In recent years, governments have become increasingly concerned about the low levels of households wealth accumulation upon retirement, and the capacity of individuals to keep the standard of living they had during their working lives. Among the reasons behind these concerns are the high relative poverty rates among elderly households, the low replacement rates provided by compulsory pension systems, and the higher responsibility placed on individuals to fund their retirement due to changes in pension systems and the increased complexity of financial instruments.

Government officials in various countries have developed a series of policies that aim at encouraging retirement savings among the population. The evaluation of the effectiveness of such policies has been a continuous objective of economists. This dissertation contributes to the public economics literature in accomplishing this objective via two cases whose analysis will hopefully inform policy makers and help better design policies geared towards improving individuals' retirement wealth accumulation.

In chapter 2, I investigate the effect of the introduction of tax free retirement accounts on the savings behavior of Mexican households. This chapter contributes empirical evidence to the debate about whether preferential tax treatment is an effective policy tool to encourage household savings. The empirical strategy is a difference-in-difference approach that utilizes an arguably exogenous change in access to tax free accounts for a well-defined set of workers. The data provide evidence of heterogeneous effects across demographic subgroups and across quantiles of the savings distribution that accord with predictions of a standard model of savings behavior. In particular, the data show an increase in the savings rate of treated workers in the year following the introduction of the accounts. The effect is driven by prime age workers and by high income workers. Among prime age workers, the lower savers experience the largest effects of the policy change. I perform multiple robustness checks on these findings, including estimating propensity score matching models and tests for potential confounding factors such as changes in retirement accounts' returns or fees, or changes in workers' income.

In chapter 3, I analyze whether information framing related to the performance of Pension Funds Administrators affects the retirement management decisions of Mexican workers. I conduct a survey to collect information on recommendations for Fund Administrator

made by Mexican workers when faced with randomly framed scenarios. The scenarios feature framing based on choice avoidance and framing exploiting loss aversion. I find evidence that reducing the number of possible choices increases the probability that individuals choose a Fund Administrator with a higher net return or with lower fees. A loss aversion framing increases the probability that individuals choose a Fund Administrator with a higher net returns. Finally, I find evidence that higher levels of financial literacy decrease the effects of framing on Fund Administrator choice.