CRIME'S FALL

Crime has fallen dramatically in the 1990s, and criminologists have trouble
explaining why.

Some clever economists may have an answer.

The star of these economists is Steven Levitt, who was at Harvard University
and is now a professor at the University of Chicago. In a series of
ingenious studies, Levitt has shown that criminals -- like almost everyone
else -- respond to costs and benefits.

That might seem obvious. But most criminologists don't find it so, and
proving them wrong is harder than you might think.

Assume, for instance, that the states that lock up criminals for long terms
suffer less crime than states that let criminals out quickly.

This might seem like good evidence that long prison terms deter crime. But
it's not, because states with long sentences might have low crime even if
long sentences don't deter criminals at all.

One possibility is that though long sentences do lower crime, they do it
only by keeping convicted criminals off the street.

Another possibility is that long sentences don't lower crime, but that long
sentences and low crime appear together because they're both caused by some
third factor, say, traditional values.

To prove that long sentences lower crime by deterring criminals, then, is
not a simple matter. Yet Levitt, in a series of statistical studies, has
done it.

Levitt hasn't stopped there. A new study, written with John Donohue of
Stanford Law School, promises to change the way we think about the causes of
crime.

The title of the study is ''Legalized Abortion and Crime.'' In it, Levitt
and Donohue argue that the legalization of abortion in the early 1970s helps
to explain the sharp drop in crime during the 1990s.

How's that?

According to Levitt and Donohue, legal abortion has reduced crime in three
ways. First, it has reduced the number of people out on the streets. With
fewer people being born - especially fewer young men - there's bound to be
less crime.

But this effect is fairly weak, say the authors, because many women who have
abortions don't end up having fewer children; they just have them at
different times.

A stronger effect of abortion on crime comes from what the authors call the
''selective abortion effect.''

The women who have abortions are not a random group. Many are teen-age,
unmarried and uneducated. Why does this matter? It is a sad fact of modern
life that many of these women's children will be at greater risk for getting
involved in crime.

A third reason that abortion lowers the crime rate is what Levitt and
Donohue call the ''improved environment effect.''

A host of factors - an uneducated mother, a teen-age mother, a single-
parent family, a mother who doesn't want her child - increase the odds that
a child will grow up to be a criminal.

If abortion prevents some of these women from having babies until they're
more prepared to be mothers, then abortion will lower the crime rate.

Taking all these factors together, say Levitt and Donohue, it appears that
the legalization of abortion has had a major effect on the U.S. crime rate.

They estimate that abortion accounts for half the drop in crime since 1991.

That's not all. Crime may well continue to drop, the authors say, as the
full effect of the growing abortion rate in the 1970s sinks in.

Is that an argument for abortion? Not at all. Abortion is a moral question,
beyond statistics. But some things in society - even things that people
don't like - can have surprising effects.