This research develops a structural equilibrium model of consumer search over product characteristics. I study the market for illicit drugs, specifically, cocaine. A consumer searches for packages of cocaine and chooses whether or not to purchase using an optimal stopping rule. She assigns a value to every potential package of cocaine (defined by its price, purity, and weight). The consumer sets her reservation value so as to maximize the value of search conditional on the distribution of package types and the cost of search. The cocaine dealer chooses the purity level, weight and price of his package type in order to maximize profit conditional on the behavior of consumers and other dealers. The cost of dealing cocaine is a function of the probability of arrest and the penalty associated with selling narcotics. Equilibrium is defined by the distribution of packages that is consistent with profit-maximizing decisions of cocaine dealers and the resulting distribution of reservation values for consumers.; I develop a strategy for estimating the structural parameters of the model given an equilibrium solution. Because the theoretical model does not allow for comparative statics. I run several simulations to illustrate how the equilibrium distributions respond to changes in the enforcement variables. I find that the relationship between enforcement variables and equilibrium distributions is non-monotonic and dependent on the initial values of enforcement.; I also develop separate models of dealer and consumer behavior that treat the other agents' behavior as exogenous. These models allow for analytical comparative statics that can be tested using available data. The data on package type is gathered from the Drug Enforcement Agency's System to Retrieve Information from Drug Evidence (STRIDE). The package data is matched to the legal penalties associated with sale and possession of cocaine in the county of observation. Data on legal penalties is collected from the National Judicial Reporting Program (NJRP) and the probabilities of arrest are estimated using data from the Uniform Crime Reports (UCR). The relationships that exist in the data are consistent with the comparative statics resulting from the partial consumer and partial dealer models.
展开▼