Decentralizing equality of opportunity and issues concerning the equality of educational opportunity

This dissertation explores issues concerning local versus global equality of opportunity, the presence of inequality for education attainment and the optimality of policies addressing such problems. The literature distinguishes between two different approaches to distributive justice: the global and the local approach. Local distributive justice analysis determines the allocation of rights, duties and resources concerning a subset of the aspects that determine individual welfare. On the other hand, global distributive justice analysis determines the allocation of rights, duties and resources concerning all aspects in a society that affect individuals welfare.; Global and local distributive justice notions have been analyzed independently in the literature and the focus has been more on global rather than local distributive justice. In the first chapter we interpret a collection of local distributive justice problems as decentralized global distributive justice problems. This permits global distributive justice analysis, mainly done by philosophers, to inform policymakers on what the local distributive justice goal should be, and also allows us to understand what the global distributive justice properties of a collection of local problems are. We then analyze the properties on local distributive justice that are necessary and sufficient to decentralize the global distributive justice notion of equality of opportunity. The second chapter provides evidence that school could potentially serve as an instrument to provide equality of educational opportunity. Using data from the UK, we find that the education process affects individuals final educational achievement, contrary to what part of the literature suggests. The last chapter is somewhat unrelated to the earlier chapters. We provide necessary and sufficient conditions on demand data for the individual to behave as if she was maximizing quasilinear preferences subject to a budget constraint. In the case where such property fails, we provide the quasilinear utility that "better" approximates the data.