Matthew S. Wilson

Disaggregation and the Equity Premium Puzzle
Journal of Empirical Finance, Sep. 2020, vol. 58

Abstract: Standard macroeconomic models cannot explain why stocks so greatly outperform bonds. However, this result depends on the use of aggregate consumption data. If markets are incomplete, then a representative agent might not exist and it is necessary to use consumption data at the household rather than aggregate level. In the household data, the equity premium puzzle is solved when the coefficient of relative risk aversion is as low as 2.7-3.8. I demonstrate that this result is robust in a very general framework and prove that many of the tests used in the literature are biased.

Full text

A Real Business Cycle Model with Money as a Sunspot Variable
Journal of Economics and Business, May-June 2020, vol. 109

Abstract: A well-known criticism of the RBC model is that it cannot match the data on money. Due to the perfect flexibility of prices and the absence of frictions, any exogenous increase in the money supply will be fully offset by wage and price increases, implying that money is neutral even in the short run. However, beliefs that money is non-neutral could become a self-fulfilling prophesy. By using money as a sunspot variable in an RBC model, I successfully replicate many of the correlations in the data, even though money does not directly affect the economy’s fundamentals. This shows that models with flexible prices are not necessarily incompatible with the monetary data and offers support for the use of sunspot variables in macroeconomics.

A non-technical summary I wrote for the Lindau Blog

Full text

There is a typo in the published version. Equation (46) should be the following.


Rationality with Preference Discovery Costs
Theory and Decision, Aug. 2018, vol. 85, no. 2, 231-251.

Abstract: Economic theory assumes that preferences are rational. However, experiments have found small violations of transitivity. This paper develops a model of rationality with preference discovery costs. Introspection is costly. Thus, agents may find it optimal to use less than full effort, even though this raises the risk of making a poor choice. This model could potentially explain the intransitivities observed in the data while retaining rationality and optimization.

Full text

Observed Altruism in Dental Students: An Experiment Using the Ultimatum Game
(with Crutchfield, Jarvis, and Olson). Journal of Dental Education, Nov. 2017, vol. 81 no. 11, 1301-1308.

This interdisciplinary collaboration was selected as one of the "10 Most Notable Articles of 2017" by the Journal of Dental Education. The editor writes, "This creative use of research methods from outside dental education sheds light on a common question: does dental students’ altruism decrease over their education? The study used a computer game that asked dental students in all four years to choose how much of their limited resources to give to recipients based on factors such as socioeconomic status. In the results, students showed higher levels of altruism than the general population, and their altruism was highest in the fourth year. The latter finding—contrary to previous studies—suggests that the often-found decline in altruism may be due not to student characteristics but to factors in the educational environment."

Abstract: The conventional wisdom in dental and medical education is that dental and medical students experience “ethical erosion” over the duration of dental and medical school. There is some evidence for this claim, but in the case of dental education the evidence consists entirely of survey research, which does not measure behavior. The aim of this study was to measure the altruistic behavior of dental students in order to fill the significant gap in knowledge of how students are disposed to behave, rather than how they are disposed to think. To test the altruistic behavior of dental students, the authors conducted a field experiment using the Ultimatum Game, a two-player game used in economics to observe social behavior. In the game, the “proposer” is given a pot of resources, typically money, to split with the “responder.” The proposer proposes a split of the pot to the responder. If the responder accepts the proposed split, both participants keep the amounts offered. If the proposal is rejected, then neither participant receives anything. In this study, the students played the proposer, and the responder was a fictional individual although the students believed they were playing the computerized game with a real person. In fall 2015, dental students from each of the four years at one university played the game. All 160 students were invited to participate, and 136 did so, for a response rate of 85%. The results showed that the students exhibited greater levels of altruism than the general population typically does. The students’ altruism was at its highest in year four and was associated with the socioeconomic status of responder. This result raises the possibility that if a decreasing ability to behave altruistically is observed during dental school, it may not be due to a general disposition of students, but rather some factor specific to the educational environment.

Full text

Disaggregation Reverses the Risk-free Rate Puzzle
Working paper

Abstract: In the macro data, it appears that people are buying too many bonds. They help agents smooth consumption, but the rate of return is very low and aggregate consumption is already very smooth. However, consumption at the household level is far more volatile. This makes bonds more appealing. Instead of solving the puzzle, the main result is reversed: households are not buying enough bonds. Under some specifications, the unconditional Euler equation is not rejected, but in those cases the forecast errors are inconsistent with rational expectations.

Social Contact During a Pandemic: Rationality vs. Heuristics
Working paper

Abstract: During a pandemic, you weigh the benefits of social contact against the risks to your own health. However, optimality includes another factor: the risk that you pose to others. I study this externality with the SEIR model from epidemiology. Unlike the previous literature, I do not rely upon death rates; they are uncertain and fail to reflect the suffering that survivors experienced. Instead, I estimate revealed preferences for health and social contact at the county level. I find that people relied on a heuristic and there is no evidence of pandemic fatigue.


How measurement error impacts the hypothesis tests. The null is still rejected about 5% of the time.


An impulse response function for the sunspot


Due to preference discovery costs, the consumer picks at random from the dotted portion of the budget lines


Offers made by the average dental student


Miami-Dade County: optimal social contact vs. social contact with laissez-faire policy and rational agents