Date of Degree


Document Type


Degree Name





Sangeeta Pratap

Committee Members

Frank Heiland

Wim Vijverberg

Subject Categories

Labor Economics | Macroeconomics | Other Economics | Public Economics


Retirement Annuity Learning Frailty Bayesian


This dissertation consists of three chapters.

Chapter 1

The discrepancy between the high demand for annuities predicted by economic theory and the empirical low holdings of these assets, known as the annuity puzzle, is still not completely understood in economic studies of retirement finance. This paper assesses the effect of individuals' mortality risk learning process on annuitization. I isolate this effect by building a life-cycle model in which individuals have imperfect information of their true survival probability distribution, and therefore have to update their beliefs about it in a Bayesian manner. Using data on subjective mortality by the Health and Retirement Study to evaluate the model, the baseline result shows that the demand for annuities can be about 40 percent lower than full annuitization solely attributable to individuals learning about their true mortality risk—a situation that does not allow for the known strong take-up for annuities to take effect. I further expand the model to have a bequest motive to show how more features that drive down annuitization can be added and interacted with this learning mechanism.

Chapter 2

Early claiming of Social Security benefits imply a reduction in the annuity value these payments offer to individuals who retire before the normal retirement age. To understand the prevalence of this behavior, in this paper I investigate the effect of mortality learning on early benefits take-up and early retirement by building a life-cycle model where individuals reduce their longevity uncertainty as they age. As individuals are more certain of their lifespans, the annuity value provided by Social Security benefits is less appealing, and consequently, early claiming can be optimal. Using data from the Health and Retirement Study to calibrate the model, I find that mortality learning is an important element in explaining this phenomena: early benefits take-up is 37.4 percent lower in a counterfactual

scenario where individuals do not learn about their mortality. The impact of this result on a basic policy aimed at discouraging early retirement is discussed.

Chapter 3

The Annuity Puzzle has been studied in the economic literature for over 50 years. This chapter provides a summary of the main findings.