Marijuana on Main Street? Estimating Demand in Markets with Limited Access

Abstract

Marijuana is the most common illicit drug with vocal advocates for legalization. Among other things, legalization would increase access and remove the stigma of illegality. Our model disentangles the role of access from preferences and shows that selection into access is not random. We find that traditional demand estimates are biased resulting in incorrect policy conclusions. If marijuana were legalized, those under 30 would see modest increases in use of 28 percent, while on average use would increase by 48 percent (to 19.4 percent). Tax policies are effective at curbing use, where Australia could raise AU$1 billion (and the United States US$12 billion). (JEL D12, H25, K14, K42)

Choice in the presence of experts: the role of general practitioners in patients' hospital choice

“…This is similar to the well-known issue of incidental truncation (Heckman (1976)) whereby decision outcomes of interest are only observed for a selected subsample and where failure to properly model the sample selection mechanism induces the estimates of the outcome relationship to be biased and inconsistent. This has also been noted by Eizenberg (2014) and Jacobi and Sovinsky (2016). Similar issues also arise in the analysis of endogenous sample attrition (Hausman and Wise (1979)).…”

A Structural Model of the Retail Market for Illicit Drugs

“…Hence, it may not be well suited to understanding the implications of different policies, such as penalties on buyers and sellers. In this strand of the literature, the closest papers include Reuter and Caulkins (2004), which discusses information issues in drug markets; Jacobi and Sovinsky (2016), which studies buyers' limited information about drug accessibility; and Adda, McConnell, and Rasul (2014), which analyzes the effects of depenalization on the cannabis market (but does not consider the role of information frictions). We build on Galenianos, Pacula, and Persico (2012), but we differ in two ways: our primary contribution is empirical, and we include some novel features in our model (recounted in Section III).…”

Forensic Econometrics: Demand Estimation When Data are Missing

“…We first discuss the data and then the framework. More details on the method are ongoing in Jacobi and Sovinsky (2018).…”

“…The data we used in Jacobi and Sovinsky (2016) are from two sources. The first are individual-level cross-section data from the Australian National Drug Strategy Household Survey (NDSHS).…”