Bayesian privacy

Modern information technologies make it possible to store, analyze, and trade unprecedented amounts of detailed information about individuals. This has led to public discussions on whether individuals' privacy should be better protected by restricting the amount or the precision of information that is collected by commercial institutions on their participants. We contribute to this discussion by proposing a Bayesian approach to measure loss of privacy in a mechanism. Specifically, we define the loss of privacy associated with a mechanism as the difference between the designer's prior and posterior beliefs about an agent's type, where this difference is calculated using Kullback–Leibler divergence, and where the change in beliefs is triggered by actions taken by the agent in the mechanism. We consider both ex post (for every realized type, the maximal difference in beliefs cannot exceed some threshold κ) and ex ante (the expected difference in beliefs over all type realizations cannot exceed some threshold κ) measures of privacy loss. Applying these notions to the monopolistic screening environment of Mussa and Rosen (1978), we study the properties of optimal privacy‐constrained mechanisms and the relation between welfare/profits and privacy levels.

A model of weighted network formation

This paper proposes a game of weighted network formation in which each agent has a limited resource to form links of possibly different intensities with other agents and to use for private purposes. We show that every equilibrium is either “reciprocal” or “nonreciprocal.” In a reciprocal equilibrium, any two agents invest equally in the link between them. In a nonreciprocal equilibrium, agents are partitioned into “concentrated” and “diversified” agents, and a concentrated agent is only linked to diversified agents and vice versa. For every link, the concentrated agent invests more in the link than the diversified agent. The unweighted relationship graph of an equilibrium, in which two agents are linked if they both invest positively in each other, uniquely predicts the equilibrium values of each agent's network investment and utility level, as well as the ratio of any two agents' investments in each other. We show that equilibria are not pairwise stable and are not efficient due to the positive externalities of investing in a link.

Evolution, heritable risk, and skewness loving

Our understanding of risk preferences can be sharpened by considering their evolutionary basis. The existing literature has focused on two sources of risk: idiosyncratic risk and aggregate risk. We introduce a new source of risk—heritable risk—in which there is a positive correlation between the fitness of a newborn agent and the fitness of her parent. Heritable risk was plausibly common in our evolutionary past and it leads to a strictly higher growth rate than the other sources of risk. We show that the presence of heritable risk in the evolutionary past may explain the tendency of people to exhibit skewness loving today.

Mechanism design with financially constrained agents and costly verification

A principal distributes an indivisible good to budget‐constrained agents when both valuation and budget are agents' private information. The principal can verify an agent's budget at a cost. The welfare‐maximizing mechanism can be implemented via a two‐stage scheme. First, agents report their budgets, receive cash transfers, and decide whether to enter a lottery over the good. Second, recipients of the good can sell it on a resale market but must pay a sales tax. Low‐budget agents receive a higher cash transfer, pay a lower price to enter the lottery, and face a higher sales tax. They are also randomly inspected.

Bounded rationality and limited data sets

Bounded rationality theories are typically characterized over exhaustive data sets. We develop a methodology to understand the empirical content of such theories with limited data, adapting the classic revealed‐preference approach to new forms of revealed information. We apply our approach to an array of theories, illustrating its versatility. We identify theories and data sets testable in the same elegant way as rationality, and theories and data sets where testing is more challenging. We show that previous attempts to test consistency of limited data with bounded rationality theories are subject to a conceptual pitfall that may lead to false conclusions that the data are consistent with the theory.

Bottleneck links, essential intermediaries, and competing paths of diffusion in networks

We investigate how information goods are priced and diffused over links in a network. A new equivalence relation between nodes captures the effects of network architecture and locations of sellers on the division of profits, and characterizes the topology of competing (and potentially overlapping) diffusion paths. Sellers indirectly appropriate profits over intermediation chains from buyers in their equivalence classes. Links within the same class constitute bottlenecks for information diffusion and confer monopoly power. Links that bridge distinct classes are redundant for diffusion and generate competition among sellers. In dense networks, competition limits the scope of indirect appropriability and intellectual property rights foster innovation.

Payoff implications of incentive contracting

In the context of a canonical agency model, we study the payoff implications of introducing optimally structured incentives. We do so from the perspective of an analyst who does not know the agent's preferences for responding to incentives, but does know that the principal knows them. We provide, in particular, tight bounds on the principal's expected benefit from optimal incentive contracting across feasible values of the agent's expected rents. We thus show how economically relevant predictions can be made robustly given ignorance of a key primitive.

Robust sequential search

We study sequential search without priors. Our interest lies in decision rules that are close to being optimal under each prior and after each history. We call these rules robust. The search literature employs optimal rules based on cutoff strategies, and these rules are not robust. We derive robust rules and show that their performance exceeds 1/2 of the optimum against binary independent and identically distributed (i.i.d.) environments and 1/4 of the optimum against all i.i.d. environments. This performance improves substantially with the outside option value; for instance, it exceeds 2/3 of the optimum if the outside option exceeds 1/6 of the highest possible alternative.