The purpose of the paper is to analyze the e¤ects of changes in policy of child benefit provision on child poverty. In particular, we examine whether the introduction of means-testing schemes in the regions has improved targeting of child benefits. We test whether the probability of being paid depends negatively or positively on income, and whether this probability increased between 1996 and 1998 for the poor part of the population. In addition to probability analysis, we use direct poverty analysis, i.e., we decompose the change in poverty into changes in child benefits, and other factors. We carry out a comparative analysis among three groups of regions which differ with respect to the approach chosen to identify the needy families with children. We pay special attention to the relative efficiency of a variety of means-testing schemes.
