Dynamic Incentives in Retirement Earnings-Replacement Benefits

64 Pages Posted: 24 Feb 2020

See all articles by Andrés Dean

Andrés Dean

Universidad de la Republica - Instituto de Economía

Sebastian Fleitas

KU Leuven

Mariana Zerpa

KU Leuven


Many defined-benefit pension systems in developed and developing countries use a small set of final years of earnings to compute pension benefits. This provides dynamic incentives to report higher earnings in the final years of the career. In this paper, we document the responses of self-employed and employed workers to these incentives, using social security administrative records and household surveys from Uruguay. We implement event studies that leverage the use of a 10-year benefit-calculation window, combined with the discrete change in the probability of retirement at the minimum retirement age. We find that reported earnings of self-employed workers and employees of small firms start increasing sharply 10 years prior to minimum retirement age, reaching a 3% increase on average. This is not the case for employees of large firms, where earnings underreporting is less prevalent. These responses are not explained by changes in total earnings or hours of work, as reported in household surveys, suggesting a change in reporting behavior. Back of the envelope calculations for the self-employed bound the cost of these responses between 1.9% and 2.6% of the total cost of pensions for this group.

Keywords: earnings replacement benefits, social security, tax compliance

JEL Classification: J26, H26, H55, O15, O17

Suggested Citation

Dean, Andrés and Fleitas, Sebastian and Zerpa, Mariana, Dynamic Incentives in Retirement Earnings-Replacement Benefits. IZA Discussion Paper No. 12982, Available at SSRN: https://ssrn.com/abstract=3542651

Andrés Dean (Contact Author)

Universidad de la Republica - Instituto de Economía ( email )


Sebastian Fleitas

KU Leuven ( email )

Oude Markt 13
Leuven, 3000

Mariana Zerpa

KU Leuven

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
PlumX Metrics