The Economics of Information,Communication, and Entertainment

The Impacts of Digital Technologyin the 21st Century

Series Editor

Darcy Gerbarg, New York NY, USA

For further volumes:http://www.springer.com/series/8276

James AllemanÁine Marie Patricia Ní-ShúilleabháinPaul N. RappoportEditors

Demand forCommunications Services -Insights and Perspectives

Essays in Honor of Lester D. Taylor

123

EditorsJames AllemanCollege of Engineering and Applied

ScienceUniversity of Colorado—BoulderBoulder, COUSA

Áine Marie Patricia Ní-ShúilleabháinColumbia Institute for Tele-InformationColumbia Business SchoolNew York, NYUSA

Paul N. RappoportDepartment of EconomicsTemple UniversityPhiladelphia, PAUSA

ISSN 1868-0453 ISSN 1868-0461 (electronic)ISBN 978-1-4614-7992-5 ISBN 978-1-4614-7993-2 (eBook)DOI 10.1007/978-1-4614-7993-2Springer New York Heidelberg Dordrecht London

Library of Congress Control Number: 2013951159

� Springer Science+Business Media New York 2014This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part ofthe material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission orinformation storage and retrieval, electronic adaptation, computer software, or by similar or dissimilarmethodology now known or hereafter developed. Exempted from this legal reservation are briefexcerpts in connection with reviews or scholarly analysis or material supplied specifically for thepurpose of being entered and executed on a computer system, for exclusive use by the purchaser of thework. Duplication of this publication or parts thereof is permitted only under the provisions ofthe Copyright Law of the Publisher’s location, in its current version, and permission for use mustalways be obtained from Springer. Permissions for use may be obtained through RightsLink at theCopyright Clearance Center. Violations are liable to prosecution under the respective Copyright Law.The use of general descriptive names, registered names, trademarks, service marks, etc. in thispublication does not imply, even in the absence of a specific statement, that such names are exemptfrom the relevant protective laws and regulations and therefore free for general use.While the advice and information in this book are believed to be true and accurate at the date ofpublication, neither the authors nor the editors nor the publisher can accept any legal responsibility forany errors or omissions that may be made. The publisher makes no warranty, express or implied, withrespect to the material contained herein.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Contents

Overview: The Future of Telecommunications,Media and Technology . . . . . . . . . . . . . . . . . . . . . . . . . . ix

Áine Marie Patricia Ní-Shúilleabháin, James Allemanand Paul N. Rappoport

Prologue I: Research Demands on Demand Research . . . . . . . . . . . . xvEli Noam

Prologue II: Lester Taylor’s Insights . . . . . . . . . . . . . . . . . . . . . . . xxvTimothy J. Tardiff and Daniel S. Levy

Part I Advances in Theory

1 Regression with a Two-Dimensional Dependent Variable . . . . . . . 3Lester D. Taylor

2 Piecewise Linear L1 Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . 17Kenneth O. Cogger

Part II Empirical Applications: Information and CommunicationTechnologies

3 ‘‘Over the Top:’’ Has Technological Change RadicallyAltered the Prospects for Traditional Media? . . . . . . . . . . . . . . . 33Robert W. Crandall

4 Forecasting Video Cord-Cutting: The Bypass of TraditionalPay Television . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Aniruddha Banerjee, Paul Rappoport and James Alleman

v

5 Blended Traditional and Virtual Seller Market Entryand Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83T. Randolph Beard, Gary Madden and Md. Shah Azam

6 How Important is the Media and Content Sectorto the European Economy? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113Ibrahim Kholilul Rohman and Erik Bohlin

7 Product Differences and E-Purchasing: An EmpiricalStudy in Spain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133Teresa Garín-Muñoz and Teodosio Pérez-Amaral

8 Forecasting the Demand for BusinessCommunications Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153Mohsen Hamoudia

9 Residential Demand for Wireless Telephony . . . . . . . . . . . . . . . . 171Donald J. Kridel

Part III Empirical Applications: Other Areas

10 Pricing and Maximizing Profits Within Corporations . . . . . . . . . 185Daniel S. Levy and Timothy J. Tardiff

11 Avalanche Forecasting: Using Bayesian Additive RegressionTrees (BART) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211Gail Blattenberger and Richard Fowles

Part IV Evidenced Based Policy Applications

12 Universal Rural Broadband: Economics and Policy . . . . . . . . . . . 231Bruce Egan

13 Who Values the Media? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255Scott J. Savage and Donald M. Waldman

14 A Systems Estimation Approach to Cost, Schedule,and Quantity Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273R. Bruce Williamson

vi Contents

Part V Conclusion

15 Fifty Years of Studying Economics . . . . . . . . . . . . . . . . . . . . . . . 291Lester D. Taylor

16 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305Áine M. P. Ní-Shúilleabháin, James Alleman and Paul N. Rappoport

Appendix: The Contribution of Lester D. TaylorUsing Bibliometrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307

Sharon G. Levin and Stanford L. Levin

Biographies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325

Contents vii

Overview: The Future of Telecommunications,Media, and Technology

Introduction

This book grew out of a conference organized by James Alleman and PaulRappoport, conducted on 10 October 2011 in Jackson Hole, Wyoming, in honor ofthe work of Lester D. Taylor. The conference lasted just one weekend, but thepapers are more durable.1

We begin with two Prologues; the first is written by Eli M. Noam. He focuseson demand analysis for media and communication firms. He notes that demandanalysis in the information sector must recognize the ‘public good’ characteristicsof media products and networks, while taking into account the effects ofinterdependent user behavior; the strong cross-elasticities in a market; as well asthe phenomenon of supply creating its own demand.

Noam identifies several challenges. The first involves privacy concerns (sincecompanies do not want to share their data). In addition, research and analyticaldata collection are falling behind. Third is the demonstrable lack of linkage ofeconomic with behavioral data. The fourth major hurdle consists in the lack ofbridges from the academic world of textbook theory involving consumer demandto the practical empirical world of media researchers.

The second Prologue by Timothy Tardiff and Daniel Levy focuses morenarrowly on Lester Taylor’s body of work, in particular its practical applicationsand usefulness in analyses of, and practices within, the ICT sector.

The remainder of this book is divided into four parts: Advances in Theory;Empirical Applications; Evidence-Based Policy Applications; and a final Conclu-sion. The contents of these Parts are discussed in detail below.

The book closes with an Appendix by Sharon Levin and Stanford Levindetailing the contributions of Professor Taylor using Bibliometrics.

Ordinarily, Festschrifts only have contributions by the honoree’s students andfollowers; however, this book is blessed by two contributions by Professor Taylor.

1 Thanks are due to the Columbia Institute for Tele-Information (CITI), as well as to theInternational Telecommunications Society (ITS) for sponsorship. Thanks are also due to MohsenHamoudia (for sponsorship on behalf of the International Institute of Forecasters (IIF)); Centris,and the authors for a riveting set of papers as summarized above, and as collected herein/below.

ix

The first is a seminal contribution to demand estimation when log–logtransformations are inappropriate because of the existence within the data set ofnegative values. The second, serving as part of the Conclusion, provides insightinto economics from Taylor’s fifty-plus years in the field.

Advances in Theory

Lester Taylor develops a procedure for dealing with variables which cannot betransformed into the traditional log–log in regression models. Rather, he suggestsrepresenting such dependent variables in polar coordinates. In this case, two-equation models can be specified with estimation proceeding in terms of functionsinvolving sines, cosines, and radius vectors. Taylor’s approach permits general-ization to higher dimensions, and can be applied in circumstances in which valuesof the dependent variable can be points in the complex plane.

Kenneth Cogger demonstrates how piecewise linear models may be estimatedwith the L1 criterion (which minimizes the sum of absolute errors, as distinct fromthe ordinary least squares (OLS) criterion). He introduces the Quantile Regressionprogram, using Mixed Integer Linear Programming. If an OLS program is desired,a Mixed Integer Quadratic Programming approach may prove useful.

Empirical Applications: Information and CommunicationTechnologies

Robert Crandall investigates the impact of recent changes in the Telecommuni-cation, Media, and Technology (TMT) sector on participants in the traditionalmedia sector. He focuses on empirical evidence on how changes in equipment andaccess to media have affected consumers’ time management. He examines theeffects of the profound changes taking place in the TMT sector on the economicprospects of the variety of firms engaged in traditional communications and media.Crandall demonstrates that while market participants may currently recognize thethreats posed to traditional media companies, the disruptions have been relativelymodest thus far—and have had little apparent effect on the financial market’sassessment of the future of media companies.

Aniruddha Banerjee, Paul N. Rappoport, and James Alleman report on efforts toforecast the effect of consumer choices on the future of video cord-cutting. Thischapter presents evidence on household ownership of OTT (Over the Top)-enabling devices as well as subscription to OTT services; this paper also forecaststhe effects of both phenomena upon traditional video. This chapter also examineshow consumers’ OTT choices are determined by household geo-demographiccharacteristics, device ownership, and subscription history. Ordered logit regres-sions are used to analyze and forecast future choices of devices and services, and

x Overview: The Future of Telecommunications, Media, and Technology

to estimate switching probabilities for OTT substitution by different consumerprofiles.

T. Randolph Beard, Gary Madden, and Md. Shah Azam focus upon analysis ofblended ‘bricks and clicks’ firms, as well as on virtual firms lacking any presenceoffline. Their study utilizes a data set of small Australian firms, and examines therelationship between the strategic motivation for entry, and the results of entry.Utilizing a bivariate ordered probit model with endogenous dummy variables, theendogeneity of firm strategic goals and implicit estimates of the parameters of thepost-entry business is analyzed. Their study finds that the goal of the firmmaterially affects subsequent performance: firms entering to expand their marketsize ordinarily succeed, but those entering to reduce costs do not. Blended firmsenjoy no strong advantages over pure online entrants.

Ibrahim Kholilul Rohman and Erik Bohlin focus upon the media and contentsub-segments of the Information and Communication Technology (ICT) sector—as defined by the OECD—and in tandem with the ISIC classification of these twocomponents. The media and content sector—by these definitions—consists of theprinting industry, motion pictures, video and television, music content, gamessoftware, and online content services. The authors aim to measure the contributionof the media and content sector in driving economic output in the Europeaneconomies, and to decompose the change of output into several sources. This paperaims to forecast the impact of a reduction in price on national GDP. The mainmethodology in this study is the Input–Output (IO) table. The study reveals thatthe price elasticity of media and content sectors to GDP is approximately 0.17 %.The impact varies across countries, but France, Sweden, and Norway were foundto have the higher elasticity coefficients. It found that price reductions mainlyaffects the financial sector, together with manufacturing of ICT products besidesthe media and content sectors themselves.

Teresa Garín-Muñoz and Teodosio Pérez-Amaral demonstrate key determi-nants of online shopping in Spain. They model how socio-demographic variables,attitudes, and beliefs toward internet shopping affect both the decision and usageof online shopping. In this chapter, three different samples of internet users aredefined as separate groups: those who purchase online (buyers); those who look forinformation online but purchase in stores (browsers); and those who do not shoponline at all (non-internet shoppers). Logit models are used to select the usefulfactors to assess the propensity to shop online.

Donald J. Kridel focuses upon residential demand for wireless telephony as itcontinues to grow at a dramatic rate. His chapter analyzes the residential demandfor wireless telephony using a sample of surveyed US households. Using a largedata set with over 20,000 observations, Kridel estimates a discrete-choice modelfor the demand for wireless telephony. Preliminary elasticity estimates indicatethat residential wireless demand is price-inelastic.

Daniel S. Levy and Timothy J. Tardiff focus upon pricing and maximizingprofits within corporations. Their chapter addresses examples of how particularbusinesses establish prices and improve profitability. Levy and Tardiff focus uponissues identified in Lester Taylor’s research such as theoretical and practical

Overview: The Future of Telecommunications, Media, and Technology xi

approaches to methodological issues including endogeneity of supply and demand;how cross-elasticities among products offered by the same company are addressedwhen improving profitability; how to deal with competing products in establishingprices; and how to define and measure costs.

Gail Blattenberger and Richard Fowles provide another focused example ofapplied empirical research. Whether or not an avalanche crosses a heavilytraveled, dangerous road to two popular ski resorts in Utah is modeled viaBayesian additive tree methods. Utilizing daily winter data from 1995 to 2010,their results demonstrate that using Bayesian tree analysis outperforms traditionalstatistical methods in terms of realized misclassification costs that account forasymmetric losses arising from Type 1 and Type 2 errors.

Mohsen Hamoudia focuses upon estimation and modeling of BusinessCommunications Services (BCS)—accounting for both demand and supply—theFrench market is used to illustrate the approach. This paper first provides a broadoverview of the demand for BCS: it briefly describes the scope of BCS productsand solutions, their structure, evolution, and key drivers. Then it presents thespecification and estimation of the demand and supply models and someillustrative forecasts.

Multi-equation models are employed, with demand variables (eight), supplyvariables (three), and independent variables (three). The models estimated includedummy variables to accommodate qualitative effects—notably market openness.Three-stage least squares—a combination of two-stage least squares andseemingly unrelated regression—is employed for estimation of supply anddemand for BCS, allowing for endogeneity of function arguments. The results ofthe analyses suggest that investigation of more complex lag structures than thoseemployed in this paper may be appropriate.

Evidence-Based Policy Applications

Bruce Egan addresses the matter of providing universal rural broadband service.His chapter discusses the efficacy of investments to date (focusing upon the case ofthe USA), and compares that with what might be achieved if coupled with rationalpolicies based upon economic welfare. Wyoming (USA) serves as a case study toillustrate the vast differences in what is achievable versus actual (or likely) results.Egan offers four main policy recommendations: eliminate entry barriers from ruralwaivers; reduce tariffs for network interconnection; target direct subsidies in atechnologically neutral fashion; and reform spectrum regulations.

Scott Savage and Donald M. Waldman examine consumer demand for theirlocal media environment (newspapers, radio, television, the internet, andSmartphone). Details regarding the purchasing habits of representative consumersare provided, distinguishing among men and women; different age groups;minorities and majorities; and their willingness to pay for community news,multiculturalism, advertising-free programs, and multiple valuation agendas.

xii Overview: The Future of Telecommunications, Media, and Technology

R. Bruce Williamson analyzes cost, schedule, and quantity outcomes in asample of major defense acquisition programs since 1965. A systems approach isapplied here with Seemingly Unrelated Regression (SUR), to explore cross-equation correlated error structures, and to compare results with those of single-equation models in analysis of defense acquisition programs. The author finds thatSUR coefficient estimates improve upon single equation regression results for costgrowth, schedule slippage, and quantity changes. These results are invoked topropose that SUR is useful for enhancing the quality and reliability of the weaponsacquisition program for the Department of Defense, USA.

Conclusion

Lester Taylor focuses as an economist upon a set of principal relationships thatcurrently face the US, and the world at large. He discusses important conceptsfrom the theory of consumer choice; unappreciated contributions of Keynes in theGeneral Theory; fluid capital and fallacies of composition; transfer problems; andlessons learned from the financial meltdown of 2007–2009.

Áine M. P. Ní-Shúilleabháin, James Alleman, and Paul N. Rappoport conclude,focusing upon future research: where we go from here.

Appendix

Sharon Levin and Stanford Levin provide a biography of Professor Taylor as well asfocusing on the extensive contributions of Professor Taylor using techniquesbeyond the tradition counting publications and citations to his influence toconsumer and telecommunications demand. As you can anticipate, his contributionsare extensive.

Áine Marie Patricia Ní-ShúilleabháinJames Alleman

Paul N. Rappoport

Overview: The Future of Telecommunications, Media, and Technology xiii

Prologue I: Research Demandson Demand Research

Overview

This should be the golden age of demand research. Many of the constraints of thepast have relaxed when it comes to data collection. Yet the methodologies ofdemand analysis created by thought leaders such as Lester Taylor (1972, 1974;Houthakker and Taylor 2010) have not grown at the same pace and are holdingback our understanding and power of prediction.

Demand research is, of course, highly important. On the macro-level, govern-ments and businesses need to know what to expect by way of aggregate or sectorialdemand, such as for housing or energy. On the micro-level, every firm wants toknow who its potential buyers are, their willingness to pay, their price sensitivity,what product features they value, and what they like about competing products(Holden and Nagle 2001).

Yet it is always difficult to determine demand. It is easy to graph a hypotheticalcurve or equation in the classroom but hard to determine the real world nature ofdemand and the factors that go into it.

Demand analysis is particularly important and difficult for media and commu-nications firms (Kim 2006; Burney et al. 2002; Green et al. 2002; Taylor andRappoport 1997; Taylor et al. 1972). They must grapple with high investment needsahead of demand, a rapid rate of change in markets and products, and an instabilityof user preferences. Demand analysis in the information sector must recognize the‘‘public good’’ characteristics of media products and networks, while taking intoaccount the effects of interdependent user behavior, the strong cross-elasticities in amarket, as well as the phenomenon of supply creating its own demand.

There is a continuous back-and-forth between explanations of whether ‘‘pow-erful suppliers’’ or ‘‘powerful users’’ determine demand in the media sector.Research in the social sciences has not resolved this question (Livingstone 1993).On one side of this debate is the ‘‘Nielsen Approach,’’ where the power is seen tolie with the audience. User preferences govern and it is up to the media companiesto satisfy these preferences (Stavitsky 1995). Demand creates supply. The otherside of the debate is the ‘‘Marketing’’ or ‘‘Madison Avenue Approach.’’ In thisview, the power to create and determine demand lies with the media

xv

communications firms themselves and the marketing messages they present(Bagdikian 2000). Supply creates demand.

In contrast to most other industries, demand measurement techniques affectfirms’ bottom lines, directly and instantly, and hence are part of public debate andcommercial disputes. When, for television, a transition from paper diaries toautomatic people meters took place in 1990, the effect of people meters on thebottom line was palpable. The introduction of people meters permanently loweredoverall TV ratings by an average of 4.5 points. Of the major networks, CBS lost2.0 points and NBC showed an average loss of 1.5 points. In New York City, Fox5, UPN 9, and WB 11 showed large drops. Ratings for cable channels showed again of almost 20 percent (Adams 1994). In 1990, each ratings point was worthapproximately $140 million/year. The decrease in ratings would have cost majornetworks between $400 and $500 million annually. Thus, demand analysis canhave an enormous impact on a business in the media and communications sector.

The forecasting of demand creates a variety of issues. One can divide theseproblems into two broad categories. ‘‘Type I Errors’’ exist when the wrong actionis taken. In medicine this is called a ‘‘false positive.’’ Human nature mercifullycontains eternal optimism but this also clouds the judgment of many demandforecasts. By a wide margin entrepreneurs overestimate the demand for productsrather than underestimate it. Eighty percent of films, music, recordings, and booksdo not break-even. Observe AT&T’s 1963 prediction that, ‘‘there will be 10million picture phones in use by US households in 1980,’’ yet in 1980 picturephones were little more than a novelty for a few and remained so for decades(Carey and Elton 2011). Another example of this type of error can be seen in theview summarized in 1998 by the Wall Street Journal that ‘‘The consensus forecastby media analysts is of 30 million satellite phone subscribers by 2006.’’ In actu-ality, their high cost and the concurrent advancements in terrestrial mobile net-works have relegated satellite phones to small niches only.

The second category of demand forecasting mistakes is a ‘‘Type II Error,’’when the correct action is not being taken. This is a ‘‘false negative.’’ There aremultiple historical examples in the media and communications sector. In 1939, theNew York Times reported that television could never compete with radio since itrequires families to stare into a screen. Thomas Watson, chairman of IBM, pro-claimed in 1943, ‘‘I think there is a world market for maybe five computers.’’Today there are two billion computers, not counting smartphones and other‘‘smart’’ devices. In 1977, Ken Olsen, President of the world’s number twocomputer firm Digital Equipment Corporation, stated, ‘‘There is no reason anyonewould want a computer in their home.’’ A 1981 McKinsey study for AT&Tforecast that there would only be 900,000 cell phones in use worldwide by the year2000. In reality, in that year there were almost one billion cell phones in use andthree billion in 2011.

xvi Prologue I: Research Demands on Demand Research

Major Stages of Demand Analysis

When it comes to demand analysis the two major stages are data collection anddata interpretation. Data used to be gathered in a leisurely fashion with interviews,surveys, focus groups, and test marketing. Television and radio audiences weretracked through paper diaries. This data sample was small yet expensive, unreli-able, and subject to manipulation. Four times a year, during the ‘‘sweeps’’ periods,the audiences of local stations were measured based on samples of 540 householdssubject to a barrage of the networks’ most attractive programs. Other media datawas collected through the self-reporting of sales. This is still practiced by news-papers and magazines, and is notoriously unreliable. For book best-seller lists,stores are sampled. This, too, has been subverted. Two marketing consultants spent30,000 dollars in purchases of their own book and were able to profitably propel itinto the bestseller list. For film consumption attendance figures are reportedweekly. According to the editor of a major entertainment magazine, these numbersare ‘‘made up—fabricated every week.’’ In parallel to these slow and unreliabledata collection methods, analytical tools were similarly time-insensitive: they werelengthy studies using methodologies that could not be done speedily. In fact, manyacademic demand models could never be applied realistically at all. They includedvariables and information that were just not available. And the methodologiesthemselves had shortcomings.

Estimation Models

The major methodological approaches to demand estimation include econometricmodeling, conjoint analysis, and diffusion models. Econometric estimations areusually created with economic variables for the elasticities for price, income, aswell as socio-demographic control variables (Cameron 2006). The price of sub-stitutes is often used. There are generic statistical problems to any econometricestimation, such as serial correlation, multicollinearity, homoscedasticity, lags, andexogeneity (Farrar and Glauber 1967). Moreover, predicting the future requires theassumption that behavior in the future is similar to behavior in the past.

One needs to choose and assume a specific mathematical model for the rela-tionship between price, sales, and the variables. If the specification is incorrect theresults will be misleading. Examples of this are several demand estimation modelsfor newsprint, the paper used by daily newspapers. This demand estimation is ofgreat importance to newspaper companies who need to know and plan for the priceof their main physical input. It is also of great importance to paper and forestrycompanies who must make long-term investments in tree farming.

Here is how the different models described the past and project the future(Hetemäki and Obersteiner 2002), and a comparison with subsequent reality. Onemodel is that of the United Nation’s Food and Agriculture Organization (FAO).

Prologue I: Research Demands on Demand Research xvii

Another is that of the Regional Planning Association (RPA). And there are sevenother models.

As one can see, the models, though using the same past data from 1970–1993,thereafter radically diverge and predict, for 2010, in a range from 16.4 million tonsto about 11 million tons. The gap widens by 2020 to over 130 %, making themessentially useless as a forecasting tool for decision makers in business and policy.On top of that, none of the models could predict the decline of newspapers due tothe internet, as shown by the ‘x’ markings on the graph. The actual figures were,for 2010, 5.4 million and for 2011, about 4.3 million—literally off the chart of theoriginal estimates. The worst of the predictions is the UN’s authoritative predictionwhich is a basic input into many countries’ policy making, as well as for globalconferences assessing pressure on resources.

Econometric models have also been employed by the film industry. They havetried to create some black-box demand models to aid in the green-lighting of filmprojects. Essentially, coefficients are estimated for a variety of variables such asgenres, e.g., science fiction; stars, e.g., Angelina Jolie; plot, e.g., happy endings,directors, e.g., Rob Reiner, and other factors. These models include the MotionPicture Intelligencer (MIP), MOVIEMOD, and others (Eliashberg et al. 2000;Wood 1997). Such models are proprietary and undisclosed, but even afteremploying them, films still bomb at an 80 % rate.

A second traditional empirical methodology for demand analysis has beenconjoint analysis (Green and Rao 1971; Allison et al. 1992). This method permitsthe researcher to identify the value (utility) that a consumer attaches to variousproduct attributes. The subject is asked to select among different bundles ofattributes and one measures the trade-off in the utility of such attributes. There isnot much theory in conjoint analysis, but it is a workable methodology.

Fig. 1 Forecasts for Newsprint Consumption in the US, 1995–2020—Various Models (Hetemakiand Obersteiner 2002)

xviii Prologue I: Research Demands on Demand Research

An example is an attribute-importance study for MP3 players. On a scale of1–10, people’s preference weights were found to be quality: 8.24; styling: 6.11;price: 2.67; user friendliness: 7.84; battery life: 4.20; and customer service: 5.66.

These weights enable the researcher to predict the price which the consumerwould pay for a product of various combinations of attributes (Nagle and Holden1995). There are computer packages that generate an optimal set of trade-offquestions and interpret results. But the accuracy of this technique is debatable.People rarely decompose a product by its features and are likely to be moreaffected by generic perspectives such as brand reputation or recommendations, notby feature trade-offs.

The third major empirical method for demand analysis is an epidemic diffusionmodel. Such a model is composed of a logistic function such as y(t) = 1/(1 + c e-kt).This technique, like the others, has its own inherent problems (Guo 2010). It isdifficult to find the acceleration point and the ‘‘saturation level.’’ Comparisons ofthe product are made and forecasted with some earlier product that is believed tohave been similar.

Thus, in the past, demand analysis was constrained by weak data and clunkyanalytical models. Recently, however, things have changed on the data collectionend. Data have ceased to be the constraint that it once was as more advancedcollection tools have emerged. First, there are now increasing ways to measurepeoples’ actual sensory perceptions to media content and to products more gen-erally. ‘‘Psycho-physiology’’ techniques measure heart rate (HR), brainwaves(electroencephalographic activity, EEG), skin perspiration (electrodermal activity,EDA), muscle reaction (electromyography, EMG), and breathing regularity(respiratory sinus arrhythmia, RSA) (see Ravaha 2000; Nacke et al. 2010). Thesetools can be used in conjunction with audience perception analyzers, which arehand-held devices linked to software and hardware that registers peoples’responses and their intensity.

Second, the technology of consumer surveying has also improved enormously.There are systems of automated and real-time metering. Radio and televisionlistening and channel-surfing can be followed in real-time. Measuring tools arecarried by consumers, such as the Passive People Meter (PPM) (Arbitron 2011;Maynard 2005). The TiVo Box and the cable box allow for instant gathering oflarge amounts of data. Music sales are automatically logged and registered; geo-graphic real-time data is collected for the use of the internet, mobile applications,and transactions (Roberts 2006; Cooley et al. 2002). Mobile Research, orM-Research, uses data gathered from cell phones for media measurement and canlink it to locations. Radio-frequency identification (RFID) chips can track productlocation (Weinstein 2005). Even more powerful is the matching of such data.Location, transaction, media consumption, and personal information can be cor-related in real-time (Lynch and Watt 1999). This allows, for example, the mea-surement in real-time of advertising effectiveness and content impact, and enablessophisticated pricing and program design.

Thus, looking ahead, demand data measurement will be increasingly real-time,global, composed of much larger samples, yet simultaneously more individualized.

Prologue I: Research Demands on Demand Research xix

This will allow for increasing accuracy in the matching of advertising, pricing, andconsumer behavior.

Of course, there are problems as data collection continues to improve (O’Leary1999). The first challenge is the coordination and integration of these data flows(Clark 2006). This is a practical issue (Deck and Wilson 2006). Companies areworking on solutions (Carter and Elliott 2009; Gordon 2007). Nielsen has laun-ched a data service (Gorman 2009), Nielsen DigitalPlus, which integrates set topbox data with People Meter data, transaction data from Nielsen Monitor Plus, retailand scanning information from AC Nielsen, and modeling and forecasting infor-mation from several databases (Claritas, Spectra, and Bases.) Nielsen intends toadd consumers’ activities on the internet and mobile devices into this mass of data.

The second challenge is that of privacy: the power of data collection has grownto an extent that it is widely perceived to be an intrusive threat (Clifton 2011;Matatov et al. 2010; Noam 1995). So there will be further legal constraints on datacollection, use, matching, retention, and dissemination.

The third problem is that when it comes to the use of these rich data streams,academic and analytical research are falling behind (Holbrook et al. 1986;Weinberg and Weiss 1986). When one looks at what economists in demandresearch do these days, judging from the articles’ citations, they still show littleconnectedness to other disciplines or to corporate demand research. There is aweak appreciation of the literatures of academic marketing studies, of informationscience on data mining (Cooley 2002), of the behavioral sciences (Ravaha et al.2008), of communications research (Zillman 1988; Vorderer et al. 2004), and evenin the recent work by behavioral economists (Camerer 2004). There is littleconnection to real-world demand applications—the work that Nielsen or Simmonsor the media research departments of networks do (Coffey 2001). Conversely, thework process of Nielsen and similar companies seems to be largely untouched bythe work of academic economists, which is damning to both sides.

The next challenge is therefore to create linkage of economic and behavioraldata. Right now there is no strong link of economic behavioral models andanalysis. Behavioral economics is in its infancy (Kahneman 2003, 2012), and itrelies mostly on individualized, traditional, slowpoke data methods of surveys andexperiments. The physiologists’ sensor-based data techniques, mentioned above,have yet to find a home in economic models or applied studies. There is also aneed to bridge the academic world of textbook theory of consumer demand withthe practical empirical work of media researchers.

Thus, the biggest challenge in moving demand studies forward is the creation ofnew research methodologies. The more powerful data collection tools will push,require, and enable the next generation of analytical tools. One should expect arenaissance in demand analysis. Until it arrives one should expect frustration.

In short: What we need today, again, is a Lester Taylor.

Eli Noam

xx Prologue I: Research Demands on Demand Research

References

Arbitron (2011) The portable people meter system. http://www.arbitron.com/portable_people_meters/thesystem_ppm.htm. Accessed 1 June 2011

Adams WJ (1994) Changes in ratings patterns for prime time before, during and after theintroduction of the people meter. J Media Econ 7(2):15–28

Allison N, Bauld S, Crane M, Frost L, Pilon T, Pinnell J, Srivastava R, Wittink D, Zandan P(1992) Conjoint analysis: a guide for Designing and Interpreting conjoint studies. Americanmarketing association, Chicago

Bagdikian BH (2000) Media monopoly. Beacon Press, BostonCamerer C (2004) Advances in behavioral economics. Princeton University Press, PrincetonCameron S (2006) Determinants of the demand for live entertainment: some survey-based

evidence. Econ Issues 11(2):51–64Carey J, Elton M (2010) When media are new: understanding the dynamics of new media

adoption and use. University of Michigan Press, MichiganCarey J, Elton M (2011) Forecasting demand for new consumer services: challenges and

alternatives. New infotainment technologies in the home: demand-side perspectives.Lawrence Erlbaum Associates, New Jersey, pp 35–57

Carter B, Stuart E (2009) Media companies seek rival for nielsen ratings. New York Times.http://www.citi.columbia.edu/B8210/cindex.htm. Accessed 8 June 2011

Clark D (2006) Ad measurement is going high tech. Wall Street Journal. http://www.utdallas.edu/*liebowit/emba/hightechmeter.htm. Accessed 1 June 2011

Clifton C (2011) Privacy-preserving data mining at 10: what’s next? Purdue University, WestLafayette. http://crpit.com/confpapers/CRPITV121Clifton.pdf

Coffey S (2001) Internet audience measurement: a practitioner’s view. J Interact Advert 1(2):13Cooley R, Deshpande M, Tan P (2002) Web usage mining: discovery and applications of usage

patterns from web data. SIGKDD Explor 18(2). http://www.sigkdd.org/explorations/issues/1-2-2000-01/srivastava.pdf. Accessed 2 June 2011

Deck CA, Wilson B (2006) Tracking customer search to price discriminate. Electronic Inquiry44(2):280–295

Eliashberg J, Jonker J, Sawhney MS, Wierenga B (2000) MOVIEMOD: an implementabledecision-support system for prerelease market evaluation of motion pictures. Mark Sci19(3):226–243

Farrar DE, Glauber, Robert R (1967) Multicollinearity in regression analysis: the problemrevisited. Rev Econ Stat 49(1):92–107

Gordon R (2007) ComScore, Nielsen report dissimilar numbers due to methodology differences.Newspaper association of America. http://www.naa.org/Resources/Articles/Digital-Edge-Pondering-Panels/Digital-Edge-Pondering-Panels.aspx. Accessed 1 June 2011

Gorman B (2009) Nielsen to have ‘‘Internet Meters’’ in place prior to 2010–2011 Seasons. TV bythe numbers. http://tvbythenumbers.com/2009/12/01/nielsen-to-have-internet-meters-in-place-prior-to-2010-11-season/34921

Guo JL (2010) S-curve networks and a new method for estimating degree distributions ofcomplex networks. Chin Phys B 19(12). http://iopscience.iop.org/1674-1056/19/12/120503/pdf/1674-1056_19_12_120503.pdf

Green J, McBurney P, Parsons S (2002) Forecasting market demand for new telecommunicationsservices: an introduction. Telematics Inform 19(3). http://www.sciencedirect.com/science/article/pii/S0736585301000041. Accessed 2 June 2011

Green PE, Rao VR (1971) Conjoint measurement for quantifying judgmental data. J Mark Res8(3):355–363

Hetemäki L, Obersteiner M (2002) US newsprint demand forecasts to 2020. University ofCalifornia, Berkeley. http://groups.haas.berkeley.edu/fcsuit/PDF-papers/LauriFisherPaper.pdf

Holden R, Nagle T (2001) The strategy and tactics of pricing: a guide to profitable decisionmaking, 3rd edn. Prentice Hall, New Jersey

Prologue I: Research Demands on Demand Research xxi

Holbrook MB, Lehmann DR, O’Shaughnessy J (1986) Using versus choosing: the relationship ofthe consumption experience to reasons for purchasing. Eur J Mark 20(8):49–62

Houthakker HS, Taylor LD (2010) Consumer demand in the United States, 1929–1970, Analysesand projections, 3rd edn. Harvard University Press, Cambridge

Kahneman D (2012) Thinking, fast and slow. Farrar, Straus, Giroux, New YorkKahneman D (2003) Maps of bounded rationality: psychology for behavioral economics. Am

Econ Rev. http://www.jstor.org/stable/3132137?seq=2Kim HG (2006) Traditional media audience measurement. Print and Broadcast MediaLivingstone SM (1993) The rise and fall of audience research: an old story with a new ending. Int

J Commun 43(4):5–12Lynch M, Watt JH (1999) Using the internet for audience and customer research. In: Malkinson

TJ ( ed.) Communication Jazz: Improvising the New International Communication Culture,IEEE International. 127 http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=799109.Accessed 1 June 2011

Maynard J (2005) Local people meters may mean sweeping changes on TV. The Washington Post.http://www.washingtonpost.com/wp-dyn/content/article/2005/04/27/AR2005042702324.html.Accessed June 2011

Matatov N, Oded M, Lior R (2010) Privacy-preserving data mining: a feature set partitioningapproach information sciences elsevier 2010. http://www.sciencedirect.com/science?_ob=MiamiImageURL&_cid=271625&_user=18704&_pii=S0020025510001234&_check=y&_origin=&_coverDate=15-Jul-2010&view=c&wchp=dGLbVlt-zSkzk&md5=960ee422099a0bd825604ce69812aa5e/1-s2.0-S0020025510001234-main.pdf

McBurney P, Parsons S, Green J (2002) Forecasting market demand for new telecommunicationsservices: an introduction. Telematics Inform 19(3):225–249

Nacke LE, Drachen A, Yannakakis G, Lee Pedersen A (2010) Correlation between heart rate,Electrodermal activity and player experience in first-person shooter games. Academia.edu.http://usask.academia.edu/LennartNacke/Papers/333839/Correlation_Between_Heart_Rate_Electrodermal_Activity_and_Player_Experience_In_First-Person_Shooter_Games. Accessed19 May 2011

Nagle TT, Holden RK (1995) The strategy and tactics of pricing: a guide to profitable decisionmaking, 2nd ed. Prentice Hall, Pearson, New Jersey

Noam E (1995) Privacy in telecommunications, Part III. New Telecommun Q 3(4):51–60O’Leary M (1999) Web measures wrestle with methodologies, each other. Online Magazine 23:

105–106Ravaha N (2004) Contributions of psychophysiology to media research: review and

recommendations. Media Psychol 6(2): 193–235Roberts JL (2006) How to count eyeballs on the web. Newsweek, New York, p. 27Stavitsky A (1995) Guys in suits with charts: audience research in U.S. public radio. J Broadcast

Electron Media 39(2):177–189Taylor LD, Weiserbs D (1972) On the estimation of dynamic demand functions. Rev Econ Stat

54(4):459–465Taylor LD, Rodney DA, Leonard W, Tsuang-Hua L, Michael DG, Copeland (1972) Telephone

communications in Canada: demand, production and investment decisions. Bell J EconManage Sci 3(1):175–219

Taylor LD (1974) On the dynamics of dynamic demand models. Recherches Économiques deLouvain/Louvain Economic Review 40(1): 21–31

Taylor LD, Rappoport PN (1997) Toll price elasticities estimated from a sample of US residentialtelephone bills. Inf Econ Policy 9(1):51–70

Vorderer P, Christoph K, Ritterfeld U (2004) Enjoyment: at the heart of media entertainment.Commun Theory 14(4):388–408

Weinstein R (2005) RFID: a technical overview and its application to the enterprise. A IEEEComputer Society. http://electricalandelectronics.org/wp-content/uploads/2008/11/01490473.pdf. Accessed 1 June 2011

xxii Prologue I: Research Demands on Demand Research

Wood D (1997) Can computer help hollywood pick hits? The Christian science monitor.http://www.citi.columbia.edu/B8210/cindex.htm. Accessed 7 June 2011

Weinberg C, Weiss D (1986) A simpler estimation procedure for a micromodeling approach tothe advertising-sales relationship. Mark Sci 5(3):269–272

Zillman D (1988) Communication, social cognition and affect. In Donhohew L, Howard S,ToryHiggens E (eds) Mood management: using entertainment to full advantage. Erlbaum,Hillsdale 147–171

Prologue I: Research Demands on Demand Research xxiii

Prologue II: Lester Taylor’s Insights

Introduction

Lester Taylor has been at the forefront of what we know about the theory andpractice of consumer demand, in general (Taylor and Houthakker 2010), andtelecommunications demand, in particular (Taylor 1994). Never content with mereabstract theories, Professor Taylor has applied a wide variety of techniques topractical problems that have evolved in unpredictable ways as technology andcompetition transform major sectors of our economy. His contributions are per-haps most widely recognized in the once pervasively regulated telecommunica-tions industry,1 where he ‘‘wrote the book’’ about what is known about consumerdemand for the products and services. And when technology and competitiontransformed that industry, he identified a research agenda to advance our knowl-edge of consumer demand.

In his 1994 update to an earlier comprehensive survey of telecommunicationsdemand and elasticity findings, Professor Taylor identified specific gaps thatresearch needed to address in order for firms to accommodate the changes that wereemerging and accelerating. Those gaps have been increasing because investing innew technologies to maintain a competitive edge requires greater understanding ofconsumer behavior than when there was little competition and products andservices were well established. Perhaps his most ironic observation was that whilethe gap between what we know and what businesses needed to know was widening,information about consumer behavior was becoming increasingly harder to find,because (1) companies were more likely to consider such information as a pro-prietary competitive advantage and (2) the once highly talented and large group ofeconomists and demand specialists that companies assembled during the regulatedmonopoly era had become somewhat of a luxury—subject to cost-cutting—as

1 Some of Professor Taylor’s elasticity estimates are being used to this day. For example, JerryHausman (2011) used the classic market elasticity of -0.72 for long-distance services to estimatea consumer benefit of approximately $5 billion that would result from lower interconnectioncharges for wireline long-distance calls.

xxv

these companies faced less regulatory scrutiny and began to prepare for morecompetition.

Despite the fact that the public knowledge of demand and the human resourcesproducing that knowledge have undoubtedly continued to shrink, Professor Tay-lor’s prescription for what is needed remains valid today.2 Specifically, with theshift in focus from justifying prices and investment decisions before industryregulators to competing successfully with new products, services, and technologiesin a largely deregulated world, analytical requirements have in turn shifted fromdeveloping industry elasticities for generally undifferentiated products and ser-vices to developing firm-specific own and cross-price elasticities for increasinglydifferentiated products.3 And because such demand information would be mostuseful to the extent that it can inform pricing, investment planning, marketing, andbudgeting decisions, it is most effectively used in conjunction with complementaryinformation on product and service costs—a combination that would facilitate theevaluation of the short-term and long-term profitability of product introduction andpricing actions. Professor Taylor’s prescient advice from 1994 is still well worthheeding both by formerly- and never-regulated telecommunications firms (Taylor1994, p. 270 (footnote omitted)).

In truth, some slimming-down is probably in order, but to allow demand analysis tolanguish because of the advent of competition and incentive regulation would be a mis-take. The challenge for demand analysis in the telephone companies in the years ahead isto forge links with marketing departments and to become integrated into company bud-geting and forecasting processes. Applied demand analysis has a strategic role to play in acompetitive environment, ranging from the conventional types of elasticity estimation intraditional markets to the identification of new markets. The possibilities are vast. It onlyrequires imagination, hard work—and some humility—on the part of economists anddemand analysts.

Professor Taylor’s Telecommunication Demand Findings

Professor Taylor initially summarized what was known about telecommunicationsdemand in a seminal 1980 book (Taylor 1980) and then updated that work 14 yearslater (Taylor 1994). The first book described research findings for an industry thatwas regulated by federal and state authorities in the United States and by theircounterparts internationally. By the time of the publication of the second book,competition had taken root in some segments of the industry—most prominentlyfor long distance services and telecommunications equipment as a result of thedivestiture of the Bell System in 1984. While competition was beginning to

2 See, in particular Taylor (1994), Chap. 11.3 Taylor (1994), pp. 266–270. In addition to the fact that data to develop firm-specific elasticitiesis often proprietary, successful estimation introduces the additional challenge of accounting forthe price responses of competing firms.

xxvi Prologue II: Lester Taylor’s Insights

emerge in other segments,4 the primary focus on publicly available demandinformation continued to reflect the concerns of a regulated industry: how wouldvolumes of services offered by ‘‘dominant’’ providers change if regulatorsapproved price changes; and how would these volume changes affect the revenues,costs, and profitability of those regulated companies?

Accordingly, Professor Taylor identified trends and regularities in the market(or industry) price elasticities (and income elasticities) for clearly delineated andwell-understood services. Indeed, he took special notice of the stability in long-distance price elasticities over the period between the publication of his two books,including the empirical regularity that customers tended to be more price-sensitive(with elasticities higher in absolute value) as the distance of the call becamelonger.

Professor Taylor also cataloged gaps in our knowledge, which by 1994 werewidening as competition grew, technologies advanced, and the formerly distinctdata, video, and voice industries converged. In partial recognition of these trends,large segments of the industry were further deregulated in the U.S. by the 1996Telecommunications Act. These trends foreshadowed the facts that (1) priceswould increasingly be the result of market outcomes, rather than regulatorymandates; (2) businesses would have to introduce new products and services,rather than continue to offer a stable portfolio of well-recognized products; and (3)those products would compete with those of other providers deploying sometimessimilar, but other times different (e.g., wireless) technologies. In technical terms,Professor Taylor recognized that the tide was shifting from a predominantemphasis on market own-price elasticities to (1) the need to understand cross-elasticities among the complementary and substitute products offered by the samefirm and (2) how the firm elasticity, not the industry, or market elasticity wasparamount as formerly ‘‘dominant’’ firms lost share to new rivals. And with regardto the need to identify and measure firm elasticities, he reminded us that it is notonly the price the firm is considering, but also the reactions of other firms to thatprice that affects consumer demand.

Despite the shift in emphasis that competitive and technological trends haverequired, there have been fundamental features of Professor Taylor’s approach thatmake it timely, if not timeless. In particular, (1) he seeks to provide a theoretically-sound underpinning to the empirical regularities he identifies and (2) he takesadvantage of advances in data sources and analytical techniques in providing anever-sounder underpinning for his results. Therefore, while the industry trends henoted may have changed some well-established empirical regularities that at onetime seemed quite solid, his research approaches and agenda retain their power.

4 For example, large businesses were taking advantage of alternatives that ‘‘bypassed’’ the localphone companies and certain services offered by these companies, e.g., short-haul long-distancecalls were not the ‘‘natural monopoly’’ services that the Bell System divestiture presumed.

Prologue II: Lester Taylor’s Insights xxvii

For example, Professor Taylor observed that the magnitude of industry elas-ticities for various categories of toll calls had been reasonably stable and that theabsolute value of the long-distance elasticity increases with the distance of thecall.5 However, since that time the former distinctions between types of calls (suchas long-distance) have blurred as both traditional telecommunications companiesand newer firms such as wireless providers offer pricing plans that make no dis-tinction between what were once viewed as clearly different types of calls, e.g.,local and long-distance. Indeed, long-distance calling—a key factor in both thedivestiture of the old Bell System in 1984 and the structure and provisions of the1996 Telecommunications Act—has ceased to exist as a stand-alone industry. Thisis a result of the acquisition of the legacy AT&T and MCI – the two largestcompanies in that industry, by the current AT&T (formerly SBC) and Verizon byearly 2006. Consequently, replicating the studies that produced such findingswould be extremely difficult (if even possible). Whether the previous relationshipbetween price sensitivity and the distance of a call would persist appeared to beproblematic.

And yet, despite these changes in industry structure and product offerings,Professor Taylor’s explanation of why the empirical regularity was observedremains relevant. According to this explanation, the reason for the lower pricesensitivity for calls of shorter distance was the likelihood that proportionately moreof these calls were related to economic activities, such as work and shopping, andhence less avoidable if price were to increase. In contrast, for calls of longerdistances, a relatively greater proportion were not economic in nature, e.g., calls tofriends and relatives, as those ‘‘communities of interest’’ tended to span greaterdistances than calls related to economic activities. In other words, the empiricalregularity had a fundamental explanation—it was the distribution of the types ofcalls that spanned particular distances, rather than distance itself, that produced theconsistent findings Professor Taylor noted. Therefore, while technological devel-opments—particularly the widespread use of the internet for all types of com-mercial and personal interactions—has most likely changed how calls of differenttypes are distributed by distance, the differential price sensitivity of different typesof calls has likely persisted.6

5 Taylor (1994) p. 260. ‘‘All considered, this probably represents the best-established empiricalregularity in telecommunications demand.’’.6 Professor Taylor’s quest for fundamental explanations of empirical regularities in consumerbehavior and demand was even deeper in the book he co-authored with Houthakker. In this work,the authors presented a framework for explaining consumption, income, and time expendituresbased on underlying brain functions. See Taylor and Houthakker (2010), Chapter 2.

xxviii Prologue II: Lester Taylor’s Insights

While Professor Taylor’s contributions to the theory, and perhaps even moreimportantly, to the practice of demand analysis are most prominent in telecom-munications, his advice on how demand practitioners could maintain and enhancetheir relevance at the outset of the technological and competitive transition in thatindustry cuts across all efforts to use demand and price optimization tools toimprove companies’ performances.

Timothy J. TardiffDaniel S. Levy

References

Hausman J (2011) Consumer Benefits of Lower Intercarrier Compensation Rates, Attachment 4to the universal service and intercarrier compensation proposal filed at the FederalCommunications Commission by AT&T, CenturyLink, FairPoint, Frontier, Verizon, andWindstream. http://fjallfoss.fcc.gov/ecfs/document/view?id=7021698772. Accessed July 29,2011

Taylor LD (1980) Telecommunications demand: a survey and critique. Ballinger, CambridgeTaylor LD (1994) Telecommunications demand in theory and practice. Kluwer, BostonTaylor LD, Houthakker HS (2010) Consumer demand in the United States: prices, income, and

consumption behavior. 3rd ed. Springer, New York

Prologue II: Lester Taylor’s Insights xxix