The Demand for Risky Assets in Retirement Portfolios Yoonkyung Yuh and Sherman D. Hanna 1. Introduction Asset allocation decisions in for retirement savings have become more important for individuals with the increase in 401(k) and related retirement saving programs and the decrease in defined benefit plans since the 1980s. Because of superior returns of stocks for the long-term, investors should consider stocks as major investments for long-term financial goals such as retirement. Chen and Hanna (1996) found that periodic investors would always have accumulated more with index stock funds than with funds with bonds or any mixed portfolios based on Ibbotson asset categories, for any 20 year period since 1926. Based on a study of nearly 200 years of returns on different asset portfolios, Siegel (1994) concluded that for long holding periods (twenty years), the risk of stock portfolios was even lower than that of Treasury bills. Thus, investing in stocks or risky assets in retirement funds would provide the best way to accumulate retirement wealth. The purpose of this study is to analyze the proportion of risky assets in retirement savings based on total financial situation of household, household characteristics and risk tolerance. In addition, having a high proportion of risky assets in the retirement savings might guarantee high accumulation of retirement wealth because of the long-run superiority of stocks. 2. Theoretical Model Choice among asset classes or asset class mixes depends on an investor s wealth, investment objectives, risk tolerance, and investment horizon (Butler & Domian, 1991). Specifically, the percentage of portfolio that investors put in stocks is dependent on both the horizon over which they plan to hold their investments and their willingness to trade off risk and return (Siegel, 1994). Investors with a very long time horizon (e.g., people in their 20s saving for retirement) are advised to allocate more of their funds to equity investments than should older savers whose retirement is imminent (Thorley, 1995). In other words, because the risks of holding stocks decreases over time, the percentage of investors portfolio committed to stock rises substantially as the investment horizon increases (Siegel, 1994). The risk-return trade-off is believed to differ by individual, with some being more risk averse than others, i.e. they require a greater increase in return for a given increase in risk than do others. Thus individual s risk tolerance level is important to portfolio decisions and financial planning. On the other hand, Malkiel (1996) distinguished between investors attitude toward risk and their capacity for risk. He asserted that the risks investors can afford to take depend on their total financial situation, including the types and sources of income exclusive of investment income. He argued that the kinds of investments that are appropriate for investors depend importantly on their sources of income other than income derived from their investment portfolio. Their earning ability outside investments, and thus capacity for risk, is usually related 1
to age. There are various explanations about effect of age on the demand for risky assets. The first argument is that the young enjoy greater labor-supply flexibility then the old and may therefore be more inclined to hold risky assets. The second is that since the young are more likely to be liquidity constrained, they may be less inclined to hold them (Guiso, Jappelli, & Terlizzese, 1996). On the other hand, some previous research found that age itself is not strongly associated with the demand for risky assets (Blume & Friend, 1975). It was found that liquid constraints had a large and significant negative effect on the demand for risky assets in the household portfolio. Households who do not have adequate financial assets to cover emergencies and have important short-term goals may not be in a position to invest in stocks (Guiso, Jappelli, & Terlizzese, 1996; Sung & Hanna, 1996). The positive effects of education, income, and net worth on the share of risky assets in total financial assets were also found (Guiso, Jappelli, & Terlizzese, 1996; Cohn, Lewellen, Lease, & Schlarbaum, 1975). Therefore, it is summarized that investment horizon, attitudes toward risk, total financial situation and age representing capacity for risk, and other household characteristics would determine the mix of assets in retirement portfolios. 3. Methods Data and Sample This study used data from the 1992 Survey of Consumer Finances. The Survey, sponsored by the Federal Reserve Board in cooperation with the Department of the Treasury, is conducted by the National Opinion Research Center at the University of Chicago (Board of Governors of the Federal Reserve System, 1992). The data were weighted to produce descriptive statistics and conduct multivariate analyses. In this study, non-retired households aged 70 years or younger were selected, resulting in a sample of 1,578 households. Dependent variables The dependent variable of this study was the proportion of retirement savings (funds) invested in risky assets of households. Because direct variable (measure) for risky assets in retirement savings was not available in the data set, a proxy for the proportion of risky assets in retirement savings was created using appropriate weight using available data. 1 The retirement savings of this study included Individual Retirement Accounts (IRA), Keogh, and all types of defined contribution pension plans in the households. The defined contribution pension plans were; 401(k) plans, 403(b) plans, ESOP(Employee Stock Option plan), SRA, thrift/ savings, stock ownership plans, profit sharing plans, deferred compensation, SEP, TIAA-CREF, money purchase plan, and tax-deferred annuity (TDA). Independent variables Independent variables to explain the share of risky assets in household retirement savings were 2
selected based on the theoretical model and literature review. Proceedings of the Academy of Financial Services, 1997 Analysis Since about thirty percent (30.6 %) of the sample had zero percent of risky assets in their retirement savings, Tobit analysis was performed. 4. Results and Discussion Table 1 shows the sample characteristics and mean and median in proportion of the risky assets share by categorical variables. People aged 30 or less had highest proportion of risky assets in their retirement savings. They tend to have more risk tolerant than other group because of their capacity for risk related to earning potentials in the future. Also, they might have more aggressive investment attitudes since substantial proportion in this group is expected to be single (never married). On the other hand, two age categories (31-40, 41-50) had a relatively lower proportion of risky assets in their retirement savings. These results reflect that people who are in this stage tend to have short-term goals such as housing or college education for children, which makes them to take more conservative attitudes in their retirement investments. The proportion of risky assets in retirement savings obviously increases with educational levels. For the household type, male-headed households had a highest proportion of risky assets while femaleheaded households had a lowest proportion of risky assets in their retirement funds. Households having retirement as a saving goal had a higher proportion of risky assets than other households. In addition, the share of risky assets in retirement savings differed by levels of risk tolerance. There are obvious differences in the proportion between people who take above average risk and people who take average risk or less. The results of the Tobit analysis on the risky assets shares in retirement savings are presented in Table 2. The Tobit analysis confirmed the above descriptive results. The effects of age, education, household type, and risk tolerance on the risky assets shares in retirement savings were significant. People aged 41-50 had lower proportion of risky assets in their retirement funds compared to younger groups. The negative sign of the estimates for all age groups indicated that their risky assets shares were lower than the youngest group aged 21-30. The youngest groups tend to have longer life expectancy and more earning power to maintain their standard of living in the face of any financial loss. The shares of risky assets in retirement funds significantly increased with the educational level. This result might support argument that financial information on more sophisticated financial assets is acquired slowly over the life cycle (Guiso, Jappelli, & Terlizzese, 1996). Compared to people who take no risk, people who take average or above average risk had significantly more risky assets in their retirement savings. This result confirms the important role of investors risk tolerance level on their portfolio decision in retirement funds. Female-headed households had a significantly lower proportion of risky assets compared to married couple households. On the other hands, effects of net worth, liquid assets, and non-investment income were not significant. 3
5. Implications (Conclusion) Currently, U.S. private pension plans are making a major shift away from defined-benefits toward defined-contributions and personal retirement plans such as IRA. Thus, investors are increasingly being charged with the critical decision of how to allocate their retirement savings between risky stock and less-risky money market funds (Thorley, 1995). Therefore, role of financial planners and accurate investment information (strategy) to give advice for clients to make informed decisions on retirement assets portfolios became more important. Financial counselors should be informed of this study to be better aware of clients demand for risky assets in retirement funds depending on their capacity for risk. The long-run potential of stocks (risky assets) should be used for retirement planning. At the very long horizons faced by U.S. investors participating in tax-privileged retirement plans, the probability of stock market underperformance is almost zero (Thorley, 1995). Therefore, investing stocks in their retirement funds should be encouraged by presenting the evidence of the long-run superiority of stocks, since planned investment horizon for retirement tends to be relatively longer than other investments. Additionally, as time passes, the asset allocation decision should be reviewed in light of the investor s changing situation (Butler & Domian, 1991). Because there is a limitation in the measurement of the shares of risky assets in this study, further research using accurate measures is needed. Also, empirical definition of risky assets and riskless assets should be clearly defined to obtain rigorous results. 2 4
Table 1. Risky assets share in retirement savings by sample characteristics Variables N = 1,578 % Mean (%) Median (%) Age 21-30 31-40 41-50 51-60 61-70 Education less than high school high school graduates some college college degree or more Race /Ethnicity White or others Black or Hispanic Household type married couple male-headed female-headed Retirement as a saving goal yes no Risk tolerance substantial risk above average risk average risk no risk 130 369 470 400 209 73 271 285 949 1449 129 1274 142 162 715 863 89 363 773 353 8.2 23.4 29.8 25.3 13.2 4.6 17.2 18.1 60.1 91.8 8.2 80.7 9.0 10.3 45.3 54.7 5.6 23.0 49.0 22.4 49.6 45.5 41.3 47.8 41.6 31.2 41.0 45.5 48.9 45.4 41.9 45.8 51.2 36.7 48.2 42.6 49.6 65.4 43.3 36.1 47.2 37.4 33.0 37.7 54.2 83.8 33.0 5
Table 2. Tobit analysis of risky assets share in retirement savings Proceedings of the Academy of Financial Services, 1997 Variables Estimate Pr > Chi Significance Age (vs. 21-30) 31-40 41-50 51-60 61-70 -6.6-11.9-2.2-7.3 0.1249 0.0084 0.6661 0.2509 ** Log (net worth) 0.3 0.6623 Log (liquid assets) -0.7 0.3188 Log (non-investment income) 1.9 0.3924 Education (vs. less than high school) high school graduates some college college degree or more 13.4 17.2 17.6 0.0277 0.0058 0.0043 * ** ** Black or Hispanic 0.4 0.9229 6
Household type (vs. married couple) male-headed female-headed 3.9-8.1 0.3774 0.0446 * Retirement as a saving goal 2.5 0.3759 Risk tolerance (vs. no risk) substantial risk above average risk average risk 12.2 33.2 7.9 0.1071 0.0001 0.0137 *** ** Intercept -1.5 0.9469 Scale (Normal scale parameter) 49.1 * p <.05, ** p <.01, *** p <.001 7
Endnotes 1. First, to get the amount of risky assets in the defined-contribution plans, variables x4234, x4334, x4434, x4834, x4934, x5034 for investment categories and variables x4226, x4326, x4426, x4826, x4926, x5026 for defined-contribution plans were used in the codebook. For defined-contribution retirement plans, five investment categories were available; mostly or all stock/ mostly or all interesting earning assets/ split/ other/ no pension. Weight of 0.9 and 0.5 were given to mostly or all stock and split respectively and weight of 0.0 were given to other categories. Thus, the amount of risky assets in defined-contribution retirement plans were defined as sum of risky assets in each defined-contribution plans in the household. Second, to get the amount of risky assets in IRA/Keogh plans, variable x3631 for investment categories and variables x3610, x3620, x3630 for IRA/Keogh plans were used in the codebook. For IRA/Keogh plans, ten investment categories were available; stock, mutual funds/ bonds, similar assets, T-bills/ CD s, bank accounts, money market/ combination of all these three types/ combination of stock and bonds/ combination of stock and money market/ life insurance, annuities/ real estate/ other/ no account. Weight of 1.0 was given to both stock, mutual funds and real estate, 0.5 was given to both second and third combinations, and 0.33 was given to the first combination. For other categories, weight of 0.0 was given. Thus, the amount of risky assets in IRA/Keogh plans were defined as sum of risky assets in each IRA/Keogh in the household. Finally, risky asset share in retirement plans was defined as following : [(risky assets in defined-contribution plans + risky assets in IRA/Keogh plans) / (total assets in defined-contribution plans and IRA/Keogh plans)]*100. 2. Empirical work in finance has often employed short-term Treasury Bills as a surrogate for the riskless asset, but those securities are not fully satisfactory since they are nor a complete hedge against changes in the absolute price level (Cohn, Lewellen, Lease, & Schlarbaum, 1975). Appendix Appendix Table 1. Distribution of risky assets shares in retirement savings of sample Proportion of risky assets (%) 0 % : n = 482 (30.6 %) > 0 % : n = 1,096 (69.4 %) Mean (N = 1,578) Quantiles 0% 25% 50% 75% 100% 45.0 0.0 0.0 90.0 100.0 Appendix Table 2. Distribution of continuous variables of sample Continuous variables Mean (N = 1,578) Quantiles 0% 25% 50% 75% 100% 8
Net worth 279,518-325,300 29,900 96,870 215,000 520,430,000 Liquid assets 17,067 0 1,000 3,500 11,200 12,546,000 Non-investment income 62,886 1,200 30,000 46,000 68,000 30,300,000 References Blume, M. E. & Friend, I. (1975). The asset structure of individual portfolio and some implications for utility functions. The Journal of Finance, 30(2), 585-603. Board of Governors of the Federal Reserve System (1992). Working Codebook for 1992 Survey of Consumer Finances (SCF). Washington, DC. Butler, K. C. & Domian, D. L. (1991). Risk, diversification, and the investment horizon. The Journal of Portfolio Management, 17(3), 41-48. Chen, P. & Hanna, S. (1996). Retirement accounts: High returns with safety, Proceedings of the Association for Financial Counseling and Planning Education, 107-116. Cohn, R. A., Lewellen, W. G., Lease, R. C. & Schlarbaum, G. G. (1975). Individual investor risk aversion and investment portfolio composition. The Journal of Finance, 30(2), 605-620. Guiso, L., Jappelli, T. & Terlizzese, D. (1996). Income risk, borrowing constraints, and portfolio choice. American Economic Review, 86(1), 158-172. Malkiel, B. G. (1996). A Random Walk Down Wall Street. New York: W.W. Norton & Company. Siegel, J. J. (1994). Stocks for the Long Run. Burr Ridge, IL: Irwin Professional Publishing. Sung, J. & Hanna, S. (1996). Factors related to risk tolerance. Financial Counseling and Planning, 7, 11-20. Thorley, S. R. (1995). The time-diversification controversy. Financial Analysts Journal, May- June, 68-76. Wiatrowski, W. J. (1993, March). Factors affecting retirement income. Monthly Labor Review, 25-33. Zhong, L. X. & Xiao, J. J. (1995). Determinants of family bond and stock holdings. Financial Counseling and Planning, 6, 107-114. 9
10