by Gabriel Lara-Ibarra, University of Maryland
In recent decades, Mexico experienced important changes in its demographic composition with the older share of the population becoming relatively larger than the younger one. The associated increase in pension costs led the federal government to adopt a private pension account system in 1997 to replace the traditional Pay-as-you-go system. Additionally, the reform required all private sector workers to choose a private institutions called an AFORE, to handle their retirement accounts. Public sector employees continued to contribute to a separate federal institution. A large share of private sector workers did not choose an AFORE until 2001 and an even smaller share were making voluntary contributions to their pension funds. In order to incentivize savings in retirement accounts the Congress passed a Tax Break law in December 2002 (effective January 2003) on voluntary contributions. This law effectively created saving vehicles for private sector workers that were similar to a hybrid between the IRA and Roth IRAs available to workers in the United States.
Proponents of the Mexican pension reform of 2003 argued that the preferential tax treatment would lead to an increase in domestic savings. Theoretically, however, the overall effect is ambiguous. The proposed study will empirically evaluate the effect of the 2003 reform on Mexican household savings. It will contribute to the ongoing debate in the empirical literature in Public Economics of whether tax deferred accounts lead to increased households savings. The policy change studied in this paper provides an exogenous change to a the availability of tax-deferred savings accounts to private sector workers that allows me to surmount many of the identification issues that have limited previous studies in the US context. I will use two sources of data. One is the National Household Survey of Income and Expenditure conducted in Mexico every two years. The other is the Mexican Family Life Survey, a panel dataset conducted in 2002 and 2005. I will employ a difference-in-difference (DID) approach using salaried private sector workers as a “treatment” group and public employees as a control. I will use three empirical methodologies as a means of getting a broad description of the impact of the policy change and testing the robustness of the DID estimates. First, I will estimate quantile regressions to estimate the impact of the policy change at different points of the savings distribution. Second, I will employ propensity score matching methods to address the non-experimental nature of the data. Third, I will apply Athey and Imbens (2006) change-in-changes model to get a non-parametric DID estimate of the tax break law. I plan to evaluate the policy’s impact for different subgroups of the population and test for heterogeneity in the effect of the policy change across age cohorts and income levels.