’, Academy of Management Journal, 29(3), pp. 562-585.

Sitkin, S.B. (1992). ‘Learning through failure: The strategy of small losses’, Research in Organizational Behavior 14, pp. 231-266.

Sole, D. and A. Edmondson (2002). ‘Situated knowledge and learning in dispersed teams‘, British Journal of Management, 13(S2), pp. S17-S34.

Sydow, J., G. Schreyögg and J. Koch (2009). ‘Organizational path dependence: Opening the black box’, Academy of Management Review, 34(4), pp. 689-709.

Teece, D.J. (2010). ‘Business models, business strategy and innovation’, Long Range Planning, 43(2), pp. 172-194.

Teece, D.J. (2007). ‘Explicating dynamic capabilities: The nature and microfoundations of (sustainable) enterprise performance’, Strategic Management Journal, 28(13), pp. 1319-1350.

Teece, D.J., G. Pisano and A. Shuen (1997). ‘Dynamic capabilities and strategic management’, Strategic Management Journal, 18(7), pp. 509-533.

Tucker, A.L. and A.C. Edmondson (2003). ‘Why hospitals Don't Learn from Failures: Organizational and Psychological Dynamics that Inhibit System Change’. California Management Review, 45(2), p. 55-72.

Uotila, J., M. Maula, T. Keil and S. Zahra (2009). ‘Exploration, exploitation, and financial performance: Analysis of S&P 500 corporations’, Strategic Management Journal, 30(2), pp. 221-231.

Van Bruggen, G., G.L. Lilien and M. Kacker (2002). ‘Informants in Organizational Marketing Research: Why Use Multiple Informants and How to Aggregate Responses’, Journal of Marketing Research, 39, pp. 469-478.

von Hippel, E. (1976). ‘The dominant role of users in the scientific instrument innovation process’, Research Policy 5, pp. 212-239.

von Hippel, E. (1998) ‘Economics of product development by users: The impact of `sticky’ local information’, Management Science, 44(5), pp. 629-644.

Vigoda-Gadot, E., A. Shoham, N. Schwabsky and A. Ruvio (2008). ‘Public sector innovation for Europe: A multinational eight-country exploration of citizen's perspectives’, Public Administration, 86(2), pp. 307-329.

37

Voss, G.B., D. Sirdeshmukh and Z.G. Voss (2008). ‘The effects of slack resources and environmental threat on product exploration and exploitation’, Academy of Management Journal, 51(1), pp. 147-164.

Walker, R.M. and G. Enticott (2004). ‘Using multiple informants in public administration: Revisiting the managerial values and actions debate’, Journal of Public Administration Research and Theory, 14, pp. 417-434.

Wawro, G. (2002). ‘Estimating dynamic panel data models in political science’, Political Analysis, 10(1), pp. 25-48.

Weick, K.E and K.M Sutcliffe (2007). Managing the Unexpected: Resilient Performance in an Age of Uncertainty. 2. ed., San Francisco: Jossey-Bass.

Wenger, E. (1998). Communities of Practice: Learning, Meaning and Identity. Cambridge: Cambridge University Press.

Winter, S.G. (2003). ‘Understanding dynamic capabilities’, Strategic Management Journal, 24(10), pp. 991-995.

Wouters, M.., M. Jansen-Landheer and C. van de Velde (2010). ‘The quality of cancer care initiative in the Netherlands’, European Journal of Surgical Oncology, 36, pp. S3-S13.

Zahra, S. and G. George (2002). ‘Absorptive capacity: a review, reconceptualization and extension’, Academy of Management Review, 27(2), pp. 213-40.

Zahra, S., H. Sapienza and P. Davidsson (2006). ‘Entrepreneurship and Dynamic Capabilities: A review, model and research agenda’, Journal of Management Studies, 43(4), pp. 917-955.

Zhao, B. (2010). ‘Learning from errors: The role of context, emotion, and personality’, Journal of Organizational Behavior, doi: 10.1002/job.696.

Zhao, B. and F. Olivera (2006). ‘Error reporting in organizations’, Academy of Management Review, 31(4), pp. 1012-1030.

Zietsma, C., M. Winn, O. Branzei and I. Vertinsky (2002). ‘The war of the woods: Facilitators and impediments of organizational learning processes’, British Journal of Management, 13(S2), pp. S61-S74.

Zollo, M. and S.G. Winter (2002). ‘Deliberate learning and the evolution of dynamic capabilities’, Organization Science, 13(3), pp. 339-351.

Zott, C. (2003). ‘Dynamic capabilities and the emergence of intraindustry differential firm performance’, Strategic Management Journal, 24(2), pp. 97-125.

Zott, C. and R. Amit (2010). ‘Business model design: An activity system perspective’, Long Range Planning, 43(2-3), pp. 216-226.

38

FIGURES AND TABLES

Figure I: The Conceptual Model

 

 

 

Table 1: A Typology of Learning Modes

Practice-Based Learning

(DUI Mode)

Science-Based Learning

(STI Mode)

Knowledge Type implicit local

know-how know-who

explicit global

know-what know-why

Practice Integration strong weak

Thematic Scope wide narrow

Internal Ownership distributed concentrated

Organizational Context informal formal

Adapted from Jensen et al. (2007)

Past Practice-based

Learning Capabilities

Operational Slack

Relative Labor Intensity

of Business Model

Practice-based

Learning Capabilities

Organizational

Performance

H3 (+)

H4 (+)

H2 (∩)

H1 (+)

39

Table 2: Descriptive Statistics and Correlations

Variable Mean SD Min Max

1. Organizational performance 100.6 12.1 55.6 134.0

2. Practice-based learning 3.4 0.1 3.1 3.7 0.03

3. Relative labor intensity 26.6 13.1 9.1 163.0 -0.05 0.10

4. Operational slack 15.4 6.2 0.3 38.5 0.06 -0.04 0.06

5. Organizational size 820.6 382.1 250.1 2512.2 0.14 * -0.09 0.01 0.09

6. University affiliation 0.2 0.4 0.0 1.0 0.39 * 0.03 0.02 0.03 0.43 *

7. Foundation trust status 0.2 0.4 0.0 1.0 0.12 0.08 -0.01 0.10 0.06 0.10

8. Environmental competitiveness 15.6 2.3 11.0 21.1 0.08 -0.12 0.05 0.37 * 0.14 * -0.01 0.19 *

9. Environmental rurality 18.8 26.7 0.0 100.0 -0.03 -0.07 0.17 * -0.02 -0.31 * -0.27 * -0.17 * 0.01

Notes: N = 459 (153 organizations over three years (April 2004 to March 2007). * p < 0.01.

8.1. 2. 3. 4. 5. 6. 7.

Table 3: GMM Analyses Explaining Practice-based Learning Capabilities

Coeff. Coeff.

Control Variables

Intercept 3.7795 (0.1174) *** 2.2421 (0.2915) ***

Organizational size -0.0571 (0.0155) *** -0.0433 (0.0125) ***

University affiliation 0.0209 (0.0160) 0.0083 (0.0117)

Foundation trust status 0.0479 (0.0122) *** 0.0258 (0.0122) **

Environmental competitiveness -0.0023 (0.0012) * -0.0025 (0.0016)

Environmental rurality -0.0005 (0.0003) * -0.0005 (0.0002) **

Time dummies Yes *** Yes ***

Main Effects

Lagged practice-based learning H1 (+) 0.4347 (0.0709) ***

Operational slack H2 (+) 0.0018 (0.0021)

Operational slack squared H2 (-) -0.0000 (0.0001)

Observations 306 306

F-Statistic 8.42 *** 15.01 ***

Notes: Dynamic panel data estimation using one-step system General Method of Moments (GMM) with heteroscedasticity and autocorrelation consistent standard errors.* p < 0.10; ** p < 0.05; *** p < 0.01.

Model 2Model 1

S.E. S.E.

40

Table 4: GMM Analyses Explaining Organizational Performance

Coeff. Coeff. Coeff.

Control Variables

Intercept 55.6441 (13.6686) *** 49.7947 (13.6146) *** 49.6545 (13.6904) ***

Lagged performance 0.4408 (0.0998) *** 0.3957 (0.1039) *** 0.3851 (0.1044) ***

Organizational size 0.2679 (1.2377) 1.3900 (1.3995) 1.5494 (1.4024)

University affiliation 6.5440 (1.9224) *** 7.0350 (1.9295) *** 7.2875 (1.9162) ***

Foundation trust status 1.7538 (1.0364) * 1.2548 (1.0685) 1.1871 (1.0651)

Environmental competitiveness -0.1777 (0.1317) -0.1339 (0.1313) -0.1451 (0.1326)

Environmental rurality 0.0354 (0.020) * 0.0571 (0.021) *** 0.0581 (0.020) ***

Time dummies Yes *** Yes *** Yes ***

Main EffectsPractice-based learning H3 (+) 14.0711 (5.371) ** 16.3354 (5.266) ***

Relative labor intensity -0.1220 (0.057) ** -0.1628 (0.055) ***

Interaction Effect

Practice-based learning x Relative labor intensity

H4 (+) 1.0441 (0.479) **

Observations 306 306 306

F-Statistic 23.62 *** 20.64 *** 18.64 ***

Notes: Dynamic panel data estimation using one-step system General Method of Moments (GMM) with heteroscedasticity andautocorrelation consistent standard errors.* p < 0.10; ** p < 0.05; *** p < 0.01.

S.E.

Model 5Model 3 Model 4

S.E. S.E.