Temperature and arrests in Prohibition-era 

South Carolina

Abstract: The positive correlation between high temperatures and crime is well established. I study how random, within-month daily high temperature and precipitation affects arrests rather than incidents. I analyze arrest records of a Prohibition Era southern city with mean summer daily high temperatures of 88ºF without modern air conditioning or weather forecasting so outdoors temperatures are salient. Nonparametric estimates from negative binomial regressions with standard errors corrected for heteroskedasticity and serial correlation reveal that all-offense arrests are 21% lower at 35º than at 70ºF. Arrests do not increase at higher at temperatures above 70º. The same pattern of fewer arrests at low temperatures emerges for violent, public order, and low-level fraud arrests, as well as for men, Whites, and Blacks.

Supplemental Online Appendix

Temperature and arrests

October 2026 (revision)

Weather to convict?

Abstract: Heat is a psychological and physiological stressor that affects high-stakes decision making. Recent studies that explore the effect of heat on modern judicial decision making generate mixed results. I consider judicial decisions in a southern US municipal court in the Prohibition Era before the introduction of air conditioning. Using parametric and non-parametric specifications, I find that neither daily high temperatures nor rainfall exert any meaningful effect on conviction probabilities. A 15ºF increase in the daily high temperature, however, leads to 7.0% lower fines and 3.0% shorter incarceration sentences. Because sentences were served on southern chain gangs in sometimes-oppressive heat and humidity, shorter sentences are consistent with utilitarian and retributive theories of punishment. A day of hard labor in oppressive heat is harsher than a day of labor in temperate weather, so shorter sentences in hot weather provide comparable deterrence and are retributively proportionate.

Misdemeanor guilty pleas in the light of trial

Abstract: The shadow-of-trial hypothesis posits that, relative to defendants who stand trial, defendants who plead guilty will receive a sentencing discount proportional to the probability of conviction. Prosecutors offer discounts in relatively routine cases to conserve the courts’ trial resources that can be used more productively in less routine, weak, or high-profile cases. Critics of plea bargaining highlight several factors that may lead from divergences from the efficient outcome, which include the relatively short shadows created by plea-dominated adjudication, prosecutorial misfeasance, and poor behavioral heuristics by defendants. This paper studies 4,000 misdemeanor cases in an early twentieth-century, trial-dominant court, conducts several tests of the shadow hypothesis, including an instrumental variables test, and finds that defendants who plead guilty pay higher fines, are more likely to be incarcerated, and serve longer periods in incarceration than defendants who stand trial. The principal reasons for this counterintuitive outcome are: (1) defendants appear without representation and without having negotiated a plea sentence prior to trial, so they make hasty decisions in the bright light of immediately preceding trials; and (2) although pleading defendants throw themselves on the mercy of the court, the court shows little mercy for defendants who admit guilt and do not offer mitigating testimony on their own behalf.

Working paper (January 2025).

Cash bail and trial outcomes in an early twentieth-century southern police court

Abstract: Studies of modern misdemeanor adjudication find that courts set bail higher than is required to reasonably assure that nonviolent defendants who pose no immediate threat to the community will appear for trial. Some defendants languish in jail for extended periods during which time they lose income, employment, and the ability to provide an effective defense for themselves. Some plead guilty just to be released on time served, but conviction may lead to the loss of certain government benefits. This paper considers the downstream consequences of bail setting in an urban, southern police court in the 1910s. Using an instrumental variables strategy, I find that defendants who were unwilling or unable to post cash bail were not more likely to be convicted or to be incarcerated than defendants who posted bail. Conditional on conviction, however, defendants who had posted bail and returned for their hearings served about 14 less days on the city’s chain gang than defendants that did not post bail. Fourteen days is 1.4 times a standard deviation in days served, so it is an economically meaningful effect.  Further, I posit that the ability to post bail was correlated with unobserved income or wealth and provide some evidence that defendants who did not post bail and served on the chain gang faced a binding cash-in-advance constraint because there were no bail bondsmen from whom they could borrow.

NBER working paper 32887 (August 2024).

Bad men, good roads, Jim Crow, and the economics of southern chain gangs

Abstract: Penology in the Jim Crow South centered on the chain gang. Gangs ostensibly served three purposes: their severity served as a deterrent; their putting convicts to work on roads and other public improvements reduced the taxpayers’ costs of infrastructure; and their discriminatory implementation reinforced the social order defined by Jim Crow. Drawing on insights from the economics of crime literature, this paper analyzes whether chain gangs reduced road maintenance costs. Using a fixed-effects design, the analysis finds that the costs of using gangs in road maintenance were marginally lower on average than using wage labor. The results are consistent with county officials choosing between convict and free labor in manner consistent with minimizing taxpayers’ costs.

Essays in Economic & Business History 41(2), 149-170 (November 2023)

Bad men, good roads, Jim Crow, and the economics of southern chain gangs

Data set