price discrimination4

7/24/2019 price discrimination4

1/20

89Journal of Marketing

Vol. 71 (July 2007), 89107

2007, American Marketing Association

ISSN: 0022-2429 (print), 1547-7185 (electronic)

Siva Viswanathan, Jason Kuruzovich, Sanjay Gosain, & Ritu Agarwal

Online Infomediaries and PriceDiscrimination: Evidence from the

Automotive Retailing SectorThis article focuses on a novel mechanism for market segmentation and price discrimination based on consumersuse of online infomediaries. Using the automotive retailing context as the setting for the study, the authors addressthe following question: Can online infomediaries serve as a viable mechanism for market segmentation and pricediscrimination? They draw on a unique and extensive data set of consumers who report on the information theyfound when using online buying services (OBS) as part of their new vehicle purchase process.The analysis of thedata set shows that consumers who obtain price information pay lower prices (for the same product), whereasconsumers who obtain product information pay higher prices. Although this points to the existence of distinctconsumer segments, this knowledge is of limited value without a viable mechanism that enables firms to identifyand target these customer segments specifically. On the basis of consumer usage patterns of OBS, the authorsuncover distinct OBS clusters and empirically demonstrate that the use of these different clusters is associatedwith predicted differences in consumer outcomes. They also show that the differential use of OBS clusters issystematically related to underlying consumer characteristics. They discuss the relevance of the findings for

automobile dealers and manufacturers and for other industries in which online infomediaries have established asignificant presence.

Siva Viswanathan is Assistant Professor of Information Systems (e-mail:[email protected]), and Ritu Agarwal is Professor and RobertH. Smith Deans Chair of Information Systems (e-mail: [email protected]), Decision and Information Technologies Department, Robert H.Smith School of Business, University of Maryland. Jason Kuruzovich isAssistant Professor of Management Information Systems, Lally School ofManagement and Technology, Rensselaer Polytechnic Institute (e-mail:[email protected]). Sanjay Gosain is a Web strategy analyst, Capital GroupCompanies Inc. (e-mail: [email protected]). The authors acknowledgethe valuable industry insights and feedback provided by Scott Weitzmann

and Dennis Galbraith. The authors are grateful to J.D. Power and Associ-ates for providing the data. They thank the Center for Electronic Marketsand Enterprises at the Robert H. Smith School of Business, University forMaryland, for financial support. They also thank seminar participants atCarnegie Mellon University, University of Connecticut, Ohio State Univer-sity, University of Texas at Dallas, and University of Maryland for theircomments on previous versions of this article. P. Rajan Varadarajanserved as guest editor for this article.

To read and contribute to reader and author dialogue on JM, visithttp://www.marketingpower.com/jmblog.

Researchers and practitioners acknowledge that effec-tive market segmentation is crucial for price dis-crimination and can play a vital role in a firms prof-

itability and survival (e.g., Bolton and Myers 2003).Traditionally, firms have attempted to target different cate-gories of consumers through product versioning, couponsand rebates, bundling, and quantity discounts. A primaryobjective of these strategies is to identify consumers or

groups of consumers with different price elasticities toenable greater surplus extraction. Over the years, simplersegmentation strategies based on demographic, lifestyle,

and socioeconomic variables have been superseded by moresophisticated benefit- and need-based segmentation. Withthe increase in the variety of marketing channels, firms haveattempted to exploit differences across channels to segmentconsumers, with the underlying logic that customers self-select into channels that differ in their costs of time, travel,shopping, and so forth (Anderson, Day, and Rangan 1997).

Recently, a few studies have examined differences in

consumer behavior across online and offline channels. Hittand Frei (2002) study online banking consumers and findspecific unobservable characteristics that make online con-sumers better customers than their offline counterparts.Scott Morton, Zettelmeyer, and Silva-Risso (2001a) findthat consumers with lower negotiation skills are more likelyto use online intermediaries, and Zettelmeyer (2000) exam-ines how firms can exploit the differences between tradi-tional and online channels to segment consumers by con-trolling the amount of information made available throughalternative channels.

Although differences across online and offline channelscan be useful in segmenting consumers, in several sectors,

such as automotive retailing, brokerages, mortgages, andinsurance, the Web is rapidly emerging as a primary sourceof information. Much of this information is provided byneutral third-party online infomediaries that have estab-lished themselves as pivotal and trustworthy informationsources. Thus, as consumers increasingly rely on the Web tosatisfy their information needs, the distinction betweenonline and offline channels as predictors of differences inconsumer characteristics is becoming moot. Acknowledg-ing the growth of the Web, researchers have begun to focuson its role in customer segmentation. For example, a few


2/20

90 / Journal of Marketing, July 2007

studies have attempted to segment online consumers on thebasis of observable information-seeking behaviors, such assites visited and time spent. Forsyth, Lavoie, and McGuire(2000) find that simplifiers, who look for readily availableproduct information, are the most profitable segment andaccount for 50% of all online transactions. In contrast, bar-gainers are primarily concerned with finding a good priceand maintaining control over their transactions. A relatedstudy by Rozanski, Bollman, and Lipman (2001, p. 3) iden-tifies occasionalization, or how online consumers behave

during different Internet sessions, as a good predictor ofconsumer segments. Although these studies do not offertangible guidelines for firms to exploit underlying differ-ences in online behaviors, in light of the growing impor-tance of the Internet as an information source, they never-theless highlight the need to understand the nature of onlinesegmentation strategies that can complement existing strate-gies in conventional channels.

This article focuses on a novel mechanism for marketsegmentation and price discrimination based on consumersuse of online infomediaries. Online infomediariesalsoknown as online buying services (OBS)such as AutobytelInc., LendingTree, and InsWeb Inc., have established them-

selves as pivotal players in the value chains of multiple sec-tors, including automobiles, financial services, insurance,and real estate. The array of services these firms provide,ranging from simple information provisioning to the facili-tation and brokering of transactions with other companies,is well documented. What is not addressed in the literature,however, is the potential role of online infomediaries inmarket segmentation and price discrimination. To the extentthat consumers differ in their information-seeking behaviorsand online infomediaries differ in the types of informationthey provide, focusing on consumers use of these infome-diaries offers a rich and novel opportunity for understand-ing online market segmentation through consumer self-

selection.Using the automotive retailing context as the setting forour study, we address the following question: Can onlineinfomediaries serve as a viable mechanism for market seg-mentation and price discrimination? We begin with a simpleanalytical model that investigates the implications of pro-viding different types of information on the prices con-sumers paid; this model motivates our empirical analyses.We draw on a unique and extensive data set of more than16,000 consumers who obtained price and product-relatedinformation from online information sources in their newvehicle purchase process to test the propositions that followfrom our analytical model. These initial findings reinforcethe results of the analytical model. They indicate the exis-

tence of consumer segments that not only seek differenttypes of information but also pay different prices for thesame product, a likely reflection of their underlying pricesensitivities. However, the key question that remains unan-swered is how to identify and target these customer seg-ments to develop actionable marketing strategies.

On the basis of consumers OBS usage patterns, weidentify distinct clusters of OBS and show that use of theseclusters, each of which provides a different mix of informa-tion, leads to significant differences in consumer outcomes.

We also find that the observed behavioral choices (i.e., con-sumers use of OBS clusters) are related to underlying dif-ferences in consumer characteristics. These findings helpestablish a crucial linkage between consumers usage of dif-ferent types of online infomediaries and the prices they payfor identical purchases. This mapping provides valuableinsights that can potentially help sellers segment consumerson the basis of their OBS usage patterns. We discuss therelevance of our findings for retailers that obtain referralsfrom these different online infomediaries, as well as for

manufacturers.This study makes several important contributions. It is

among the first to characterize the information provisionstrategies of online automotive retailing infomediaries andto investigate their potential as a mechanism for traditionalfirms to implement segmentation and price discriminationstrategies. Whereas prior research has largely ignored dis-tinctions among different online infomediaries, we showthat there are theoretically meaningful and empiricallyrobust clusters of OBS firms that are clearly demarcated interms of consumer usage patterns and further differentiatedon the type of information provided. A key implication isthat as long as the cost of providing different categories of

information is low, online infomediaries can serve not onlyas an effective but also as an efficient market segmentationand price discrimination mechanism for dealers. Our find-ings also have some notable implications for dealers part-nerships with online infomediaries. In contrast to Chen,Iyer, and Padmanabhan (2002), who show that an exclusivereferral arrangement between an online infomediary andone of the many competing traditional dealers in a givengeographical area is optimal, our study suggests that whenchoosing among several competing online infomediaries, atraditional dealer can benefit from using these differentcategories of infomediaries as complementary referralmechanisms.

We organize the rest of the article as follows: The nextsection describes the different types of infomediaries andexamines the impact of their information provision strate-gies on consumer outcomes. We then provide a briefdescription of the online automotive retailing landscape anddescribe the data and measures used in the study. Followingthe preliminary analyses, we empirically identify distinctclusters of online infomediaries based on consumer usagepatterns and present the predictions related to the use ofthese clusters on consumer outcomes along with the associ-ated empirical analysis and results. We then discuss thefindings, their managerial implications, and the limitationsof our study. The final section contains concluding remarks.

Online Infomediaries and TheirImpacts

Online infomediaries position themselves by making strate-gic choices about the type of information they will provide.Two varieties of infomediaries have been identified by pre-vious studiesthose that focus primarily on price and thosethat also provide significant product-related information.With respect to price information, some prior research hasexamined outcomes associated with price-comparison


3/20

Online Infomediaries and Price Discrimination / 91

1In a study of automotive retailing infomediaries, Scott Morton,Zettelmeyer, and Silva-Risso (2001b) find that consumers who useonline automotive retailing infomediaries save approximately 2%($450) compared with those who negotiate directly with dealers.In another study of the life insurance sector, Brown and Goolsbee(2002) find that price-comparison Web sites led to a 8%15% dropin term life prices, and though price dispersion increased initially,it decreased with increased Internet usage.

engines, or shopbots. For example, Baye and Morgan(2001) find that establishing a market for price informationleads to more competitive pricing by firms. A subsequentstudy by Baye, Morgan, and Scholten (2003) finds that con-sumers using a price infomediary (Shopper.com) save anaverage of 16% on purchases compared with nonusers. Inan analysis of consumer choice data from another priceinfomediary, Dealtime.com, Smith and Brynjolfsson (2001)find similar results of lower average prices, but their find-ings suggest that retailers use several strategic options to

mitigate price pressures. Studies of automotive retailinginfomediaries (Chen, Iyer, and Padmanabhan 2002) uncoversimilar dynamics; findings show that online referral inter-mediaries lead to greater price competition among tradi-tional dealers and improve consumer welfare by transfer-ring surplus from competing dealers to online consumers.As might be expected, most of these studies focus on theimplications of price transparency enabled by these infome-diaries,1 and their findings echo the conventional wisdomthat electronic markets lead to lower average prices andimproved efficiency.

Although the implications of price information avail-ability are relatively straightforward, the impact of product

information on consumer welfare is more complex. Forexample, studies by Alba and colleagues (1997) and Lynchand Ariely (2000) find that reducing product-related searchcosts can increase consumers willingness to pay. Studies ofinformation provided by certification intermediaries(Lizzeri 1999) and expert reviewers (Eliashberg and Shugan1997) also suggest that such information can have a positiveimpact on sellers revenues. Grossman and Shapiro (1984)find that informative advertising about product characteris-tics can lead to greater substitutability between horizontallydifferentiated products and lower prices. Thus, dependingon the specifics of the context (e.g., Von der Fehr and Stevik1998) and, more important, on whether information accen-

tuates or attenuates the differences, the availability of prod-uct information could lead to higher or lower prices for con-sumers. Given the focus of infomediaries on providing priceand product information, it is important to understand theimpact of each of these information categories on consumeroutcomes. We begin our investigation with a stylized ana-lytical model that explores the ramifications of differenttypes of information made available to consumers.

Impact of Price-Related Information

Consumers typically seek information on multiple productattributes during the purchase process. In the case of high-involvement and highly differentiated products, such ascars, comparative price (rather than product) informationcan be difficult to obtain. Consider a situation in which allconsumers have full access to information on the attributes

2This simplifying assumption helps maintain analyticaltractability and enables us to capture the effects of product infor-mation availability in the case of complex and infrequently pur-chased products with real differentiation.

of competing products but not all consumers have access toprice information. To examine the impact of an increase inthe availability of price information, consider the extremecase in which there is no price information and all con-sumers are uninformed about the relative prices of compet-ing offerings. Furthermore, consumers are unable to inferprices from information about the product offerings. Lack-ing additional information about relative prices that firmscharge, all uninformed consumers have symmetric expecta-tions about the firms prices. In other words, uninformed

consumers have no a priori expectations that one sellercharges a higher/lower price than the other. Because con-sumers have product-related information, lacking informa-tion about relative prices, they purchase the product thatoffers the best fit. Consequently, each firm acts as a monop-olist in its market because no consumer switches firms.

Let a fraction of consumers have costless access to theprices charged by the two firms, whereas the rest (1 ) areuninformed about firms prices. It is easy to comprehendthe role of increasing access to price information. Whensome consumers are informed of prices (i.e., > 0), in mak-ing their choices, the informed segment accounts for notonly the relative fit of the two products but also their rela-

tive prices. With increasing availability of price informa-tion, increases, and firms increasingly compete on pricesfor these informed consumers. Thus:

P1: As consumers obtain price information about competingofferings, the average price they pay is lower inequilibrium.

Impact of Product-Related Information

Third-party infomediaries typically provide informationabout the characteristics of competing products and enableconsumers to compare competing offerings. Unlike com-parison advertising, which attempts to reduce the value ofcompeting brands relative to the advertised brand (Aluf and

Shy 2001), information provided by third-party infomedi-aries might not always increase demand for a single firm;rather, such information enables consumers to learn abouttheir personal fit with a product. Consequently, suchunbiased information enables some consumers to learn thata particular sellers offering is not suited to their tastes andothers to realize that it is. More precisely, product informa-tion in this setting enables the population of consumers todiscover their true underlying distribution of preferences,leading to better fit. Thus, the impact of product informa-tion provided by neutral third-party infomediaries leads toan increase in the variance (represented by 1/ in the prooffor P2 in Appendix A) of the distribution of consumer pref-

erences, whereas the means of the altered distributionsremain the same; that is, the distribution of consumer pref-erences ex ante and ex post differs in mean-preservingspreads.2

The assumption of increasing variance (increasing 1/)in consumer preferences resulting from product-related


4/20


3As we noted previously, product information could have differ-ent impacts depending on the product category and purchase con-text. In the case of products with perceived (rather than real) dif-ferentiation, such as branded versus generic drugs, productinformation can reduce dispersion of consumer preferences asconsumers become aware of the lack of any real differentiationbetween the competing products (see Johnson and Myatt 2004).

information is particularly suited to complex and infre-quently purchased products, such as automobiles, for whichconsumers are unaware a priori of their true preferences onseveral dimensions that are critical to their purchase deci-sion. Specifically, for the case of products with real differ-entiation, providing information about the characteristics ofthe competing products is more likely to increase dispersionof consumer preferences.3 In other words, a more hetero-geneous distribution (higher values of 1/) is obtained byrelocating consumers from the center of a homogeneous

distribution (smaller values of 1/) toward the tails. As con-sumers become more sensitive to the differences among thecompeting products (i.e., as they move closer to one of thecompeting offerings), the market becomes less competitive.Thus, given the availability of price information, providingcomparative product information about competing offeringshas the same effect as increasing product differentiation,leading to greater market power (higher prices) for the com-peting firms.

P2: As consumers obtain product-related information, theaverage price they pay is higher in equilibrium.

In summary, we assert that differential availability of

price and product information is systematically related tothe prices consumers pay. Appendix A describes a simpleanalytical model that captures the impacts of increasingprice and product information and contains the proofs forP1 and P2. As a next step, we test these broad propositions,grounding our preliminary analyses in the specific contextof automotive retailing. This preliminary analysis forms thebasis for contextualized hypotheses that yield implicationsfor segmentation strategies linked to infomediary usage.

Research Context and DataDescription

Research ContextEstimated at approximately $1 trillion a year, automotiveretailing is the biggest retailing sector in the United States.Dominated by more than 22,600 new car dealerships, theautomotive retailing sector is unique in several respects.There is no national brand, the 10 largest dealers enjoy lessthan a 6% market share, and sales are largely localized. Thepresence of franchise laws serves to strengthen the positionof dealers in the retail value chain. Unlike commodityitems, significant differences exist among cars in terms ofperformance and features as well as in the pricing strategiesof dealers, making the car-buying process a complex deci-sion for most consumers. The need for finance, insurance,

warranties, and spare parts and service, among other issues,adds to this complexity. Given that search and comparison

4According to consumer surveys (J.D. Power and Associates2002), 62% of all new vehicle shoppers research their purchasesonline, and the average consumer visits approximately seven auto-motive sites before buying.

5In addition to OBS, dealers and manufacturers have establishedtheir own Web sites. However, unlike the independent online info-mediaries, most have failed to establish a significant presenceonline.

shopping are costly for consumers, dealers typically resortto aggressive sales tactics to close deals with consumerswho visit their dealerships.

With the advent of the Web, several online intermedi-aries have emerged to improve overall market efficiencythrough better information availability.4 Independent OBS,the pioneers of online automotive retailing, perform severalfunctions. Primary among such functions is to provide analternative channel for dealers to acquire customers, or toact as a referral intermediary. In addition to their role as

referral intermediaries, OBS also serve as infomediaries byproviding information related to various facets of the pur-chase. For example, OBS, such as Autobytel, facilitate side-by-side comparisons of different makes and models, pro-vide pricing information, and offer financing and insuranceservices to consumers, none of which are easily available inthe traditional setting. The number of online automotiveretailing infomediaries has risen significantly in the pastfive years, and given the plethora of services and informa-tion that various online intermediaries offer, there appearsto be some differentiation in their value proposition for con-sumers or for dealers.5

Data and MeasuresWe use an extensive secondary data set constructed from asurvey of new vehicle purchasers conducted by a leadingmarket research organization. This data set has several dis-tinguishing features. It is representative of the overall U.S.automobile consumer population and captures consumerdemographics, psychographics, and online informationsources that consumers use in their purchase process. Theresults we report herein reflect data collected for20032004 new vehicle purchases. In this market researchsurvey, the sample was randomly selected on the basis ofregistration data, and the sampling process was designedwith quota sampling and a sales-weighting scheme to

reflect the overall market for new cars accurately. Two ver-sions of the questionnaire were used with a different orderof response elicitation to check for bias due to respondentfatigue or programmed responses. A response rate ofapproximately 24% was achieved for 116,317 surveysmailed out. The data set consists of both traditional con-sumers and consumers who used the Internet as part of theirpurchase process. We restrict our analyses here to con-sumers who reported using Internet information sources toaid their automobile purchase process. In addition, to con-trol for vehicle fixed effects adequately, the analysis isbased on the 140 most popular vehicles purchased. Of a


5/20


total sample of 26,361 consumers, 16,188 reported using 1or more of 30 different online infomediaries (OBS) as partof their automobile purchase process.

Consumers answered questions related to their usage ofspecific online OBS, information found online, price paid(excluding tax, license, and trade-in fees), vehicle choice,and satisfaction. They indicated which sites they had visitedfrom a list of the top 30 automobile-related OBS andanswered a set of questions about the specific type of infor-mation they found online. Of these, five items represented

price-related information (e.g., information about rebatesand special offers), and three items represented product-related information (e.g., reliability ratings of vehicles). Themeasures indicate good convergent validity (price = .83,product = .84, see Appendix B, Panel 1) and discriminantvalidity for product and price dimensions (for factor struc-ture, see Appendix B, Panel 3).

Individual characteristics used as controls in the analy-sis include consumer demographics, psychographics, Inter-net usage, and technical competence measures. Consumerdemographics measures included race, gender, age, educa-tion, and income. Price sensitivity and involvement with theproduct were two key psychographic dimensions (related to

consumers utility for information) that captured the valueconsumers place on prices and product choice, respectively(Laurent and Kapferer 1985; Nunnally and Bernstein 1994;Zettelmeyer, Scott Morton, and Silva-Risso 2004). Thesewere measured through items that consumers indicated toreflect their values. Involvement was assessed throughresponses to items such as I want a vehicle that stands outfrom the crowd, and price sensitivity was measuredthrough responses to items such as I will shop as manydealers as it takes to get the absolute lowest price. Mea-sures of psychographics traits of involvement and price sen-sitivity indicate convergent validity (involvement = .77,price = .62) and discriminant validity (for factor structure,

see Appendix B, Panel 2). The pairwise correlation matrixappears in Appendix C. Descriptive statistics and a list of allmeasures appear in Appendix B.

Preliminary Empirical AnalysisOur preliminary analysis tests P1 and P2 by examining therelationship between the type of information the consumerobtained and the resulting price he or she paid, controllingfor individual (demographics, psychographics, Internetusage, and competence) and vehicle characteristics. Weaccount for the possibility that the information obtained isitself driven by consumer characteristics through structuralestimation. We used the three-stage least squares systems

estimator to derive the parameters of the full systembecause we used endogenous variables (information found)in the first two equations of the model as explanatoryvariables for price paid. Furthermore, there is a possibilityof correlation among error terms across regression equa-tions because the same unobserved variables may affectboth product and price information retrieval. We control forvehicle characteristics by coding make, model, and triminformation using the vehicle identification number. Thespecific system of equations we tested is as follows:

(1) Found price information = 01

+ 0414(Individual controls) + ,

(2) Found product information = 01

+ 0414(Individual controls) + , and

(3) Price paid = 01 + 02(Found price information)

+ 03(Found product information)

+ 0414(Individual controls)

+ 15154(Vehicle controls) + .

As Table 1 shows, the results support P1 and P2. Onaverage, finding more price information results in the con-sumer paying a significantly lower price ( = .427, p 1, the higher the value of , the more

V X[ ]( )

.=+

1

4 2 1

E X[ ] ,=1

2

f xx x dx

B( )

( )

[ , ],=

1 11

( ) *A p t

Fp p

t

fp p

t

A

B A

B A

2 22

1

2

2

1

2

=

+

+

=

+

+

,

*

and

p t

Fp p

t

f p p

t

B

B A

B A

2

12

1

2

2

1

2

.

A AB A

B BB A

p Fp p

t

p F p pt

=

+

= +

2

1

2

12

, and

112

.


16/20


peaked is the distribution, and the more homogeneous arethe consumer preferences. With identical firms and a sym-metric distribution of consumers, the optimal prices are

where

Specifically, for a symmetric beta distribution, we have

When consumers have little product-related informa-tion, they are indifferent to the competing offerings andmake their purchase decisions based solely on the relativeprices. This can be represented by a distribution with lowvariance (high ), with consumers being clustered towardthe mean. Because consumers obtain more product-related

p p tA B* *

[ ].= =

+

2

41

2

F

x x dx

x x dx

1

2

1

1

11

0

1

2

11

0

=

( )

( )

11

1 1

1

1

2

1

2

1

2

1

=

, and f

dx

x

xx dx( ) 1

0

1.

p p t

F

fA B* * ,= =

2

1

2

1

2

information, their preferences become well defined, andthey discover that their preferences are more closely alignedwith one of the competing offerings. In other words, thevariance of the distribution of consumer preferencesincreases as consumers discover their true preferences.Thus, increasing product information causes an increase inthe variance (lower ) of the distribution of consumer pref-erences. Figure A1 illustrates how the optimal prices varywith.

FIGURE A1

Optimal Prices with Product Information


17/20


APPENDIX B

Descriptive Statistics and Measures

Panel 3: Factor Analysis for Information Found Dimensions

Found PriceInformation

Found ProductInformation

Dealer cost/invoice of new vehicles .739 .016Options and features information (e.g., descriptions, prices) .527 .076Tool to show how much dealers are discounting vehicles .658 .078Tool to calculate manufacturer suggested retail price with the options you want .644 .015

Information about rebates and special offers .656 .055Road test and reviews about vehicles by automotive writers .096 .706Safety information (e.g., crash test results) .002 .830Reliability ratings of vehicles .018 .872

Notes: The main results are indicated in bold.

Panel 2: Factor Analysis for Consumer Psychographic Dimensions

Involvement Price Sensitivity

I want a vehicle that stands out from the crowd. .856 .014What you drive says a lot about you. .832 .027Getting the lowest price is more important to me than finding a dealer that provides

customer service. .005 .677I will shop as many dealers as it takes to get the absolute lowest price. .037 .754I would gladly travel another 50 miles to buy from a dealer that could save me an

additional $300..030 .664

Panel 1: Construct Operationalization

Metaconstruct Constructs M SD Operationalized as

Consumerpsychographics

Involvement(Cronbachs = .77)

3.26 1.15 Average of agreement (five-point scale) withstatements for items corresponding to factors

Price sensitivity(Cronbachs = .62)

2.82 1.05

Consumerdemographics

RaceGender

.121.38

.33

.490 = white, and 1 = nonwhite

1 = male, and 2 = femaleAge 56.12 13.27 Year of birth

Education 5.93 1.57 Eight-point scaleIncome 8.27 3.71 Fifteen-point scale

Information found Found product information(Cronbachs = .84)

.60 .41 Average of found information online (dichotomousscale) for items corresponding to each factor

Found price information(Cronbachs = .83)

.69 .29

Searchcompetency andeffort

Internet expertise

Internet usage

2.20

6.69

.65

6.50

Overall level of Internet experience (three-pointscale)

Hours per week

Outcomes Vehicle price paid 2.93 1.29 Total price (excluding tax, license, trade-in)/$10thousandSatisfaction with search

process6.76 2.10 Overall experience using the Internet to

research/shop for vehicle (ten-point scale)Impact on vehicle choice 2.08 .767 How much of an impact did your Internet research

have on what make/model to purchase/lease?(three-point scale)

Extent of physical dealervisits

1.46 1.06 How many physical dealers visited to look atalternate vehicles?


18/20


APPENDIXC

PairwiseCorrelationMatrix

1

2

3

4

5

6

7

8

9

10

11

12

13

1

4

15

16

17

1.Race

2.Gender

.02

3.Age

.13

.17

4.Education

.01

.09

.06

5.Income

.07

.18

.14

.36

6.Firstvehicle

.06

.02

.15

.00

.11

7.Technicalcompetence

.03

.11

.22

.16

.12

.04

8.Internetusage

.04

.05

.00

.00

.03

.02

.27

9.Involvement

.02

.01

.14

.00

.11

.02

.07

.03

10.Pricesensitivity

.09

.01

.12

.08

.17

.04

.03

.05

.03

11.PortalOBSclusterusage

.07

.01

.02

.02

.05

.02

.01

.05

.01

.06

12.Foundpriceinformation

.01

.07

.02

.09

.07

.01

.15

.07

.03

.06

.03

13.Foundvehicleinformation

.02

.02

.07

.10

.06

.00

.18

.09

.02

.01

.09

.49

14.ProductOBSclusterusage

.03

.03

.04

.08

.06

.01

.13

.07

.02

.00

.13

.12

.27

15.PriceOBSclusterusage

.04

.10

.06

.10

.08

.00

.17

.08

.01

.06

.05

.26

.20

.18

16.Purchaseprice

.01

.13

.14

.17

.53

.06

.03

.01

.22

.16

.04

.02

.01

.01

.02

17.Vehiclechoiceimpact

.07

.00

.16

.07

.01

.04

.17

.11

.05

.04

.05

.21

.26

.17

.14

.08

18.Satisfaction

.06

.01

.17

.04

.02

.04

.23

.14

.06

.07

.02

.32

.26

.10

.21

.01

.41


19/20


Alba, Joseph, John Lynch, Barton Weitz, Chris Janiszewski,Richard Lutz, Alan Sawyer, and Stacy Wood (1997), Inter-active Home Shopping: Consumer, Retailer, and ManufacturerIncentives to Participate in Electronic Marketplaces, Journalof Marketing, 61 (July), 3853.

Aluf, Yana and Ozy Shy (2001), Comparison Advertising andCompetition, working paper, Department of Economics, Uni-versity of Haifa, Israel.

Anderson, Erin, George S. Day, and V. Kasturi Rangan (1997),Strategic Channel Design, Sloan Management Review, 38(Summer), 5969.

Ansari, Asim, Nicholas Economides, and Avijit Ghosh (1994),Competitive Positioning with Nonuniform Preferences,Mar-keting Science, 13 (Summer), 24873.

Ayres, Ian and Peter Siegelman (1995), Race and Gender Dis-crimination in Bargaining for a New Car,American Economic

Review, 85 (June), 304321.Bagnoli, Mark and Ted Bergstrom (2004), Log-Concave Proba-

bility and Its Applications, Economic Theory, 26 (August),44569.

Baye, Michael R. and John Morgan (2001), Information Gate-keepers on the Internet and the Competitiveness of Homoge-neous Product Markers, American Economic Review, 91(June), 45475.

, , and Patrick Scholten (2003), The Value ofInformation in an Online Consumer Electronics Market,Jour-nal of Public Policy & Marketing, 22 (Spring), 1725.

Belkin, Nicholas J. and W. Bruce Croft (1992), Information Fil-tering and Information Retrieval: Two Sides of the Same Coin,Communications of the ACM, 35 (December), 2938.

Bolton, Ruth N. and Mathew B. Myers (2003), Price-BasedGlobal Market Segmentation for Services,Journal of Market-ing, 67 (July), 108128.

Boudette Neal E. (2005), Test Drives Get a New Spin; to WooBuyers, Auto Makers Stage Elaborate Trial Runs, The WallStreet Journal, (February 3), B1.

Brown, Jeffrey and Austan Goolsbee (2002), Does the InternetMake Markets More Competitive? Evidence from the LifeInsurance Industry,Journal of Political Economy, 110 (June),

481507.Chen, Yuxin, Ganesh Iyer, and V. Padmanabhan (2002), Referral

Infomediaries, Marketing Science, 21 (Fall), 41234.Clemons, Eric K., Il Horn Hann, and Lorin M. Hitt (2002), Price

Dispersion and Differentiation in Online Travel: An EmpiricalInvestigation,Management Science, 48 (April), 53449.

Economides, Nicholas (1989), Symmetric Equilibrium Existenceand Optimality in Differentiated Products Markets,Journal of

Economic Theory, 47 (1), 17894.Eliashberg, Jehoshua and Steven M. Shugan (1997), Film Critics:

Influencers or Predictors? Journal of Marketing, 61 (April),6878.

Forsyth, John E., Johanne Lavoie, and Tim McGuire (2000),Managing Expectations for Value, The McKinsey Quarterly,4, 1219.

Furse, David H., Girish N. Punj, and David W. Stewart (1984), ATypology of Individual Search Strategies Among Purchasers ofNew Automobiles, Journal of Consumer Research, 10(March), 41731.

Grossman, Gene M. and Carl Shapiro (1984), Informative Adver-tising with Differentiated Products, The Review of EconomicStudies, 51 (January), 6381.

Hitt, Lorin M. and Francis X. Frei (2002), Do Better CustomersUtilize Electronic Distribution Channels? The Case of PCBanking,Management Science, 48 (June), 73248.

J.D. Power and Associates (2002), New Autoshopper.com Study,management report, Agoura Hills, Calif.

Jobson, J. Dave (1992), Applied Multivariate Data Analysis. NewYork: Springer-Verlag.

Johnson, Justin P. and David P. Myatt (2004), On the Simple Eco-nomics of Advertising, Marketing, and Product Design,Oxford University Economics Discussion Paper No. 185.

Katz, Michael and Carl Shapiro (1985), Network Externalities,Competition and Compatibility, American Economic Review,75 (3), 42440.

Ketchen, David J., Jr., and Christopher L. Shook (1996), TheApplication of Cluster Analysis in Strategic ManagementResearch: An Analysis and Critique, Strategic Management

Journal, 17 (June), 44159.Laurent, Gilles and Jean-Noel Kapferer (1985), Measuring Con-

sumer Involvement Profiles, Journal of Marketing Research,22 (February), 4253.

Lizzeri, Alessandro (1999), Information Revelations and Certifi-cation Intermediaries, Rand Journal of Economics, 30 (Sum-mer), 21431.

Lynch, John G., Jr., and Dan Ariely (2000), Wine Online: SearchCosts Affect Competition on Price, Quality, and Distribution,

Marketing Science, 19 (Winter), 83104.Moorthy, K. Sridhar (1984), Market Segmentation, Self-Selec-

tion, and Product Line Design, Marketing Science, 4 (Fall),288307.

Nunnally, Jum C. and Ira H. Bernstein (1994), PsychometricTheory. New York: McGraw-Hill.

Olshavsky, R.W. and W. Wymer (1995), The Desire for NewInformation from External Sources, in Proceedings of theSociety for Consumer Psychology, S. Mackenzie and R. Stay-man, eds. Bloomington, IN: Printmaster, 1727.

Podsakoff, Philip M. and Dennis W. Organ (1986), Self-Reportsin Organizational Research: Problems and Prospects,Journalof Management, 12 (Winter), 53144.

Rozanski, Horacio D., Gerry Bollman, and Martin Lipman (2001),Seize the Occasion: The Seven-Segment System for OnlineMarketing, Strategy and Business, 24 (3), 4253.

Scott Morton, Fiona M., Florian Zettelmeyer, and Jorge Silva-Risso (2001a), Consumer Information and Discrimination:Does the Internet Affect the Pricing of New Cars to Women

and Minorities? Quantitative Marketing and Economics, 1(March), 6592.

, , and (2001b), Internet Car Retailing,Journal of Industrial Economics, 69 (December), 501519.

Sinha, Indrajit and Wayne S. DeSarbo (1998), An IntegratedApproach Toward the Spatial Modeling of Perceived CustomerValue,Journal of Marketing Research, 35 (May), 23649.

Sireci, Stephen G., Frederic R. Robin, and Thanos Patelis (1999),Using Cluster Analysis to Facilitate Standard Setting,

Applied Measurement in Education, 12 (3), 301323.Smith, Michael D. and Erik Brynjolfsson (2001), Consumer

Decision-Making at an Internet Shopbot: Brand Still Matters,Journal of Industrial Economics, 49 (December), 54158.

Von der Fehr, Nils-Henrik M. and Kristin Stevik (1998), Persua-sive Advertising and Product Differentiation, Southern Eco-

nomic Journal, 65 (July), 11326.Ward, Joe H., Jr. (1963), Hierarchical Grouping to Optimize an

Objective Function,Journal of the American Statistical Asso-ciation, 58 (March), 23644.

Zettelmeyer, Florian (2000), Expanding to the Internet: Pricingand Communications Strategies when Firms Compete on Mul-tiple Channels, Journal of Marketing Research, 37 (August),292309.

, Fiona M. Scott Morton, and Jorge Silva-Risso (2004),Cowboys or Cowards: Why Are Internet Car Prices Lower?working paper, Haas School of Business, University of Califor-nia at Berkeley.

REFERENCES


20/20

price discrimination4

Documents