A Time-Series Analysis of Crime, Deterrence, and Drug Abuse in New York City

Abstract

Since Gary S. Becker's (1968) groundbreaking work on the economics of crime, economists have expanded upon both the theory and the empirical analysis of crime (e.g., Isaac Ehrlich, 1973; M. K. Block and J. M. Heineke, 1975; Ehrlich, 1975; Ann Dryden Witte, 1980). According to the standard theoretical framework, optimizing individuals engage in criminal activities depending upon the expected payoffs of the criminal activity, the return to legal labormarket activity, tastes, and the costs of criminal activity, such as those associated with apprehension, conviction, and punishment. Excellent reviews of the literature appear in Daniel Nagin (1978), Sharon Long and Witte (1981), Richard Freeman (1983), and Theodore G. Chiricos (1987). While some studies reported evidence that increases in criminal-justice sanctions reduce criminal activity (Ehrlich, 1975; Witte, 1980; Stephen K. Layson, 1985; Jeffrey Grogger, 1991; Steven D. Levitt, 1997), others found either a weak relationship, or none at all between the two (Samuel L. Myers, Jr., 1983; James Peery Cover and Paul D. Thistle, 1988; Christopher Cornwell and William N. Trumbull, 1994). Contradictory results can be explained, at least in part, by the empirical problems inherent in crime research, the most significant being the simultaneity between crime and criminal-justice sanctions.' Thus, after 30 years of empirical research there is no consensus on the impact of police and arrests on criminal activity. The purpose of this study is to provide new, and potentially more refined, evidence on the crime-deterrence relationship using a unique data set, which consists of monthly observations in New York City for nearly 30 years. This is the only data set of its kind, based on highfrequency observations of five different crimes, the corresponding arrests, the size of the police force, and a poverty indicator, spanning decades of experience in one city.2 Consequently, this is the first paper that employs high-frequency time series of individual crime categories to circumvent many problems found in studies that employ cross-sectional or low-frequency (e.g., annual) time-series data sets. We also use recent advances in time-series econometrics to test and correct for problems that may have contaminated the results of previous time-series analyses of crime. We find robust evidence for the deterrent effects of arrests and police on most categories of serious felony offenses. Another unique feature of the study is the addition of drug-use proxies. In the 1980's and into 1990 the media focused much attention on drug abuse, crime control, and the criminaljustice system. It had been claimed that in* Corman: Department of Economics, Rider University 2083 Lawrenceville Road, Lawrenceville, NJ 08648, and National Bureau of Economic Research; Mocan: Department of Economics, University of Colorado-Denver, Campus Box 181, P.O. Box 173364, Denver, CO 80217, and National Bureau of Economic Research. This research is supported by a grant from the National Institute of Drug Abuse to the National Bureau of Economic Research (Grant No. 1-R03-DA06764). An earlier version of this paper was presented at the 1996 American Economic Association Meetings in San Francisco, CA. Ofira Schwartz, Keith Amadio, Joseph Bucs, and Ronald Teodoro helped in data collection. Timothy Potter, Jennifer Giellis, Erdal Tekin, Melissa Anderson, Paul Niemann, and Danny Rees provided assistance in data analysis. John Lott, Jody Overland, and especially Michael Grossman provided very valuable suggestions. We thank two anonymous referees for helpful comments. Any opinions expressed here are those of the authors, and should not be assumed to be those of the granting agency, Rider University, University of ColoradoDenver, or NBER. ' Franklin Fisher and Daniel Nagin's (1978) article describing the problem is a classic in the field. Recent literature suggests several new approaches to the simultaneity problem: using careful empirical analyses of individual rather than aggregate data (Grogger, 1991; Helen Tauchen et al., 1994), and finding better exogenous instruments for identification (Levitt, 1996, 1997). 2 Among the advantages of using just one city is the fact that there is one unit defining and collecting crime and deterrence data, which prevents inconsistencies across observations.