Research

This paper explores the relationship between market accessibility and various participants’ welfare in an intermediated directed-search market. For a general class of meeting technologies, we provide a necessary and sufficient condition under which efficiency requires imperfect accessibility, such that each seller’s listing is only observed by some but not all buyers. We show that the platform optimally implements the efficient outcome, but fully extracts surplus from the transactions it intermediates. We also find that in general, buyers prefer to minimize market accessibility, while sellers prefer a weakly greater accessibility level than that which is socially efficient. The efficiency of imperfect accessibility is robust to the introduction of a second chance for unmatched buyers to search. 

We study how acquisitions by a dominant incumbent affect entry and R\&D incentives in markets with multiple startups. We show that acquisitions can create a kill zone, which suppresses entry and distorts the innovation direction of non-targeted startups. The resulting reduced threat of entry may lead the incumbent to shelve the acquired technology. The kill zone effect strengthens targeted startups’ incentives to enter primarily to be bought out, and makes acquisitions more attractive to the incumbent than in-house R\&D. To balance kill zone distortions to innovation against the potential synergies from acquisitions, a consumer welfare–maximizing merger policy may involve blocking some, but not all, acquisitions.

Sender maximizes how many receivers buy a widget of uncertain quality. She targets receivers to privately communicate information about the widget’s quality. After the target chooses whether to buy the widget, other receivers observe the target’s choice with probability increasing in the target’s popularity. We prove two results that hold remarkably generally: Sender optimally communicates with targets as if no other receiver exists; a target's popularity is a double-edged sword for Sender.  We fully characterize the optimal choice of a single target under multiple communication protocols and establish that Sender can benefit from protocols that constrain her communication.

A sender conveys scarce information to a number of receivers to maximize the sum of receiver payoffs. Each receiver’s payoff depends on the state of the world and an action she takes. The optimal action is state contingent. Under mild regularity conditions, we show that the payoff of each receiver is convex in the amount of information she receives. Thus, it is optimal for Sender to target information to a single receiver. We then study three extensions in which interior information allocations are optimal.

We characterize Pareto-efficient allocations of a stock of capital among workers who may acquire it at a cost between receiving a firm’s offer and deciding whether to accept. Capital raises a worker's value not only to the firm, but also upon rejecting, leading to a moral hazard problem. This problem implies an upper bound on total welfare that decreases in workers' welfare. We show that an allocation is Pareto-efficient if and only if it achieves this bound. This bound is achievable if and only if workers’ welfare lies below a cutoff. Subsidizing acquisition costs sharply increases the bound and cutoff.

We discuss how post-acquisition fears of killer acquisitions can be overstated, especially when targets are engaging in product rather than process innovations. We also discuss how the prospect of buyouts throughout acquisition may not always steer startups’ entry and pre-acquisition innovation directions toward the more disruptive innovation. To balance short-run competitive loss against long-run innovation gains, a consumer-welfare merger policy may involve approving some, but not all, acquisitions. We draw practical implications for merger control.

A principal incentivizes a team of symmetric agents to work on a project by offering them success-contingent bonuses. I assume the equilibrium selected under an incentive scheme is the limit of Level-K thinking. Agents' initial belief about others working, called culture, determines equilibrium selection. I show that a minimum cost incentive scheme which implements work offers a subset of workers the largest bonuses identically, and the remaining workers smaller bonuses. A more hardworking culture reduces the cost of implementing work and increases the number of agents offered the largest bonuses identically. This also implies a shorter bonus hierarchy when bonuses must be public, and more transparent incentives when bonuses can be private. For most cultures, the minimum-cost incentive scheme discriminates between agents both in terms of bonuses and information. Whether the minimum-cost incentive  public or private scheme discriminates more among agents depends on culture.

I study information design in symmetric binary action supermodular games. Agents choose to invest or not, and the designer always prefers higher aggregate investment. I assume the equilibrium selected under an information structure is the limit of Level-K thinking. The initial belief about aggregate investment, called the benchmark, determines equilibrium selection. Designer-preferred and adversarial selection coincide with extreme benchmarks. I show that all outcomes are implemented by information structures that send a fraction of agents identical signals, and others distinct signals. I construct a designer-optimal information structure from this class. I show that under it, raising the benchmark always increases aggregate investment and the fraction of agents receiving identical signals, but often has an ambiguous effect on agents' payoffs. Finally, the designer's loss when limited to public information is lower under higher benchmarks. My work elucidates on how payoff-irrelevant factors, captured by the benchmark, shape optimal information disclosure.

We analyse the progressivity and redistributive effects of India’s income tax system utilizing Income Tax Department data for 2011–18. By fitting Lorenz and tax concentration curves to these data, we find that despite exhibiting high levels of progressivity, the redistributive effects of income taxes remain modest amongst tax assessees and miniscule within the adult population. We also find that plugging the gap between statutory and actual average tax rates will do little to improve redistributive effects, and lowering income thresholds for top marginal tax rates offers greater redistributive and revenue potential than reducing exemption limits or increasing top marginal tax rates.