An ongoing challenge for demographers is to know the number of international migrants of a country. While the lack of records and the presence of undocumented migrants hinders the numbers of people who come to a country – the in-migrants – at least they are present in the territory, offering possibilities to account for them. On the other hand, provide estimates for those leaving a country – the out-migrants –  is a much more complicated process.

It is argued here the possibilities of use of population censuses to calculate out-migration based on questions about former residents of the country living abroad. It is believed that the population censuses continue to be a major source of information on international migration, especially for the territorial scope of this type of survey. Since migration is usually a rare event, with a high spatial concentration, it is difficult to produce general information based on surveys that not cover the whole territory of a country.

Based on the census, the number of out-migrants can be measured directly or indirectly. The «direct» information are those drawn from the answers given by respondents to the census questions. «Indirect» data are derived from answers to one or more questions of from other data sources and, usually, undergoing demographic modeling to produce estimates. 

One strategy has been used in several Latin American countries to generate «direct» information about out-migrants. This is based on questions about ex-residents living abroad. It is normally used with two approaches: the first one, proposed by Somoza in 1977, draws upon get information with the mothers about the place of residence of their children; the second approach, proposed by Hill in 1979, investigates the place of residence of the brothers of interviewed people. Recently, Brazil was attempted in a strategy that does not require that there is a family relationship between respondent and emigrant, but only an ex-member of the household living abroad.

The method that investigates the total surviving children residing abroad (SOMOZA, 1977) can be considered as an extension of the requirements used to calculate the fecundity and mortality by Brass Methods. It requires information about the number of surviving children had by women with fertile ages. The investigation of the migration is done by adding one question in that topic, asking whether the place of residence of the surviving children is located in the country or abroad. Thus, from the perspective of information gathering, the attractiveness of this method lies in the small change required in the questionnaires that already have the questions of fertility and infant mortality. In the method proposed by Hill (1979), out-migrants is obtained based on information of the brothers living abroad. This approach requires that new questions are placed at the census questionnaires exclusively to investigate the topic. This may be the reason for this little used in recent decades. At the Brazilian Census of 2010 were questions about people who had lived with the respondent that was residing abroad, regardless of the relationship that existed between them. In terms of methodology, the questions used in the Brazilian Census differ from approaches of Somoza and Hill exactly by this point: it did not require that there was a family relationship between the informant and the emigrant.

The biggest limitation of these techniques probably stems from the assumption of independence between migration and family structure. We know that migration is strongly correlated to the family structure of individuals, as those with few family ties have different propensity to migrate in comparison whit those living in large family households.

Thus, total migration estimation based on information from relatives lose accuracy, mainly when family structures change in time. Analyses carried out in recent years show that any of these is able to measure accurately the number of international migrants in a country.

However, they make it possible to know very important aspects of international migration, as the main places of destination, the regions of origin of the migrants in the country, provide information about their families, who remained in the origin country, among other information that can be inserted in the census.

Then, this topic contributes to discussion about the census as a data source to study the migration process beyond the numbers or if this instrument is targeted only to estimated the stock of migrants of a country.