Estimating earnings in an information-poor market: the case of crack cocaine.

Li Way Lee [*]

Department of Economics, FAB 2131, Wayne State University, Detroit, MI 48202 USA

Abstract

From a procedure that uses information on price and transaction behavior at the retail level, the paper estimates that the typical crack cocaine dealer in 1988 earned $35, 000 from dealing. Due to biases in the procedure, the estimate should be regarded as the upper bound on the average earnings. The analysis sheds light on other issues about illicit drugs. By focusing on actual behavior in the market, the analysis leads to a better understanding of the relations between risks, price, and dealer behavior. Also, the analysis shows a way of estimating earnings in other markets where financial information is equally poor if available at all. (c) 1999 Elsevier Science Inc. All rights reserved.

JEL classifications: D2. D8, J3, K4, L8

Keywords: Estimation; Earnings; Dealers; Cocaine; Transaction; Imperfect information

1. Introduction

The press popularizes the perception that street-level cocaine dealers make an extraordinary amount of money fast and spend it fast. This perception no doubt has attracted many inner-city teenagers to drug trafficking and therefore has had enormous social consequences.

But it has been difficult to say if the perception is realistic. In the cocaine market, earnings are not reported to authorities and, to further evade detection, records usually are not kept.

Obviously we could simply find drug dealers and ask them about their earnings. Indeed this is the method used in a remarkable RAND study of drug dealers (Reuter, MacCoun, and Murphy, 1990). The study interviewed 186 drug distributors on probation in the District of Columbia in the second half of 1988. From those interviews, the study concludes that, in 1988, the typical cocaine dealer made $22,000 (ibid., Table 4.10). Although this is a significant income for teenage dealers, it falls far short of the earnings reported in popular press.

The authors of the RAND study point out, however, that their estimate is susceptible to two types of sampling bias (ibid., Appendix C). First, because the dealers interviewed were on probation, the study omitted those who avoided probation. Those were dealers whose offenses were deemed too serious for probation or who managed to evade the criminal justice system. Those dealers probably had above-average acumen and earnings. The omission of those dealers then created a downward bias in the reported earnings. The other bias of the RAND estimate arises from the …