Essays on the Economics of Open-Source Software

Abstract

This dissertation comprises three essays analyzing various economic questions relating to open-source software development. The common thread linking these essays is the long-term sustainability of the open-source software development model, which is largely built on unpaid contributions from individual developers scattered across the world. The first essay develops a theoretical analysis of the market for operating systems as two-sided platforms, modeling the effects of competition and compatibility between a proprietary platform developed by a profit-maximizing firm, and an open platform (a public good developed by volunteers). Looking at the impacts on the proprietary platform firm, and application developer firms and users of both platforms, I find that under certain circumstances, a proprietary platform can find it profitable to become compatible with the open platform. However, it is always optimal in terms of social welfare to have compatibility between platforms. The second essay uses a laboratory experiment to examine how these characteristics and levels of motivations that are heterogeneous across individuals interact to result in sustainable, non-zero levels of contribution to open source software. There is a pronounced “leadership effect,” with subjects playing in the first position invariably contributing more frequently than those in the second position, and so on. Heterogeneity preserves the leadership effect, but increases contributions across the board, and eliminates the pattern of declining individual and total group contributions over time frequently observed in public goods experiments. The third essay studies the micro-foundations of open-source software contributions and provides an empirical examination of developer motivations using survey data. If open-source contributions and education are both signals of ability, then their impact on income is likely to be linked. They may be complements if open source contributions reinforce the signal from education by showing that one stands out from the crowd, or they might be substitutes if open-source development replaces expensive education in honing programming skills by offering more immediate feedback. Using an instrumental variables framework to deal with the endogeneity of the education and contribution choices, I find that leading an open-source project and completing college are complementary practices, so that the signaling and reputation-building aspect dominates.