Academic journal article Economia

Homicides and the Age of Criminal Responsibility: A Density Discontinuity Approach

Academic journal article Economia

Homicides and the Age of Criminal Responsibility: A Density Discontinuity Approach

Article excerpt

Crime is a prominent public concern in urban areas worldwide. Violent crime participation and victimization are known to be highest among young males, especially those from disadvantaged neighborhoods. Although the problem is not restricted to developing countries, it takes dramatic contours in Latin America. In the region, more than a hundred cities exceed twenty murders per 100,000 inhabitants annually. Additionally, public safety is listed in opinion polls as the top public concern in almost every Latin American country.1 As part of an effort to reduce youth involvement in violent crime, lawmakers and the general public have considerably debated the age of criminal responsibility, that is, the age above which a young offender is subject to the stricter legal treatment of adults.2

An analysis of criminal behavior and the deterrent effects of punitive policies is part of a long literature that, among economists, has strong roots in Becker's rational model of crime and punishment.3 The general focus has been on understanding the responses of potential criminals to incentives, in particular to severe or more likely punishments. The legislation regarding violent crimes in many countries and U.S. states determines that only people older than a particular age cutoff—the age of majority—are processed by the adult justice system. People younger than this cutoff are subject to the juvenile justice system, usually with more lenient punishments. The Beckerian model would imply that the severity of potential punishments faced by people after they reach the age cutoff would reduce their propensity for involvement in violent criminal acts.4

We provide novel empirical evidence of the crime-deterrent effects of stronger punishments using data from a country in which sixteen per thousand people under the age of nineteen were murdered in 2013 alone: namely, Brazil.5 Our approach complements previous studies by using a new proxy variable for criminal involvement in a natural quasi-experiment design.

The common difficulty for every empirical study of criminal behavior is that individual propensities to participate in crimes are, by nature, not directly observable. Thus, we need to resort to proxy variables. The most natural and commonly used proxies lie in criminal records, such as arrest and conviction data. Our main concern is that too many obstacles lie between a criminal offense and an official record identifying the offender.6 Importantly, the severity of most of these obstacles depends on enforcement efforts and the discretion of authorities, both of which may vary with the offender's characteristics or the type of crime, leading to potential biases.7

For policy evaluation purposes, the main threat is that, conditional on a crime committed, the propensity of the criminal justice system to generate a formal arrest or conviction record is likely to depend on the procedures and potential sanctions the law itself prescribes. For example, in responding to a possible offense, a public agent might issue only an informal warning if the offender is a minor, subject to an unlikely conviction; whereas if the offender were an adult, the agent might instead choose to arrest him. In such cases, the likelihood of a crime committed by a minor showing up in arrest records is lower than if the same crime is committed by an adult. Therefore, proxies based on criminal records have a potential underreporting bias of unknown magnitude that would compromise any age-based estimates. We first provide new evidence of this bias and its magnitude by studying the comprehensive police arrest records from Rio de Janeiro, the third largest Brazilian state.

To circumvent this differential underreporting bias of criminal records, we propose a novel proxy for an individual's involvement in violent crime: the change in the number of violent deaths by age.8 We base our proxy on the well-documented fact from the criminology literature that criminal offending and violent victimization partially overlap. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.