The Price of Cocaine, Heroin and Marijuana, 1981–1993

• Author: Raymond Hyatt,Paul Scheiman,William Rhodes
• Date: 01 July 1994
• DOI: 10.1177/002204269402400302
• Published date: 01 July 1994
• Subject Matter: Article

The Journal of Drug Issues 24(3), 383-402 1994

THE PRICE OF COCAINE, HEROIN AND MARIJUANA,

1981-1993

William

Rhodes

Raymond

Hyatt

Paul

Scheiman

The analysis reported in this article provides a consistent price series for

cocaine. heroin, and marijuana based on over twelve years

of

data from the

Drug Enforcement Administration's System to Retrieve Information from

Drug Evidence (STRIDE). Cocaine and heroin prices fell sharply through

most

of

the 1980s, but seem to have stabilized in the early I 990s. In

contrast, marijuana prices rose markedly through most

of

the 1980s, but

seem to havefallen in 1992 and 1993.

Prices

of illicit drugs vary widely over time, from place to place, and even

from transaction to transaction. This variation is important to monitor because

trends in drug prices are an indicator of the success of national policies to reduce

the availability of illicit substances. Furthermore, the relationship between illicit

drug prices and drug use is an important research topic, whose study is hindered

by the lack of suitable price series for cocaine, heroin and marijuana.

For

example, we have used price series similar to those described in this

article to show that the consumption of cocaine by heavy users is sensitive to its

price (Hyatt and Rhodes n.d.) and that recent decreases in the price of heroin have

been associated with an increased rate of insufflation of heroin (Hunt and Rhodes

1993). Others have used their own price series to analyze the relationship

between heroin prices and crime (Brown and Silverman 1974), cocaine prices and

consumption decisions by high school students (DiNardo 1992), and price mark-

ups across illicit drug distribution networks (Caulkins and Padman 1993).

This essay discusses some reasons why trends in illicit drug prices are difficult

to measure. It advances methods to overcome these difficulties and provides

national and representative local estimates of trends in prices of cocaine, heroin,

and marijuana from 1981 through the first quarter of 1993.

WilliamRhodes, Ph.D., is an expert in

quantitative

analysis

and

computer

applications.

Trained

in

econometrics,

he has

taught

graduate-level

applied

statistics

at two

universities

andhas beensenior

scientist

at Abt

Associates

since

1988.

He is a past

research

directorat the U.S.

Sentencing

Commission,

senior

economist

at the

Institute

for Law

and

Social

Research.

and

assistant

professor

at

Florida

State

Universiry.

RaymondR. Hyall, Jr. is a senior

analyst

at Abt

Associates.

He has eight years

experience

teaching

both

mathematics

and

computer

science

at the

college

level.has served as chair of a

computer

science

department,

andwas

consulting

senior

research

methodologist

in the

Centerfor

Applied

Social

Research

at

Northeastern

University.

Paul M. Scheiman has

worked

on Abt's

contract

with the Office of

National

Drug

Control

Policy

(ONDCP)

since

1990.

He

analyzed

the data for a

forthcoming

National

Institute

of Justice report. Send reprint

requests

to

William

Rhodes,

Ph.D., ABT

Associates

lnc., 55

Wheeler

Street,

Cambridge,

MA

02138.

©Journal of Drug Issues, Inc. 002-0426/94/03/383-402 $1.00

383

RHODES, HYATT, SCHEIMAN

Problems in Determining the Market Price for Illicit Drugs

The principal difficulties when estimating the market prices of cocaine, heroin,

and marijuana are similar to those when estimating prices in any illegal market.

Transactions are clandestine; there are few advertised prices, and market surveys

sponsored by groups representing the interest of dealers or buyers are rare.

Information about prices comes from researchers and police who observe drug

transactions and sometimes make drug buys. The data in this analysis result from

drug purchases reported in the Drug Enforcement Administration's System to

Retrieve Information from Drug Evidence (STRIDE).

STRIDE reports the prices paid by Drug Enforcement Administration (DEA)

agents (as well as other federal agents and some state agents) for cocaine, heroin,

and marijuana. These prices vary depending on the amount of drug purchased;

obviously, larger transactions have higher costs. Illicit drugs also vary in quality.

Although quality has many dimensions, the most important for buyers and sellers

is the drug's purity - the amount of pure cocaine or heroin contained in a

purchase exclusive of diluents. In the case of marijuana, it is the amount of the

chemical THC. Therefore, to afford meaningful comparisons when bulk amounts

and purities vary widely, calculations are based on price

per

pure gram for

cocaine and heroin, and price per bulk gram for marijuana. 1

However, the price per pure gram also varies with distribution level. Retail

buyers pay more per pure unit of drug than retailers pay to wholesalers, and

retailers pay more per unit of drug than wholesalers pay to importers. To monitor

prices over time and across cities, prices must be compared at equivalent

distribution levels. Because they are not readily identified in the STRIDE data,

distribution levels have been inferred from the amount of drug exchanged - pure

amount for cocaine and heroin and bulk amount for marijuana.

This method of identifying distribution levels is imperfect, because the multiple

distributions levels have no clear demarkation. Nevertheless, transactions greater

than some specified threshold for each drug type are unlikely to be retail

purchases. For example, it is unlikely that a transaction of five ounces or more of

pure cocaine occurs at the retail level, while it is likely that a high percentage of

cocaine transactions involving less thanone pure gram are retail-level transactions.

Average prices based on data truncated at a selected threshold are imperfect

reflections of prices paid at retail, however. Any threshold that affords a

sufficient number of observations to support the computation of reliable estimates

will necessarily mix transactions from different distribution levels. That is, from

period to period, the sample of transactions will contain varying percentages of

sales at the retail and wholesale levels, and apparent price trends will be sensitive

to changes in this overlap. In one year more purchases may have been in the

upper range of transaction sizes, where price per pure gram tends to be relatively

low. In another year, purchases may have been at the lower end of the range,

where price per pure gram tends to be relatively high. Thus, observed changes in

average prices from one year to the next may arise from differences in transaction

levels represented in the data rather than from changes in prices.

A second approach to taking the distribution level into account is to use

regression analysis to control for the distribution level when estimating prices.

Later, we describe a regression-based approach that we used to estimate a

standardized price for cocaine, heroin, and marijuana. The standardized price is

the price per pure gram (bulk for marijuana) assuming a fixed transaction size.

The reported standardized price of cocaine is the estimated average price per pure

384 JOURNAL OF DRUG ISSUES