Economists should be forward rather than backward-looking, so I will consider Steve’s new paper—"Measuring the Impact of Crack Cocaine," co-authored with Roland Fryer, Paul Heaton, and Kevin M. Murphy.
The contributions of this paper are clear:
1. Crack cocaine enters the U.S. market in a big way in the mid-1980s.
2. It is possible to construct a meaningful time series for crack usage. This is the paper’s most significant contribution.
3. Murder rates rise dramatically when crack comes. If I recall Steve’s talk correctly, some young black homicide rates nearly tripled, of course starting from an already-high rate. This is a startling statistic.
4. Today crack continues to be consumed at high levels but mostly by longer-term addicts. This kind of crack consumption causes much less crime. Such a causal story, however, is closer to speculation than a proven result; it cannot be read off the numbers.
A central question is why crack has caused so many more murders than have other drugs. I can think of a few possible (and possibly false) arguments:
1. Heroin and pot make you sleepy. Crack gets you riled up.
2. Crack was a new drug when it hit the market. Gangs were competing to hook new buyers. This is a far more violent activity than serving established drug clientele.
3. In dollar terms crack was a "bigger" drug than ever before. The gross and the profit margins were bigger. The resulting turf wars over profits led to murders. It is not worth killing people over a few marijuana sales. (Yet still I find this puzzling. Falling prices have taken profits out of the market; the gangs must either have had an extraordinarily high discount rate or they behaved irrationally in killing each other. In the latter case we have no economic explanation at all for the hike in crime.)
4. Perhaps you buy other drugs from your friends, but you buy crack from dealers. (Most people get Ecstasy from their friends, and this market is not very violent.) The new question is then why this might be. Could crack somehow require less personal certification from trusted acquaintances?
These market features might have implications for drug policy. Most of the discussion focuses on either how to stamp out drugs or why we should legalize drugs. We should consider a new approach based on differential features of drug markets. Let us find out why crack differed from other drugs so much. Then watch for when these critical features resurface. If "another crack" comes along, what should we do to ameliorate the crime problem? Encourage people to buy it from their friends? Encourage exclusive territories for the new drug?
Steve’s paper has given us considerable food for thought. But I would like to see him take the next step of pinning down the relevant policy questions. Why has the crack market become safer rather than more dangerous? Which features of drugs lead to associated crime waves? Can we make drug markets—yes "black" ones—less dangerous?