- Email: [email protected]
First-mover advantages in regimes of weak appropriability: the case of financial services innovations
Journal of Business Research 55 (2002) 997 – 1005
First-mover advantages in regimes of weak appropriability: the case of financial services innovations Luis E. Lo´peza,*, Edward B. Robertsb a
b
INCAE, Graduate School of Business, Apartado 960-4050, Alajuela, Costa Rica Alfred Sloan School of Management, Massachusetts Institute of Technology, Cambridge, MA, USA Received 21 January 2000; accepted 31 October 2000
Abstract A well-established stream of research indicates that there are advantages to early market entry demonstrated in several industries, particularly consumer products. Such empirical regularity has not been extensively tested in regimes of weak appropriability, such as the financial services industry. In this study, we use historical methods to analyze the effects of order of market entry on market share in the financial services industry in an international setting. Analyzing three lines of financial products, we find important market share advantages to early entry in financial services innovations. The traditional models of order of market entry do not utilize all the information that is revealed by a detailed qualitative analysis of longitudinal data, and they do not distinguish nearly simultaneous entries or entries separated by long periods of time. We find that a model using elapsed time since first entry renders stronger results and better interpretations. D 2002 Elsevier Science Inc. All rights reserved. Keywords: First-mover advantage; Early-mover advantage; Financial services; Appropriability; Product innovation
1. Introduction Pioneering new products can give firms competitive advantage (Lieberman and Montgomery, 1988). Numerous empirical findings report market share advantages to pioneers and early entrants (Kalyanaram and Urban,1992; Robinson and Fornell, 1985; Robinson, 1988; Whitten, 1979; Lambkin, 1988). Robinson et al. (1994), Kalyanaram et al. (1995), and VanderWerf and Mahon (1997) review this literature. Other studies, however, do not report advantages from pioneering new products (Golder and Tellis, 1993; Lilien and Yoon, 1990). Lilien and Yoon (1990), for instance, found that the likelihood of success for first and second entrants was lower than the likelihood of success for the third and fourth entrants in a sample of 112 new industrial products from 52 French firms. The contrasting results could stem from several reasons: methodological inconsistencies, entrant firms having diverse resources and capabilities (Kerin et al., 1992), dissimilar product characteristics among studies, or industry idiosyncrasies. Such causes are not clearly established in extant literature, because most * Corresponding author. Tel.: +506-437-2372/2389; fax: +506-4339101. E-mail address: [email protected] (L.E. Lo´pez).
studies are based on data from U.S. manufacturing firms and consumer packaged products. Studies with data from other industries in international settings are scarce. More studies are needed to test the relevance of first-mover results in different national locations (Lieberman and Montgomery, 1998) and different industries. In one of the few studies that deals with services, Tufano (1989) found, for a sample of 58 financial innovations, that there were pioneering advantages in the form of lower costs of trading, underwriting, and marketing. He showed that pioneers capture shares, which are ‘‘. . . almost 2.5 times as large as the followers in those markets’’ (Tufano, 1989, p. 231). These findings have not been confirmed with additional studies. The need to explore these issues is particularly interesting in services because there are differences between services and consumer or industrial products along many dimensions (intangibility, perishability, heterogeneity, and simultaneity; Bitran and Lojo, 1993), and service industries may have different underlying dynamics. In financial services, for example, two parameters are highly relevant to the dynamics of order of market entry: increased environmental volatility and, most importantly, a weak appropriability regime. Environmental volatility (frequent and abrupt changes in prices, the globalization of markets, regulatory changes, technological advances, and advances in financial
0148-2963/02/$ – see front matter D 2002 Elsevier Science Inc. All rights reserved. PII: S 0 1 4 8 - 2 9 6 3 ( 0 1 ) 0 0 2 0 0 - 4
998
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
theory) has induced the proliferation of many new financial products (Finnerty, 1988). The augmented changes in the speed, frequency, and magnitude of price variations require the creation of new financial schemes to manage risk. This results in a more rapid product diffusion and a reduced duration of the transient monopoly allowed to innovators. In this regard, Tufano (1989) reports in his study that rivals enter the market less than 1 year after the pioneer in 60% of the cases. Moreover, developing new financial products requires large financial outlays. Clearly, financial innovation, which implies the commitment of a great deal of resources into the development process, is undertaken as a profit-maximizing activity worth protecting from imitation (Tufano, 1989). Levin et al. (1987) argue that firms use alternative means to protect the competitive advantages of new or improved processes and products. They identify six mechanisms firms use to appropriate the returns of innovative activities: (1) patents to prevent duplication, (2) patents to secure royalty income, (3) secrecy, (4) lead time, (5) moving quickly down the learning curve, and (6) sales or service efforts. In financial services, appropriability is feeble. Little protection is allowed either in the form of copyright or trade secret laws. Very few patents have been awarded to financial products. When Merrill Lynch was awarded a patent for its Cash Management Account, the company was able to obtain it on the grounds that it was a novel computer program. In these weak appropriability regimes, new services are copied and improved after very short periods of time. Moreover, certain types of financial products, particularly consumeroriented products, once released, are relatively easy to imitate. Bitran and Lojo (1993, p. 272) report that: In the financial services industry, whenever an institution introduces a new type of account, others not only copy it very quickly, but often introduce an improved service since they have some time to assess the strengths and weaknesses of the original service. In order to be a leader in the service industry in terms of new product introduction, therefore, it is important to have the next service almost ready by the time the first one is introduced.
Tufano (1989) estimates that replicating a financial product costs 50 – 75% less than the innovator’s original investment. In the case of financial services in the U.S., the Securities and Exchange Commission (SEC) in many cases forces ‘‘inventors’’ to disclose detailed information about the products they create (Tufano, 1989). Even products that require intricate technological or institutional networking, which may render them a bit more difficult to replicate, are not necessarily difficult to copy. As Morone and Berg (1993, p.125) say, ‘‘. . . the relative benefits of pioneering vs. fast following are particularly open to question since information technology can often be quickly copied, and later entrants often enjoy the benefits of lower-cost hardware.’’ Hence, adding complexity to products through a more intensive use of information technology or other
means will not necessarily cause delays in product diffusion or in competitors’ movement down the learning curve. In spite of this, numerous financial instruments are constantly created and marketed. As Miller (1986, p. 459) indicates: ‘‘. . . the word revolution is entirely appropriate for describing the changes in financial institutions and instruments that have occurred in the past 20 years.’’ These innovations include securities (zero-coupon bonds and junk bonds), consumer-type financial instruments (debit cards and credit cards), process innovations (ATMs, automated teller machines; CMAs, cash management accounts; and EFTs, electronic fund transferring), and financial strategies (LBOs, leveraged buyouts; swaps; and corporate restructuring) (Finnerty, 1988; Miller, 1986). Given the relative inefficiency of some forms of appropriation in certain contexts, firms probably use a combination of means of appropriation, and they balance the relative inefficiency of some appropriation methods by placing a greater emphasis on other alternative methods. Since in financial services patent-related protection and secrecy are ineffective, and moving down the learning curve occurs almost instantaneously, then lead time and a quick response in introducing new products may be the most effective methods of appropriating the returns of innovation efforts. Hence, we would expect that, at least in financial services, pioneering should be an important form of appropriation if not the dominant one. Thus, financial services provide an interesting quasiexperimental setting for the study of innovation and, particularly, to look at the sustainability of firms’ competitive advantages over time in regimes with weak appropriability. Yet, the literature on pioneering advantages has primarily concentrated on U.S. industries, particularly packaged goods aimed at consumer markets, and it sheds little light upon the problem, as it applies to services in general and financial services in particular. The more traditional literature also contains a number of problematic aspects: (a) The dynamics of order of entry is sometimes evaluated disregarding the characteristics of firms that launch the new products. Thus, whether products fail because they lack strategic relation with other products in the firm’s portfolio, or for other reasons not related to order of entry, is seldom studied. (b) The first entrant is sometimes identified as a product that first ‘‘achieves national distribution’’ (e.g., Urban et al., 1986) or requires that the product be supported by national advertising (e.g., Whitten, 1979). ‘‘Hence, if a firm does not have national advertising or national distribution in a national market, it does not qualify as an entrant’’ (Robinson et al., 1994, p. 2). This might be problematic because indirectly it means that the product that qualifies as an entrant has been launched by a firm that has an important set of complementary assets (i.e., being able to nationally advertise or distribute), thus precluding many products launched by smaller firms. By using this methodology,
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
one runs the risk of confounding ‘‘entry’’ with financial muscle. The method also censors entrants that have failed to stay in the market or grow to national scope, thus overstating the effect of entry on the survivors. (c) Units of analysis vary. Some studies center their attention on products while others analyze organizations. Making generalizations from the literature as a whole is difficult. (d) Censoring: Many studies cannot or do not consider the products’ entire history, thus overestimating or underestimating pioneering. (e) A rapidly changing market may defy order of entry dominance, especially if subsequent entrants in a very volatile environment are able to incorporate into their products characteristics that take such changes into account. (f) Finally, premature market conditions (i.e., pioneering ‘‘too early’’) may also obscure the effect of order of entry. Pioneers may end up bearing the weight of educating future consumers or creating network externalities yet unconsciously level the field for others to invade it. As a result, though empirical results appear strong, empirical generalization is controversial. Some of this controversy is discussed by Golder and Tellis (1993) who, using an historical approach, argue that moving first achieves no advantage. On the contrary, they show for a variety of products that almost all of those who have moved first have died in their youth. Golder and Tellis do not require the pioneer to reach any specified scale of commercialization and many of those pioneers do not have a meaningful share of the market in the long run. In their methodology, Golder and Tellis gather information from articles some as far back as 1842 to prove their point. Though this historical approach solves the problem of reverse causality (i.e., high market share firms being identified quasiautomatically as pioneers), it can also overlook other entries into the market that are not well documented. Such aspects should be taken into account when choosing an appropriate methodology to study the phenomenon in financial services. In this study, we investigate the relationship between order of market entry and market share for a set of financial services innovations in an international setting. We propose that early entry advantages will be negligible and, if there are any, they will erode quickly; much more rapidly than they would in manufactured consumer packaged goods.
2. Methods The study is based primarily on historical analyses. Product histories for several financial products were assembled and reconstructed from archival records. These records were available in published sources of information and from other sources such as regulatory institutions. A manual search of all the issues of the most important local business, financial, and economic journals since 1985 (and a selective search of some older issues) was coupled to both electronic and manual searches of the countries’ most
999
important newspapers. We also searched books, papers, and monographs. Over 100 articles or printed pieces of information, which referenced the product lines under scrutiny, were assembled. The historic information was validated with data from institutional informants and industry experts. Contradictory information was resolved through a third piece of confirmatory evidence or with clarification from industry experts. We avoided overreliance on single informants within organizations who might overstate pioneering status or whose recollections might not be entirely accurate. Multiple measures were collected both from organizational informants and from secondary sources, allowing us to cross-validate data from different origins (scholars, experts, and journalists). The sample chosen is a set of financial services launched in Costa Rica during a 15-year period (1980 –1995). Thus, the sample of products was circumscribed purposefully to a small market under progressive deregulation. This was also done with the intention of (a) correctly identifying pioneers, (b) avoiding censoring, and (c) choosing products launched in roughly similar time-frames (to control for potentially different market effects). Historical information could be obtained, both from public records and from people, to permit an accurate determination of pioneering and subsequent entries. Although historical analysis is little used in studies of pioneering carried out in the marketing field, it has been widely used in studies of technological innovation (Cooper and Schendel, 1976; Meyer and Roberts, 1986; Sanderson and Uzumeri, 1995; Utterback, 1994). An important disadvantage with the chosen sample of products is that we may in fact be right-censoring the dependent variable, since it is plausible that not enough time has elapsed for assessing the survival and performance of the different market entrants. In this case, the long term may not yet be long enough. Another concern that might arise with the chosen sample is the extent to which any finding can be generalized to other industries or countries. The sample was chosen for convenience. It allowed us to gather large amounts of detailed data for a relatively extensive and meaningful period of time. Using this methodology, it was possible to circumvent many of the mentioned methodological concerns often associated with much of the literature on pioneering. The purpose, thus, is not to make sweeping claims about the generalizability of our findings but to check whether early-entry advantages erode faster in regimes of weak appropriability. To do so, the sample and setting chosen become a convenient quasiexperimental site. Ultimately, whether these findings can be generalized can only be determined by performing similar studies in other settings.
3. Variables The dependent variable in this study is the total market share of the nth entrant at the time of the study, measured as
1000
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
the percentage of total customers using a particular financial instrument. The order of market entry is used as an independent variable. Pioneers are operationalized as the first entrant. An additional independent variable is the elapsed time delay from the first entry until the nth entry.
financial services (from relatively sophisticated corporate financial services to many retail products, which included, at the time, telephone banking that would later evolve to internet banking, extensive ATM networks, linked cash management accounts, and investment services such as money market mutual funds, retirement accounts, etc.).
4. Results
4.1.1. Pension funds Pension funds in this market had traditionally been administered by the state. Economically active people contributed a fixed percentage of their salaries to a national fund. In such environments, private pension funds tend to be successful if the state-operated pension system gives small pensions at high retirement ages. Problems with the stateadministered fund opened an opportunity for other institutions to enter the market and offer complementary pensions. This industry operated with little supervision until July 1995. In fact, before that time, private pension funds could not operate openly as such but only as providers of additional coverage to that given by the state. Even so, the first private pension funds operator entered the market in August 1988. In 1995, the funds were authorized to operate as such and also received an additional impulse in the form of a tax break by which the equivalent of a portion of the funds saved could be deducted from income tax. Ten more operators entered the market between 1991 and September 1996. Different types of institutions entered this market. The most prominent participants were banks, but there was also one insurance company and one firm that operated pension funds as its sole product. Fig. 1 shows a longitudinal history of total market shares (measured as percentage of total customers) for all pension funds participants in the period under study. The pioneer in this market operated alone for a total of 23 months. Firms’ market shares start leveling off after approximately 4 years after the first entry. According to these graphs and data, an early pension funds entrant gained a maximum market share
4.1. Qualitative data analysis Costa Rica has a small economy with a GDP of US$24 billion and per capita income of US$6700, both 1998 purchasing power-adjusted figures, a population of 3.7 million, and an area slightly smaller than West Virginia. The Costa Rican financial system underwent a modernization process that started in the mid-1980s. Before that time, the financial landscape was entirely dominated by stateowned financial intermediaries. Private banks were not permitted by law to operate. The Central Bank intervened establishing credit limits for different economic activities, limits to resource allocation by type of economic activity, and interest rates. The effect of this intervention was to hamper product and process innovation in financial institutions. Banks, for instance, limited themselves to lending once and again to a limited customer base. The little innovation that actually occurred was done with the purpose of circumventing the restrictive legislation. The reforms to the Costa Rican financial system sought to increase competition and promote market-driven prices for financial services. The deregulation process that started in the 1980s permitted the operation of new banks and financial institutions. Thus, the relative importance of private institutions increased. The participation of private banks, for instance, in total lending went from 15.3% in 1986 to 29.1% in 1990 (Gonza´les and Mesalles, 1993). By 1995, there were 24 banks operating in the country, offering a wide array of
Fig. 1. Order of market entry vs. market share for pension funds.
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
over time that is approximately 50% of the pioneer’s share. In this case, firms that entered the market within an elapsed time delay between 27 and 57 months eventually averaged 57% of the pioneer’s share. Very late entrants appear to have stabilized with market shares that represent approximately 5 –10% of the pioneer’s share (8% in this case). When percentage of market share is measured in terms of portfolio size, the pioneer also appears to retain an important portion of the market. Spearman-rank correlation coefficients showed a strong negative relationship between order of market entry and market share. 4.1.2. Credit cards The first credit card plan in the financial system under study was created by a commercial bank in 1976 (a relatively early entrance even by U.S. standards). The second entrant appeared in the market after 108 months. In this environment, pioneering advantages may have been obtained in terms of network externalities. Card operators derive income from each customer’s interest payments and also from the commissions paid by vendors, thereby making it crucial to develop a large network of affiliated businesses. Fig. 2 displays market shares for the credit card market measured as a percentage of the pioneer’s share (as percentage of total customers). As shown, the very early first entry does not sustain an above average market share over time. The second entrant (108 months after) captures approximately 45% of the market and retains it consistently. In general, however, the data seem to support a negative relationship between order of market entry and market share. Notwithstanding the pioneer’s period of approximately 7 years operating as a monopolist, it fails to reap long-term market share advantages. In this case, the second entrant became (and stayed) the market leader. Later, another three competitors entered the market during a 3-year period. The
1001
pioneer retains the number of customers (in fact increases its number of customers somewhat consistently) despite losing market share. Spearman-rank correlation coefficients showed a strong negative relationship between elapsed time since first entry and market share in the credit card business. 4.1.3. Debit cards Debit cards were introduced into the financial system later than the other two products under study. This instrument requires its holder to have a checking account with the issuing institution, with the debit card serving to automatically and electronically deduct payments from the checking account. The product requires an extensive network of affiliated businesses. Moreover, debit cards become more useful if they can also be used to obtain cash from ATMs or from bank offices. The first debit card entered the market in August 1987. Six additional operators had entered the market by 1996. All debit card institutions are commercial banks, and there is a fairly short elapsed time between the pioneer and the next two entrants. A wave of new entrants occurred in 1996, caused by regulatory changes that allowed all banks to provide regular checking accounts. This change made it unnecessary for banks to circumvent this regulatory constraint, which they often did through alternative and more creative means. No longitudinal data could be obtained for this product. A brief examination of available descriptive data showed that, strictly speaking, the pioneer does not have any advantage. It is the third entrant who is able to capture market share. The pioneer’s market share, in this case, is approximately one-sixth of the third entrant’s market share. A late entrant (with a total elapsed time of 106 months after the pioneer’s entry) appears to capture approximately 50% of the pioneer’s share in a relatively short period of time. Here, market share is measured as percentage of customers
Fig. 2. Market shares for credit cards.
1002
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
for all participants in the debit card industry at the end of the time period studied. Notice that the effect of entry may be influenced by the time elapsed since the pioneer’s entry. The absolute order of market entry could hide nearly simultaneous entries or entries, which are separated by long periods of time. Fig. 3 shows total market share at the end of the time period studied against elapsed time since the pioneer’s entry. Two groups can be recognized clearly in the graph. One group of early entrants introduced debit cards into the market within a total elapsed time of 17 months. The next entry occurred 58 months later, and there is a second group of institutions entering the market more than 8 years after the pioneer’s entry. Hence, although we cannot speak, in a strict sense, of pioneering advantages, there are advantages to early entry. The data on debit cards indicate the potential inconveniences of utilizing strict order of market entry as an independent variable. It might be preferable to moderate this strict order by coding entries as early entries, early followers, and so on, or alternatively, time elapsed since first entry could be used as an independent variable. This notwithstanding, there is, for the case of debit card introduction, a statistically significant relationship between order of market entry and long-term market share. 4.2. Quantitative data analysis To complement the qualitative data analysis, we constructed a data set with the data points collected in all three product categories. This data set included observations for a total of 36 entries in three product categories with the following variables: ENTRY (numerical variable containing the order of market entry), SHARE (percentage of market share for nth entry measured as percentage of total customers), ELAPSED (elapsed time in months for particular entry since first entry in the particular category), SIZE (ratio that gives relative size of entrant with respect to the largest entrant in the product category, calculated based on total
number of employees), INNOV (measures degree of dispersion of product offerings for that particular institution), and REGUL (dichotomous variable that indicated if the product was driven by regulatory changes). Our goal in this part of the paper was to build a baseline model for the dependent performance variable SHARE, controlling for institution size, the effect of the institution’s total product offerings, and the effects of regulation. We then constructed a model to include question predictors ENTRY and ELAPSED. After examining the raw data set and the distributions of all variables, inspecting bivariate scatterplots, and studying bivariate correlations (estimated through Pearson correlation coefficients), a taxonomy of multiple regression models was fitted. The variables SIZE (transformed to account for curvilinearities), INNOV, and REGUL were classified as control predictors. ENTRY and ELAPSED were established as question predictors. Three regression models were subsequently fitted with the intention of creating an appropriate baseline model. First, SHARE was regressed on ENTRY. Second, SHARE was regressed on ELAPSED, and, finally, both variables were used as predictors in the regression analysis. After carrying out sensitivity analyses and performing a variety of tests, and after validating the assumptions about normality of the residuals and independence of errors, we were left with a somewhat unexpected result: our best and chosen model was one that simply regressed SHARE against ENTRY. We chose that model as the final model (intercept= 1.6, P<.001; coefficient of ENTRY= .24, P<.001; r2=69.67%; df=32) for the purpose of discussion. We then proceeded, in a similar fashion, to fit a model in which SIZE, INNOV, and REGUL were included as control predictors and ELAPSED as a question predictor, controlling also for interactions among these variables. The second final and most parsimonious fitted model regresses SHARE on ELAPSED (intercept= 1.82, P<.001; coefficient of ELAPSED= .015, P<.001). Data limitations did not permit a more exhaustive empirical analysis. The results obtained nevertheless provide some interesting insights that complement our qualitative observations.
5. Summary and conclusion
Fig. 3. Elapsed time since pioneer’s entry in months vs. total market share for debit cards.
We wanted to determine whether innovation initiatives in the development of new financial services might gain advantages from early entry. We expected that in a regime of such weak appropriability, no such advantage would be accrued or, if any advantage was attained at all, that it would erode very quickly. We obtained two groups of results. Our quantitative analysis, however limited, turned out to be in line with most of the prior research work that has been performed in other industries, particularly consumer products. We found that a baseline model in which SHARE was regressed against order of market entry (ENTRY) explained almost 70% of the variability in performance. Including
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
Fig. 4. Results of final model of order of market entry vs. market share for a prototypical financial services case.
control variables such as the firm’s size and the amplitude of its product offerings did not have a sizable effect upon our baseline model. Fig. 4 is a visual display of our model results for different orders of market entry. Our findings indicate a negative relationship between order of market entry and long-term market share. A second entrant is likely to capture approximately 78% of the pioneer’s eventual market share. A 10th entrant will only capture 11% on average. The results appear to be in line with those found in other industries. Table 1 establishes such a comparison. This rather conventional analysis leads us to conclude primarily that there are important advantages to early entry in financial service innovations. This result extends established generalizations to other industries and settings in a field that ‘‘relies heavily on North American manufactured goods that are included in either the Urban et al. Assessor data or the PIMS data’’ (Kalyanaram et al., 1995, p. 9). Such results, however, need to be qualified. We see that using order of market entry alone might render somewhat equivocal results because nearly simultaneous entries or entries separated by large periods of time are not evident. Results need to be moderated by a sense of real time between entries and, for such purpose, interpreting entry utilizing real elapsed time is very illustrative. As we can see
1003
in Fig. 5, if subsequent entrants invade the market within short periods of time after the pioneer, pioneering advantages are scant. Shorter lead times mean pioneers have less time to establish their brand names and other distribution advantages. This means that the earlier competitive rivalry starts, the larger later entrants’ market shares. From the pioneer’s perspective, lengthening lead time by 1 year could represent, according to our data, as much as 2.5% of overall market share. Use of elapsed time since first entry also adds to a managerial interpretation of these models. For our sample of products, we see that entrants, which follow shortly after the pioneer, may obtain substantial market shares relative to the pioneer, a result that is not obvious by looking at order of market entry alone. We were not able to collect reliable data on product profitability. It is entirely possible that certain institutions capture a large market share and not add value to the organization. Unfortunately, no detailed data that could provide an indication of product profitability was available. Subsequent studies should address this limitation. Insofar as our proposition goes, early entry advantages do not seem to erode any quicker in this regime of low appropriability than they would in other industries. These results coincide with the work of Makadok (1998) who found that early movers and pioneers enjoy both a highly sustainable pricing advantage and a moderately sustainable market share advantage in the U.S. Money Market Mutual Fund Industry. Although no claim of generalizability can be made, the results we obtained here appear to also hold in the different quasiexperimental setting, in a G7 locale (with a similar appropriability regime) such as the one explored by Makadok. These results, however, need to be examined in the light of the qualitative data. If we look at individual products, establish their longitudinal histories, and operationalize pioneering as the first entrant, we find that the pioneers, strictly speaking, do not necessarily retain an appreciable market share advantage. In fact, no important pioneering advantage is observed in two of the three product lines explored. This is consistent with the exposition of Golder and Tellis (1993) although in our case the product histories are more concise
Table 1 Order of market entry and market share for consumer packaged goods, prescription antiulcer drugs, and financial services innovations Entry order
Forecasted market share relative to the pioneering brand Consumer packaged goods
First Second Third Fourth Fifth Sixth
Prescription antiulcer drugs
Financial products
Urban et al. (1986)
Kalyanaram and Urban (1992)
Berndt et al. (1994)
This study
1.00 0.71 0.58 0.51 0.45 0.41
1.00 0.76 0.64 0.57 0.53 0.49
1.00 0.70 0.57 0.49 0.44 0.40
1 0.79 0.62 0.49 0.38 0.30
Modified from Kalyanaram et al. (1995).
1004
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
Fig. 5. Results of final model of elapsed time since first entry vs. market share for a prototypical financial services case.
because of the youth of the product lines studied. On the other hand, when our product sample is aggregated, an important early entry effect is observed. We get results that are in line with much other literature on first-mover advantages. Evidently, by aggregating the data, we can only affirm the advantages of early entry and not those of pioneering, in the strict sense of the word. History reveals a more detailed picture that cannot be captured completely in cross-sectional studies. Moreover, this raises the possibility that studies that use aggregated information from extant databases could have a built-in bias that is methodologically difficult to identify and isolate based solely on such information. It is also of interest to speculate why the first entrant does not retain a long-term market share advantage in two of the three products studied but does sustain its advantage in pension funds. To do so, we carried on additional interviews with several industry experts who gave us two categories of possible causes. One set of explanations has to do with how firms approach their new product launches. Another set of explanations is associated with environmental contingencies. Some experts argued that a pioneer could have an advantage in acquiring and developing resources. Given enough lead time, a first-mover may spend more on acquiring resources (personnel, brand equity, technological abilities, and the like), thus preempting or relegating subsequent entrants to weak positions. Such preeminence might be tightly linked to the time elapsed between first and subsequent entries. The longer the time elapsed, the longer the cumulative effect of spending on advertising, technology, and training. Moreover, early entrants are likely to attain enough volume (earlier) that will trigger scale effects that will eventually render an insurmountable cost advantage. This appears to be the case of pension funds. Another key aspect that, according to industry experts, permits sustaining early entry advantages is access to an established base of existing customers and to an array of complementary services. Customers may prefer to do business with an organization that offers an ample array of products. Thus, they may be inclined to switch to, or
reluctant to switch from, certain institution’s products because of the added convenience of carrying on all businesses with one organization. Hence, one product’s early entry advantage should be evaluated in the light of how it fits within the complete product portfolio of the institution. Our data do not permit us to formally test this. However, the data for pension funds do provide some insight. The only standalone pension fund was unable to sustain an early entry advantage and was overcome, for instance, by a sixth entrant, which was a large bank. Evidently, although the financial product may be very easy to imitate, the existing customer base of certain institutions, and the opportunity to create unique bundles of products, will provide the basis for sustaining an early-entry advantage. As some industry experts indicated, complete product platforms are much harder to imitate than a single product. The internal procedure for researching and developing the new product was also mentioned by industry experts we consulted. To investigate this, we asked several bank executives to determine how pioneering products had been developed within their institutions and to rate their success. We listed a total of 46 entries in several product categories. These products included some, but not all, of those used in the previously analyzed 36 entries into pension funds, debit cards, and credit cards. We observed that pioneering products were developed differently. Some (about 24% of the cases listed) were generated in an orderly and formalized manner with a somewhat structured process of market research and a subsequent development process aimed at satisfying certain market needs. Most products (44%) were adaptations of ideas that had been seen operating in other countries. A third large category of products stemmed from regulatory changes (13%). We asked executives to rate the success of these products. They indicated that more formalized processes of research and development do have an important impact upon performance, particularly when products are adequately conceptualized as part of a wellstructured product strategy and product platform. Another explanation deals with environmental contingencies, particularly technological change. In the case of credit cards, the second entrant entered very aggressively and with a technology that clearly leapfrogged earlier investments. This permitted this entrant to develop a network of affiliated businesses that eventually lured customers who then found it too costly to switch to other services. Similarly, the long-term sustainability within the debit card business was driven by scale effects and network externalities. For the debit card sample, there is a statistically significant negative relationship between entry and two measures of institution size (Spearman-rank=.86, P<.01 for SIZE1 and Spearman-rank= .96, P<.001) and there is a positive relationship between a measure of network externalities (number of available ATMs and number of clients (Spearman-rank=.88, P<.01). Finally, it appears to be useful and important to consider elapsed time since first entry when evaluating these models.
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
These effects have been, to the best of our knowledge, explored only in two studies. Huff and Robinson (1994) determined that increasing the years of competitive rivalry (i.e., reducing the pioneer’s lead time) tended to reduce the pioneer’s market share advantage. Furthermore, Hurwitz and Caves (1988) examined 56 pharmaceutical products and their performance after patent expiration, reporting that longer patent life (i.e., longer pioneer lead time) renders appreciable market share advantages.
Acknowledgements We appreciate funding assistance provided by INCAE, the MIT International Center for Research on the Management of Technology (ICRMOT), and the MIT Center for Innovation in Product Development (National Science Foundation Grant #EEC-9529140). We thank Professors Donald Lessard, Scott Stern and two anonymous reviewers for their many thoughtful comments and suggestions.
References Berndt ER, Bui LT, Reilly DH, Urban GL. The roles of marketing, product quality, and price competition in the growth and composition of the U.S. anti-ulcer drug industry. Working Paper 19-94, Massachusetts Institute of Technology, 1994. Bitran GR, Lojo M. A framework for analyzing service operations. Eur Manage J 1993;11(3):271 – 82. Cooper AC, Schendel D. Strategic responses to technological threats. Bus Horiz 1976;19(1):61 – 9. Finnerty JD. Financial engineering in corporate finance: an overview. Financ Manage 1988;17:14 – 33 (Winter). Golder PN, Tellis GJ. Pioneer advantage: marketing logic or marketing legend? J Mark Res 1993;30:158 – 70 (May). Gonza´lez C, Mesalles L. La economı´a polı´tica de la nacionalizacio´n bancaria en Costa Rica: 1948 – 1990. In: Gonza´les C, Vargas T, editors. Reforma financiera en Costa Rica, perspectivas y propuestas. San Jose´, Costa Rica: Imprenta LIL, 1993. p. 3 – 54. Huff LC, Robinson WT. The impact of leadtime and years of competitive rivalry on pioneer market share advantages. Manage Sci 1994;40(10): 1370 – 7. Hurwitz MA, Caves RE. Persuasion or information? Promotion and the shares of brand name and generic pharmaceuticals. J Law Econ 1988; 31:299 – 320 (October). Kalyanaram G, Urban GL. Dynamic effects of the order of entry on market share, trial penetration, and repeat purchases for frequently purchased consumer goods. Mark Sci 1992;11(3):235 – 50 (Summer).
1005
Kalyanaram G, Robinson WT, Urban GT. Order of market entry: established empirical generalizations, emerging empirical generalizations, and future research. Mark Sci 1995;14(3):G212 – 21 (Summer). Kerin RA, Varadarajan PR, Peterson RA. First-mover advantage: a synthesis, conceptual framework, and research propositions. J Mark 1992; 56:33 – 52 (October). Lambkin M. Order of entry and performance in new markets. Strategic Manage J 1988;9:127 – 40 (Summer). Levin RC, Klevorick AK, Nelson RR, Winter SG. Appropriating the returns from industrial research and development. Brookings Pap Econ Act 1987;3:783 – 820. Lieberman MB, Montgomery DB. First-mover advantages. Strategic Manage J 1988;9:41 – 58 (Summer). Lieberman MB, Montgomery DB. First-mover (dis)advantages: retrospective and link with the resource-based view. Strategic Manage J 1998; 19(12):1111 – 25. Lilien G, Yoon E. The timing of competitive market entry: an exploratory study of new industrial products. Manage Sci 1990;36(5): 568 – 85 (May). Makadok R. Can first-mover and early-mover advantages be sustained in an industry with low barriers to entry/imitation? Strategic Manage J 1998; 19(7):683 – 96. Meyer M, Roberts E. New product strategy in small technology-based firms: a pilot study. Manage Sci 1986;32(7):806 – 22. Miller MH. Financial innovation: the last twenty years and the next. J Financ Quant Anal 1986;21(4):459 – 71. Morone J, Berg D. Management of technology in the service sector: practices in the banking industry. J High Technol Manage Res 1993;4(1): 123 – 37. Robinson WT. Sources of market pioneer advantages: the case of industrial goods industries. J Mark Res 1988;25(1):87 – 94. Robinson WT, Fornell C. Sources of market pioneer advantages in consumer goods industries. J Mark Res 1985;22:305 – 17 (August). Robinson WT, Kalyanaram G, Urban GL. First-mover advantages from pioneering new markets: a survey of empirical evidence. Rev Ind Organ 1994;9:1 – 23. Sanderson S, Uzumeri M. Managing product families: the case of the Sony Walkman. Res Policy 1995;24(4):761 – 82. Tufano P. Financial innovation and first-mover advantages. J Financ Econ 1989;25(2):213 – 40. Urban GT, Carter T, Gaskin S, Mucha Z. Market share rewards to pioneering brands: an empirical analysis and strategic implications. Manage Sci 1986;32(6):645 – 59. Utterback JM. Mastering the dynamics of innovation. Boston, MA: Harvard Business School Press, 1994. VanderWerf PA, Mahon JF. Meta-analysis of the impact of research methods on findings of first-mover advantage. Manage Sci 1997;43(11): 1510 – 9. Whitten IT. Brand performance in the cigarette industry and the advantage to early entry, 1913 – 74. Staff Report to the Federal Trade Commission 1979; June.
First-mover advantages in regimes of weak appropriability: the case of financial services innovations Luis E. Lo´peza,*, Edward B. Robertsb a
b
INCAE, Graduate School of Business, Apartado 960-4050, Alajuela, Costa Rica Alfred Sloan School of Management, Massachusetts Institute of Technology, Cambridge, MA, USA Received 21 January 2000; accepted 31 October 2000
Abstract A well-established stream of research indicates that there are advantages to early market entry demonstrated in several industries, particularly consumer products. Such empirical regularity has not been extensively tested in regimes of weak appropriability, such as the financial services industry. In this study, we use historical methods to analyze the effects of order of market entry on market share in the financial services industry in an international setting. Analyzing three lines of financial products, we find important market share advantages to early entry in financial services innovations. The traditional models of order of market entry do not utilize all the information that is revealed by a detailed qualitative analysis of longitudinal data, and they do not distinguish nearly simultaneous entries or entries separated by long periods of time. We find that a model using elapsed time since first entry renders stronger results and better interpretations. D 2002 Elsevier Science Inc. All rights reserved. Keywords: First-mover advantage; Early-mover advantage; Financial services; Appropriability; Product innovation
1. Introduction Pioneering new products can give firms competitive advantage (Lieberman and Montgomery, 1988). Numerous empirical findings report market share advantages to pioneers and early entrants (Kalyanaram and Urban,1992; Robinson and Fornell, 1985; Robinson, 1988; Whitten, 1979; Lambkin, 1988). Robinson et al. (1994), Kalyanaram et al. (1995), and VanderWerf and Mahon (1997) review this literature. Other studies, however, do not report advantages from pioneering new products (Golder and Tellis, 1993; Lilien and Yoon, 1990). Lilien and Yoon (1990), for instance, found that the likelihood of success for first and second entrants was lower than the likelihood of success for the third and fourth entrants in a sample of 112 new industrial products from 52 French firms. The contrasting results could stem from several reasons: methodological inconsistencies, entrant firms having diverse resources and capabilities (Kerin et al., 1992), dissimilar product characteristics among studies, or industry idiosyncrasies. Such causes are not clearly established in extant literature, because most * Corresponding author. Tel.: +506-437-2372/2389; fax: +506-4339101. E-mail address: [email protected] (L.E. Lo´pez).
studies are based on data from U.S. manufacturing firms and consumer packaged products. Studies with data from other industries in international settings are scarce. More studies are needed to test the relevance of first-mover results in different national locations (Lieberman and Montgomery, 1998) and different industries. In one of the few studies that deals with services, Tufano (1989) found, for a sample of 58 financial innovations, that there were pioneering advantages in the form of lower costs of trading, underwriting, and marketing. He showed that pioneers capture shares, which are ‘‘. . . almost 2.5 times as large as the followers in those markets’’ (Tufano, 1989, p. 231). These findings have not been confirmed with additional studies. The need to explore these issues is particularly interesting in services because there are differences between services and consumer or industrial products along many dimensions (intangibility, perishability, heterogeneity, and simultaneity; Bitran and Lojo, 1993), and service industries may have different underlying dynamics. In financial services, for example, two parameters are highly relevant to the dynamics of order of market entry: increased environmental volatility and, most importantly, a weak appropriability regime. Environmental volatility (frequent and abrupt changes in prices, the globalization of markets, regulatory changes, technological advances, and advances in financial
0148-2963/02/$ – see front matter D 2002 Elsevier Science Inc. All rights reserved. PII: S 0 1 4 8 - 2 9 6 3 ( 0 1 ) 0 0 2 0 0 - 4
998
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
theory) has induced the proliferation of many new financial products (Finnerty, 1988). The augmented changes in the speed, frequency, and magnitude of price variations require the creation of new financial schemes to manage risk. This results in a more rapid product diffusion and a reduced duration of the transient monopoly allowed to innovators. In this regard, Tufano (1989) reports in his study that rivals enter the market less than 1 year after the pioneer in 60% of the cases. Moreover, developing new financial products requires large financial outlays. Clearly, financial innovation, which implies the commitment of a great deal of resources into the development process, is undertaken as a profit-maximizing activity worth protecting from imitation (Tufano, 1989). Levin et al. (1987) argue that firms use alternative means to protect the competitive advantages of new or improved processes and products. They identify six mechanisms firms use to appropriate the returns of innovative activities: (1) patents to prevent duplication, (2) patents to secure royalty income, (3) secrecy, (4) lead time, (5) moving quickly down the learning curve, and (6) sales or service efforts. In financial services, appropriability is feeble. Little protection is allowed either in the form of copyright or trade secret laws. Very few patents have been awarded to financial products. When Merrill Lynch was awarded a patent for its Cash Management Account, the company was able to obtain it on the grounds that it was a novel computer program. In these weak appropriability regimes, new services are copied and improved after very short periods of time. Moreover, certain types of financial products, particularly consumeroriented products, once released, are relatively easy to imitate. Bitran and Lojo (1993, p. 272) report that: In the financial services industry, whenever an institution introduces a new type of account, others not only copy it very quickly, but often introduce an improved service since they have some time to assess the strengths and weaknesses of the original service. In order to be a leader in the service industry in terms of new product introduction, therefore, it is important to have the next service almost ready by the time the first one is introduced.
Tufano (1989) estimates that replicating a financial product costs 50 – 75% less than the innovator’s original investment. In the case of financial services in the U.S., the Securities and Exchange Commission (SEC) in many cases forces ‘‘inventors’’ to disclose detailed information about the products they create (Tufano, 1989). Even products that require intricate technological or institutional networking, which may render them a bit more difficult to replicate, are not necessarily difficult to copy. As Morone and Berg (1993, p.125) say, ‘‘. . . the relative benefits of pioneering vs. fast following are particularly open to question since information technology can often be quickly copied, and later entrants often enjoy the benefits of lower-cost hardware.’’ Hence, adding complexity to products through a more intensive use of information technology or other
means will not necessarily cause delays in product diffusion or in competitors’ movement down the learning curve. In spite of this, numerous financial instruments are constantly created and marketed. As Miller (1986, p. 459) indicates: ‘‘. . . the word revolution is entirely appropriate for describing the changes in financial institutions and instruments that have occurred in the past 20 years.’’ These innovations include securities (zero-coupon bonds and junk bonds), consumer-type financial instruments (debit cards and credit cards), process innovations (ATMs, automated teller machines; CMAs, cash management accounts; and EFTs, electronic fund transferring), and financial strategies (LBOs, leveraged buyouts; swaps; and corporate restructuring) (Finnerty, 1988; Miller, 1986). Given the relative inefficiency of some forms of appropriation in certain contexts, firms probably use a combination of means of appropriation, and they balance the relative inefficiency of some appropriation methods by placing a greater emphasis on other alternative methods. Since in financial services patent-related protection and secrecy are ineffective, and moving down the learning curve occurs almost instantaneously, then lead time and a quick response in introducing new products may be the most effective methods of appropriating the returns of innovation efforts. Hence, we would expect that, at least in financial services, pioneering should be an important form of appropriation if not the dominant one. Thus, financial services provide an interesting quasiexperimental setting for the study of innovation and, particularly, to look at the sustainability of firms’ competitive advantages over time in regimes with weak appropriability. Yet, the literature on pioneering advantages has primarily concentrated on U.S. industries, particularly packaged goods aimed at consumer markets, and it sheds little light upon the problem, as it applies to services in general and financial services in particular. The more traditional literature also contains a number of problematic aspects: (a) The dynamics of order of entry is sometimes evaluated disregarding the characteristics of firms that launch the new products. Thus, whether products fail because they lack strategic relation with other products in the firm’s portfolio, or for other reasons not related to order of entry, is seldom studied. (b) The first entrant is sometimes identified as a product that first ‘‘achieves national distribution’’ (e.g., Urban et al., 1986) or requires that the product be supported by national advertising (e.g., Whitten, 1979). ‘‘Hence, if a firm does not have national advertising or national distribution in a national market, it does not qualify as an entrant’’ (Robinson et al., 1994, p. 2). This might be problematic because indirectly it means that the product that qualifies as an entrant has been launched by a firm that has an important set of complementary assets (i.e., being able to nationally advertise or distribute), thus precluding many products launched by smaller firms. By using this methodology,
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
one runs the risk of confounding ‘‘entry’’ with financial muscle. The method also censors entrants that have failed to stay in the market or grow to national scope, thus overstating the effect of entry on the survivors. (c) Units of analysis vary. Some studies center their attention on products while others analyze organizations. Making generalizations from the literature as a whole is difficult. (d) Censoring: Many studies cannot or do not consider the products’ entire history, thus overestimating or underestimating pioneering. (e) A rapidly changing market may defy order of entry dominance, especially if subsequent entrants in a very volatile environment are able to incorporate into their products characteristics that take such changes into account. (f) Finally, premature market conditions (i.e., pioneering ‘‘too early’’) may also obscure the effect of order of entry. Pioneers may end up bearing the weight of educating future consumers or creating network externalities yet unconsciously level the field for others to invade it. As a result, though empirical results appear strong, empirical generalization is controversial. Some of this controversy is discussed by Golder and Tellis (1993) who, using an historical approach, argue that moving first achieves no advantage. On the contrary, they show for a variety of products that almost all of those who have moved first have died in their youth. Golder and Tellis do not require the pioneer to reach any specified scale of commercialization and many of those pioneers do not have a meaningful share of the market in the long run. In their methodology, Golder and Tellis gather information from articles some as far back as 1842 to prove their point. Though this historical approach solves the problem of reverse causality (i.e., high market share firms being identified quasiautomatically as pioneers), it can also overlook other entries into the market that are not well documented. Such aspects should be taken into account when choosing an appropriate methodology to study the phenomenon in financial services. In this study, we investigate the relationship between order of market entry and market share for a set of financial services innovations in an international setting. We propose that early entry advantages will be negligible and, if there are any, they will erode quickly; much more rapidly than they would in manufactured consumer packaged goods.
2. Methods The study is based primarily on historical analyses. Product histories for several financial products were assembled and reconstructed from archival records. These records were available in published sources of information and from other sources such as regulatory institutions. A manual search of all the issues of the most important local business, financial, and economic journals since 1985 (and a selective search of some older issues) was coupled to both electronic and manual searches of the countries’ most
999
important newspapers. We also searched books, papers, and monographs. Over 100 articles or printed pieces of information, which referenced the product lines under scrutiny, were assembled. The historic information was validated with data from institutional informants and industry experts. Contradictory information was resolved through a third piece of confirmatory evidence or with clarification from industry experts. We avoided overreliance on single informants within organizations who might overstate pioneering status or whose recollections might not be entirely accurate. Multiple measures were collected both from organizational informants and from secondary sources, allowing us to cross-validate data from different origins (scholars, experts, and journalists). The sample chosen is a set of financial services launched in Costa Rica during a 15-year period (1980 –1995). Thus, the sample of products was circumscribed purposefully to a small market under progressive deregulation. This was also done with the intention of (a) correctly identifying pioneers, (b) avoiding censoring, and (c) choosing products launched in roughly similar time-frames (to control for potentially different market effects). Historical information could be obtained, both from public records and from people, to permit an accurate determination of pioneering and subsequent entries. Although historical analysis is little used in studies of pioneering carried out in the marketing field, it has been widely used in studies of technological innovation (Cooper and Schendel, 1976; Meyer and Roberts, 1986; Sanderson and Uzumeri, 1995; Utterback, 1994). An important disadvantage with the chosen sample of products is that we may in fact be right-censoring the dependent variable, since it is plausible that not enough time has elapsed for assessing the survival and performance of the different market entrants. In this case, the long term may not yet be long enough. Another concern that might arise with the chosen sample is the extent to which any finding can be generalized to other industries or countries. The sample was chosen for convenience. It allowed us to gather large amounts of detailed data for a relatively extensive and meaningful period of time. Using this methodology, it was possible to circumvent many of the mentioned methodological concerns often associated with much of the literature on pioneering. The purpose, thus, is not to make sweeping claims about the generalizability of our findings but to check whether early-entry advantages erode faster in regimes of weak appropriability. To do so, the sample and setting chosen become a convenient quasiexperimental site. Ultimately, whether these findings can be generalized can only be determined by performing similar studies in other settings.
3. Variables The dependent variable in this study is the total market share of the nth entrant at the time of the study, measured as
1000
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
the percentage of total customers using a particular financial instrument. The order of market entry is used as an independent variable. Pioneers are operationalized as the first entrant. An additional independent variable is the elapsed time delay from the first entry until the nth entry.
financial services (from relatively sophisticated corporate financial services to many retail products, which included, at the time, telephone banking that would later evolve to internet banking, extensive ATM networks, linked cash management accounts, and investment services such as money market mutual funds, retirement accounts, etc.).
4. Results
4.1.1. Pension funds Pension funds in this market had traditionally been administered by the state. Economically active people contributed a fixed percentage of their salaries to a national fund. In such environments, private pension funds tend to be successful if the state-operated pension system gives small pensions at high retirement ages. Problems with the stateadministered fund opened an opportunity for other institutions to enter the market and offer complementary pensions. This industry operated with little supervision until July 1995. In fact, before that time, private pension funds could not operate openly as such but only as providers of additional coverage to that given by the state. Even so, the first private pension funds operator entered the market in August 1988. In 1995, the funds were authorized to operate as such and also received an additional impulse in the form of a tax break by which the equivalent of a portion of the funds saved could be deducted from income tax. Ten more operators entered the market between 1991 and September 1996. Different types of institutions entered this market. The most prominent participants were banks, but there was also one insurance company and one firm that operated pension funds as its sole product. Fig. 1 shows a longitudinal history of total market shares (measured as percentage of total customers) for all pension funds participants in the period under study. The pioneer in this market operated alone for a total of 23 months. Firms’ market shares start leveling off after approximately 4 years after the first entry. According to these graphs and data, an early pension funds entrant gained a maximum market share
4.1. Qualitative data analysis Costa Rica has a small economy with a GDP of US$24 billion and per capita income of US$6700, both 1998 purchasing power-adjusted figures, a population of 3.7 million, and an area slightly smaller than West Virginia. The Costa Rican financial system underwent a modernization process that started in the mid-1980s. Before that time, the financial landscape was entirely dominated by stateowned financial intermediaries. Private banks were not permitted by law to operate. The Central Bank intervened establishing credit limits for different economic activities, limits to resource allocation by type of economic activity, and interest rates. The effect of this intervention was to hamper product and process innovation in financial institutions. Banks, for instance, limited themselves to lending once and again to a limited customer base. The little innovation that actually occurred was done with the purpose of circumventing the restrictive legislation. The reforms to the Costa Rican financial system sought to increase competition and promote market-driven prices for financial services. The deregulation process that started in the 1980s permitted the operation of new banks and financial institutions. Thus, the relative importance of private institutions increased. The participation of private banks, for instance, in total lending went from 15.3% in 1986 to 29.1% in 1990 (Gonza´les and Mesalles, 1993). By 1995, there were 24 banks operating in the country, offering a wide array of
Fig. 1. Order of market entry vs. market share for pension funds.
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
over time that is approximately 50% of the pioneer’s share. In this case, firms that entered the market within an elapsed time delay between 27 and 57 months eventually averaged 57% of the pioneer’s share. Very late entrants appear to have stabilized with market shares that represent approximately 5 –10% of the pioneer’s share (8% in this case). When percentage of market share is measured in terms of portfolio size, the pioneer also appears to retain an important portion of the market. Spearman-rank correlation coefficients showed a strong negative relationship between order of market entry and market share. 4.1.2. Credit cards The first credit card plan in the financial system under study was created by a commercial bank in 1976 (a relatively early entrance even by U.S. standards). The second entrant appeared in the market after 108 months. In this environment, pioneering advantages may have been obtained in terms of network externalities. Card operators derive income from each customer’s interest payments and also from the commissions paid by vendors, thereby making it crucial to develop a large network of affiliated businesses. Fig. 2 displays market shares for the credit card market measured as a percentage of the pioneer’s share (as percentage of total customers). As shown, the very early first entry does not sustain an above average market share over time. The second entrant (108 months after) captures approximately 45% of the market and retains it consistently. In general, however, the data seem to support a negative relationship between order of market entry and market share. Notwithstanding the pioneer’s period of approximately 7 years operating as a monopolist, it fails to reap long-term market share advantages. In this case, the second entrant became (and stayed) the market leader. Later, another three competitors entered the market during a 3-year period. The
1001
pioneer retains the number of customers (in fact increases its number of customers somewhat consistently) despite losing market share. Spearman-rank correlation coefficients showed a strong negative relationship between elapsed time since first entry and market share in the credit card business. 4.1.3. Debit cards Debit cards were introduced into the financial system later than the other two products under study. This instrument requires its holder to have a checking account with the issuing institution, with the debit card serving to automatically and electronically deduct payments from the checking account. The product requires an extensive network of affiliated businesses. Moreover, debit cards become more useful if they can also be used to obtain cash from ATMs or from bank offices. The first debit card entered the market in August 1987. Six additional operators had entered the market by 1996. All debit card institutions are commercial banks, and there is a fairly short elapsed time between the pioneer and the next two entrants. A wave of new entrants occurred in 1996, caused by regulatory changes that allowed all banks to provide regular checking accounts. This change made it unnecessary for banks to circumvent this regulatory constraint, which they often did through alternative and more creative means. No longitudinal data could be obtained for this product. A brief examination of available descriptive data showed that, strictly speaking, the pioneer does not have any advantage. It is the third entrant who is able to capture market share. The pioneer’s market share, in this case, is approximately one-sixth of the third entrant’s market share. A late entrant (with a total elapsed time of 106 months after the pioneer’s entry) appears to capture approximately 50% of the pioneer’s share in a relatively short period of time. Here, market share is measured as percentage of customers
Fig. 2. Market shares for credit cards.
1002
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
for all participants in the debit card industry at the end of the time period studied. Notice that the effect of entry may be influenced by the time elapsed since the pioneer’s entry. The absolute order of market entry could hide nearly simultaneous entries or entries, which are separated by long periods of time. Fig. 3 shows total market share at the end of the time period studied against elapsed time since the pioneer’s entry. Two groups can be recognized clearly in the graph. One group of early entrants introduced debit cards into the market within a total elapsed time of 17 months. The next entry occurred 58 months later, and there is a second group of institutions entering the market more than 8 years after the pioneer’s entry. Hence, although we cannot speak, in a strict sense, of pioneering advantages, there are advantages to early entry. The data on debit cards indicate the potential inconveniences of utilizing strict order of market entry as an independent variable. It might be preferable to moderate this strict order by coding entries as early entries, early followers, and so on, or alternatively, time elapsed since first entry could be used as an independent variable. This notwithstanding, there is, for the case of debit card introduction, a statistically significant relationship between order of market entry and long-term market share. 4.2. Quantitative data analysis To complement the qualitative data analysis, we constructed a data set with the data points collected in all three product categories. This data set included observations for a total of 36 entries in three product categories with the following variables: ENTRY (numerical variable containing the order of market entry), SHARE (percentage of market share for nth entry measured as percentage of total customers), ELAPSED (elapsed time in months for particular entry since first entry in the particular category), SIZE (ratio that gives relative size of entrant with respect to the largest entrant in the product category, calculated based on total
number of employees), INNOV (measures degree of dispersion of product offerings for that particular institution), and REGUL (dichotomous variable that indicated if the product was driven by regulatory changes). Our goal in this part of the paper was to build a baseline model for the dependent performance variable SHARE, controlling for institution size, the effect of the institution’s total product offerings, and the effects of regulation. We then constructed a model to include question predictors ENTRY and ELAPSED. After examining the raw data set and the distributions of all variables, inspecting bivariate scatterplots, and studying bivariate correlations (estimated through Pearson correlation coefficients), a taxonomy of multiple regression models was fitted. The variables SIZE (transformed to account for curvilinearities), INNOV, and REGUL were classified as control predictors. ENTRY and ELAPSED were established as question predictors. Three regression models were subsequently fitted with the intention of creating an appropriate baseline model. First, SHARE was regressed on ENTRY. Second, SHARE was regressed on ELAPSED, and, finally, both variables were used as predictors in the regression analysis. After carrying out sensitivity analyses and performing a variety of tests, and after validating the assumptions about normality of the residuals and independence of errors, we were left with a somewhat unexpected result: our best and chosen model was one that simply regressed SHARE against ENTRY. We chose that model as the final model (intercept= 1.6, P<.001; coefficient of ENTRY= .24, P<.001; r2=69.67%; df=32) for the purpose of discussion. We then proceeded, in a similar fashion, to fit a model in which SIZE, INNOV, and REGUL were included as control predictors and ELAPSED as a question predictor, controlling also for interactions among these variables. The second final and most parsimonious fitted model regresses SHARE on ELAPSED (intercept= 1.82, P<.001; coefficient of ELAPSED= .015, P<.001). Data limitations did not permit a more exhaustive empirical analysis. The results obtained nevertheless provide some interesting insights that complement our qualitative observations.
5. Summary and conclusion
Fig. 3. Elapsed time since pioneer’s entry in months vs. total market share for debit cards.
We wanted to determine whether innovation initiatives in the development of new financial services might gain advantages from early entry. We expected that in a regime of such weak appropriability, no such advantage would be accrued or, if any advantage was attained at all, that it would erode very quickly. We obtained two groups of results. Our quantitative analysis, however limited, turned out to be in line with most of the prior research work that has been performed in other industries, particularly consumer products. We found that a baseline model in which SHARE was regressed against order of market entry (ENTRY) explained almost 70% of the variability in performance. Including
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
Fig. 4. Results of final model of order of market entry vs. market share for a prototypical financial services case.
control variables such as the firm’s size and the amplitude of its product offerings did not have a sizable effect upon our baseline model. Fig. 4 is a visual display of our model results for different orders of market entry. Our findings indicate a negative relationship between order of market entry and long-term market share. A second entrant is likely to capture approximately 78% of the pioneer’s eventual market share. A 10th entrant will only capture 11% on average. The results appear to be in line with those found in other industries. Table 1 establishes such a comparison. This rather conventional analysis leads us to conclude primarily that there are important advantages to early entry in financial service innovations. This result extends established generalizations to other industries and settings in a field that ‘‘relies heavily on North American manufactured goods that are included in either the Urban et al. Assessor data or the PIMS data’’ (Kalyanaram et al., 1995, p. 9). Such results, however, need to be qualified. We see that using order of market entry alone might render somewhat equivocal results because nearly simultaneous entries or entries separated by large periods of time are not evident. Results need to be moderated by a sense of real time between entries and, for such purpose, interpreting entry utilizing real elapsed time is very illustrative. As we can see
1003
in Fig. 5, if subsequent entrants invade the market within short periods of time after the pioneer, pioneering advantages are scant. Shorter lead times mean pioneers have less time to establish their brand names and other distribution advantages. This means that the earlier competitive rivalry starts, the larger later entrants’ market shares. From the pioneer’s perspective, lengthening lead time by 1 year could represent, according to our data, as much as 2.5% of overall market share. Use of elapsed time since first entry also adds to a managerial interpretation of these models. For our sample of products, we see that entrants, which follow shortly after the pioneer, may obtain substantial market shares relative to the pioneer, a result that is not obvious by looking at order of market entry alone. We were not able to collect reliable data on product profitability. It is entirely possible that certain institutions capture a large market share and not add value to the organization. Unfortunately, no detailed data that could provide an indication of product profitability was available. Subsequent studies should address this limitation. Insofar as our proposition goes, early entry advantages do not seem to erode any quicker in this regime of low appropriability than they would in other industries. These results coincide with the work of Makadok (1998) who found that early movers and pioneers enjoy both a highly sustainable pricing advantage and a moderately sustainable market share advantage in the U.S. Money Market Mutual Fund Industry. Although no claim of generalizability can be made, the results we obtained here appear to also hold in the different quasiexperimental setting, in a G7 locale (with a similar appropriability regime) such as the one explored by Makadok. These results, however, need to be examined in the light of the qualitative data. If we look at individual products, establish their longitudinal histories, and operationalize pioneering as the first entrant, we find that the pioneers, strictly speaking, do not necessarily retain an appreciable market share advantage. In fact, no important pioneering advantage is observed in two of the three product lines explored. This is consistent with the exposition of Golder and Tellis (1993) although in our case the product histories are more concise
Table 1 Order of market entry and market share for consumer packaged goods, prescription antiulcer drugs, and financial services innovations Entry order
Forecasted market share relative to the pioneering brand Consumer packaged goods
First Second Third Fourth Fifth Sixth
Prescription antiulcer drugs
Financial products
Urban et al. (1986)
Kalyanaram and Urban (1992)
Berndt et al. (1994)
This study
1.00 0.71 0.58 0.51 0.45 0.41
1.00 0.76 0.64 0.57 0.53 0.49
1.00 0.70 0.57 0.49 0.44 0.40
1 0.79 0.62 0.49 0.38 0.30
Modified from Kalyanaram et al. (1995).
1004
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
Fig. 5. Results of final model of elapsed time since first entry vs. market share for a prototypical financial services case.
because of the youth of the product lines studied. On the other hand, when our product sample is aggregated, an important early entry effect is observed. We get results that are in line with much other literature on first-mover advantages. Evidently, by aggregating the data, we can only affirm the advantages of early entry and not those of pioneering, in the strict sense of the word. History reveals a more detailed picture that cannot be captured completely in cross-sectional studies. Moreover, this raises the possibility that studies that use aggregated information from extant databases could have a built-in bias that is methodologically difficult to identify and isolate based solely on such information. It is also of interest to speculate why the first entrant does not retain a long-term market share advantage in two of the three products studied but does sustain its advantage in pension funds. To do so, we carried on additional interviews with several industry experts who gave us two categories of possible causes. One set of explanations has to do with how firms approach their new product launches. Another set of explanations is associated with environmental contingencies. Some experts argued that a pioneer could have an advantage in acquiring and developing resources. Given enough lead time, a first-mover may spend more on acquiring resources (personnel, brand equity, technological abilities, and the like), thus preempting or relegating subsequent entrants to weak positions. Such preeminence might be tightly linked to the time elapsed between first and subsequent entries. The longer the time elapsed, the longer the cumulative effect of spending on advertising, technology, and training. Moreover, early entrants are likely to attain enough volume (earlier) that will trigger scale effects that will eventually render an insurmountable cost advantage. This appears to be the case of pension funds. Another key aspect that, according to industry experts, permits sustaining early entry advantages is access to an established base of existing customers and to an array of complementary services. Customers may prefer to do business with an organization that offers an ample array of products. Thus, they may be inclined to switch to, or
reluctant to switch from, certain institution’s products because of the added convenience of carrying on all businesses with one organization. Hence, one product’s early entry advantage should be evaluated in the light of how it fits within the complete product portfolio of the institution. Our data do not permit us to formally test this. However, the data for pension funds do provide some insight. The only standalone pension fund was unable to sustain an early entry advantage and was overcome, for instance, by a sixth entrant, which was a large bank. Evidently, although the financial product may be very easy to imitate, the existing customer base of certain institutions, and the opportunity to create unique bundles of products, will provide the basis for sustaining an early-entry advantage. As some industry experts indicated, complete product platforms are much harder to imitate than a single product. The internal procedure for researching and developing the new product was also mentioned by industry experts we consulted. To investigate this, we asked several bank executives to determine how pioneering products had been developed within their institutions and to rate their success. We listed a total of 46 entries in several product categories. These products included some, but not all, of those used in the previously analyzed 36 entries into pension funds, debit cards, and credit cards. We observed that pioneering products were developed differently. Some (about 24% of the cases listed) were generated in an orderly and formalized manner with a somewhat structured process of market research and a subsequent development process aimed at satisfying certain market needs. Most products (44%) were adaptations of ideas that had been seen operating in other countries. A third large category of products stemmed from regulatory changes (13%). We asked executives to rate the success of these products. They indicated that more formalized processes of research and development do have an important impact upon performance, particularly when products are adequately conceptualized as part of a wellstructured product strategy and product platform. Another explanation deals with environmental contingencies, particularly technological change. In the case of credit cards, the second entrant entered very aggressively and with a technology that clearly leapfrogged earlier investments. This permitted this entrant to develop a network of affiliated businesses that eventually lured customers who then found it too costly to switch to other services. Similarly, the long-term sustainability within the debit card business was driven by scale effects and network externalities. For the debit card sample, there is a statistically significant negative relationship between entry and two measures of institution size (Spearman-rank=.86, P<.01 for SIZE1 and Spearman-rank= .96, P<.001) and there is a positive relationship between a measure of network externalities (number of available ATMs and number of clients (Spearman-rank=.88, P<.01). Finally, it appears to be useful and important to consider elapsed time since first entry when evaluating these models.
L.E. Lo´pez, E.B. Roberts / Journal of Business Research 55 (2002) 997–1005
These effects have been, to the best of our knowledge, explored only in two studies. Huff and Robinson (1994) determined that increasing the years of competitive rivalry (i.e., reducing the pioneer’s lead time) tended to reduce the pioneer’s market share advantage. Furthermore, Hurwitz and Caves (1988) examined 56 pharmaceutical products and their performance after patent expiration, reporting that longer patent life (i.e., longer pioneer lead time) renders appreciable market share advantages.
Acknowledgements We appreciate funding assistance provided by INCAE, the MIT International Center for Research on the Management of Technology (ICRMOT), and the MIT Center for Innovation in Product Development (National Science Foundation Grant #EEC-9529140). We thank Professors Donald Lessard, Scott Stern and two anonymous reviewers for their many thoughtful comments and suggestions.
References Berndt ER, Bui LT, Reilly DH, Urban GL. The roles of marketing, product quality, and price competition in the growth and composition of the U.S. anti-ulcer drug industry. Working Paper 19-94, Massachusetts Institute of Technology, 1994. Bitran GR, Lojo M. A framework for analyzing service operations. Eur Manage J 1993;11(3):271 – 82. Cooper AC, Schendel D. Strategic responses to technological threats. Bus Horiz 1976;19(1):61 – 9. Finnerty JD. Financial engineering in corporate finance: an overview. Financ Manage 1988;17:14 – 33 (Winter). Golder PN, Tellis GJ. Pioneer advantage: marketing logic or marketing legend? J Mark Res 1993;30:158 – 70 (May). Gonza´lez C, Mesalles L. La economı´a polı´tica de la nacionalizacio´n bancaria en Costa Rica: 1948 – 1990. In: Gonza´les C, Vargas T, editors. Reforma financiera en Costa Rica, perspectivas y propuestas. San Jose´, Costa Rica: Imprenta LIL, 1993. p. 3 – 54. Huff LC, Robinson WT. The impact of leadtime and years of competitive rivalry on pioneer market share advantages. Manage Sci 1994;40(10): 1370 – 7. Hurwitz MA, Caves RE. Persuasion or information? Promotion and the shares of brand name and generic pharmaceuticals. J Law Econ 1988; 31:299 – 320 (October). Kalyanaram G, Urban GL. Dynamic effects of the order of entry on market share, trial penetration, and repeat purchases for frequently purchased consumer goods. Mark Sci 1992;11(3):235 – 50 (Summer).
1005
Kalyanaram G, Robinson WT, Urban GT. Order of market entry: established empirical generalizations, emerging empirical generalizations, and future research. Mark Sci 1995;14(3):G212 – 21 (Summer). Kerin RA, Varadarajan PR, Peterson RA. First-mover advantage: a synthesis, conceptual framework, and research propositions. J Mark 1992; 56:33 – 52 (October). Lambkin M. Order of entry and performance in new markets. Strategic Manage J 1988;9:127 – 40 (Summer). Levin RC, Klevorick AK, Nelson RR, Winter SG. Appropriating the returns from industrial research and development. Brookings Pap Econ Act 1987;3:783 – 820. Lieberman MB, Montgomery DB. First-mover advantages. Strategic Manage J 1988;9:41 – 58 (Summer). Lieberman MB, Montgomery DB. First-mover (dis)advantages: retrospective and link with the resource-based view. Strategic Manage J 1998; 19(12):1111 – 25. Lilien G, Yoon E. The timing of competitive market entry: an exploratory study of new industrial products. Manage Sci 1990;36(5): 568 – 85 (May). Makadok R. Can first-mover and early-mover advantages be sustained in an industry with low barriers to entry/imitation? Strategic Manage J 1998; 19(7):683 – 96. Meyer M, Roberts E. New product strategy in small technology-based firms: a pilot study. Manage Sci 1986;32(7):806 – 22. Miller MH. Financial innovation: the last twenty years and the next. J Financ Quant Anal 1986;21(4):459 – 71. Morone J, Berg D. Management of technology in the service sector: practices in the banking industry. J High Technol Manage Res 1993;4(1): 123 – 37. Robinson WT. Sources of market pioneer advantages: the case of industrial goods industries. J Mark Res 1988;25(1):87 – 94. Robinson WT, Fornell C. Sources of market pioneer advantages in consumer goods industries. J Mark Res 1985;22:305 – 17 (August). Robinson WT, Kalyanaram G, Urban GL. First-mover advantages from pioneering new markets: a survey of empirical evidence. Rev Ind Organ 1994;9:1 – 23. Sanderson S, Uzumeri M. Managing product families: the case of the Sony Walkman. Res Policy 1995;24(4):761 – 82. Tufano P. Financial innovation and first-mover advantages. J Financ Econ 1989;25(2):213 – 40. Urban GT, Carter T, Gaskin S, Mucha Z. Market share rewards to pioneering brands: an empirical analysis and strategic implications. Manage Sci 1986;32(6):645 – 59. Utterback JM. Mastering the dynamics of innovation. Boston, MA: Harvard Business School Press, 1994. VanderWerf PA, Mahon JF. Meta-analysis of the impact of research methods on findings of first-mover advantage. Manage Sci 1997;43(11): 1510 – 9. Whitten IT. Brand performance in the cigarette industry and the advantage to early entry, 1913 – 74. Staff Report to the Federal Trade Commission 1979; June.