Deslxrrollo y s
Septiembke ы iw»
at a much higher level, which favors the hypothesis that more inequality
leads to more crime. But this may also be due to a change in definitions.
The CK and the US are practically the only developed countries where what
may be called a drastic change in inequality took place over the last 20
years or so12. It is interesting that in both countries there is evidence of a
concomitant increase in criminality. Inequality also increased significantly
in Sweden and in the Netherlands since the mid 1980s35. Crime rates
apparently did not change much. However, it is interesting that inequality
in both countries increased more because of changes at the top rather than
at the bottom of the distribution. According to the theoretical arguments
above, it is the latter which is supposed to matter for the evolution of crime.
Crime and Inequality: Cross-sectional Evidence
As it is impossible to find other countries with reasonably good time series
both on erime and on inequality and having experienced substantial
changes in inequality further evidence on the relationship between crime
and inequality can only be found in cross-sectional studies. Probably the
most comprehensive Studyof this type is that of Fajnzy lber, Lederman and
Loayza (fll) (already'referenced). It is based on the UN data complemented
for a few countries by homicide rate series obtained from cause of death
statistics compiled by the World Health Organization (who). As discussed
above, this data base is very imperfect but it is unfortunately the only one
available on a sufficiently large scale. An interesting feature, though, is
that it is both cross-sectional and longitudinal. Tb some extent, this permits
minimizing the effect of cross-sectional measurement errors which are
likely to be the most serious source of bias. Inequality data are taken from
Deininger and Squire (1996). They are not themselves without problem.
In particular, they are not available for all countries and all points of time,
which still reduces the data sample used by FLL.
Standard cross-sectional analysis on mean robbery and homicide rates for
the period 1970-1994 are based on samples of 50 to 60 countries depending
'3 IncquaJiw has also increased significantly in Sweden and in the Netherlands since the mid∙l 980s.
Crime rates apparently did not change much. However, it is interesting that in both countries
inequality increased more because of distributional changes at the top rather than at the bottom
of the distribution.
ʒʒ See Gottschalk and Smeeding (1998).
85