The impact of superstar and non-superstar software on hardware sales: the moderating role of hardware lifecycle


In the context of two-sided markets, we propose hardware lifecycle as a key moderator of the impact of superstar and non-superstar software on hardware adoption. A hardware’s earlier adopters are less price sensitive and have a higher preference for exciting and challenging software. In contrast, later adopters are more price sensitive and prefer simplicity in software. Superstar software tend to be more expensive and more complex compared to non-superstars. Therefore, earlier (later) adopters prefer superstars (non-superstars), which leads to higher impact of superstars (non-superstars) on hardware adoption in the early (later) stages of the hardware lifecycle. Using monthly data over a 12-year timeframe (1995–2007) from the home video game industry, we find that both superstar and non-superstar software impact hardware demand, but they matter at different points in the hardware lifecycle. Superstars are most influential when hardware is new, and this influence declines as hardware ages. In contrast, non-superstar software has a positive impact on hardware demand later in the hardware lifecycle, and this impact increases with hardware age. Findings reveal that eventually the amount of available non-superstar software impacts hardware adoption more than the amount of available superstar software. We provide several managerial implications based on these findings.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5


  1. 1.

    We acknowledge that certain types of software may act as platforms too, e.g., operating systems.

  2. 2.

    We follow this convention throughout the paper and refer to the platform as “hardware” and the content as “software.

  3. 3.

    We thank an anonymous reviewer for this suggestion.

  4. 4.

    These prices are significantly different at p < 0.01 using a standard T-test. Average price calculated for each game adjusting for inflation using the Consumer Price Index for Urban Consumers (2015 = 100).

  5. 5.

    Emphasis ours.

  6. 6.

    We note that games are physically distinct from consoles in that no console is preloaded with gaming software.

  7. 7.

    While we do not differentiate between exclusive and non-exclusive games in the main econometric analysis in the paper, we provide similar analysis using only exclusive games and non-exclusive games in Web Appendix B.

  8. 8.

    Though results are robust to classifying games as superstars according to sales criteria.

  9. 9.

    For face validity, we identify games in our dataset that sold 1 million copies or more. We find the average quality of these games (81.77) is significantly higher than the average quality of games that do not achieve this threshold (58.44). These averages are significantly different with >99% confidence using standard t-test.

  10. 10.

    Our concern is that quality ratings by users may be influenced by sales performance, given that users enter their ratings after the game is released. As a robustness check, we also used a quality measure based on the average of the three ratings and found similar results to those presented below.

  11. 11.

    Metacritic touts this policy on their website ( as a way to protect critics from outside influences that may pressure them to change their scores.

  12. 12.

    This is generated in a manner similar to Fig. 1 in Binken and Stremersch (2009).

  13. 13.

    We perform a series of robustness checks using different quality ratings to classify games as superstars. All estimations from robustness checks are available from the authors upon request.

  14. 14.

    The outside option is the fraction of people in the market per period who do not buy any console. The market in any period is defined as the number of U.S. homes with TV sets who have not purchased a video game console. The number of U.S. homes with a TV set was obtained from The Nielsen Company ( We estimate the number of households that do not own a video game console from our data—in each generation we calculate total generational sales up to the period in question and subtract from the number of U.S. homes with a TV set. Importantly, we depreciate generational sales by 90% per year to account for the fraction of consumers who purchase multiple consoles in each generation. In other words, we assume consumers can buy only one console a month, but some fraction returns to the market to make another purchase later. This approach has been used in Clements and Ohashi (2005), Gretz (2010), and Kretschmer and Claussen (2016). We should mention that our results are robust to various depreciation rates for total generational sales.

  15. 15.

    Market share of the focal console is calculated as the sales of the focal console relative to the total sales of all consoles in the same console generation as the focal console. Sales of previous versions of the console are not included in the market share calculation of the focal console.

  16. 16.

    Using the logit demand model derived from the structural approach to model hardware demand has the advantage of controlling for the effect of competing product characteristics in a tractable way without entering them directly into the estimation equation (McFadden 1973). Otherwise, as Berry (1994) notes “a system of N goods gives N2 elasticities to estimate” in order to account for competitive effects, which quickly becomes impractical to estimate as degrees of freedom decrease quadratically with the addition of competing products. However, as a robustness check we estimate the model for hardware demand using the natural log of console sales for console k in period t as the dependent variable and obtain qualitatively similar results to those presented below. These results are available from the authors upon request.

  17. 17.

    We thank an anonymous reviewer for suggesting this operationalization of backward compatible superstars. While we obtain similar results to those presented below using the number of backward compatible superstars (rather than the share), this variable was highly correlated with the number of backward compatible games which lead to collinearity concerns in our estimations. Using the share of backward compatible superstars alleviates this issue.

  18. 18.

    We use the levels of the number of available superstar and non-superstar games as the dependent variables in Equations 2 and 3 instead of the natural log as suggested by Gretz (2010) to be consistent with how these variables enter the hardware demand specification in Equation 1. Similarly, we use the level of installed base rather than the natural log. However, our results are qualitatively similar if we use a log-log or linear-log specification for software supply.

  19. 19.

    We treat the number of backward compatible superstar and non-superstar games as exogenous, since they are more likely influenced by the market share of new buyers for the previous generation console they were originally designed for rather than the backward compatible, next-generation console.

  20. 20.

    VIF tables are available from the authors upon request.

  21. 21.

    We cluster on console and use the optimal cluster robust weighing matrix to obtain the single equation GMM estimations. A similar cluster-robust weighing matrix is not feasible in the joint GMM estimation because there are more moments in the system than clusters (i.e., the cluster robust weighing matrix is not invertible). Instead we use a heteroskedastic and autocorrelation consistent weighing matrix with the lag order optimally selected using the Newey and West (1994) algorithm. Standard errors robust to clustering at the console level are presented in every estimation.

  22. 22.

    Restricting the coefficients on the interaction terms in Column 4 to zero yields χ2(2) = 10.54, p < 0.01, implying that the model with interactions has a higher explanatory power.

  23. 23.

    Testing restrictions in Columns 3 and 4 that the effect of superstars and non-superstar games are equal yields χ2(1) = 4.06, p < 0.05 for Column 3 and χ2(1) = 5.92, p < 0.05 for Column 4. Thus, in both cases we reject the null hypothesis that superstar and non-superstar games have equivalent effects on console market share.

  24. 24.

    Restricting the coefficients on the interaction terms in Estimation 4 to be the same yields χ2(1) = 4.22, p < 0.05, implying that the two interaction terms do not have the same magnitude.

  25. 25.

    This is found by solving 0.685–0.009 × \( \mathrm{Console}\ {\mathrm{Age}}_t^k \) = −0.048 + 0.001 × \( \mathrm{Console}\ {\mathrm{Age}}_t^k \).

  26. 26.

    A joint test of the hypothesis that the interaction terms are zero in Column 8 in Table 5 and Column 12 in Table 6 yields χ2(2) = 184.13, p < 0.01.

  27. 27.

    A test of the hypothesis that the coefficients on \( {\mathrm{IB}}_t^k \) are equal in Column 8 in Table 5 and Column 12 in Table 6 yields χ2(1) = 156.53, p < 0.01.

  28. 28.

    These elasticities are calculated from the linear models of Equations 2 and 3. We do not calculate elasticities for the non-linear model presented in Equation 1 given the risk of misinterpretation (Ai and Norton 2003) and note this as a limitation of our current level of analysis.

  29. 29.

    We note that these elasticities could be quite different if we use different variable values (Ai and Norton 2003).

  30. 30.

    We also test to see if elasticities are the same over time within the same software type (separately for both superstars and non-superstar games). We reject the null hypothesis of similar coefficients in every case with p < 0.01. We suppress the test statistics given the number of individual tests conducted. The results are available from authors upon request.

  31. 31.

    For the hardware demand estimations, the Durbin-Wu-Hausman test comparing OLS and column (1) in Table 4 yields χ2(3) = 14.15, p < 0.01 while comparing OLS and column (2) yields χ2(5) = 15.65, p < 0.01. Similarly, for the non-superstar supply estimations, the Durbin-Wu-Hausman test comparing OLS and column (5) results in χ2(1) = 147.38, p < 0.01 while comparing OLS and (6) yields χ2(2) =86.62, p < 0.01. For the superstar supply estimations, the Durbin-Wu-Hausman test comparing OLS and column (9) yields χ2(1) = 138.29, p < 0.01 while comparing OLS and column (10) yields χ2(2) = 119.33, p < 0.01.

  32. 32.

    We thank an anonymous reviewer for this suggestion.

  33. 33.

    We thank an anonymous reviewer for this suggestion.

  34. 34.

    We thank an anonymous reviewer for pointing out this issue.


  1. Ai, C., & Norton, E. C. (2003). Interaction terms in logit and probit models. Economics Letters, 80(1), 123–129.

    Article  Google Scholar 

  2. Allenby, G., Fennell, G., Huber, J., Eagle, T., Gilbride, T., Horsky, D., et al. (2005). Adjusting choice models to better predict market behavior. Marketing Letters, 16(3–4), 197–208.

    Article  Google Scholar 

  3. Avnet, T., & Higgins, E. T. (2006). How regulatory fit affects value in consumer choices and opinions. Journal of Marketing Research, 43(1), 1–10.

    Article  Google Scholar 

  4. Basuroy, S., Chatterjee, S., & Ravid, S. A. (2003). How critical are critical reviews? The box office effects of film critics, star power, and budgets. Journal of Marketing, 67(4), 103–117.

    Article  Google Scholar 

  5. Bernbeck, S. (2015). What drives a review score? Accessed 1 June 2018.

  6. Berry, S. T. (1994). Estimating discrete-choice models of product differentiation. The Rand Journal of Economics, 25(2), 242–262.

    Article  Google Scholar 

  7. Berry, S., Levinsohn, J., & Pakes, A. (1995). Automobile prices in market equilibrium. Econometrica, 63(4), 841.

    Article  Google Scholar 

  8. Binken, J. L., & Stremersch, S. (2009). The effect of superstar software on hardware sales in system markets. Journal of Marketing, 73(2), 88–104.

    Article  Google Scholar 

  9. Chatterjee, R. A., & Eliashberg, J. (1990). The innovation diffusion process in a heterogeneous population: A micromodeling approach. Management Science, 36(9), 1057–1079.

    Article  Google Scholar 

  10. Clements, M. T. (2004). Direct and indirect network effects: Are they equivalent? International Journal of Industrial Organization, 22(5), 633–645.

    Article  Google Scholar 

  11. Clements, M. T., & Ohashi, H. (2005). Indirect network effects and the product cycle: Video games in the U.S., 1994-2002. The Journal of Industrial Economics, 53(4), 515–542.

    Article  Google Scholar 

  12. Corts, K. S., & Lederman, M. (2009). Software exclusivity and the scope of indirect network effects in the U.S. home video game market. International Journal of Industrial Organization, 27(2), 121–136.

    Article  Google Scholar 

  13. Coughlan, P. J. (2001). Note on home video game technology and industry structure. Harvard Business School Case.

  14. Cox, J. (2014). What makes a blockbuster video game? An empirical analysis of US sales data. Managerial and Decision Economics, 35(3), 189–198.

    Article  Google Scholar 

  15. Day, G. S. (1981). The product life cycle: Analysis and application issues. Journal of Marketing, 45, 60–67.

    Article  Google Scholar 

  16. Dranove, D., & Gandal, N. (2003). The dvd-vs.-divx standard war: Empirical evidence of network effects and preannouncement effects. Journal of Economics & Management Strategy, 12(3), 363–386.

    Article  Google Scholar 

  17. Dubé, J. H., Hitsch, G. J., & Chintagunta, P. K. (2010). Tipping and concentration in markets with indirect network effects. Marketing Science, 29(2), 216–249.

    Article  Google Scholar 

  18. Fennell, G. (1978). Consumers' perceptions of the product. use situation. The Journal of Marketing, 38–47.

  19. Gandal, N., Kende, M., & Rob, R. (2000). The dynamics of technological adoption in hardware/software systems: The case of compact disc players. The Rand Journal of Economics, 43–61.

  20. Gaston, M. (2014). Titanfall launch sees Xbox One sales almost double in the UK. Accessed 1 June 2018.

  21. Germann, F., Ebbes, P., & Grewal, R. (2015). The chief marketing officer matters! Journal of Marketing, 79(3), 1–22.

    Article  Google Scholar 

  22. Gilbert, B. (2017). The PlayStation 4 is selling about twice as fast as the Xbox One,. Accessed 1 June 2018.

  23. Golder, P. N., & Tellis, G. (2004). Growing, growing, gone: Cascades, diffusion, and turning points in the product life cycle. Marketing Science, 23(2), 207–218.

    Article  Google Scholar 

  24. Gooding, C., & Stephenson, E. F. (2017). Superstars, uncertainty of outcome, and PGA tour television ratings. Journal of Sports Economics, 18(8), 867–875.

    Article  Google Scholar 

  25. Gretz, R. T. (2010). Hardware quality vs. network size in the home video game industry. Journal of Economic Behavior and Organization, 76(2), 168–183.

    Article  Google Scholar 

  26. Gretz, R. T., & Basuroy, S. (2013). Why quality may not always win: The impact of product generation life cycles on quality and network effects in high-tech markets. Journal of Retailing, 89(3), 281–300.

    Article  Google Scholar 

  27. Hagiu, A., & Wright, J. (2015). Multi-sided platforms. International Journal of Industrial Organization, 43, 162–174.

    Article  Google Scholar 

  28. Hausman, J. A., & Leonard, G. K. (1997). Superstars in the national basketball association: Economic value and policy. Journal of Labor Economics, 15(4), 586–624.

    Article  Google Scholar 

  29. Hofmann, J., Clement, M., Völckner, F., & Hennig-Thurau, T. (2017). Empirical generalizations on the impact of stars on the economic success of movies. International Journal of Research in Marketing, 34(2), 442–461.

    Article  Google Scholar 

  30. Jahanmir, S. F., & Lages, L. F. (2015). The lag-user method: Using laggards as a source of innovative ideas. Journal of Engineering and Technology Management, 37, 65–77.

    Article  Google Scholar 

  31. Jahanmir, S. F., & Lages, L. F. (2016). The late-adopter scale: A measure of late adopters of technological innovations. Journal of Business Research, 69(5), 1701–1706.

    Article  Google Scholar 

  32. Katz, M. L., & Shapiro, C. (1994). Systems competition and network effects. Journal of Economic Perspectives, 8(2), 93–115.

    Article  Google Scholar 

  33. Kharpal, A. (2017). Sony profits rise 346% thanks to strong performance from PlayStation gaming unit. Retrieved February 18, 2018, from

  34. Kim, J. H., Prince, J., & Qiu, C. (2014). Indirect network effects and the quality dimension: A look at the gaming industry. International Journal of Industrial Organization, 37, 99–108.

    Article  Google Scholar 

  35. Kretschmer, T., & Claussen, J. (2016). Generational transitions in platform markets – The role of backward compatibility. Strategy Science, 1(2), 90–104.

    Article  Google Scholar 

  36. Lee, R. S. (2013). Vertical integration and exclusivity in platform and two-sided markets. The American Economic Review, 103(7), 2960–3000.

    Article  Google Scholar 

  37. Lee, Y., & O’Connor, G. C. (2003). New product launch strategy for network effects products. Journal of the Academy of Marketing Science, 31(3), 241–255.

    Article  Google Scholar 

  38. Liu, Y. (2006). Word of mouth for movies: Its dynamics and impact on box office revenue. Journal of Marketing, 70(3), 74–89.

    Google Scholar 

  39. Malshe, A. V., Colicev, A., & Mittal, V. (2017). How main steet drives wall street: Customer satisfaction, short sellers, and shareholder value. Accessed 1 June 2018.

  40. Marchand, A. (2016). The power of an installed base to combat lifecycle decline: The case of video games. International Journal of Research in Marketing, 33(1), 140–154.

    Article  Google Scholar 

  41. Marchand, A., & Hennig-Thurau, T. (2013). Value creation in the video game industry: Industry economics, consumer benefits, and research opportunities. Journal of Interactive Marketing, 27(3), 141–157.

    Article  Google Scholar 

  42. McFadden, D. (1973). Conditional logit analysis of qualitative choice behavior.

  43. McIntyre, D. P., & Srinivasan, A. (2017). Networks, platforms, and strategy: Emerging views and next steps. Strategic Management Journal, 38(1), 141–160.

    Article  Google Scholar 

  44. Moorhead, P. (2013). Xbox One: buy now, later, or never? Accessed 1 June 2018.

  45. Nair, H., Chintagunta, P. K., & Dubé, J. P. (2004). Empirical analysis of indirect network effects in the market for personal digital assistants. Quantitative Marketing and Economics, 2(1), 23–58.

    Article  Google Scholar 

  46. Nevo, A. (2000). A practitioner’s guide to estimation of random-coefficients logit models of demand. Journal of Economics and Management Strategy, 9(4), 513–548.

    Article  Google Scholar 

  47. Newey, W. K., & West, K. D. (1994). Automatic lag selection in covariance matrix estimation. The Review of Economic Studies, 61(4), 631–653.

    Article  Google Scholar 

  48. Papies, D., & van Heerde, H. J. (2017). The dynamic interplay between recorded music and live concerts: The role of piracy, unbundling, and artist characteristics. Journal of Marketing, 81(4), 67–87.

    Article  Google Scholar 

  49. Pearce, C. (2002). Emergent authorship: The next interactive revolution. Computers & Graphics, 26(1), 21–29.

    Article  Google Scholar 

  50. Phan, M. H., Keebler, J. R., & Chaparro, B. S. (2016). The development and validation of the game user experience satisfaction scale (GUESS). Human Factors, 58(8), 1217–1247.

    Article  Google Scholar 

  51. Reddy, S. K., Swaminathan, V., & Motley, C. M. (1998). Exploring the determinants of Broadway show success. Journal of Marketing Research, 370–383.

  52. Rietveld, J., & Eggers, J. P. (2018). Demand heterogeneity in platform markets: Implications for complementors. Organization Science, 29(2), 304–322.

    Article  Google Scholar 

  53. Rogers, E. M. (2003). Diffusion of innovations. Simon and Schuster.

  54. Rosen, S. (1981). The economics of superstars. The American Economic Review, 71(5), 845–858.

    Google Scholar 

  55. Rossi, P. E. (2014). Invited paper—Even the rich can make themselves poor: A critical examination of IV methods in marketing applications. Marketing Science, 33(5), 655–672.

    Article  Google Scholar 

  56. Rysman, M. (2009). The economics of two-sided markets. Journal of Economic Perspectives, 23(3), 125–143.

    Article  Google Scholar 

  57. Schreier, J. (2015). Metacritic matters: how review scores hurt video games. Accessed 1 June 2018.

  58. Shankar, V., & Bayus, B. L. (2003). Network effects and competition: An empirical analysis of the home video game industry. Strategic Management Journal, 24(4), 375–384.

    Article  Google Scholar 

  59. Shih, E., & Schau, H. J. (2011). To justify or not to justify: The role of anticipated regret on consumers’ decisions to upgrade technological innovations. Journal of Retailing, 87(2), 242–251.

    Article  Google Scholar 

  60. Song, R., Kim, H., Lee, G. M., & Jang, S. (2017). Does deceptive marketing pay? The evolution of consumer sentiment surrounding a pseudo-product-harm crisis. Journal of Business Ethics, 1–19.

  61. Squire, K., & Jenkins, H. (2002). The art of contested spaces. In Game On (pp. 97–102). New York.

  62. Stanko, M. A., Bonner, J. M., & Calantone, R. J. (2007). Building commitment in buyer–seller relationships: A tie strength perspective. Industrial Marketing Management, 36(8), 1094–1103.

    Article  Google Scholar 

  63. Steiner, M., Wiegand, N., Eggert, A., & Backhaus, K. (2016). Platform adoption in system markets: The roles of preference heterogeneity and consumer expectations. International Journal of Research in Marketing, 33(2), 276–296.

    Article  Google Scholar 

  64. Stock, J. H., & Watson, M. H. (2007). Introduction to econometrics. Boston: Pearson-Addison Wesley.

    Google Scholar 

  65. Stock, J. H., & Yogo, M. (2002). Testing for weak instruments in linear IV regression. Cambridge. Retrieved February 1, 2018 from

  66. Stock, J. H., Wright, J. H., & Yogo, M. (2002). A survey of weak instruments and weak identification in generalized method of moments. Journal of Business and Economic Statistics, 20(4), 518–529.

    Article  Google Scholar 

  67. Stremersch, S., Tellis, G. J., Franses, P. H., & Binken, J. L. (2007). Indirect network effects in new product growth. Journal of Marketing, 71(3), 52–74.

    Article  Google Scholar 

  68. Sun, L., Rajiv, S., & Chu, J. (2016). Beyond the more the merrier: The variety effect and consumer heterogeneity in system markets. International Journal of Research in Marketing, 33(2), 261–275.

    Article  Google Scholar 

  69. Totilo, S. (2012, July 9). The difference between a good video game and a bad one. Kotaku. Accessed 29 April 2018.

  70. Usher, W. (2015). Destiny’s poor Metacritic score could cost Bungie millions. Retrieved February 1, 2018, from

  71. Wooldridge, J. (2010). Econometric analysis of cross section and panel data. MIT press.

  72. Zeelenberg, M. (1999). Anticipated regret, expected feedback and behavioral decision making. Journal of Behavioral Decision Making, 12(2), 93–106.

    Article  Google Scholar 

  73. Zeithaml, V. A. (1988). Consumer perceptions of Price, quality, and value: A means-end model and synthesis of evidence. Journal of Marketing, 52(3), 2–22.

    Article  Google Scholar 

  74. Zhu, F., & Zhang, X. (2010). Impact of online consumer reviews on sales: The moderating role of product and consumer characteristics. Journal of Marketing, 74(2), 133–148.

    Article  Google Scholar 

Download references


This research benefited from generous support provided by the Carolan Research Institute and a Bradley University Foster College of Business Administration Faculty Development Grant. The authors would like to thank participants at the 2017 American Marketing Association Winter Educators’ Conference and the 2015 American Marketing Association Winter Educators’ Conference. Additionally, the authors thank BJ Allen for helpful comments on an earlier draft and four anonymous referees, an associate editor, and the editor for insightful comments throughout the review process. Lastly, the authors thank Katlyn Brinkley and Aaron Gleiberman for their copyediting prowess. The usual disclaimer applies.

Author information



Corresponding author

Correspondence to Carlos Bauer.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material


(DOCX 716 kb)


Appendix 1

Endogeneity and Instruments, GMM Diagnostics

This study mainly utilizes cost side instrumental variables to address endogeneity in all three of the estimation equations. Cost side instruments are appropriate to address endogeneity issues in a variety of demand estimations (Nevo 2000) and similar instruments have been employed in several studies within our context (Clements and Ohashi 2005; Dubé et al. 2010; Gretz and Basuroy 2013; Kretschmer and Claussen 2016). The key identification assumption is that cost shifters are likely considered by managers when choosing strategies such as price or number of available games but are likely not related to unobservable variables that impact consumer decisions.

There are two main endogeneity issues in the hardware demand equation. First, according to Zhu and Zhang (2010), price is set by managers who likely use information not available to the researcher. Given this, price is likely correlated with the error term in the estimation. Second, new console adopters likely influence the number of superstar and non-superstar games available on a console. This results in a reverse causality issue – the availability of non-superstar games and superstars influences market share, but market share also influences the number of these two types of games available (Clements and Ohashi 2005).

In the software supply equations, we are concerned that shocks in software supply during past periods may influence the installed base of the current period (Clements and Ohashi 2005).

To address these issues, we employ a series of cost side instruments used in other studies of the video game industry (Dubé et al. 2010; Gretz and Basuroy 2013). We use producer price indexes for Game Software Publishing (GSP), Computer Storage Device Manufacturing (CSDM), Electronic Computer Manufacturing (ECM), Magnetic and Optical Recording Media Manufacturing (MORMM), and Audio and Video Equipment Manufacturing (AVEM) obtained from the U.S. Bureau of Labor Statistics ( as cost shifters. Additionally we use various measures of competitor characteristics similar to Germann et al. (2015) and Malshe et al. (2017). We discuss each instrument, why we believe the validity and relevance assumptions hold in each of our estimations, and show the first-stage estimations in Web Appendix A. Briefly, all first-stage F-statistics are well over 10 indicating the instruments are strong (Wooldridge 2010), the signs and significance of the instruments are generally in line with our expectations, and the overidentification restrictions are satisfied in every estimation.

Diagnostic tests for GMM superiority

First, the Durbin-Wu-Hausman test (Wooldridge 2010) is statistically significant when comparing OLS and single equation GMM estimations for hardware demand and software supply equations, suggesting endogeneity is present.Footnote 31 Therefore, OLS estimates are likely biased. Second, GMM estimates are preferred if the instruments are exogenous and correlated with the endogenous variables. The non-significant Hansen J-statistics in all the GMM estimations supports the former condition, while the large (over 10) and significant first-stage F-stats support the latter (Rossi 2014; Stock and Watson 2007; Stock and Yogo 2002; Wooldridge 2010). Third, GMM estimation is more efficient than traditional Two-Stage Least Squares estimation (or Three-Stage Least Squares in the case of joint estimation) in the presence of arbitrary heteroskedasticity (Stock et al. 2002). Lastly, joint estimation of hardware demand and superstar/non-superstar software supply improves efficiency if errors are correlated across equations.

Appendix 2

Further Limitations

Future research should also consider differences in the utility derived from hedonic (e.g., video games) vs utilitarian (e.g., computers) products. We argue that the phase of the hardware lifecycle will moderate the impact of different software types on hardware sales in either setting. However, this moderation depends on the characteristics of superstars and non-superstars in relation to preferences of earlier hardware adopters compared to later adopters within the industry setting.Footnote 32

We note that we do consider the complexities associated with relating hardware adoption to expectations of future software (superstar or non-superstar) availability. Expectations of the number of future software available and hardware success may have an influence on current hardware adoption (Steiner et al. 2016). This is especially important when considering the role of superstars in launching new hardware to compete against an established incumbent. Research has shown that anticipated regret may influence willingness to purchase in a variety of settings (Zeelenberg 1999) especially when technology is fast evolving (Shih and Schau 2011). We expect this to have a limited impact in our setting where new generations of video game consoles are introduced every 5 to 7 years, but may be more relevant in industries like smartphones which usually have yearly introductions of new hardware. However, understanding how quality signals of superstars may influence anticipated regret or heuristics early adopters may employ when making decisions early in the hardware lifecycle is a useful area of future research in platform and two-sided markets.Footnote 33

Also, we do not consider possible inefficiencies caused by the compensation strategies we recommend to encourage superstar and non-superstar development over the hardware lifecycle. Software developers and hardware managers often have exclusive deals in the video game industry (e.g. 70.7% of games are released exclusively to a single console in our sample) which means our strategy suggestions can be implemented by hardware managers in negotiation with exclusive software providers. However, there may be inefficiencies or inequities that could occur for software developers that serve multiple hardware firms with the same game – especially when each hardware firm would prefer a different level of game quality.Footnote 34 Speaking to these possible inefficiencies, or potential remedies, is beyond the scope of the current paper and we leave it to future research.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Gretz, R.T., Malshe, A., Bauer, C. et al. The impact of superstar and non-superstar software on hardware sales: the moderating role of hardware lifecycle. J. of the Acad. Mark. Sci. 47, 394–416 (2019).

Download citation


  • Indirect network effect
  • Superstars
  • Two-sided markets
  • Lifecycle theory
  • Relationship marketing