78
Looking at the shares of straight reelection of mayors, the net numbers show
about a 40% rate, as Table 6.6 shows. In order to avoid the mentioned different-N bias, I
checked the rates over different samples, calculating them for districts with information
in more than three, four or five periods in the database. In all of the cases, variation did
not exceed 2%. So, overall, effective reelection rates at the municipal level are about 40%.
Considering that information about candidacies is still imprecise, but assuming that no
single incumbent will automatically win; it can be thought that the rate of pursuit of
reelection is higher than that 40%. In order to verify whether there is any first-sight
variation in the reelection rates by municipality, a common intuition is to try to
distinguish districts over the basis of their relevance. Many proxies could be useful to
make that differentiation. Population is one of those. However, given the variations in
inhabitants across provinces, and also the dissimilar municipal structures, population
might over represent the effects of districts of bigger provinces. As alternatives,
economic indicators such as budgets, public employment or financial autonomy vis-à-
vis the province might be good measures of relevance, but they are not available. Thus,
the best indicator I came up with is the percentage of provincial population living in a
given municipality. I am aware that this proxy might be imprecise (municipalities of
provinces with 10 units are likely to concentrate a higher share than those of provinces
with 400 ιmits), but it has been the most reliable measure I could find so far. As a proxy,
I created a measure that could be called the effective number of municipalities (enm). The
mathematical formula is the same as that used by Laakso and Taagepera (1979) and