I am a behavioral economist, using theory and lab experiments to explore information transmission topics such as price-quality inference. I enjoy teaching and searching for the perfect graph to communicate complex data. Outside of academia, I love spending time in nature with my wife and 3-year-old in Charlottesville, Virginia.

CV      JMP

Working Papers

• “Competition and Price Informativeness: An Experiment”

  Consumers often rely on price as a guide to the quality of a product. If consumers are unable to observe quality directly, price might not perfectly communicate product quality. In this context, competition might increase or decrease the informativeness of prices. I study how competition impacts the informativeness of prices theoretically and in a lab experiment. While theory leaves the question open, I find in the experiment that competition leads both high- and low-quality firms to decrease prices, but the price reduction is larger for high-quality firms who are more likely to price high in the absence of competition. Thus, prices become a less reliable guide to quality when there are more sellers in the market.

• “The Directional Difference Test” (joint with Charlie Holt and Sean Sullivan)

  When analyzing data from experiments involving more than two treatments that can be ranked in terms predicted effect, it is natural to turn to statistical tests with directional alternative hypotheses. This paper uses Monte Carlo simulations to evaluate the power characteristics of the Directional Difference test, which has a test statistic defined as the sum of treatment differences for all pairs of sample observations in different treatments. When sample data are distributed according to familiar distributions like the normal, uniform, and logistic distributions, we find that the Directional Difference test is better able to detect treatment effects than the commonly used Jonckheere-Terpstra test. The Jonckheere-Terpstra test still performs best for heavy-tailed distributions. Experimenters wishing to exploit the informational content of multiple-treatment samples should consider replacing the Jonckheere-Terpstra test with the Directional Difference test in appropriate circumstances.

In Progress

• “The Value of Medicaid Auto-Enrollment: Evidence from the Oregon Health Insurance Experiment” (joint with Sasha Ruby)

  Is voluntary enrollment an effective strategy to limit Medicaid participation? This paper provides new evidence, using data from the Oregon Health Insurance Experiment, where Medicaid expansion was implemented through random assignment. We compare the average recipient value for Medicaid and average recipient cost to the government under voluntary enrollment to the average recipient value for Medicaid and average recipient cost to the government under automatic enrollment. We find that the average recipient under automatic enrollment not only benefits more from Medicaid but also imposes lower costs on the government. This result stems from the fact that Medicaid recipients under voluntary enrollment, despite having lower incomes, have greater access to alternative forms of health insurance, which diminishes their relative value for Medicaid.

• “Cursedness in Simultaneous and Sequential Voting Games” (Conference presentation at ESA 2023)

  Agents often fail to accurately infer others’ private information from their behavior, formalized as “cursedness” (Eyster and Rabin, 2005). Inferring others’ types is even more difficult when agents must condition on behavior that is not directly observable. In a voting game, a voter should condition their behavior on the information set where their vote is pivotal, even if they do not yet know whether they will turn out to be pivotal. Here I examine whether voters are more cursed when they do not directly observe how likely they are to be pivotal. I examine voting game experiments conducted by Anderson et al. (2022) to compare simultaneous and sequential voting outcomes. I fit quantal response to capture noise in the agents’ choices, and cursedness to capture agents’ failure to infer others’ types. I find that agents are much more cursed when they must infer others’ types based on the hypothetical event in which they are pivotal, and less cursed when they see from others’ behavior that they are in fact likely to be pivotal. Moreover, quantal response and cursedness work together to rationalize a smooth information cascade as agents’ uncertainty about their pivotality is gradually realized.

• “Can Price Inform Quality when Verification is Costly? The Grossman-Stiglitz Paradox in the Lab” (Conference presentation at ESA 2022)

  A product’s price will reflect its quality, if that quality is known to consumers. But many experiments have shown that consumers believe high prices signal high quality, even in situations where the quality is not easy to observe. I run a lab experiment where prices are chosen endogenously by sellers, and buyers may exert costly effort to become informed about quality. While theory leaves open how much information will be conveyed by equilibrium prices, I find in the lab that prices convey about as much information as theoretically possible. Consistent with the Grossman-Stiglitx paradox, subjects acquire information up until the return to information offsets its costs. Subjects also have behavioral deviations from Nash that follow quantal response intuition: subjects trust the market prices too much when it is costly to acquire information themselves, and trust the market prices too little when becoming informed is easy.