Direct and Indirect Effects of Public Policies

Direct and Indirect Effects of Public Policies by Giacomo De Giorgi Chair: Richard Blundell This dissertation analyzes two im portant recent public policies and sheds light on the direct and indirect impacts of both; and more in general on the need of sound evaluations to carefully consider possible spillovers as a fundamental component of programs. In the first two essays I focus on the “New Deal for Young People (NDYP) in the UK” , a multiple treatm ents policy launched in 1998. Such analysis is conducted on two main margins: i. employment; and ii. job quality as wage returns. I discuss possible indirect effects as substitution of workers and general equilibrium effects on wages. The main results are tha t the policy enhances the (re)employment proba­ bility of participants by 5% and the effect lasts over several cohorts. Further, no evidence of displacement as well as general equilibrium effects is found. On the rela­ tive effectiveness of the different treatments: IV estimates of the treatm ent effect of the subsidized employment option clearly point towards a negative and significant penalty of roughly 20%, due to stigma or worsening of the matching function. The other treatments do not seem to have a significantly differential impact. Chapter IV analyzes an aid policy implemented in rural Mexico: PROGRESA. The focus of the chapter is the identification of the spillovers of the policy to those households who just happen to live in treated communities, but are not eligible for the program. A formal definition of the Indirect Treatment Effect ( I T E ) param eter is given. Focusing on consumption the mechanisms generating such indirect effects are discussed. Those mechanisms need not be limited to the specific policy. In particular, in developing Countries policies are likely to affect all residents of the areas where they are implemented, especially when village economies and social networks create strong links between a limited number of households. The large liquidity injections into small communities increase the consumption of the non-treated through changes in the credit and insurance markets. Thus, the total effect of the policy is larger than its effect on the treated. Further, the results confirm th a t a key identifying assumption tha t the program has no effect on non-treated individuals (SUTVA)is likely to be violated in similar policy designs.