Title: Essays in Industrial Organization
- Mar Reguant, Northwestern University
- Heski Bar-Isaac, University of Toronto
- Johannes Horner, University of Yale
- Stefan Ambec, INRAE & TSE
- Mathias Reynaert, UT1 Capitole – TSE
- Jean Tirole, Honorary Chairman TSE & IAST, UT1 Capitole – TSE
This thesis consists of three independent chapters borrowing concepts and methods from the field of industrial organization to address some of the challenges and concerns that have arisen with the development of information technologies. In the first two chapters, I develop theoretical frameworks to study societal questions related to the collection and use of personal data. In particular, I explore the idea that one can encourage certain behavior by exploiting personal data - rather than financial incentives. The first chapter is motivated by the growing demand for better control over who can access personal data and for which purposes - in particular concerning the spreading practice, both by public authorities and private companies, of scoring individuals. While disclosing information makes individuals accountable for their actions and can protect others from un- desirable encounters, it also increases ostracism and costly avoidance strategies. I study the efficient disclosure of information by a principal about an agent who has reputation concerns vis-à-vis multiple audiences with heterogeneous valuations for information. Disclosing in- formation to an audience jeopardizes the agent’s reputation and thus incentivizes effort and disclosing more information increases effort. To maximize effort under a participation constraint it is optimal to disclose undifferentiated information across the audiences who value information the most and to protect the agent’s reputation vis-à-vis the other ones. When the principal internalizes the audiences’ payoffs, he trades-off providing more information to the most interested audiences with protecting the agent’s reputation vis-à-vis the other ones. When the principal internalizes the audiences’ payoffs, information is no longer undifferentiated and he discloses all the more information so as it is useful to the audience. In return, he better protects the agent’s reputation vis-à-vis the other audiences. The results can rationalize existing laws protecting workers and consumers and inform the debate regarding social scoring or the sharing of information between firms and within multi-sided platforms. In the second chapter, I study how a platform can exploit personal information to decide who to accept and who meets who and encourage pro-social behavior. I build a model with the assumption that some individuals are better match partners than others - for instance because they behave more altruistically - and ask who to match them with. There are no monetary transfers to provide incentives but, instead, the platform can use the best match partners as a reward for good ratings. I characterize the matching mechanism maximizing welfare on the platform. This mechanism is not entry-proof, though, and competition between platforms leads to stratification whereby the best match partners are matched together, a phenomenon called positive assortative matching (PAM) in the literature. Thus, platform entry creates a hold-up problem which reduces individuals’ provision of effort. This issue is exacerbated with data portability because it allows en entrant to screen individuals more efficiently. The framework developed is relevant for the regulation of algorithms developed by platforms of the gig economy and can be adapted to study the effect of urban planning on community harmony. In the third chapter, co-authored with Kevin Remmy, we study empirically the introduction of real-time pricing (RTP) in New Zealand - a new electricity tariff relying on smart meters which record detailed consumption data - and our goal is to identify barriers to its adoption. While economic theory predicts that introducing RTP in an economy with rational and perfectly informed agents will lead the retail market to unravel, less than 1.25% of residential consumers adopted this tariff more than seven years after its introduction. Under this tariff, consumers are exposed to half-hourly varying spot prices, which are uncertain and volatile. We show that when ongoing spot prices spike, prospective adopters forego adoption and re- cent adopters switch to another tariff or reduce their electricity consumption - which is a sign of present bias. However, with experience, consumers on real-time pricing gradually become less sensitive to ongoing spot prices. This suggests that large unexpected prices spikes can jam the unraveling process and therefore that the timing of adoption matters. While uncertainty cannot be fully eliminated certain periods are riskier than others and should be avoided - such as before a harsh winter in regions with widespread electric heating. Furthermore, the fact that experience matters suggests that present bias is due to a lack of information about long-term payoffs and we propose remedies to help consumers make rational and informed decisions such as using consumption data from smart meters to estimate individual costs and benefits from adoption.