Research
Publications:
Costly Search with Adverse Selection: Solicitation Curse vs. Acceleration Blessing with Kyungmin (Teddy) Kim, RAND Journal of Economics, Vol. 48, No. 2 (May 2017), pp. 1756-2171
We consider a dynamic trading model in which a seller, with private information about the quality of her good, can increase the frequency of strategic price quotes through more intensive search (or advertising). A low-quality seller benefits more from trade, and therefore searches more intensively than a high-quality seller. We identify two opposing effects of endogenous search intensity on buyers' inferences. On the one hand, a low-quality seller is more likely to find a buyer than a high-quality seller, and thus a seller's contact carries negative information (solicitation curse). On the other hand, a low-quality seller leaves the market even faster than a high-quality seller because she is willing to accept lower offers, and thus a seller's availability becomes a stronger indicator of high quality (acceleration blessing). We study how these two effects manifest themselves and interact with each other in both stationary and non-stationary environments. In the stationary environment, the two effects exactly offset each other for any strategy profile, and reducing search costs is weakly beneficial to the seller. In the non-stationary environment, the relative strengths of the two effects vary over time, generating a unique form of trading dynamics, and reducing search costs can be detrimental to the seller.
Optimal Information Design for Search Goods with Michael Choi and Kyungmin (Teddy) Kim, AEA Papers and Proceedings, Volume 109, May 2019, Pages 550-556.
We consider a monopoly pricing problem in which a consumer with an uncertain valuation of a search good receives a signal of value before deciding whether to visit the seller. She discovers her true value upon visit and before purchase. We characterize the consumer-optimal and seller-worst signals in such an environment and deliver two main insights. First, both the consumer-optimal and seller-worst signals generate a unit-elastic demand. Second, the two signals coincide if and only if visitation costs are sufficiently small.
Shopping for Information: The Implications of Consumer Learning for Optimal Pricing and Product Design, Journal of Industrial Economics, Volume 71, Issue 3 (September 2023), pp. 883-923.
I study a seller's pricing problem in which consumers perform costly product research about value before purchase. They buy the product when sufficiently optimistic about value and cease research when sufficiently pessimistic. I find that the seller encourages product research when prior belief about value is high, even though he could sell immediately for a high price. The prior affects both expected value and how additional information changes consumers' beliefs. I show that an increase in research cost affects equilibrium price non-monotonically. Finally, when the seller chooses price and product value dispersion, the optimal level of dispersion need not be extremal.
Blind Disclosure with Aaron Kolb, Daniel W. Sacks, and Joshua Quick, AEJ: Microeconomics, Vol. 15, No. 2 (May 2023), pp. 41-79.
We develop and test a theory of blind disclosure. A sender chooses whether to disclose information based on a preliminary, private signal. In the unique equilibrium, contrary to the literature's canonical unraveling result, senders disclose only if their preliminary signal exceeds a cutoff. This cutoff rule leads to partial unraveling in environments with either risk aversion or moral hazard, and disclosure decreases with uncertainty. Using unique administrative data on disclosed and undisclosed grades in a large university, we find that the model is consistent with student choices during Spring 2020 to conceal letter grades by switching to optional pass-fail grades.
Unemployment Duration under Flexible Information Acquisition with Jeong Ho (John) Kim and Kyungmin Kim, International Economic Review, Vol. 65, No. 1 (February 2024), pp.471-503.
We consider a worker’s job search problem in which firms arrive sequentially, observe the worker’s unemployment duration, and conduct an interview to learn about her unobservable productivity. Firms engage in fully flexible information acquisition subject to a uniformly posterior-separable cost function. We provide a closed-form characterization of equilibrium job search dynamics and demonstrate that endogenous information amplifies the “stigma” effect of long unemployment duration relative to exogenous information. We also show that lowering firms’ information-acquisition costs has ambiguous implications for a worker’s unemployment duration.
Working Papers:
In the Interest of Full Disclosure: Consequences of the Grand Bargain in Patenting with Mike Andrews and Rajkamal Vasu
We consider a model where an innovator chooses how much to disclose about their invention before Cournot competing. More disclosure lets the follower copy more, but also signals strength and increases the innovator's probability of winning an infringement suit. We find policies that increase damages due to copying lead to universally more disclosure, while policies that increase winning probabilities induce less disclosure from large inventions and more from small inventions. We validate our predictions using two court decisions; one increased damages, the other winning probabilities. We conclude that some pro-patent policies are counterproductive, reducing disclosure for the largest inventions.
Safety, In Numbers with Mark Whitmeyer
We introduce a way to compare actions in decision problems. An action is safer than another if the set of beliefs at which the decision-maker prefers the safer action increases in size (in the set-inclusion sense) as the decision-maker becomes more risk averse. We provide a full characterization of this relation and show that it is equivalent to a robust concept of single-crossing. We discuss applications to investment hedging, security design, and game theory.
Inverting Akerlof: Quantifying Private Information in Markets for Patents with Mike Andrews and Rajkamal Vasu
Private information causes markets to fail. Markets for innovative technologies are especially fascinating from an information perspective because both buyers and sellers may possess private information about the quality of an innovative technology. Additionally, the quality of innovations are highly uncertain, and uncertainty can exacerbate market failure by increasing the value of private information. In this paper, we develop a model that incorporates private information on the part of both sellers and buyers. Importantly, key parameters of our model map to observable outcomes. This allows us to "invert" the model to obtain estimates of parameters that govern private information. We apply the model to U.S. patent data. We present estimates for private information for a baseline sample of patents. We then conduct counterfactual exercises to see how welfare changes under alternative private information regimes. We find that given the distribution of patent quality, eliminating sellers' private information has little effect on welfare, while eliminating buyers' private information would lead to a substantial welfare increase.
Information Acquisition with Spillovers with Mark Whitmeyer
We consider a model of sequential information acquisition with strategic spillovers. Two players engage in fully flexible information acquisition subject to a uniformly posterior-separable cost function. The first player (leader) chooses how much to learn about the unknown state of the world and then takes an action visible to the second player (follower). The follower then chooses how much information to acquire and takes an action of his own. Players' actions can be strategic complements or substitutes. We characterize how actions and information are governed by the complementarity of actions. In particular, we find that complementarity in actions leads to substitutabilty in learning and visa versa; if actions are complements, the follower is unlikely to acquire information of his own. If actions are substitutes, however, the follower would like to acquire additional information. We also show that the leader is only tempted to lie and mislead the follower after receiving a high signal if actions are extremely substitutable and after receiving a low signal if actions are extremely complementary.
Advertising Auctions with Kyungmin (Teddy) Kim
We introduce informative advertising into an optimal auction problem with endogenous entry. Specifically, we consider an independent-private-values environment with costly participation and allow the seller to choose not only the reserve price, but also how much product information to provide for bidders before they enter. We prove that the revenue-maximizing information structure takes a binary-cutoff form; the seller informs each bidder whether his value is above or below a threshold. We study how the level of this threshold and the optimal reserve price depend on the entry cost. We show that the seller can extract full surplus if and only if the entry cost is sufficiently large and that the optimal reserve price necessarily decreases in the entry cost if and only if the number of potential bidders is sufficiently small.