This paper presents the Proxy Means Test as an alternative method of selection of the target population to social policies in Brazil. The application of this method requires the employment of statistical methods on primary data collected by sampling household surveys. Based on PNAD 2003 we built a logistic regression model relating the probability of a household to be poor with its physical characteristics and other socioeconomic variables concerning the residents. The probabilities estimated by the model are the propensity scores to poverty and can be used as a norm to include the household in social programs. The estimates of the propensity scores and a measure of efficiency in the target population has been calculated with data from PNAD 2003.
target population and public policy; proxy means test; target population indicator