This paper develops direct tests for evaluating the performance of two types of emergency employment programs put in place in Chile since 1999. Our results suggest: First, decentralized and market-driven programs (subsidies for hiring and training) are more efficient in terms of productivity, but are targeted to people that are less vulnerable to unemployment. Second, direct employment programs result in moderate increases of the income of the households of the participants. This increase may be outweighted by the costs (in present value) associated with higher school drop-out and participation rates. Third, if analyzed at a municipality level (comuna), centralized programs do not target municipalities with higher unemployment, increased vulnerability to unemployment, or even lower median income levels, but are strongly correlated with the political affiliation of its major. Finally, our results suggest that the population targeted in direct employment programs is not more vulnerable to unemployment than the actually unemployed.