The Importance of the Age Structure in Explaining Homicide Rates in the United States

• In Event: Reducing Homicide and Gun Violence: Evidence on Schools, Jobs, and Community Strategies

Abstract

Understanding why homicide rates rise and fall is central to crime policy, yet the role of population age structure in shaping aggregate homicide trends remains unsettled. While the relationship between criminal tendencies and age at the individual level is one of the most well-established and robust patterns in the literature on crime, age-crime curves based on individual-level data do not necessarily translate into a relationship between aggregate age composition and overall crime. Changes in age structure may affect crime not only through age-specific criminal propensity, but also through shifts in crime opportunities, political demand for law enforcement, institutional capacity, and economic competition in the formal labor market.

Empirical work linking aggregate age structure to crime has yielded mixed results, in part because demographic changes may be endogenous to crime. In particular, migration patterns can respond to local crime conditions and systematically vary by age, confounding estimates that rely on observed changes in population composition. This paper addresses this challenge by isolating plausibly exogenous variation in county-level age structure. We implement an instrumental variables strategy—novel in this literature—that exploits the predicted within-county evolution of cohort sizes based on historical age structure to identify the causal effect of age composition on homicide rates.

Using U.S. county-level data from 1983 to 2023, we find that increases in the share of adults aged 25–34 have a disproportionately large impact on homicide rates. A one percentage point increase in the 25–34 population share increases homicides by approximately 0.80 per 100,000 residents, an effect that is both statistically significant and economically meaningful. The 15–24 age group is also positively associated with homicide, though the magnitude is notably smaller. In contrast, shifts toward older age groups (35–44, 45–54, 55–64, and 65+) produce smaller and less consistent effects.Our results show that variation in age structure alone strongly predicts long-run homicide trends over the past four decades, highlighting its relevance for understanding broad crime dynamics. At the same time, age composition is considerably less predictive of whether homicides involve firearms, suggesting potential substitution across homicide types that is not well-explained by changes in the age structure.

Taken together, these results underscore the importance of demographic structure—particularly the size of the young adult population—in shaping long-run homicide rates, whereas decisions about weapon choice may be more influenced by firearm availability, gun laws, and criminal justice policies.

Author

• Rosanna Smart, RAND Corporation
  Presenting Author