The advent of the Internet and the on-line book trade: the role of managerial cognition and the liabilities of experience.

Weinstein, Marc ; Standifird, Stephen S.

INTRODUCTION

Few technological innovations have created as much popular furor as the commercialization of the Internet. The boom in IPOs, twenty-something year-old millionaires, and venture capital icons not only touched the public imagination but also prompted many business schools to hastily revamp their curricula to sate the demand for new courses for the New Economy. The 2000 crash of the NASDAQ, the wave of paper riches to paper rag stories of young entrepreneurs, and the collapse of the IPO market vindicated the skeptical who had argued that old economic rules still applied in the New Economy. Underpinning this lively discourse are more fundamental questions: Does the Internet constitute a radical technological discontinuity? If so, will firms with more industry experience be able to act upon changes in the economic environment resulting from the advent of the Internet?

This paper addresses these issues by extending and testing theoretical work in managerial cognition to the study of the relationship between length of industry experience and organizational decision making during periods of technological discontinuity. Specifically, we explore how the mental models of managers contributed to an industry-wide dominant logic guiding the behavior of experienced proprietors in the used and out-of-print book trade. This research finds that the logic that served the proprietors of these small firms so well during periods of stability may prevent them from recognizing, evaluating, and responding to fundamental changes in the competitive environment.

In what follows, we review theoretical and empirical research contributing to our understanding of the role of cognition and its relationship to industry experience in organizational decision-making during periods of disruptive impact on the market. We then extend this analysis to the study of the used and out-of-print book trade by demonstrating that the Internet constitutes a disruptive impact on the retail book market that has undermined the validity of the dominant logic on which experienced proprietors have relied. This leads to the formulation of a number of hypotheses concerning the relationship between years of experience and organizational practices and the impact of these practices on sales. These hypotheses are tested in the subsequent section of the paper where we analyze an original data set that matches recently collected survey responses from on-line booksellers with independently collected archived sales and inventory data made available from the company processing the credit card transactions and electronic inventories of these same firms.

THEORY

There are a number of assumptions underpinning the study of managerial cognition that highlight the importance of decision processes of individual mangers. The first is that a large part of what managers contribute to organizational performance is their ability to filter, sort, absorb, and process complex and ambiguous information (Mintzberg, Raisinghani and Theoret, 1976; McCall and Kaplan, 1985; Starbuck and Milliken, 1988). Interchangeably referred to as knowledge constructs (Walsh, 1995), mental templates (Neisser, 1976; El Sawy and Pauchant, 1986), frames of reference (March and Simon, 1958; Shivastava and Mitroff, 1983; Dunn and Ginsberg, 1986), cognitive maps (Axelrod, 1976; Bougon, Weick, and Binkhorst, 1977; Weick and Bougon, 1986; Barr, Stimpert, and Huff, 1992; Calori, Johson, and Sarnin, 1992), and schemata (Fiske and Dyer, 1985), these mental models help managers interpret reality and formulate strategic responses. Second, the mental models of managers not only help managers determine what information is relevant but these models also guide action (Nisbett and Ross, 1980). Third, managers are assumed to have numerous cognitive limitations that inhibit their ability to filter this information (Miller, 1956; March and Simon, 1958). Thus, to process information efficiently, managers rely upon mental models which simultaneously filter "irrelevant data," and provide a template to analyze the data made salient by the knowledge structure. For the purposes of this paper, we follow Walsh (1995) and refer to knowledge structures to describe the theoretical frameworks that individuals use to process information in a deductive, top-down process (Ablelson and Black, 1986).

Knowledge structures are an essential part of a deductive, theory-driven approach to information processing necessitated by cognitive constraints of individuals and the abundance of information sources and data. Knowledge structures speed problem solving (Mischel, 1981) and help managers evaluate conflicting information (Gioia, 1986), leading to efficiencies in problem solving (Thorngate, 1980). Where knowledge structures can provide efficiencies during relatively static periods, they can lead to a number of errors during period of rapid change.

To the extent that a particular knowledge structure is shared throughout an organization, managers may come to reply upon a "dominant logic" (Bettis and Prahalad, 1995; Prahalad and Bettis, 1986). A dominant logic exists when a particular knowledge structure is found throughout the organization and thus, "governs decision-making processes throughout the firm by producing a mind-set or world view that can be shared across all of its business units" (Lampel and Shamsie 2000: 593-594). Although dominant logic is similar to the notion of routines developed through organizational learning (Levitt and March, 1988), it is distinct from organizational learning in that the logic permeates the entire organization. Whereas organizational learning can occur in specific reservoirs within the organization (Argote, McEvily & Regans, 2003), a dominant logic exists as a result of a unifying view of the organization and its associated behaviors (Bettis and Prahalad, 1995; Prahalad and Bettis, 1986). As a widely accepted framework for understanding organizational behavior, a dominant logic is not susceptible to problems of knowledge transfer and depreciation (Argote, McEvily & Regans, 2003; Darr, Argote and Epple, 1995). Instead, the dominant logic becomes a unifying understanding of how the organization should function when faced with specific problems of a repeat nature (Lampel and Shamsie 2000).

During periods of technological discontinuity decision-making premised on a dominant logic can lead to two types of errors. First, basic shifts in the competitive environment can result in managers discarding important data as irrelevant. Second, even when new data are considered, the dominant logic can lead to a misinterpretation of the meaning of this new information. Firms make sense of information collected through a self-referring process where prior behavior is used as a framework for understanding new information (Lampel and Shamsie, 2000; Cote, Langley and Pasquero, 1999; von Krogh and Roos, 1996). Insofar as previous behavior feedback loops lose their predictive power during periods of discontinuity, the continued reliance on traditional modes of sensemaking can lead to basic errors in interpretation of new data.

Managers sharing a dominant logic are likely to come to very similar conclusions when attempting to interpret a common set of data (von Krogh and Slocum, 1994). Each manager selects and makes sense of the available data within the framework of a knowledge structure and its industry specific content. Managers influenced by the dominant logic adopt a "one right way of doing things" mentality (Lyles and Schwenk, 1992). Thus, the existence of a dominant logic may inhibit the ability of the firm to respond to significant changes in its environment (von Krogh and Roos, 1996).

While the dominant logic has been primarily used to characterize behavioral patterns of individual firms, a dominant logic can be common across firms within a particular industry as well. Managers of firms within similar competitive niches share knowledge structures that shape a shared understanding of competitive boundaries and conditions (Lant and Baum, 1994; Reger and Huff, 1993; Porac, Thomas, and Baden-Fuller, 1989) even when environmental conditions change (Fombrun and Zajac, 1987). The contents of knowledge constructs have their origins in both the industries and firms in which the managers operate (Levinthal and March, 1993). Entrenched routines and practices are reinforced by the firm's past success in a particular industry setting (Levitt and March, 1988). This leads to the construction of content that provides managers within a given industry a common interpretative scheme.

In the case of the book industry, the dominant logic governing the buying and selling of books has remained remarkably stable for most of the twentieth century (Everitt, 1952; Epstein, 2002; Weinstein, 1999). Sellers of out-of-print and antiquarian books gained knowledge about book valuation slowly over a long period of time. This enabled experienced sellers to buy undervalued sleepers from their competitors. The mistake of underpricing books made by less experienced and less knowledgeable sellers provided the experienced sellers the valuable opportunity of buying under-priced books and re-selling them to established customers at the "true" market value. Moreover, the market had shown experienced sellers that out-of-print books gained value over time, providing them an incentive to hold on to books until the right buyer happened to come into their establishment. Thus, there existed within the retail book trade a dominant logic that was pervasive throughout the industry as a result of shared industry experience.

The Disruptive Impact of the Internet in the Retail Book Industry

When the Internet first emerged in the 1990s, there was a tendency to overstate its universal impact on the business environment. What is clear, in retrospect, is that the Internet has affected commerce differently depending on the nature of the industry (Globerman, Roehl and Standifird, 2001). In some cases, the Internet constitutes no more than a new supply channel or new information resource. In other cases, the Internet has had a transformative impact by changing the fundamental structure of the marketplace (Rothaermel and Sugiyama, 2001; Standifird, 2001a). As detailed below in our discussion of the retail out-of-print book industry, the advent of the Internet in the book industry has been disruptive.

To assess whether the Internet has changed the competitive environment of the retail book trade, we conducted extensive field research between August 1998 and October 2000 including formal interviews with 21 on-line booksellers, correspondence via e-mail with over 100 on-line booksellers, and informal discussions with another 27 on-line booksellers. In considering the impact of the Internet on the retail book trade, we use Afuah and Tucci's (2000) criteria for determining the disruptive impact of the Internet on a particular industry: 1) the extent to which the Internet creates new value for customers; 2) the extent to which the Internet disrupts the industry's value chain; and 3) the extent to which the Internet destroys the primary basis for success. Afuah and Tucci's framework incorporates and builds upon earlier theoretical discussions of disruptive technologies and applies them to the Internet. The focus on creating new value for consumers is consistent with Christensen's (1997) notion of disruptive technology. Christensen (1997) identifies a disruptive technology as a technology that fundamentally creates new value for consumers often in ways that are not immediately obvious to mainstream consumers. As detailed below, an example of this might be the out-of-print and antiquarian retail book industry, where the value to consumers has been the increase in the availability of information about book supply and valuation and the ability to quickly locate copies of out-of-print books. While these advantages seem obvious with the benefit of hindsight, they were not readily apparent in an industry that had been stable for nearly one hundred years. The focus on industry value chain and the primary basis for success builds upon Henderson and Clark's (1990) discussion of component and architectural changes respectively. The disruption of value chain activities such as new ways scarce books initially enter into the market place is consistent with component change. Changes in how knowledge in the industry is obtained and deployed is consistent with architectural change as discussed by Henderson and Clark. The advantage of the Afuah and Tucci framework is that it provides a comprehensive framework for understanding the depth of disruption associated specifically with the advent of Internet technologies. Thus, we use the Afuah and Tucci framework to formulate and test hypotheses concerning the relationship between organizational practices and experience and the impact of these practices on sales as a result of the introduction of Internet technologies within the retail book trade.

Value Creation for the Customer

Consumers of books have been the unquestioned beneficiaries of the advent of the Internet. What was once an uncertain and time-consuming process is, with the Internet, efficient, convenient, and free of cost. In a matter of seconds, the search for most books via the Internet is complete. Rather than being generally restricted to a local area when searching a particular title, the consumer can simultaneously search the stocks of over 10,000 booksellers globally who list their inventories on any one of several services such as abebooks.com, alibris.com, TomFolico.com, and Amazon Marketplace and Z-Shops. The book-buying public has the benefit of being able to conduct a search of these various inventory services simultaneously using meta-search engines as bookfinder.com and addall.com. The end result is that consumers not only can locate specific titles within seconds, but also they typically can choose from a range of editions and prices.

The emergence of these search engines and websites has also led to new dynamics of supply and demand. Whereas supply and demand were once defined by local markets, they are now determined internationally. Many "scarce" collectible titles have proved to be scarce only in the sense that they were hard to find in local bookstores. Consumers now have access to the inventories of over 10,000 on-line booksellers, and titles that have been collected for decades have proved, in many cases, to be not as scarce as once assumed by virtue of changes in taste and the relatively large number of preserved copies. For these titles, there has been considerable downward pressure on prices. In the case of popular titles with mass publication, the ultimate price paid by consumers a few years after initial publication, even with shipping costs, is below that of what many titles commanded before the Internet due their vast supply. In case of the small minority of titles in which strong demand and scarce supply have exerted upward pressure on prices, most collectors are pleased to have ready access to these titles even at higher prices since previously their ability to purchase these books was either subject to the vagaries of the local market place or an arduous effort. In this way, as well, the introduction of Internet technologies has disrupted the differentiation advantage historically enjoyed by those with significant industry experience.

Disruption of the Traditional Value Chain

Sellers of out-of-print and antiquarian books have traditionally relied upon their ability to buy books from the book-buying public at large and book collectors, in particular. Aside from garage and estate sales, there were, prior the Internet, few other ways for consumers to bypass bookstores and sell directly to the public. With the emergence of websites and services like abebooks.com and electronic auctions, this is no longer the case. For this reason, individuals with little or no retail experience in second hand and out-of-print books can sell books in the same marketplace where experienced retailers compete. This effectively has contributed to the lowering of barriers of entry for inexperienced book sellers. The costs of beginning an Internet bookstore are exceedingly low. The price of computer hardware is less than $1,000. Subscription fees to list books on line are as low as zero with a 15 percent commission on sales. Starting inventories can be created from rummaging garage sales and thrift stores. Except for the most established, high-end antiquarian bookstores, most Internet bookstores are undifferentiated as there books appear on the site of web intermediaries side by side with well established businesses. Nor is there a transaction cost advantage accrued by experience sellers. Indeed, established businesses with storefronts may often have higher costs than small Internet sellers who work from their home. Other free services to potential sellers have further reduced the cost barriers for inexperienced retailers. For instance, automated features available through freeware, such as automated author and title fill-in features for books that have International Standard Book Numbers (ISBN) mitigate cost advantages often associated with economies of scale. This system lowers the cost of data entry and therefore allows sellers to list individual items in less than 30 seconds. Thus, while there is a labor-cost advantage of having multiple titles with only one listing, in practice this advantage is negligible, particularly when weighed against the cost of carrying an inventory with multiple titles.

Disruption of the Traditional Basis of Competitiveness

The traditional basis of success in the out-of-print and antiquarian book industry has been specialized knowledge that enabled proprietors to recognize and value scarce books. Prior to the advent of the Internet, there was no quick or easy way for booksellers to acquire this knowledge. Reference books did not have information concerning the valuation of books priced less than $100, and the price guides that were available were frequently dated, incomplete, and expensive. Knowledge about books was acquired slowly over long periods of time. Booksellers might begin in the lower value-end of the market and learn about antiquarian books from other booksellers. Sometimes this was a difficult education, as competing booksellers would scout each other's inventories, in the hope of finding under-priced items that they could buy for resale. In other cases, individuals gained knowledge and experience about the book trade by working in other bookstores and then leaving to venture on their own.

The Internet provides easy access to this previously unavailable information. To learn if a book is in print, it is sufficient to simply check its availability from any one of several large Internet booksellers such as Amazon.com. To check the value of out-of-print and rare books, booksellers (and book buyers) simply need to go to a web intermediary like abebooks.com, which contains the inventories of thousands of booksellers. In the short time it takes to search a particular title, a user can learn not only the price of the book, but can often deduce the points that distinguish a rare first edition from a more common later printing. In this way, the advent of the Internet has turned upside down the received wisdom concerning the prolonged nature of knowledge acquisition about the value of books.

In short, the Internet has dramatically increased the value to customers, disrupted the traditional value network of the industry, and has undermined the distinctive competency of traditional booksellers. At the same time that the Internet has restructured the wholesale and retail market for used, out-of-print, and collectible books, it has also called into question the value of traditional knowledge structures and the associated dominant logic guiding the actions of the more experienced firms in the industry. The net effect of these changes has been a fundamental change in the competitive environment of used and out-of-print book industry. To assess the extent to which the dominant logic that has guided the industry may hinder experienced booksellers in managing the disruptive changes in the environment, we test a number of hypotheses related to information gathering, sense-making, and sales behavior.

HYPOTHESES

Expertise in book valuation and pricing in the out-of-print book trade, in general, and in the rare book trade, in particular, has traditionally been the distinguishing characteristic of the seasoned bookseller. Uneven knowledge among booksellers created opportunities for arbitrage and wealth creation. Everitt 's (1952: 118) characterization is that of a:

"... disorganized and smooth-running machine ... booksellers are apparently always taking in each other's ware, finding a sleeper here, selling it to another dealer, back and forth, as if private customers were the last thing in their minds. But eventually the book reaches the dealer who has the customer."

This centuries-old competitive practice has been eroded as the Internet imposes order on this "disorganized machine" (Epstein, 2001). Indeed so rapid has been the increased transparency and liquidity of the book market that we should anticipate that booksellers fully socialized in this industry through years of experience would be slow to recognize the value of the data produced by a simple Internet book search. This is consistent with the observations of Nisbett and Roos (1980) and Starbuck and Milliken (1988) that knowledge structures provide the basis for determining which information is relevant and which information should be discarded. As a result of the industry's dominant logic, experienced sellers will be slower to recognize the Internet as a potentially valuable information resource for establishing book valuation. Thus, experienced sellers may be less likely to use the Internet as a resource to gauge current supply and price data for selected book titles than their less experienced counterparts.

[H.sub.1a]: Years of industry experience will be negatively correlated to use the Internet to check book prices prior to listing them on line.

The question remains as to whether the existing dominant logic provides experienced sellers a source of competitive advantage. In other words, these sellers may be correct in not considering the new information currently available via the Internet. Insofar that new market information available as result of the Internet provides important information about change in the marketplace, this same dominant logic might adversely impact the sales of sellers who do not consult the Internet when pricing books. Indeed, the prima facie evidence indicates that the Internet has created a virtual spot market for books and such pricing information provided by the Internet is a critical determinant of sales. Booksellers willing to embrace Internet technologies can accelerate the process of acquiring the specialized knowledge needed to recognize and value books to be sold on line. The specialized knowledge developed by checking book prices on-line should have a direct impact on the ability to sell books successfully on line. Thus,

[H.sub.1b]: The use of the Internet to check book prices prior to listing them on line will be positively correlated to relative sales.

Another error that may result from the existing dominant logic may concern how those with experience in an industry make sense of information when it is available (Lampel and Shamsie, 2000). Experienced booksellers have been conditioned over the years to expect a continued appreciation in book values. Once books go out of print, there is a maximum fixed supply, and over time these books will become scarcer, and, assuming relatively stable demand, will increase in price (Basbanes, 1995). Thus, experienced sellers may feel no particular need to move inventory quickly, but rather will accept a slower movement of inventory as a natural part of the industry. Some obscure books may simply have to sit on a shelf for some time before the right buyers happens along. Less experienced sellers, by contrast, are more likely to reject this dominant logic for a number of reasons. First, they may more easily accept the dynamism of the new spot market for books. Many will have come to age as booksellers in the era of electronic commerce. Second, these inexperienced sellers may have a different view of inventory turnover and appropriate profit margins and they may be more interested in moving inventory quickly than their more experienced and patient counterparts. The dominant logic of experienced sellers concerning inventory turnover will have a direct impact on pricing in that experienced sellers will be far less likely to reduce prices merely to stimulate turn over. Consequently,

[H.sub.2a]: Years of industry experiences will be negatively correlated to the practice of pricing books below prevailing prices.

As with the use of the Internet to check prices, the consequence of the dominant logic is that experienced sellers will have different pricing practices than inexperienced sellers and this may have an impact on sales. For sellers to compete effectively in a nearly complete market with higher priced books, these sellers must offer titles or services that are sufficiently differentiated to justify a premium. This is difficult for the same reasons that pricing cutting behavior can occur. The Internet is a technology that has had a disruptive impact on the out-of-print book industry in that it has increased the effective supply of scarce books and has devalued traditional sources of competitiveness--knowledge of book valuations. With only a handful of antiquarian bookstores possessing complementary assets in the form of elite reputation, there is reason to expect market competitive pricing practices will lead to higher sales.

[H.sub.2b]: Pricing books below prevailing prices will be positively related to relative sales.

While information filtering and data interpretation capture specific errors associated with an outmoded industry knowledge construct, research in management cognition indicates the potential for variation in knowledge structures across a variety of domains (Walsh, 1995). The longer a manager's tenure in the industry, the more immersed that individual will be in the knowledge structure and content of this industry (Lurgio and Carroll, 1985, Wagner, 1987) and the less likely the individual will be able to respond to discontinuous change in the competitive environment. In the book industry, this could manifest itself in a number of ways that could erode the effectiveness of on-line sales efforts. For instance, with less faith in the relevance of the Internet, experienced retailers will focus on past investments in inventory and continue these investment patterns whereas inexperienced retailers, unencumbered by the existing dominant logic, are more likely to adjust their inventory based on feedback from on-line sales. The ability to adjust offerings in response to on-line demand should have a direct impact on the return on value of their on-line inventories for inexperienced retailers as compared to experienced retailers. Thus,

[H.sub.3]: Booksellers' tenure in the industry will be negatively related to relative sales.

METHODS

A unique data set was compiled in collaboration with abebooks.com to examine the business practices and sales of on-line booksellers. In July 2000, abebooks.com sent an e-mail notification to their 6,300 vendors explaining the goals of the research project and inviting them to visit our research website. This site provided background on the study and a link to an on-line survey. We then matched the data collected through our research website with data on the inventory and sales of individual vendors housed at abebooks.com.

Over a seven-week period we received responses from 1,926 on-line booksellers (30.6 percent response rate). Of these we had valid user identifiers for 1,488 firms, which allowed us to match the vendor-provided data to data archived at abebooks.com. Vendors from outside of North America constituted 10.3 percent of respondents.

For the purposes of this analysis, we limited the study to U.S. and Canadian companies that have been abebooks.com vendors for at least 100 days and have at least 1,000 books on-line. We limited the study to North American vendors since only a handful of non-North American firms processed their credit card transactions through abebooks.com at the time of the study. Similarly, we limited the study to booksellers with at least 100 days of experience as abebooks.com vendors because of the need for sufficient sales data from these sellers. We limited the analysis to vendors with more than 1,000 titles on-line since this filtered out hobbyist, collectors, and vendors just beginning to pursue on-line sales.

Based on our analysis of abebooks.com archival data, we calculated that 2,771 firms fit our sampling criteria, of which 882 completed the web-based survey, yielding an effective response rate of 31.8 percent. Ofthese, 755 vendors processed credit card transactions through abebooks.com. The firms in the sample are small, with an average on-line retail inventory value of $126,576. Aside from the proprietor, the average firm employs .98 full-time equivalent employees and 6.2 part-time employees.

Though always a concern, there are a number of reasons to believe that sampling bias does not pose a critical threat to the study's validity. An analysis of abebooks.com archival data indicates non-respondents have a higher average book price than non-respondents, but have virtually the same relative sales (Table 1). There remains a question as to whether vendors listing with abe.com are different from Internet book vendors in general. At the time of the data collection, abe.com was the pre-eminent web intermediary, with an estimated 60 percent of all vendors listing on abe.com, and interviews with abe.com management indicated that there seemed no patterns of participation on abe.com among U.S. sellers.

Dependent Variables

Use of the Internet as knowledge resource

Vendors indicated on a five-point scale--never, rarely, sometimes, usually, and always--the extent to which they utilized the Internet to check pricing and availability about individual titles on line.

Check Prices

This variable, Check Prices, is the dependent variable in the test of [H.sub.1a].

Pricing behavior

Cut Price

The variable Cut Price was created to represent vendors who regularly set prices for their books below the prevailing price on line. This variable is coded as "1" when vendors indicated that they set prices below prevailing prices and was coded as "O" when they did not; this latter measure includes vendors who check prices on line but do not cut prices as well as those vendors who simply do not consult the Internet for pricing information. This measure is the dependent variable in the test of [H.sub.2a].

Relative sales

To measure the consequence of business practices and the sales of on-line vendors we created a measure of relative sales. This figure was calculated by dividing the value of quarterly online sales processed through abebooks.com by the calculated value of the on-line inventory. The numerator in this ratio only includes credit card purchases transacted through abebooks.com. However, with the possible exception of antiquarian booksellers who may bypass web intermediaries in vendor-to-vendor transactions, the percentage of total on-line sales transacted through abebooks.com or other web intermediaries are assumed to be roughly the same. To account for sales outside of abebooks.com, a number of variables are included in the analysis to control for the use of other on-line sales transactions, adding further precision to our relative sales estimates.

The relative sales estimate has both strengths and limitations. Among its strengths is its precision. Both terms in this ratio are from transactions for which abebooks.com, the vendors, and a bank service verify accuracy. While not a measure of profitability, this measure does reflect an outcome desired in a fixed price market--the sale of goods. The extent to which this measure may be related to the long-term viability of these business and/or the emergence of a new business model are discussed in the subsequent analysis. This measure is the dependent variable in the tests of [H.sub.1b], [H.sub.2b], and [H.sub.3].

Experimental and Control Variables

Vendor Experience

The experience of vendors is measured by the number of years vendors report being self-employed booksellers. This measure captures the extent to which a proprietor has been immersed in the dominant logic that emerged over time in the used and out-of-print book industry.

Alternative Distribution Outlets

Four dummy variables were created to represent whether vendors cross-listed their inventories with each of five web intermediaries other than abebooks.com. The reason for including this statistical control is that some search "bots" pick up multiple listings of vendors, allowing consumers to hyperlink to vendors via services other than abebooks.com. The creation of these dummy variables is intended to control for this eventuality.

Market Segment

Control variables for market segment were created from the self-report of vendors as to whether they have a general, antiquarian, or specialty on-line bookstore. Generalists offer a multitude of largely popular books. Antiquarian booksellers focus on rare books of historical value, while Specialists focus on a particular type of books (e.g., science fiction). Each represents a distinct strategic type. These variables control for the various strategies pursued by individual book sellers. It may be that a particular strategy involves a stronger commitment (Ghemawat, 1991) and, as such inhibits the flexibility of the seller in future endeavors. The inclusion of these variables helps control for this possibility. In addition, these variables control for the possibility that some antiquarian sellers may engage in more vendor-to-vendor transactions that bypass abebooks.com credit card transactions than do other bookstores.

Days with ABE

We included a measure of the number of days booksellers have been abebooks.com vendors. Ceteris paribus the aging book inventories associated with those firms having been working with abebooks.com for a longer period of time should be negatively related to relative sales for three reasons. First, unless otherwise requested by consumers, books most recently listed on line appear first when a consumer's search is initiated. Second, there is substantial price cutting behavior among sellers. Books listed on line, even if price competitive at the time of their original listing, often become uncompetitive on the basis of price over time as new sellers undercut prevailing prices. Finally, vendors tend to list their most sellable items first and these items sell early, often providing booksellers with a large boost in sales during the first months of originally posting their inventories. The mean and standard deviation of the variables used in the analysis are listed in Table 1.

Model Specification

To estimate the probability that an Internet book seller will use the Internet as a resource to check prices, we use a linear probability model for an ordered categorical dependent variable:

Y = [[alpha].sub.1] + [[alpha].sub.2] + [[alpha].sub.3] + [[alpha].sub.4] + [[beta].sub.1](Years of Experience) + [[beta].sub.2[a,b,c,d]](Type of Business) + [epsilon]

In this model, where the dependent variable has four ranked-ordered levels ranging from "never" to "always," the model has four constants represented by [[alpha].sub.1] .. [[alpha].sub.4]. The primary experimental variable "Years of Experience" is continuous and denotes the number of years the respondent has worked in the used and out-of-print book trade. Type of business is a categorical variable, represented by dummy variables denoting specialty, general or antiquarian bookstore. Business type is included in the model as a control variable since businesses in the same market niche encounter comparable market conditions that might influence business practices and sales. The error term is represented by [epsilon].

To estimate the probability that an Internet book seller will set book prices below the prevailing market rate, we use a logistic regressions model for a dependent variable:

Y = [[alpha].sub.1] + [[beta].sub.1](Years of Experience) + [[beta].sub.2[a,b,c]](Type of Business) + [epsilon]

The specifications for the independent variables in this model are the same as those in the preceding model.

To estimate the relationship between sales and business practices and experience we use an ordinary least squares regression model:

Y = [[alpha].sub.1] + [[beta].sub.1](Years of Experience) + [[beta].sub.2][(Years of Experience).sup.2] + [[beta].sub.3](Check Prices) + [[beta].sub.4](Cut Prices) + [[beta].sub.5[a,b,c]](Type of Business) + [[beta].sub.6[a,b,c,d]](Use of Particular Sales Outlet) + [epsilon]

In this model, the dependent variable is a continuous value in dollars and represents the value of credit card transactions of books sold on the Internet. To model the empirical observation that the impact of experience on sales diminishes as experience term increases, we include a squared term for experience. To model the impact of business practices we include two dummy variables. The first indicates whether the vendor checks prices on the Internet prior to pricing books, and the second indicates whether the vendor sets prices below the prevailing market rate. As in the previous models, type of business is represented by dummy variables. Finally, we include dummy variables to represent the use of five web intermediaries or sales outlets as control variables since the use of these are likely to have a secular impact on sales of each vendor.

RESULTS

We conducted the data analysis in two steps. First, we determined whether there were systematic differences based on experience in the use of the Internet to research book prices (Check Prices) and to set prices below the prevailing market price (Cut Price). In the second stage of the analysis, we examined the relationship among these two practices, years of experience, and relative sales of individual vendors.

In Table 2 we present the results of the ordered logistic regression in which the five-level ordinal variable Check Prices is regressed on experience and two dummy variables representing specialty and general market segments; antiquarian book dealers are represented in the base of the equation. This model indicates a negative relationship between years of experience and the use of the Internet as an information resource to check availability and pricing, thus providing support for [H.sub.1a]. In this analysis, we also find that specialty store proprietors are less likely to check prices than antiquarian and general bookstore owners.

In Table 3, we present the results of the logistic regression analysis in which the dichotomous variable Cut Price is regressed on experience and the market segment dummy variables. In this case, as well, we find a statistically significant relationship between the length of time the proprietors have been self-employed booksellers and the dependent variable, providing support for [H.sub.1b]. The magnitude of these effects are represented in Figure 1 in which the anti-log of coefficient for Cut Price, multiplied by various intervals of experience measured in years, are plotted against experience level in years. Further extrapolation of the data indicates that a proprietor with two years of experience is 7.53 percent more likely to cut prices than a proprietor with five years of experience and is 37.5 percent more likely to cut prices than a proprietor with 15 years of experience.

[FIGURE 1 OMITTED]

To examine the effects of these pricing practices as well as the general relationship between experience and relative sales, we regressed relative sales on Check Prices, Cut Price, experience, two market segment dummy variables, and four dummy variables representing alternative distribution channels. Further analysis of the relationship between experience and relative sales indicated a non-linear relationship characterized by the flattening of a negative slope with an increase in experience. To account for this, we included a squared term for experience in the model.

In the OLS model presented in Table 4, we find support for [H.sub.1b], [H.sub.2b], and [H.sub.3]. With regards to [H.sub.1b], Check Prices was found to be positively correlated to relative sales (p < .01). The estimate from the OLS model indicates that vendors who engage in price-cutting have 24.7 percent higher relative sales, ceteris paribus, than vendors who do not consult the Internet prior to listing books online. The effect is compounded when vendors price items below the prevailing market price since price-cutting behavior is also positively related to relative sales (p <.01). Independent of the specific business practices represented in the model, experience is negatively related to relative sales. The relationship between experience and relative sales is presented in Figure 2. Further extrapolation of the data indicate that vendors with 5 years experience have a 12.7 percent lower relative sales than vendors with two years experience; vendors with 15 years experience of have a 42.7 percent lower relative sales than vendors with two years of experience.

[FIGURE 2 OMITTED]

As anticipated, a number of our control variables are related with relative sales. Specialty and general bookstores have higher relative sales than antiquarian dealers (in the base of the OLS equation). There are two likely explanations for this. First, antiquarian bookstores are more likely to be less concerned with relative sales per se, since replacing antiquarian books inventory is difficult. Second, antiquarian bookstores engage in more vendor-to-vendor transactions that would typically bypass abebooks.com credit card processing. Consistent with this, alternative sales distribution outlet number three is negatively related to relative sales, indicating that vendors who use this outlet to process a number of transactions via this web intermediary do so by substituting sales that would otherwise be processed through abebooks.com. Finally, as anticipated, days with abebooks.com is negatively related with relative sales (p <.05).

DISCUSSION

The empirical results from this analysis provide support for the theoretical propositions concerning managerial cognition and decision-making during periods of discontinuous change in the competitive environment of firms. Whereas knowledge constructs and the dominant logic of firms contribute to efficiency in decision-making during periods of stability, the same rules and logic may lead both to filtering out important data and/or the misinterpretation of data during periods of rapid change. In our study, experienced vendors who have been schooled in the stable and fragmented market that characterized the used and out-of-print book industry prior to the advent of the Internet have been slow to appreciate the relevance of market data available through the Internet. Moreover, even when experienced vendors scrutinized data from the Internet, they were prone to interpret the information according to the rules that governed the pre-Internet marketplace. The dominant logic of the used and out-of-print book industry led experienced vendors to behave as if new data from the environment was either irrelevant or erroneous.

The question remains as to whether the different business practices contributing to higher relative sales of less experienced vendors are sustainable and will ultimately lead to the enhanced economic performance of less experienced vendors. While only longitudinal data analysis will provide a definitive answer to this question, there may be some indirect evidence to suggest that the knowledge structures of experienced proprietors not only shape cognitive processes but may also be associated with adverse long-term economic consequences. The same factors that have made the Internet so disruptive for the used and out-of-print book industry have also disrupted traditional advantages of experienced vendors. As described earlier in our analysis of this industry, the Internet has undermined the value of bookseller experience, which was premised on expert valuation of scarce books and exclusive access to scarce books and markets. At the same time, the Internet has limited the most likely mistake made by less experienced sellers--under pricing scarce books. While inexperienced sellers are more likely to undercut prevailing prices, they are likely not to misprice a scarce book completely now that information about these items is so widely available. In the past, such errors created numerous opportunities for arbitrage for experienced sellers who could easily find sleepers in the inventories of less knowledgeable sellers. In the current market, these inexperienced sellers will verify the value of the item which effectively limits the extent of errors resulting from under valuation. If the item sells quickly, it does so because of competitive pricing practices. While it remains plausible that the relative sales of experienced sellers could be offset by the higher margins of high-end niche sellers, we should note that proprietors in specialist market niches also enjoy, ceteris paribus, higher relative sales.

The extent to which the reliance on an outdated dominant logic proves to have enduring consequences remains an empirical question that we can answer over time. The extent to which the cognitive errors of the proprietors in this study are indicative of the general attributes and liabilities of top-down information processing also is an empirical question only to be answered by more field research on managerial decision making. Booksellers' early adoption of electronic commerce allows us to investigate the initial determinants of economic performance in this industry. Nonetheless, this industry is still inchoate and additional research, particularly longitudinal analysis of experience and firm mortality, will provide a more precise assessment of the returns to experience in this industry.

A critical question concerns the generalizability of the findings of this study. To the extent that the business practices studied here emerge from the cognitive processes of individual managers, there is no basis to assume that experienced managers in large enterprises are any less prone to filter and misinterpret market data that do not fit neatly in their knowledge structures than are the proprietors who are the subject of this study. This is not to suggest that the Internet poses the same challenges to all industries--such is not the case. The cognitive errors of experienced booksellers are directly related to the dramatic impact that the Internet has had on this industry. At the same time, the book industry is by no means unique.

Any industry where the Internet has the effect of enhancing consumer value through increased transparency and/or market reach and disrupting the traditional sources of success by redefining the nature of industry knowledge is likely to experience cognitive errors of mangers similar to those observed in our study. An example here might include the travel industry where the use of web intermediaries such as Travelocity.com and Expedia.com has substantially increased transparency and, in the process, eroded the traditional value added by individual travel agents. An additional example might include the market for antiques and collectables where web intermediates such as eBay.com has dramatically increased market reach and subsequently reduced the traditional source of success based on the ability to acquire and price scare items. Given the high concentration of small proprietor-operated businesses in these sectors, the consequences of business failure go well beyond narrow financial

measures.

A potentially important implication of the analysis is the finding of a dominant logic that appears to be pervasive throughout the used and out-of-print book industry, suggesting that a dominant logic can emerge even in the absence of specific interorganizational connections. Although institutional influences assume a certain level of "connectedness" within a particular industry (Standifird, 2001b), a dominant logic does not assume connectedness but, instead, emerges as a result of a shared experience of successful behaviors. Thus, even in the absence of overarching normative or cognitive influences (DiMaggio and Powell, 1983), we may see a certain level of isomorphic behavior based largely on the shared experiences of firms. The retail book trade has experienced relative stability for decades prior to the advent of the Internet technologies. For this reason, we do not find it surprising that experienced sellers should share similar notions about their competitive environment. A question remains as to whether a dominant industry logic might permeate less entrenched industries; this is a potentially interesting question that can be only answered through further research.

In this particular analysis, we sought to explore how knowledge structures and content have contributed to a dominant logic guiding the behavior of experienced proprietors in the used and out-of-print book trade. While the dominant logic has been primarily used to characterize behavioral patterns of individual firms, we argue that a dominant logic can be common across firms within a particular industry as well. Through the analysis of field interviews and a multi-sourced data set, we find that the dominant logic that served the proprietors of the used and out-of-print book trade so well during periods of stability may prevent them from recognizing, evaluating, and responding to fundamental changes in the competitive environment.

Distinctive in its own right, retail booksellers are far from alone among companies seeking to expand their businesses to the wake of technological change. The relevance of the early experience to these on-line book retailers for other business is a theoretical and empirical question yet to be answered. Still, the results of this particular analysis suggest that the existence of an industry-wide dominant logic can have a significant impact on the ability of experienced firms to recognize, evaluate, and respond to fundamental changes in the competitive environment. Thus, the existence of an industry-wide dominant logic can have a significant impact on the ability of organizations within a particular industry to navigate the winds of technological change.

REFERENCES

Ablelson, R. P. and Black, J. B. (1986). 'Introduction'. In J. A. Galambos, R. P. Abelson and J. B. Black (eds.), Knowledge Structures: 1-18. Hillsdale, NJ: Lawrence Erlbaum Associates.

Afuah, A. and Tucci, C. (2000). A Model of the Internet as Creative Destroyer. working paper #BCES-00-21, Berkley Center for Entrepreneurial Studies, New York University.

Argote, L., McEvily, B. and Reagans, R. (2003). 'Managing knowledge in organizations: An integrative framework and review of emerging themes'. Management Science, 49: 571-582.

Axelrod, R. (1976). The Structure of Decision: The Cognitive Maps of Political Elites. Princeton, NJ: Princeton University Press.

Barr, P. S., Stimpert, J. L. and Huff, A. S. (1992). Cognitive change, strategic action, and organizational renewal. Strategic Management Journal, 13: 15-36.

Basbanes, N. (1995). A Gentle Madness: Bibliophies, Bibliomanes, and the External Passion for Books. New York, NY: Henry Holt & Co.

Bettis, R. A. and Prahalad, C. K. (1995). 'The Dominant Logic: Retrospective and Extension'. Strategic Management Journal, 16: 5-14.

Bougon M. G., Weick, K. and Binkhorst, D. (1977). 'Cognition in organizations: An analysis of the Utrecht Jazz Orchestra'. Administrative Science Quarterly, 22: 606-639.

Calori, R., Johnson, G. and Sarnin, P. (1992). 'French and British top managers' understanding of the structure and dynamics of their industry: A cognitive analysis and comparison'. British Journal of Management, 3: 61-78.

Christensen, C. M. (1997). The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail. Boston, MA: Harvard Business School Press.

Cote, L., Langley, A. and Pasquero, J. (1999). 'Acquisition strategy and dominant logic in an engineering firm'. Journal of Management Studies, 36: 919-952.

Darr, E. L., Argote, L. and Epple, D. (1995). 'The acquisition, transfer and depreciation of knowledge in service organizations: Productivity in franchises'. Management Science, 41: 1750-1762.

Day, D. V. and Lord, R. G. (1992). 'Expertise and problem categorization: The role of expert processing in organizational sense-making'. Journal of Management Studies, 29: 35-47.

Dimaggio, P. and Powell, W. (1983). 'The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields'. American Sociological Review 48: 147-160.

Dunn, W. A. and Ginsberg, A. (1986). 'A sociocognitive network approach to organizational analysis'. Human Relations, 11: 955-976.

El Sawy, O. A. and Pauchant, T. C. (1986). 'Triggers, templates and twitches in the tracking of emerging strategic issues'. Strategic Management Journal, 9: 455-473.

Epstein, J. (2001). Book Business: Publishing, Past, Present, and Future. New York NY: Norton.

Everitt, C. (1952). The Adventures of a Treasure Hunter. Boston, MA: Little, Brown, & Co.

Fiske, S. T. and Dyer, L. M. (1985). 'Structure and development of social schemata: Evidence from positive and negative transfer effects'. Journal of Personality and Social Psychology, 48: 839-852.

Fombrun, C. J. and Zajac, E. J. (1987). 'Structural and perceptual influences on intraindustry stratification'. Academy of Management Journal, 30: 33-50.

Ghemawat, P. (1991). Commitment. NY: The Free Press

Gioia, D. A. (1986). Conclusion: The state of the art in organizational social cognition. In H. P. Sims and D. A. Gioia (eds.), The Thinking Organization: Dynamics of Organizational Social Cognition: 336-356. San Francisco, CA: Jossey-Bass.

Globerman, S., Roehl, T. W. and Standifird, S. S. (2001). 'Globalization and electronic commerce: Inferences from retail brokering'. Journal of International Business Studies, 4: 749-768.

Henderson, R. M. and Clark, K. B. (1990). 'Architectural innovation: The reconfiguration of existing product technologies and the failure of established firms'. Administrative Science Quarterly, 35: 9-30

Lampel, J. and Shamsie, J. (2000). 'Probing the unobtrusive link: Dominant logic and the design of joint ventures at General Electric'. Strategic Management Journal, 21: 593-602.

Lant, T. K. and Baum. J. A. C. (1994). 'Cognitive sources of socially constructed competitive groups: Examples from the Manhattan hotel industry'. In W. R. Scott and S. Christensen (eds.), The Institutional Construction of Organizations: International and Longitudinal Studies: 15-38. Thousand Oaks, CA: Sage.

Levinthal, D.A., and March, J. G. (1993). 'The myopia of learning', Strategic Management Journal, 14 (Special Issue) 95-112.

Levitt, B. and March, J. G. (1988). 'Organizational learning', Annual Review of Sociology, 14: 319-340.

Lurgio, A. J. and Carroll, J. S. (1985). 'Probation officers' schemata of offenders: Content, development, and impact on treatment decisions'. Journal of Personality and Social Psychology, 48: 1112-1126.

Lyles, M. A. and Schwenk, C. R. (1992). 'Top management, strategy and organizational knowledge structures'. Journal of Management Studies, 29: 155-174.

March, J. G. and Simon, H. A. (1958). Organizations. NY: Wiley.

McCall, M. W. and Kaplan, R. E. 1985. Whatever it takes: Decision makers at work. Englewood Cliffs, NJ: Prentice Hall, Inc.

Miller, G. A. (1956). 'The magic number seven plus or minus two: Some limits on our capacity for processing information'. Psychological Review, 64: 81-97.

Mintzberg, H., Raisinghani, D. and Theoret, A. (1976). 'The structuring of 'unstructured' decision making'.

Administrative Science Quarterly, 21: 246-275.

Mischel, W. (1981). Personality assessment. NY: Wiley.

Neisser, U. (1976). Cognition and Reality: Principles and Implications of Cognitive Psychology. San Francisco, CA: Freeman.

Nisbett, R. and Ross, L. (1980). Human Inference: Strategies and Shortcomings of Social Judgement. Englewood Cliffs, NJ: Prentice-Hall.

Porac, J. F., Thomas, H. and Baden-Fuller, C. (1989). 'Competitive groups as cognitive communities: The case of Scottish knitwear manufacturers'. Journal of Management Studies, 26: 397-416.

Prahalad. C. K. and Bettis R. A. (1986). 'The dominant logic: A new linkage between diversity and performance'. Strategic Management Journal, 7: 485-501.

Reger, R. K. and Huff, A. S. (1993). 'Climate and culture: Interaction and qualitative differences in organizational meaning'. Journal of Applied Psychology, 75: 668-681.

Rothaermel, F. T. and Sugiyama, S. (2001). 'Virtual internet communities and commercial success: Individual and community-level theory grounded in the atypical case of TimeZone.com'. Journal of Management, 27: 297-312.

Shivastava, P. and Mitroff, I. (1983). 'Frames and reference managers use'. In R. Lamb (ed.), Advances in Strategic Management, vol 1. Greenwich, CT: JAI Press.

Standifird, S. S. (2001a). 'Reputation and e-commerce: eBay auctions and the asymmetrical impact of positive and negative ratings'. Journal of Management, 27: 279-296.

Standifird, S. S. (2001b). 'Conceptual bonds: Network analysis as a way of understanding institutional rigidity'. Emergence, 3: 7-21.

Starbuck, W. H. and Milliken, F. J. (1988). 'Executives' perceptual filters: What they notice and now they make sense'. In D. C. Hambrick (ed.), The Executive Effect: Concepts and Methods for Studying Top Managers: 35-65. Greenwich, CT.

Thorngate, W. (1980). 'Efficient decisions heuristics'. Behavioral Science, 25: 219-225.

von Krogh, G. and Roos, J. (1996). 'A tale of the unfinished'. Strategic Management Journal, 17: 729-737.

von Krogh, G. and Slocum, K. (1994). 'An essay on corporate epistemology'. Strategic Management Journal, 15 (Summer Special Issue): 53-71

Wagner, R. K. (1987). 'Tacit knowledge in everyday intelligent behavior'. Journal of Personality and Social Psychology, 52: 1236-1247.

Walsh, J. P. (1995). 'Managerial and organizational cognition: Notes from a trip down memory lane'. Organization Science, 6: 280-321.

Weick, K. E. and Bougon, M. G. (1986). 'Organizations as cognitive maps: Charting ways to success and failure'. In H. P. Sims and D. A. Gioia (eds.) The Thinking Organization: 102-135. San Francisco, CA: Jossey-Bass.

Weinstein, M. (1999). 'Railbirds, scouts, and Independent booksellers: Towards a constructivist interpretation of new business creation', Irish Marketing Review, 12(5): 1-13.

Marc Weinstein, Barry University

Stephen S. Standifird, University of San Diego

Table 1: Variable Definitions and Means
Standard Deviations in Parentheses

Variable               Non-respondents    Respondents
                          (n= 1889)          (n=882)

Relative sales         .00720 (.00914)   .00726 (.00767)
(abe.commerce
sales/total value of
inventory)

Years of experience                        9.20 (8.23)
as a self-employed
bookseller

Vendor checks prices                          4.07
before listing them                          (0.97)
on-line--five-point
scale (1=never, 5=
always)

Percentage of vendor                          45.1
usually undercuts
prevailing price

Percentage of                                 21.7
vendors with
speciality store
(e.g. science
fiction,
horticulture)

Percentage of                                 65.4
vendors with
speciality store

Percentage of                                 12.9
vendors with general
store

Percentage of                                 15.2
vendors
cross-listing with
web intermediary #1

Percentage of                                 33.2
vendors
cross-listing with
web intermediary #2

Percentage of                                 70.7
vendors
cross-listing with
web intermediary #3

Percentage of                                 29.3
vendors
cross-listing with
web intermediary #4

Average book price      42.82 (117.2)     25.42 (35.89)

Days with              783.79 (336.11)   799.57 (333.95)
abebooks.com

Table 2: Ordinal Regression Analysis

                                Model 1
                              Check Prices

Constant--Threshold 1           -5.317
                               (.383) (1)

Constant--Threshold 2         -3.628 ***
                                (0.270)

Constant--Threshold 3          -2.089***
                                (0.237)

Constant--Threshold 4            -.402
                                (0.223)

Experience                    -.0668 ***
                                (0.009)

General Bookstore                -.150
                                (0.213)

Specialty Bookstore            -.858 ***
                                 (.240)

N                                 758

Log Likelihood                   651.79

Model Chi-square               79.891 ***

Cox & Snell Pseudo R-square       0.1

(1) Standard errors in parentheses

* p <.05 ** p <.01, *** p <.001

Table 3: Logistic Regression Analysis

                           Model 2
                          Cut Price

Constant                   -.073
                         (.340) (1)

Experience                 -.024 *
                           (.006)

General Bookstore           -.223
                           (.240)

Specialty Bookstore         .491
                           (.279)

N                            751

Log Likelihood             998.59

Model Chi-square         23.292 ***

Standard errors in parentheses
* p <.05 ** p <.01, ***p <.001

Table 4: Regression Analysis of
Relative Sales (1)

Determinant               Model 3

Constant                 .0793 ***
                          (7.561)

Experience               -.376 ***
                          (-3.516)

Experience-Squared        .243 *
                          (2.385)

Check Prices              .106 **
                          (2.753)

Cut Price                 .116 **
                          (3.109)

Speciality Store          .176 **
                          (3.106)

General Store              .117 *
                          (2.096)

Alt Sales 1                -0.048
                          (-1.288)

Alt Sales 2                 .008
                           (.205)

Alt Sales 3              -.164 ***
                          (-4.297)

Alt Sales 4                -.051
                          (-1.357)

Days with abebooks        -.090 *
                          (-2.270)

F-Statistic Model        11.260 ***

Adj. R-square              0.148
N                           648

(1) Standardized coefficients, T-statistics in parentheses

 p <.10, * p <.05 **, p <.01, *** p < .001