Use of survey design for the evaluation of social programs: The PNAD and PETI

The structure of some household survey allows the evaluation of social programs which are implemented gradually by municipality and whose objectives are measurable by survey variables. Such evaluations do not require over sampling of areas in which the program was implemented, nor the application of additional questionnaires while providing baseline data and non-experimental comparison groups. We use the PNAD survey to evaluate the impact of the Program for the Eradication of Child Labor on child labor, schooling, and income for municipalities which entered the program from 1997-1999. We present results both from a reflexive comparison and from matching municipalities to form a comparison group and measuring the difference in differences (D in D). Only the reduction of child labor is robust to the D in D analysis, while the reflexive results also demonstrate a significant increase in school attendance. We find the program to be more effective in smaller municipalities as suggested by Rocha (1999).