Utilizando o relacionamento de bases de dados para avaliação de políticas públicas: uma aplicação para o Programa Bolsa Família

Abstract

The social programs form, since the last decade, one of the most frequent responses to the problems of social inequality. In Brazil, the Bolsa Familia Program (BFP) has acquired a broad national relevance because at aims at reducing the poverty and inequality of today and of tomorrow. The efficacy and the quality of the BFP can only be measured via evaluation mechanisms. To assure an appropriate evaluation of the impact of the BFP it is most important to have available a reliable and opportune information that identifies in a visible manner the treatment and comparison groups, with a small fold of simple selection, which are similar in all aspects, being solely different as to the participation in the program. Taking into consideration the characteristics of those benefited by the BFP, the research of the Impact Evaluation of Bolsa Família (IEBF) carried out in 2005 was not able to effect an experimental evaluation of the program, making an option for the elaboration of a domicile base line research, previously executing a screening operation, to categorize the domiciles as per the benefit received. Although the information obtained in the screening is considered adequate for analysis in the IEBF, it is possible that the responses be influenced by the subjective aspects; even in the field research the collection of data has followed a set of rules. However, it is important to emphasize that using the administrative records of the CadÚnico enables to check and evaluate the targeted classifications, because these are information used by the supervisors of the monitoring of the BFP. In view of the importance of the evaluation and of the methodology that cover the process of an evaluation to establish the limits of the analysis and of the description of the results, this dissertation has as objective to explore the sole possibilities that are generated with the record linkage the bases to analyze the sensibility of the results of impact of the social programs for transfer of income, when are analyzed two types of sources of information for allocation of the families in the groups of treatment and comparison. To effect the comparison of the results, two sources of information were used: the data bases obtained from the field research of the IEBF and the one of the administrative records of the CadÚnico. In accordance with the characteristics of these data bases, two record linkage strategies were used: the deterministic and the probabilistic. As result of this record linkage it was possible to measure the effects of the impact upon the education in the BFP for the population between age 7 and age 14, when the families are allocated in the comparison groups, as per the field research of the EIBF and as per the administrative records of the CadÚnico. To find the results of the evaluation of impact it was used the non-experimental method propensity score matching (PSM); besides, using a special form to identify the potentially beneficiary and non-beneficiary groups of the BFP, it was used the Regression Discontinuity method, an exercise that would not be feasible using one only source of information. The results of the work suggest that with the record linkage of the bases the number of listed families was considered satisfactory to analyze the variations or sensibilities of the results of impact with the two sources of information. On their turn, the results of the comparative analysis evidence differentials that are not relevant if it is considered the allocation of the families by the field research of the IEBF, but that become significant with the allocation derived from the CadÚnico. Besides, there are indications that these results depend largely on the educative characteristics proper of the children belonging to the families that were allocated in other groups.