Search

We present an approach for automatically generating and testing, in silico, social scientific hypotheses. This automation is made possible by recent advances in large language models (LLM), but the key feature of the approach is the use of structural causal models. Structural causal models provide a language to state hypotheses, a blueprint for constructing LLM-based agents, an experimental design, and a plan for data analysis. The fitted structural causal model becomes an object available for prediction or the planning of follow-on experiments. We demonstrate the approach with several scenarios: a negotiation, a bail hearing, a job interview, and an auction. In each case, causal relationships are both proposed and tested by the system, finding evidence for some and not others. We provide evidence that the insights from these simulations of social interactions are not available to the LLM purely through direct elicitation. When given its proposed structural causal model for each scenario, the LLM is good at predicting the signs of estimated effects, but it cannot reliably predict the magnitudes of those estimates. In the auction experiment, the in silico simulation results closely match the predictions of auction theory, but elicited predictions of the clearing prices from the LLM are inaccurate. However, the LLM's predictions are dramatically improved if the model can condition on the fitted structural causal model. In short, the LLM knows more than it can (immediately) tell.

There is a strong association between the quality of the writing in a resume for new labor market entrants and whether those entrants are ultimately hired. We show that this relationship is, at least partially, causal: a field experiment in an online labor market was conducted with nearly half a million jobseekers in which a treated group received algorithmic writing assistance. Treated jobseekers experienced an 8% increase in the probability of getting hired. Contrary to concerns that the assistance is taking away a valuable signal, we find no evidence that employers were less satisfied. We present a model in which better writing is not a signal of ability but helps employers ascertain ability, which rationalizes our findings.

Newly-developed large language models (LLM) -- because of how they are trained and designed -- are implicit computational models of humans -- a homo silicus. These models can be used the same way economists use homo economicus: they can be given endowments, information, preferences, and so on and then their behavior can be explored in scenarios via simulation. I demonstrate this approach using OpenAI's GPT3 with experiments derived from Charness and Rabin (2002), Kahneman, Knetsch and Thaler (1986) and Samuelson and Zeckhauser (1988). The findings are qualitatively similar to the original results, but it is also trivially easy to try variations that offer fresh insights. Departing from the traditional laboratory paradigm, I also create a hiring scenario where an employer faces applicants that differ in experience and wage ask and then analyze how a minimum wage affects realized wages and the extent of labor-labor substitution.

The sharp devaluation of the ruble in 2014 increased the real returns to Russians from working in a global online labor marketplace, as con- tracts in this market are dollar-denominated. Russians clearly noticed the opportunity, with Russian hours-worked increasing substantially, primarily on the extensive margin -- incumbent Russians already active were fairly inelastic. Contrary to the predictions of bargaining models, there was little to no pass-through of the ruble price changes in to wages. There was also no evidence of a demand-side response, with buyers not posting more "Russian friendly" jobs, suggesting limited cross-side externalities. The key findings -- a high extensive margin elasticity but low intensive margin elasticity; little pass-through into wages; and little evidence of a cross-side externality -- have implications for market designers with respect to pricing and supply acquisition.

We model social media as collections of users producing and consuming content. Users value consuming content, but doing so uses up their scarce attention, and hence they prefer content produced by more able users. Users also value receiving attention, creating the incentive to attract an audience by producing valuable content, but also through attention bartering -- users agree to become each others' audience. Attention bartering can profoundly affect the patterns of production and consumption on social media, explains key features of social media behavior and platform decision-making, and yields sharp predictions that are consistent with data we collect from EconTwitter., Comment: 36 pages, 9 figures