Academic journal article
By Model, Karyn E.
Journal of the American Statistical Association , Vol. 88, No. 423
Drug control policy in the United States is generally formulated under the assumption that such policies are effective in influencing drug use. This is accomplished through traditional economic channels; the fact that the possession or sale of an illegal substance carries the risk of arrest and conviction serves to increase the cost of these goods. As penalties increase, the cost of the substance may rise as sellers require greater compensation for dealing in illegal goods, while buyers must also incorporate possible penalties into their full costs of use. Thus U.S. policy proceeds on the assumption that the demand for drugs will vary inversely with the severity of penalties, because these penalties serve to increase costs.
The decriminalization of marijuana by 12 states between 1973 and 1978 is one of the few instances where penalties for an illegal substance were actually reduced. Decriminalization maintained marijuana's illegality but designated first-offense possession as a civil (not criminal) offense. Prison sentences were abolished and replaced with small fines (typically $100) for first offense possession of (usually) less than 1 ounce of marijuana. The reduced penalties also signalled a decreased priority for marijuana possession arrests and thus served to divert police attention away from these offenses. States that decriminalized marijuana experienced a dramatic reduction in possession arrests (National Governor's Conference 1977). Marijuana use thus became "cheaper" as a result of both the decline in probability of apprehension as well as the reduction in possible penalties imposed.
Empirical research concerning the effect of decriminalization on drug use is scarce. Analysis of data from Monitoring the Future, an annual survey of the values and life-styles of high school seniors, found that seniors in decriminalized states reported using no more marijuana than did their counterpart in control states (Johnston, O'Malley, and Bachman 1981). This study did not, however, control for demographic characteristics of subjects and did not examine the use of substances other than marijuana. Also, high school youth represent a sample that may not respond to new legislation in the same manner as the general population.
Standard economic analysis would predict that legal changes impact the use of marijuana by changing its full use price. Also, one must consider the possibility of related effects on the use of substitute or complement goods. Changes in marijuana policy may affect not only marijuana use, but also the use of other intoxicating substances.
DiNardo and Lemieux (1992) also used Monitoring the Future data to analyze the impact of increased minimum drinking ages (which can similarly be interpreted as imposing increased costs on seniors' alcohol consumption on both alcohol and marijuana use. They found that higher drinking ages are associated with reduced alcohol consumption, but increased marijuana use. This result lends support to the existence of substitution in consumption between marijuana and alcohol.
This article analyzes the effects of marijuana decriminalization on drug mentions in hospital emergency room (ER) episodes between 1975 and 1978. The outcome studied here is an ER drug mention. A drug mention occurs when medical staff at a reporting ER detect drug use in an ER patient (though substance abuse need not be the cause of, or even related to, the ER visit). Section 1 describes the medical and demographic data assembled for use in this study. Section 2 outlines a simple statistical model used to identify the effect of marijuana decriminalization on hospital emergency room drug episodes. Section 3 presents the results of estimation of this model. All regression specifications confirm that decriminalized cities experienced a statistically significant increase in marijuana mentions as well as a significant reduction in the mention of other drugs relative to nondecriminalized standard metropolitan statistical areas (SMSA's). …