Research
Publications
“Do People Trust Humans More Than ChatGPT?” (2024) with William Hickman. Journal of Behavioral and Experimental Economics, 112: 102239. (featured on Marginal Revolution) (Top Downloaded Paper on SSRN for Behavioral Research)(featured on the AI in Education podcast) Summary here.
Abstract: We explore whether people trust the accuracy of statements produced by large language models (LLMs) versus those written by humans. While LLMs have showcased impressive capabilities in generating text, concerns have been raised regarding the potential for misinformation, bias, or false responses. In this experiment, participants rate the accuracy of statements under different information conditions. Participants who are not explicitly informed of authorship tend to trust statements they believe are human-written more than those attributed to ChatGPT. However, when informed about authorship, participants show equal skepticism towards both human and AI writers. Informed participants are, overall, more likely to choose costly fact-checking. These outcomes suggest that trust in AI-generated content is context-dependent.
"ChatGPT Hallucinates Nonexistent Citations: Evidence from Economics" (2024) with Stephen Hill and Olga Shapoval. The American Economist, 69(1), 80-87. (coverage from the Wall Street Journal)
Abstract: In this study, we generate prompts derived from every topic within the Journal of Economic Literature to assess the abilities of both GPT-3.5 and GPT-4 versions of the ChatGPT large language model (LLM) to write about economic concepts. ChatGPT demonstrates considerable competency in offering general summaries but also cites non-existent references. More than 30% of the citations provided by the GPT-3.5 version do not exist and this rate is only slightly reduced for the GPT-4 version. Additionally, our findings suggest that the reliability of the model decreases as the prompts become more specific. We provide quantitative evidence for errors in ChatGPT output to demonstrate the importance of LLM verification.
"Willingness to be Paid: Who Trains for Tech Jobs?" (2022), Labour Economics, Vol 79, 102267. (as featured in Politico Morning Tech and Marginal Revolution; previously a "Top Ten" manuscript at SSRN)
Abstract: Having a larger high-skill workforce increases productivity, so it is useful to understand how workers self-select into high-paying technology (tech) jobs. This study examines how workers decide whether or not to pursue tech, through an experiment in which subjects are offered a short programming job. Despite the persistent gender gap in computer science fields, female subjects in this incentivized experiment are equally confident and willing to program. The most important determinants of the “reservation wage” for workers to do computer programming are whether they enjoy it and their opportunity cost of time. Two treatments were introduced to increase confidence. Randomly assigned extra information about the programming task has no effect. In some circumstances, an encouraging message raises the reservation wage for programming.
Related Policy Review Article published by the GCO: "The Slow Adjustment in Tech Labor: Why Do High-Paying Tech Jobs Go Unfilled?"
“How dictators use information about recipients,” (2022) with Laura Razzolini, Journal of Behavioral Finance, 23:4, 408-426. Summary here
Abstract: This paper explores the extent to which altruism is influenced by the salient features of the beneficiaries. We investigate how information presented to senders affects their perception of the recipient in a dictator game. In this environment, the starting endowment of a recipient can be inferred from choices the recipient made. Dictators give the same amount to all recipients regardless of the choices they made, despite the revealed preference to send more money to recipients who started with lower endowments. Dictators give very little when they are explicitly told that a recipient started with a high endowment. However, when dictators are only told that the recipient made a choice that indicates they started with a high endowment, dictators do not incorporate that information. Dictators are not more generous to others who made a similar financial choice to themselves, through in-group bias. An implication from our results is that charitable donors are influenced by salient information about recipients and do not try to infer deservingness beyond what is explicitly presented.
"Other people's money: preferences for equality in groups," with Gavin Roberts, European Journal of Political Economy, Vol. 73. (SSRN link or summary here)
Abstract: Economic policy decisions often involve a trade-off between equality and efficiency implemented through income redistribution. We test whether people are less likely to purchase equality with their own money versus transferring someone else’s money to a low-income group member which reduces inequality at the cost of group efficiency. We synthesize Andreoni and Miller (2002) and Engelmann and Strobel (2004) by using an experiment that allows us to measure willingness-to-pay for equality both with own income and other people’s money. Subjects are more likely to purchase equality with others’ money at the cost of group efficiency. The average individual prefers an outcome with more own income and more inequality. Willingness-to-pay for equality is greater than zero when using others’ money, while subjects are sensitive to prices even when making purchases with others’ money. When the cost in terms of group efficiency is very high, subjects usually do not choose to reduce inequality. We find the same outcome for majority-rule decisions and when subjects decide as a dictator.
"If Wages Fell During a Recession," (2020) with Daniel Houser, Journal of Economic Behavior and Organization, Vol. 200, 1141-1159. Working paper here
Abstract: Many economies exhibit downward wage rigidity. Surveys of managers indicate that employers hold wages rigid because they believe morale will suffer after a wage cut. Otherwise, there is little evidence for how employers' beliefs contribute to wage rigidity and whether those beliefs are accurate. Using an experiment, we demonstrate that effort falls after workers experience a wage cut. Despite this partial confirmation of the Bewley (1999) morale theory, half of the employers in our experiment cut wages and lose money as a result. Under nominal inflation, real wage cuts do not have a significant effect on effort.
"My Reference Point, Not Yours," (2020) Journal of Economic Behavior and Organization, 171: 297-311.
Abstract: Reference points formed by initial endowments influence individual decisions. This experiment tests whether an individual can predict the behavior of other people who have different reference points. Despite financial incentives for being correct, players fail to imagine themselves in another person’s shoes. A low endowment player generally cannot predict the behavior of those who were assigned high endowments, and vice versa, when asked about group behavior. Instead of considering the perspectives of others, a low endowment player predicts that all others will act as if they all had low endowments. This controlled experiment helps explain why it is difficult to understand the perspectives of other people, while also demonstrating that it is possible when a player is specifically prompted to consider an individual in a different circumstance.
"Implementation of a New Data Analytics Curriculum: A Case Study," (2019), with Jennings B. Marshall, Steven T. Jones, Alan Blankley, Chad Carson, Cynthia Lohrke, Matt Mazzei, and Kevin Pan. Proceedings of the Academy of Economics and Finance, 42: 25-30.
"Smile, Dictator, You’re On Camera,” (2017), with Matthew McMahon, Matthew Simpson and Bart Wilson. Southern Economic Journal, 84:1, 52-65.
Abstract: We investigate the degree to which people in a shopping mall express other-regarding behavior in the dictator game. Whereas many studies have attempted to increase the social distance between the dictator and experimenter and between the dictator and recipient, we attempt to minimize that social distance between random strangers by video recording the decisions, with the permission of the dictators, to display their image on the Internet. Offers made by dictators are high, relative to other experiments, and a nontrivial number give the entire experimental windfall away. However, a nontrivial number of people keep everything as well.
"Economics of the World Beyond the Classroom: Two Methods for Teaching," (2017), with Steven Gjerstad, Journal of Economics and Finance Education, 16:1, 1-6.
“Information Effects in Multi-Unit Dutch Auctions,” (2016), with Steven Gjerstad and David Porter. Southern Economic Journal 83:1, 126-145.
“An Experiment on Protecting Intellectual Property,” (2014), with Bart Wilson. Experimental Economics, 17:4, 691-716.
Abstract: We conduct a laboratory experiment to explore whether the protection of intellectual property (IP) incentivizes people to create non-rivalrous knowledge goods, foregoing the production of other rivalrous goods. In the contrasting treatment with no IP protection, participants are free to resell and remake non-rivalrous knowledge goods originally created by others. We find that creators reap substantial profits when IP is protected and that rampant pirating is common when there is no IP protection, but IP protection in and of itself is neither necessary nor sufficient for generating wealth from the discovery of knowledge goods. Rather, individual entrepreneurship is the key.
“Underwater Recession,” with Steven Gjerstad and Vernon L. Smith. The American Interest, May/June 2012.
I am pleased to have received the Faculty Research Award from the Brock School of Business in 2023 (pictured below).
