Desarrollo y s∣
Seftiembre de 1999
'Λ∣A
is imperfectly represented by the estimated model, the preceding figure
becomes still larger when the continuity of the urbanization process is taken
into account. For instance, in a Countrywhere the urbanization rate would
be increasing by 0.5 percentage point per year, a reasonable order of
magnitude in view of the experience of many developing countries over the
last two or three decades, then, other things being equal, the nationwide
robbery rate would increase by approximately 60 per cent in 20 years. If
nothing else were changing, this figure would essentially reflect urban-rural
differences in crime rates. However, other variables affecting crime are likely
to change simultaneously with the urbanization process. In particular, the
estimates reported in Ihble 1 suggest that economic growth tends to offset
the adverse effects of urbanization. In other words, it is only if urbanization
proceeded without sufficiently rapid economic growth that crime would
develop as suggested by the preceding figure. This is in agreement with the
theoretical argument recalled above of the push and pull factors of
urbanization. Overall, FLL results thus describe a rather complex combination
of forces which together contribute to possible changes in crime rates or,
alternatively, permit maintaining it steady in the course of development.
Another interesting feature of the preceding equations is the substantial
histeresis they suggest in the evolution of criminality. Here again, and as
could be expected, it is more pronounced for robberies than for homicides.
In the former case, a simple calculation made on the basis of the coefficients
shown in Table 1 suggests that a major recessions leading to a sudden and
once for all 5 per cent drop in CiDP would produce an instantaneous 50 per
cent increase in the robbery rate. However, the hysteresis effect is such
that the crime rate would still be 10 per cent above its pre-recession level
7 years after the economy resumed normal growth36. Again, these orders
of magnitude are only indicative. But such recessions arc not uncommon
in developing countries and the preceding figures show that the lasting
increase in crime caused by a temporary surge in poverty may add very
much to their social cost.
It might be thought that because they are based on a restricted number
of observations and countries, the preceding results are not truly
representative. It turns out that fixed effects arc less of a problem in the
36 Because of the symmetry built in the model this effect would disappear in ease the initial loss were
fully compensated by faster growth in the following years
89