Risk preferences in defined contribution plan investing: A study of State of Florida employees.

摘要

With a growing number of pension plans switching from defined benefit to defined contribution, it is becoming increasingly important that individuals understand the risk and rewards of investing and diversification. Educating plan participants and understanding how they influence one another is critical for furthering that goal. Understanding that not all participants view risk similarly, and that targeted education may be needed to address financial knowledge deficiencies will help plan administrators aid their participants in their investing process. This dissertation hopes to add to the knowledge by study the roles of race, gender and peers in investment preferences. This research should shed light on how differences among participants drive their investment choices and what role participant interaction plays in the process.;Using a unique matched panel data set that combines detailed demographic information from the Florida Department of Education's annual survey of school districts with investment information from the Florida State Board of Administration for 2002--2008, this dissertation is able to examine the investment preferences of plan participants at a level of detail not available with many publically available data sets.;Chapter 2 looks specifically at the roles of gender and race in investment preferences. The analysis shows that women and Blacks invest more conservatively than men and Whites, respectively. It is also shows that despite similar outcomes, the route which Black men and White women take differs. Black men, like men in general in this study, take a more active approach to their investments, while women tend to default into more conservative investments. Black women are shown to be the most conservative, with an expected return 5.5% lower than White men, which, assuming 30 year investment horizon, returns of 7.5% and steady real wages, yields a final portfolio value ratio of .919. This is a considerable difference in wealth which is yet to be justified through explanatory variables.;Chapter 3 looks for the presence of peer effects in the panel data. The regression analysis uses both pooled and first differenced observations to study the effects. In general, the pooled regressions do not seem to adequately account for the peer effect, especially at the district level where measured coefficients lose magnitude and significance once district fixed effects are included. The first differenced regressions, however, measure how participants react to changes in their peer group. This method accounts for time-constant unobservables in both the participant and the peer group and better reflects the true peer effect. The analysis also concluded that choosing the appropriate peer group is critical to the results. The results of the analysis were strengthened with falsification tests to show that the results do not occur with random pairings.