“Winners Take Some”Standards Competition in the Presence of Digital Conversion Technology - An Empirical Analysis of the Flash Memory Market Charles Z. Liu, University of Pittsburgh Chris F. Kemerer, University of Pittsburgh Michael D. Smith, Carnegie Mellon University
IT Standards Competition • Classic “Winner-Takes-All” markets: • VHS vs. Betamax VCR standards war 1987 - VHS has nearly 100% of the market 1979 – nearly equal market shares • Operating Systems: Windows vs. OS/2 • DVD Adoption: DVD vs. Divx • Blu-Ray vs. HD DVD
But, some recent counter examples: Digital media formats: WMV, RM, QuickTime, etc… Digital photography: JPEG, GIF, TIFF, PNG etc. Flash Memory Cards: • Compact Flash (CF) • SmartMedia (SM) • Secure Digital (SD) • Memory Stick (MS) • Multimedia Card (MMC) • xD Picture • Flash memory revenue is expected to reach $18.7 billion in 2010, up from a record $10.6 billion in 2005 (IDC 2006). 3
Flash Memory Cards Market Share: 01/2003 to 08/2006 Flash Memory: No Winner-Takes-All Result
Empirical Study – Research Questions What drives the lack of standards convergence in the flash memory card market? Network Effects? Conversion Technology? • Impact on Network Effects? • Impact on Different Standards? • Impact on Market Concentration? 5
Prior Literature Network Effects Price Premium (Hedonic Price Regression) (Greenstein, 1993, Gandal 1994, 1995, Brynjolfsson & Kemerer 1996) Conversion Technology Higher product price (Matutes & Regibeau, 1988, Economides, 1989, Farrell, 1992) Longer period of co-existence (Choi 1996, 1997) Reduced first-mover advantage (Liu et al. 2007) 6
Capacity, Speed, Format, Brand H1 (+) Network Effects (Installed Base) Price Premium H3 (–) H2 (+) Digital Conversion Adoption Market Concentration H4 (–) Price Variation Research Model
Data – Flash memory market • Time Frame: Jan 2003~ Aug 2006 • Data (multiple sources) • NPD Group - POS data collected from retail stores: • Flash memory cards and converters: monthly price, units sold, capacity, brand, at the product model level • Product attributes data from Amazon.com by an automatic software agent • Specifications data collected from manufacturers' websites and corresponding flash memory standard associations • Sample Size: 15,091 (consists of 706 products across 44 months)
Econometric Model – Model 1 • Econometric Issues: • Multicollinearity • Heteroskedasticity • Panel specific autocorrelation • Endogeneity • Solutions: • Centering interacting variables • GLS adjustments • 2SLS
Results – Model 1 † N=15,091, 6 Formats, 45 brands, 12 sizes, 706 panels across 44 months. ♀ Standard errors in parentheses. *: p<5%; **: p<1%.
Results – Marginal Effects • H3 is supported • Coefficient α3 is highly significant (p<0.001) • An F test between the restricted model (the model without the interaction term) and the unrestricted model is also significant.
Interaction Plot 1 Flash Memory Card Price Premium 30 Leading Format 20 10 Average Format 0 -10 Minority Format Converter Adoption Level -20 Figure: Flash Memory Card Price Premium at Different Converter Adoption Levels
Interaction Plot 2 Flash Memory Card Price Premium 30 Low Converter Adoption Level Average Converter Adoption Level 20 10 0 High Converter Adoption Level -10 Flash Memory Install Base -20 Figure: Flash Memory Card Price Premium at Different Installed Base Levels
Econometric Model 2 (Hypothesis 4) • Market concentration: Herfindahl-Hirschman Index (HHI) • Results H4 is supported
Implications • Consumers: Converters and digital media alleviate consumer dilemma when making a technology choice. • Firms: Pricing strategies • Market: Intensified competition • Society: An alternative mechanism to foster technology adoption.