498 R. Amit and C. Zott resources and capabilities that had to be built structural arguments to explore the importance over time due to factor market imperfections, of governance mechanisms such as trust(e.g and hence enable the preservation of value. The Lorenzoni and Lipparini, 1999), and the impor- prospect of value preservation or sustainability is tance of resources and capabilities(e.g, Gulati an important incentive for value creation. In a 1999), especially those of suppliers and customers networked economy, however, there is an alterna-(Afuah, 2000), for value creation. For example, tive to ownership or control of resources and in their study of the Canadian biotechnology capabilities (either through building or acquiring industry, Baum, Calabrese, and Silverman(2000) them). Accessing such resources through part- found that biotech start-ups can improve their nering and resource sharing agreements is more performance by configuring alliances into net- viable in virtual markets yet the preservation works that enable them to tap into the capabilities of value, and hence its creation becomes more and information of their alliance partners. In challenging, because rivals may have easy access addition to enabling access to information, mar- to substitute resources as well kets, and technologies( Gulati et al., 2000),stra- tegic networks offer the potential to share risk Strategic networks generate economies of scale and scope(Katz and Shapiro, 1985; Shapiro and Varian, 1999), share Strategic networks are stable interorganizational knowledge, and facilitate learning(Anand and ties which are strategically important to participat- Khanna, 2000: Dyer and Nobeoka, 2000; Dyer ing firms. They may take the form of strategic and Singh, 1998), and reap the benefits that alliances, joint ventures, long-term buyer-supplier accrue from interdependent activities such as partnerships, and other ties'( Gulati, Nohria, and workflow systems(Blankenburg Holm, Eriksson Zaheer, 2000: 203). The main questions that stra- and Johanson, 1999). Other sources of value in tegic network theorists seek to answer are as strategic networks include shortened time to mar follows: (1)Why and how are strategic networks ket (Kogut, 2000), enhanced transaction of firms formed?(2)What is the set of interfirm efficiency, reduced asymmetries of information, elationships that allows firms to compete in the and improved coordination between the firms marketplace?(3) How is value created in net- involved in an alliance( Gulati et al., 2000) works (for example, through interfirm asset co- The network perspective is clearly relevant for specialization )? and (4)How do firms'differential understanding wealth creation in e-business positions and relationships in networks affect because of the importance of networks of firms their performance? suppliers, customers, and other partners in the Traditionally, network theorists with a back- virtual market space( Shapiro and Varian, 1999 ground in sociology or organization theory have Prahalad and Ramaswamy, 2000). However, it focused on the implications of network structure may not fully capture the value creation potential for value creation. The configuration of the net- of e-businesses that enable transactions in new work in terms of density and centrality(Freeman, and unique ways. For example, strategic network 1979), for example, has been considered an theory and the formal tools provided by network important determinant of network advantages, analysis (e.g, notions of network density,cen- such as access, timing, and referral benefits(Burt, trality, network externalities) only partiall 1992). Moreover, the size of the network and the explain the value creation potential of a company heterogeneity of its ties have been conjectured to such as Priceline. com. This business, which has have a positive effect on the availability of valu- established stable interorganizational ties, for able information to the participants within that example, with airline companies, credit card com etwork( Granovetter, 1973) The appearance of networks of firms in which System, is fundamentally anchored in the inno- market and hierarchical governance mechanisms vation of its transaction mechanism-namely, the coexist has significantly enhanced the range of introduction of reverse markets in which cus- possible organizational arrangements for value tomers post desired prices for sellers'accep- creation(Doz and Hamel, 1998; Gulati, 1998). tance-by which items such as airline tickets are Consequently, strategic management and sold over the Internet. Priceline. com has even entrepreneurship scholars have moved beyond been granted a business method patent on their Copyright o 2001 John Wiley Sons, Ltd, sma. Mgmt. J.22:493-520(2001)
498 R. Amit and C. Zott resources and capabilities that had to be built over time due to factor market imperfections, and hence enable the preservation of value. The prospect of value preservation or sustainability is an important incentive for value creation. In a networked economy, however, there is an alternative to ownership or control of resources and capabilities (either through building or acquiring them). Accessing such resources through partnering and resource sharing agreements is more viable in virtual markets yet the preservation of value, and hence its creation becomes more challenging, because rivals may have easy access to substitute resources as well. Strategic networks Strategic networks are ‘stable interorganizational ties which are strategically important to participating firms. They may take the form of strategic alliances, joint ventures, long-term buyer–supplier partnerships, and other ties’ (Gulati, Nohria, and Zaheer, 2000: 203). The main questions that strategic network theorists seek to answer are as follows: (1) Why and how are strategic networks of firms formed? (2) What is the set of interfirm relationships that allows firms to compete in the marketplace? (3) How is value created in networks (for example, through interfirm asset cospecialization)? and (4) How do firms’ differential positions and relationships in networks affect their performance? Traditionally, network theorists with a background in sociology or organization theory have focused on the implications of network structure for value creation. The configuration of the network in terms of density and centrality (Freeman, 1979), for example, has been considered an important determinant of network advantages, such as access, timing, and referral benefits (Burt, 1992). Moreover, the size of the network and the heterogeneity of its ties have been conjectured to have a positive effect on the availability of valuable information to the participants within that network (Granovetter, 1973). The appearance of networks of firms in which market and hierarchical governance mechanisms coexist has significantly enhanced the range of possible organizational arrangements for value creation (Doz and Hamel, 1998; Gulati, 1998). Consequently, strategic management and entrepreneurship scholars have moved beyond Copyright 2001 John Wiley & Sons, Ltd. Strat. Mgmt. J., 22: 493–520 (2001) structural arguments to explore the importance of governance mechanisms such as trust (e.g., Lorenzoni and Lipparini, 1999), and the importance of resources and capabilities (e.g., Gulati, 1999), especially those of suppliers and customers (Afuah, 2000), for value creation. For example, in their study of the Canadian biotechnology industry, Baum, Calabrese, and Silverman (2000) found that biotech start-ups can improve their performance by configuring alliances into networks that enable them to tap into the capabilities and information of their alliance partners. In addition to enabling access to information, markets, and technologies (Gulati et al., 2000), strategic networks offer the potential to share risk, generate economies of scale and scope (Katz and Shapiro, 1985; Shapiro and Varian, 1999), share knowledge, and facilitate learning (Anand and Khanna, 2000; Dyer and Nobeoka, 2000; Dyer and Singh, 1998), and reap the benefits that accrue from interdependent activities such as workflow systems (Blankenburg Holm, Eriksson and Johanson, 1999). Other sources of value in strategic networks include shortened time to market (Kogut, 2000), enhanced transaction efficiency, reduced asymmetries of information, and improved coordination between the firms involved in an alliance (Gulati et al., 2000). The network perspective is clearly relevant for understanding wealth creation in e-business because of the importance of networks of firms, suppliers, customers, and other partners in the virtual market space (Shapiro and Varian, 1999; Prahalad and Ramaswamy, 2000). However, it may not fully capture the value creation potential of e-businesses that enable transactions in new and unique ways. For example, strategic network theory and the formal tools provided by network analysis (e.g., notions of network density, centrality, network externalities) only partially explain the value creation potential of a company such as Priceline.com. This business, which has established stable interorganizational ties, for example, with airline companies, credit card companies, and the Worldspan Central Reservation System, is fundamentally anchored in the innovation of its transaction mechanism—namely, the introduction of reverse markets in which customers post desired prices for sellers’ acceptance—by which items such as airline tickets are sold over the Internet. Priceline.com has even been granted a business method patent on their
Value Creation in E-Business 499 innovative transaction method. This method have focused on the ways in which investment distinguishes the firm from an ordinary, online in information technology can reduce coordination travel agency and poises the firm to tap the costs and transaction risk( Clemons and Row more traditional, well-known sources of value 1992). In general, organizations that economize in networks discussed above. As this example on transaction costs can be expected to extract indicates, virtual markets, with their unprec- more value from transactions dented reach, connectivity, and low-cost infor- One of the main effects of transacting over the mation processing power, open entirely new Internet, or in any highly networked environment, possibilities for value creation through the struc- is the reduction in transaction costs it engenders turing of transactions in novel ways. These new (Dyer, 1997). Hence, the transaction cost transaction structures are not fully captured by approach critically informs our understanding of network theol value creation in e-business. Transaction costs include the time spent by managers and Transaction cost economics employees searching for customers and suppliers, communicating with counterparts in other com- The central question addressed by transaction cost panies regarding transaction details,. the costs economics is why firms internalize transactions of travel, physical space for meetings, and proc- that might otherwise be conducted in markets essing paper documents, as well as the costs of (Coase, 1937). The main theoretical framework production and inventory management(Lucking was developed by Williamson (1975, 1979, Reiley and Spulber, 2001). In addition to decreas- 1983). He suggests that 'a transaction occurs ing these direct costs of economic transactions, when a good or service is transferred across a e-businesses may also reduce indirect costs, such technologically separable interface. One stage of as the costs of adverse selection, moral hazard processing or assembly activity terminates, and and hold-up. This may result from an increased another begins'(Williamson, 1983: 104). Willi- frequency of transactions( because of open stan- anson identified bounded rationality coupled with dards, anyone can interact with anyone else),a uncertainty and complexity, asymmetric infor- reduction in transaction uncertainty(by providing mation,and opportunism in small-numbers situ- a wealth of transaction-specific information), and ations as conditions under which transactional a reduction in asset specificity (for example inefficiencies may arise that vary with the adopted through lower site specificity-the next site is governance mechanism(Williamson, 1975). At only one click away). The small-numbers bar- its core, then, transaction cost theory is concerned gaining condition may be relieved in the virtual with explaining the choice of the most efficient market situation because of the possibility for governance form given a transaction that is large numbers of previously unconnected parties embedded in a specific economic context. Critical(e.g, buyers and sellers)to interact dimensions of transactions influencing this choice Nonetheless, the emphasis of transaction cost are uncertainty, exchange frequency, and the economics on efficiency may divert attention from specificity of assets enabling the exchange(Klein, other fundamental sources of value such as inno- Crawford, and Alchian, 1978; Williamson, 1979). vation and the reconfiguration of resources Transaction costs include the costs of planning, (Ghoshal and Moran, 1996). The theory also adapting, executing, and monitoring task com- focuses on cost minimization by single parties pletion(Williamson, 1983) and neglects the interdependence between Transaction cost economics identifies trans- exchange parties and the opportunities for joint action efficiency as a major source of value, as value maximization that this presents(Zajac and enhanced efficiency reduces costs. It suggests that Olsen, 1993). In addition, governance modes value creation can derive from the attenuation of other than hierarchies and markets(e.g, joint uncertainty, complexity, information asymmetry, ventures) receive relatively little attention, which and bargaining conditions contrasts Importance of strategIc net (Williamson, 1975). Moreover, reputation, trust, works in e-business. Finally, Williamson (1983) and transactional experience can lower the cost implies that a transaction is a discrete event that of idiosyncratic exchanges between firms is valuable by itself, as it reflects the choice of (Williamson, 1979, 1983). Recently, researchers the most efficient governance form and hence can Copyright o 2001 John Wiley Sons, Ltd, sma. Mgmt. J.22:493-520(2001)
Value Creation in E-Business 499 innovative transaction method. This method distinguishes the firm from an ordinary, online travel agency and poises the firm to tap the more traditional, well-known sources of value in networks discussed above. As this example indicates, virtual markets, with their unprecedented reach, connectivity, and low-cost information processing power, open entirely new possibilities for value creation through the structuring of transactions in novel ways. These new transaction structures are not fully captured by network theory. Transaction cost economics The central question addressed by transaction cost economics is why firms internalize transactions that might otherwise be conducted in markets (Coase, 1937). The main theoretical framework was developed by Williamson (1975, 1979, 1983). He suggests that ‘a transaction occurs when a good or service is transferred across a technologically separable interface. One stage of processing or assembly activity terminates, and another begins’ (Williamson, 1983: 104). Williamson identified bounded rationality coupled with uncertainty and complexity, asymmetric information, and opportunism in small-numbers situations as conditions under which transactional inefficiencies may arise that vary with the adopted governance mechanism (Williamson, 1975). At its core, then, transaction cost theory is concerned with explaining the choice of the most efficient governance form given a transaction that is embedded in a specific economic context. Critical dimensions of transactions influencing this choice are uncertainty, exchange frequency, and the specificity of assets enabling the exchange (Klein, Crawford, and Alchian, 1978; Williamson, 1979). Transaction costs include the costs of planning, adapting, executing, and monitoring task completion (Williamson, 1983). Transaction cost economics identifies transaction efficiency as a major source of value, as enhanced efficiency reduces costs. It suggests that value creation can derive from the attenuation of uncertainty, complexity, information asymmetry, and small-numbers bargaining conditions (Williamson, 1975). Moreover, reputation, trust, and transactional experience can lower the cost of idiosyncratic exchanges between firms (Williamson, 1979, 1983). Recently, researchers Copyright 2001 John Wiley & Sons, Ltd. Strat. Mgmt. J., 22: 493–520 (2001) have focused on the ways in which investment in information technology can reduce coordination costs and transaction risk (Clemons and Row, 1992). In general, organizations that economize on transaction costs can be expected to extract more value from transactions. One of the main effects of transacting over the Internet, or in any highly networked environment, is the reduction in transaction costs it engenders (Dyer, 1997). Hence, the transaction cost approach critically informs our understanding of value creation in e-business. Transaction costs include ‘the time spent by managers and employees searching for customers and suppliers, communicating with counterparts in other companies regarding transaction details … the costs of travel, physical space for meetings, and processing paper documents,’ as well as the costs of production and inventory management (LuckingReiley and Spulber, 2001). In addition to decreasing these direct costs of economic transactions, e-businesses may also reduce indirect costs, such as the costs of adverse selection, moral hazard, and hold-up. This may result from an increased frequency of transactions (because of open standards, anyone can interact with anyone else), a reduction in transaction uncertainty (by providing a wealth of transaction-specific information), and a reduction in asset specificity (for example, through lower site specificity––the next site is only ‘one click away’). The small-numbers bargaining condition may be relieved in the virtual market situation because of the possibility for large numbers of previously unconnected parties (e.g., buyers and sellers) to interact. Nonetheless, the emphasis of transaction cost economics on efficiency may divert attention from other fundamental sources of value such as innovation and the reconfiguration of resources (Ghoshal and Moran, 1996). The theory also focuses on cost minimization by single parties and neglects the interdependence between exchange parties and the opportunities for joint value maximization that this presents (Zajac and Olsen, 1993). In addition, governance modes other than hierarchies and markets (e.g., joint ventures) receive relatively little attention, which contrasts with the importance of strategic networks in e-business. Finally, Williamson (1983) implies that a transaction is a discrete event that is valuable by itself, as it reflects the choice of the most efficient governance form and hence can
500 R. Amit and C. Zott be a source of transactional efficiencies. However, inquiry. The analysts wrote up the answers to in the context of virtual markets, considering any the questions using information gathered from given exchange in isolation from other exchanges multiple data sources, writing up to several para- that may complement or facilitate that exchange graphs in response to each question makes it difficult to assess the value created by Our research design was based on multiple a specific economic exchange. This is evident cases and multiple investigators, thereby allowing from the absence of direct empirical validation for replication logic (Yin, 1989). That is, we of the relationship between exchange attributes treated a series of cases like a series of experi and market and firm performance(Poppo and ments. Each case served to test the theoretical Lenger, 1998), and the absence of estimates of insights gained from the examination of previous transaction costs themselves(see Shelanski and cases, and to modify or refine them. This repl Klein, 1995, for a review) cation logic fosters the emergence of testable theory that is free of researcher bias(Eisenhardt, Summary 1989),and allows for a close correspondence between theory and data (Glaser and Strauss, Each theoretical framework discussed above 1967). Such a grounding of the emerging theory nakes valuable suggestions about possible in the data can provide a new perspective on an sources of value creation. As we have seen, many already researched topic (e.g, Hitt et al, 1998) of the insights gained from cumulative research However, it is especially useful in the early stages in entrepreneurship and strategic management are of research on a topic, when it is not clear yet applicable to e-business. However, the multitude to what extent the research question is informed of value drivers suggested in the literature raises by existing theories( for a recent example of such the question of precisely which sources of value an inductive study, see Galunic and Eisenhardt, are of particular importance in e-business, and 2001). Both motivations hold in the context of whether unique value drivers can be identified in e-business. Furthermore, using case studies is a the context of e-business. We have also drawn good research strategy for examining a contem- attention to the fact that each theoretical frame- porary phenomenon in its real-life context, ork that might explain value creation has limi- especially when the boundaries between phenom- tations when applied in the context of highly enon and context are not clearly evident'(Yin, interconnected electronic markets. We believe that 1981: 59). This difficulty is present in the e- this reinforces the need for an identification and business context prioritization of the sources of value creation in e-business We begin this process by grounding population of e-business firms a model of the sources of value creation in e- business in using data on e-business firms We define an e-business firm as one that derives a significant proportion(at least 10%)of its revenues from transactions conducted over the DATA AND METHOD Internet. This definition of an e-business firm is Research strategy quite broad. It includes, for example, Internet Service Providers(e.g, European ISP Freeserve), A lack of prior theorizing about a topic makes and companies that have not aligned all of their the inductive case study approach an appropriate internal business processes with the Internet but hoice of methodology for developing theory that use the Internet solely as a sales channel Eisenhardt, 1989). Hence, to gain a deeper (e.g, companies such as the speech recognition understanding of value creation in e-business, we software provider Lernout and Hauspie). On the conducted in-depth inquiries into the sources of other hand, it excludes providers of Internet value creation of 59 e-business firms. Our related hardware or software. that is. firms that esearch analysts, two of our former MBA stu- facilitate e-business but that do not engage in dents carefully selected from a pool of applicants the activity themselves (e.g, a backbone switch based on their sound understanding of e-business manufacturer, such as Packet Engines Inc) transactions, investigated each firm using approxi- Companies that derive all of their revenues mately 50 open-ended questions to guide their from e-business(so-called pure plays)are rela Copyright o 2001 John Wiley Sons, Ltd, sma. Mgmt. J.22:493-520(2001)
500 R. Amit and C. Zott be a source of transactional efficiencies. However, in the context of virtual markets, considering any given exchange in isolation from other exchanges that may complement or facilitate that exchange makes it difficult to assess the value created by a specific economic exchange. This is evident from the absence of direct empirical validation of the relationship between exchange attributes and market and firm performance (Poppo and Zenger, 1998), and the absence of estimates of transaction costs themselves (see Shelanski and Klein, 1995, for a review). Summary Each theoretical framework discussed above makes valuable suggestions about possible sources of value creation. As we have seen, many of the insights gained from cumulative research in entrepreneurship and strategic management are applicable to e-business. However, the multitude of value drivers suggested in the literature raises the question of precisely which sources of value are of particular importance in e-business, and whether unique value drivers can be identified in the context of e-business. We have also drawn attention to the fact that each theoretical framework that might explain value creation has limitations when applied in the context of highly interconnected electronic markets. We believe that this reinforces the need for an identification and prioritization of the sources of value creation in e-business. We begin this process by grounding a model of the sources of value creation in ebusiness in using data on e-business firms. DATA AND METHOD Research strategy A lack of prior theorizing about a topic makes the inductive case study approach an appropriate choice of methodology for developing theory (Eisenhardt, 1989). Hence, to gain a deeper understanding of value creation in e-business, we conducted in-depth inquiries into the sources of value creation of 59 e-business firms. Our research analysts, two of our former MBA students carefully selected from a pool of applicants based on their sound understanding of e-business transactions, investigated each firm using approximately 50 open-ended questions to guide their Copyright 2001 John Wiley & Sons, Ltd. Strat. Mgmt. J., 22: 493–520 (2001) inquiry. The analysts wrote up the answers to the questions using information gathered from multiple data sources, writing up to several paragraphs in response to each question. Our research design was based on multiple cases and multiple investigators, thereby allowing for replication logic (Yin, 1989). That is, we treated a series of cases like a series of experiments. Each case served to test the theoretical insights gained from the examination of previous cases, and to modify or refine them. This replication logic fosters the emergence of testable theory that is free of researcher bias (Eisenhardt, 1989), and allows for a close correspondence between theory and data (Glaser and Strauss, 1967). Such a grounding of the emerging theory in the data can provide a new perspective on an already researched topic (e.g., Hitt et al., 1998). However, it is especially useful in the early stages of research on a topic, when it is not clear yet to what extent the research question is informed by existing theories (for a recent example of such an inductive study, see Galunic and Eisenhardt, 2001). Both motivations hold in the context of e-business. Furthermore, using case studies is a good research strategy for examining ‘a contemporary phenomenon in its real-life context, especially when the boundaries between phenomenon and context are not clearly evident’ (Yin, 1981: 59). This difficulty is present in the ebusiness context. Population of e-business firms We define an e-business firm as one that derives a significant proportion (at least 10%) of its revenues from transactions conducted over the Internet. This definition of an e-business firm is quite broad. It includes, for example, Internet Service Providers (e.g., European ISP Freeserve), and companies that have not aligned all of their internal business processes with the Internet but that use the Internet solely as a sales channel (e.g., companies such as the speech recognition software provider Lernout and Hauspie). On the other hand, it excludes providers of Internetrelated hardware or software, that is, firms that facilitate e-business but that do not engage in the activity themselves (e.g., a backbone switch manufacturer, such as Packet Engines Inc.). Companies that derive all of their revenues from e-business (so-called ‘pure plays’) are rela-
Value Creation in E-Business 501 tively easy to identify using publicly available States or in Europe,(b) be publicly quoted on a descriptions of their major lines of business(e.g, stock exchange, and (c) involve individual con Amazon. com). In other instances, however, it is sumers in some of the electronic transactions it more difficult to establish whether a firm derives enables. The international scope of our study significant revenues from e-business. This is the not only reflects the decreasing importance of case for many incumbents (e.g, the British geographic boundaries in virtual markets, it also retailer Iceland ) It is often impossible to assert strengthens our theory development. Theory if this criterion has been met since companies building on value creation in e-business from seldom report their e-business revenues as a sep- inductive case studies is less idiosyncratic if one arate category. In these cases, we used other allows for cases from different economic environ- information to determine the company's fit with ments. 9 our target population. For example, we checked We chose to include only public companies in whether at least two trade publications such as our sample to ensure the availability and accuracy the Wall street Journal and the financial Times of information we are aware that this limits the eferred to the company as an e-business, or a scope of our analysis, as there are many private pioneer or early innovator in the virtual market firms with interesting business ideas. However space unlike private firms, publicly traded companies provide a wealth of data that can be collected Sample organized, and analyzed. At this point, it is unclear whether or not this choice introduces a For the United States, we created a list of e- large-company bias into our sample, and hence businesses that went public between 2 April 1996 into our conceptual development, because there (Lycos and 15 October 1999(Women. com are many large, private e-business operations, and Networks) using information available on several large, public firms not included in our www.hoovers.com This list includes about 150 sample(.g, AOL and Yahoo) firms, most of which are pure plays. Our initial Including only public companies in our sample subsample of 30 U.S. e-business companies was may bias it towards surviving companies. While then taken at random from this list on the basis of limitations on the availability of data prevent us a uniform probability distribution over all sample from broadening the sample to firms that failed companies. The U.S. subsample represents a (according to some definition of failure), we do broad cross-section of firms(see Appendix). By not believe that the survival bias affects the theo- contrast, the challenge in creating the European retical development. First, some of the firms we sub-sample was in identifying public e-businesses. studied will likely fail eventually. Second, the The number of European firms engaged in e- argument can be made for theoretical rather than business, as well as the development of indicators random sampling of cases, and for studying of Internet usage and e-business activity in Eu- extreme situations and polar types in which the rope, have lagged behind the corresponding fig ures in the United States in recent years(Morgan Stanley Dean Witter, 1999). Despite these 9 The decision to include U.S. as well as European firms in difficulties, we established a sample of 29 public our sample has several implications. E-business activity in Appendix). Companies were found on all major United States, an ed less by start-ups, as is the case in the European e-businesses (also listed in the Europe is dominate more by established companies(Morgan Stanley Dean Witter, 1999). For example, the United King European exchanges, as well as on new venture dom's Freeserve is a spin-off of Dixons, a large"bricks-and- markets(such as Germany's Neuer Markt) mortar'retailer, and Spains Terra Networks is a spin-off of To be eligible for inclusion in our sample Telefonica, a large telecommunication firm. An affiliation(past or present)with established companies probably influences the e-business had to(a) be based either in the Un particular business models of respective e-business firms. For of their parent companies, while others may suffer from The principal reason for choosing 2 April 1996(date of imposed organizational constraints. However, a possible sam- few days later by Yahoo's ple bias toward(mostly former) subsidiaries of established IPO) start date for was that this date marked companies should not affect our ability to develop a general the beginning of a peri multiple IPOs of e-business framework for evaluating the value creation potential of e- companies that occurred accession. This enabled us business firms. In fact, such a general framework should be to create a data set of sufficient size and breadth independent of the mode of Copyright o 2001 John Wiley Sons, Ltd, sma. Mgmt. J.22:493-520(2001)
Value Creation in E-Business 501 tively easy to identify using publicly available descriptions of their major lines of business (e.g., Amazon.com). In other instances, however, it is more difficult to establish whether a firm derives significant revenues from e-business. This is the case for many incumbents (e.g., the British retailer Iceland). It is often impossible to assert if this criterion has been met since companies seldom report their e-business revenues as a separate category. In these cases, we used other information to determine the company’s fit with our target population. For example, we checked whether at least two trade publications such as the Wall Street Journal and the Financial Times referred to the company as an e-business, or a pioneer or early innovator in the virtual market space. Sample For the United States, we created a list of ebusinesses that went public between 2 April 1996 (Lycos)8 and 15 October 1999 (Women.com Networks) using information available on www.hoovers.com. This list includes about 150 firms, most of which are ‘pure plays.’ Our initial subsample of 30 U.S. e-business companies was then taken at random from this list on the basis of a uniform probability distribution over all sample companies. The U.S. subsample represents a broad cross-section of firms (see Appendix). By contrast, the challenge in creating the European sub-sample was in identifying public e-businesses. The number of European firms engaged in ebusiness, as well as the development of indicators of Internet usage and e-business activity in Europe, have lagged behind the corresponding figures in the United States in recent years (Morgan Stanley Dean Witter, 1999). Despite these difficulties, we established a sample of 29 public European e-businesses (also listed in the Appendix). Companies were found on all major European exchanges, as well as on new venture markets (such as Germany’s Neuer Markt). To be eligible for inclusion in our sample, an e-business had to (a) be based either in the United 8 The principal reason for choosing 2 April 1996 (date of Lycos’s IPO, which was followed a few days later by Yahoo’s IPO) as a start date for sampling was that this date marked the beginning of a period of multiple IPOs of e-business companies that occurred in quick succession. This enabled us to create a data set of sufficient size and breadth. Copyright 2001 John Wiley & Sons, Ltd. Strat. Mgmt. J., 22: 493–520 (2001) States or in Europe, (b) be publicly quoted on a stock exchange, and (c) involve individual consumers in some of the electronic transactions it enables. The international scope of our study not only reflects the decreasing importance of geographic boundaries in virtual markets, it also strengthens our theory development. Theory building on value creation in e-business from inductive case studies is less idiosyncratic if one allows for cases from different economic environments.9 We chose to include only public companies in our sample to ensure the availability and accuracy of information. We are aware that this limits the scope of our analysis, as there are many private firms with interesting business ideas. However, unlike private firms, publicly traded companies provide a wealth of data that can be collected, organized, and analyzed. At this point, it is unclear whether or not this choice introduces a large-company bias into our sample, and hence into our conceptual development, because there are many large, private e-business operations, and several large, public firms not included in our sample (e.g., AOL and Yahoo). Including only public companies in our sample may bias it towards surviving companies. While limitations on the availability of data prevent us from broadening the sample to firms that ‘failed’ (according to some definition of failure), we do not believe that the survival bias affects the theoretical development. First, some of the firms we studied will likely fail eventually. Second, the argument can be made for theoretical rather than random sampling of cases, and for studying ‘extreme situations and polar types in which the 9 The decision to include U.S. as well as European firms in our sample has several implications. E-business activity in Europe is dominated less by start-ups, as is the case in the United States, and more by established companies (Morgan Stanley Dean Witter, 1999). For example, the United Kingdom’s Freeserve is a spin-off of Dixons, a large ‘bricks-andmortar’ retailer, and Spain’s Terra Networks is a spin-off of Telefo´nica, a large telecommunication firm. An affiliation (past or present) with established companies probably influences the particular business models of respective e-business firms. For example, some spin-offs may benefit from the alliance network of their parent companies, while others may suffer from imposed organizational constraints. However, a possible sample bias toward (mostly former) subsidiaries of established companies should not affect our ability to develop a general framework for evaluating the value creation potential of ebusiness firms. In fact, such a general framework should be independent of the mode of business creation
502 R. Amit and C. Zott process of interest is transparently observable' which is available to the public online. Data on (Eisenhardt, 1989: 537) companies included in the data base adhere to a As implied by ling criterion (), we single, U.S. standard set by the SEC. In Europe focused our study on e-business firms that enabled however, there is no central data depository. In transactions in which individual consumers were addition, company reporting requirements vary involved. These companies are hereafter collec- across European countries, ranging from strict tively referred to as with-C companies. For (e.g, the United Kingdom) to relatively lax(e.g example, our sample included so-called 'B-to- Italy). European firms also vary widely in their C(business-to-consumer) companies, which are accounting and disclosure practices, making com companies that directly and exclusively engage parisons across firms difficult. This made the use in transactions with individual customers. We of multiple sources of information particularl did not sample businesses that solely engaged in importan commercial activities with other businesses(so- called 'B-to-B. or "business-to-business companies). We made this choice based primarily Data analysis on the fact that the quality of data available for In inductive studies, data analysis is often hard with-C' firms was higher than that available for to distinguish from data collection since building B-to-B firms at the time this research project theory that is grounded in the data is an iterative was launched o process in which the emergent frame is compared systematically with evidence from each case Data collection (Eisenhardt, 1989). Some researchers argue for a deliberate process of joint data collection and We gathered detailed data on our sample com- analysis (e.g, Glaser and Strauss, 1967). We panies mainly from publicly available sources: employed this joint process by frequently moving IPO prospectuses (our major source), annual between the data and the emerging theory as we eports, investment analysts'reports, and com- developed our model. The value driver categories panies' web sites. A structured questionnaire was derived from our preliminary analysis of the used to collect information about:(a) the com- initial data clearly influenced the design of the pany(e.g, founding date, size, lines of business, subsequent questionnaire that we used for further products and services provided, and some finan- data collection. cial data);(b)the nature and sequence of trans- We used standard techniques for both within actions that the firm enables (e.g, questions case analysis and cross-case analysis(Eisenhardt, included: 'What is the company's role in consum- 1989: Glaser and Strauss, 1967; Miles and Huber mating each transaction? 'and 'Who are the other man, 1984: Yin, 1989). Within-case evidence was players involved?);(c)potential sources of value acquired by taking notes rather than by writing creation(e.g, questions included: How important narratives. For this purpose, research analysts are complementary products or services?' and answered the questions enumerated in the ques- Are they part of the transaction offering?); and tionnaire, integrating and triangulating facts from (d)the firms strategy (e.g, questions included: the various data sources mentioned above. As How does the company position itself vis-a- observed by Yin(1981: 60),The final case vis competitors?). Most of the approximately 50 studies resembled comprehensive examinations questions enumerated in the questionnaire were rather than term papers. The authors then ana- open-ended, which was consistent with our pri- lyzed these products sequentially and indepen mary objective of developing a conceptual frame- work that was informed by empirical evidence Much high-quality data about U.S. firms was started with an initial version of the questionnaire that obtained from the SeC's edgar data base yas intended to bring focus and clarity to the questions This initial questionnaire had been pretested on several cases. Subsequently, we modified, added, and dropped ques- o We do not believe that our focus on with- c' firms seriously about 2 months into the research project, and made fects the theory development. The value driver categories revisions again about I month later. After every dentified in the analysis shou apply to'B-to-B model on, all cases that had hitherto been examined were albeit perhaps with different weights. Copyright o 2001 John Wiley Sons, Ltd, sma. Mgmt. J.22:493-520(2001)
502 R. Amit and C. Zott process of interest is transparently observable’ (Eisenhardt, 1989: 537). As implied by sampling criterion (c), we focused our study on e-business firms that enabled transactions in which individual consumers were involved. These companies are hereafter collectively referred to as ‘with-C’ companies. For example, our sample included so-called ‘B-toC’ (business-to-consumer) companies, which are companies that directly and exclusively engage in transactions with individual customers. We did not sample businesses that solely engaged in commercial activities with other businesses (socalled ‘B-to-B,’ or ‘business-to-business’ companies). We made this choice based primarily on the fact that the quality of data available for ‘with-C’ firms was higher than that available for ‘B-to-B’ firms at the time this research project was launched.10 Data collection We gathered detailed data on our sample companies mainly from publicly available sources: IPO prospectuses (our major source), annual reports, investment analysts’ reports, and companies’ web sites. A structured questionnaire was used to collect information about: (a) the company (e.g., founding date, size, lines of business, products and services provided, and some financial data); (b) the nature and sequence of transactions that the firm enables (e.g., questions included: ‘What is the company’s role in consummating each transaction?’ and ‘Who are the other players involved?’); (c) potential sources of value creation (e.g., questions included: ‘How important are complementary products or services?’ and ‘Are they part of the transaction offering?’); and (d) the firm’s strategy (e.g., questions included: ‘How does the company position itself vis-a`- vis competitors?’). Most of the approximately 50 questions enumerated in the questionnaire were open-ended, which was consistent with our primary objective of developing a conceptual framework that was informed by empirical evidence. Much high-quality data about U.S. firms was obtained from the SEC’s EDGAR data base, 10 We do not believe that our focus on ‘with-C’ firms seriously affects the theory development. The value driver categories identified in the analysis should also apply to ‘B-to-B’ models, albeit perhaps with different weights. Copyright 2001 John Wiley & Sons, Ltd. Strat. Mgmt. J., 22: 493–520 (2001) which is available to the public online. Data on companies included in the data base adhere to a single, U.S. standard set by the SEC. In Europe, however, there is no central data depository. In addition, company reporting requirements vary across European countries, ranging from strict (e.g., the United Kingdom) to relatively lax (e.g., Italy). European firms also vary widely in their accounting and disclosure practices, making comparisons across firms difficult. This made the use of multiple sources of information particularly important. Data analysis In inductive studies, data analysis is often hard to distinguish from data collection since building theory that is grounded in the data is an iterative process in which the emergent frame is compared systematically with evidence from each case (Eisenhardt, 1989). Some researchers argue for a deliberate process of joint data collection and analysis (e.g., Glaser and Strauss, 1967). We employed this joint process by frequently moving between the data and the emerging theory as we developed our model. The value driver categories derived from our preliminary analysis of the initial data clearly influenced the design of the subsequent questionnaire that we used for further data collection.11 We used standard techniques for both withincase analysis and cross-case analysis (Eisenhardt, 1989; Glaser and Strauss, 1967; Miles and Huberman, 1984; Yin, 1989). Within-case evidence was acquired by taking notes rather than by writing narratives. For this purpose, research analysts answered the questions enumerated in the questionnaire, integrating and triangulating facts from the various data sources mentioned above. As observed by Yin (1981: 60), ‘The final case studies resembled comprehensive examinations rather than term papers.’ The authors then analyzed these products sequentially and indepen- 11 We started with an initial version of the questionnaire that reflected a working framework we had already constructed. This was intended to bring focus and clarity to the questions asked. This initial questionnaire had been pretested on several cases. Subsequently, we modified, added, and dropped questions about 2 months into the research project, and made similar revisions again about 1 month later. After every revision, all cases that had hitherto been examined were updated accordingly