Title not available in English.

Abstract

Tabulated Census information on the situation of previous place of residence and of length of residence in the present state or municipality is not often used to measure migration. The problem is that only the last movement of the individual is indicated, ignoring possible multiple flows and not struturally considering return flows. Our proposal is to estimate rates of annual transitions, based on the hypothesis that the decision to migrate in a given year depends on the residence condition in that year only and not on some other point in the past. That is lo say that the migratory process is a first order Markovian process order. These estimates allow a better approximation of actual migration, as opposed to measures of net migration obtained through indirect methods.