Abstract
Households are increasingly responsible for making their own decisions about matters that involve risk. For example, they need to decide how much to save for retirement. Since households often have multiple retirement savings accounts, one interesting question is how savings to one account crowds out savings to other accounts. In Chapter 1, I conduct an event study to estimate how an increase in the contribution rate to one mandatory retirement plan crowds out the total contribution rate to other savings plans, and to this end I examine a novel panel data set for retirement plan contributions of employees at a large public university. Of the particular interest is the fact that the university increased the total contribution rate to a mandatory retirement plan from 10.4% to 13.9% for employees hired on or after July 1, 2010. I find that compared to those hired before the policy change, those hired after the change decreased their contribution rates to voluntary plans by about 2.23 percentage points on average during the first month of employment, but they gradually reached a rate that is not statistically different from the average contribution rate among those hired before the policy change. My results suggest that although new hires respond to the policy change, their response disappears over the long run. One explanation for this phenomenon is that when employees determine their voluntary contribution rate, they follow a rule of thumb that never takes mandated plans into account.
In Chapter 2, I use the same administrative data to test how employees adjusted their contributions to retirement savings plan during the Great Recession. I find that compared to new faculty hired before the Great Recession, those hired during the Great Recession are 11.8 percentage points less likely to participate in the mandatory DC plan, and this estimate is statistically significant at the 1% level. With regard to the voluntary TDA, I find that compared to new faculty hired before the Great Recession, those hired during the Great Recession are 9.56 percentage points more likely to own a voluntary TDA (significant at the 1% level), but they contribute 1.91 percentage points less to the voluntary TDA (significant at the 1% level). Compared to new staff hired before the Great Recession, those hired during the Great Recession are 9.59 percentage points more likely to own a voluntary TDA (significant at the 1% level), and they contribute 0.41 percentage points more to the voluntary TDA (significant at the 5% level). The voluntary TDA participation rate and the contribution rate among new employees hired after the Great Recession, however, are not statistically different from those hired during the Great Recession.
Chapter 3 estimates the distribution of the risk aversion level from employees' Flexible Savings Accounts (FSAs). An FSA is a tax-preferred financial account into which the employee and the employer (if applicable) put money that the employee can use to pay for medical expenses not covered by the health insurance plan, such as deductible, copayments, and co-insurance. I find that the distribution of the risk aversion coefficient is not a normal distribution and is not a lognormal distribution, and within the same individual it varies a lot across time. This suggests that the constant absolute risk aversion utility function and the constant relative risk aversion utility function should not be used to explain people's choices under uncertainty. Consequently, it may be worthwhile to apply some alternative models, such as the loss aversion model or the probability weighting model, when analyzing the choices made by individual who face risk.
