Resumo:In this paper we propose a method to estimate externality effects in cash transfer programmes, even in cases when the benefit is not randomly assigned. Externality is assessed through the decomposition of the average treatment effect on the treated (ATT) into participation (direct) effect and externality (indirect) effect. We also suggest a non parametric decomposition to investigate whether changes in household outcomes are caused by the income transfer itself or by the other non-monetary components of the programme, such as conditionalities and family support services. We apply all these decompositions on the effect of a conditional cash transfer (CCT) programme on household consumption and savings in Paraguay. This was possible because of the presence of two distinct comparison groups, one within the village and potentially exposed to the externality, and another in a different village not affected by the programme. Furthermore, the evaluation survey collected information on both income and consumption. The results indicate that the programme has a small impact on consumption and a considerable impact on savings. In the absence of externality, however, the programme would have a higher effect on consumption, mostly associated with the cash transfer, and a lower effect on savings. Moreover, the impact on the pattern of consumption is significantly related to a substitution effect and is not related to the increase in income. (...)