See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228980285

Does Risk Tolerance Decrease With Age?

Article in Journal of Financial Counseling and Planning · January 1998

DOI: 10.2139/ssrn.95489

CITATIONS READS
167 4,894

2 authors, including:

Sherman D. Hanna
The Ohio State University
266 PUBLICATIONS 3,982 CITATIONS

SEE PROFILE

All content following this page was uploaded by Sherman D. Hanna on 02 May 2018.

The user has requested enhancement of the downloaded file.

 Does Risk Tolerance Decrease With Age?
Hui Wang1 and Sherman Hanna2
This study examines the effect of age on risk tolerance. The life-cycle investment hypothesis is tested
using the 1983-89 panel of the Survey of Consumer Finances. Household wealth is defined as the sum
of human capital and net worth. Risk tolerance is measured by the ratio of risky assets to total wealth.
Risk tolerance increases with age when other variables are controlled.
Key Words: Risk tolerance, Risk aversion, Investment, Survey of Consumer Finances
The proportion of households in the U.S. headed by literature, but risk aversion can be thought of as inversely
someone 65 and older is expected to rise from the 1996 related to the financial planning concept of risk
level of 22% (U. S. Bureau of the Census, 1997, Table tolerance.
70) to as high as 40% by the year 2040 (Lumsdaine &
Wise, 1990). This dramatic increase in the proportion of Review of Literature
elderly households will result in part from the aging of Previous researchers have studied the relationship
the baby-boom generation, born between 1946 and 1962. between age and the holding of risky assets. Morin and
Consequently, the adequacy of retirement income is Suarez (1983) investigated the effect of age on the
becoming a matter of national concern. holding of risky assets using 1970 Canadian Survey of
Consumer Finance data. Risky assets were defined as the
The buying power of most pensions is reduced over time sum of stocks, bonds, mutual funds, real estate other than
due to inflation. In 1989-90, less than 5% of full-time owner-occupied home, equity in own business, and loans.
employees covered by a defined benefit pension were Morin and Suarez (1983) concluded that on average, risk
provided cost-of-living adjustments (Wiatrowski, 1993). aversion increased with age. For those at the low levels
Therefore, private household savings should become an of net worth, risk aversion increased with age. In
increasing component of retirement income. contrast, for households with high net worth, risk
aversion decreased with age. The authors concluded that
Stocks offer a higher return than do less volatile both net worth and age influenced risk aversion.
investments such as Treasury bills (Ibbotson Associates,
1997). However, Malkiel (1996) suggested that the Based on the capital asset pricing model, McInish,
proportion of a portfolio devoted to stocks should Ramaswami, and Srivastava (1993) studied the
decrease as one approaches retirement. Hanna and Chen relationship between net worth and risk aversion. They
(1997) concluded that the proportion of stocks in a assumed that the investment choice along the risk/return
portfolioa should decrease with age, assuming that risk line depended on the investor’s attitude toward risk.
tolerance does not change with age. Thus, more risk-averse investors should hold less risky
portfolios, which would lead to lower levels of wealth.
Bakshi and Chen (1994) discussed the Life-cycle Risk In addition, McInish, et al. (1993) investigated the effect
Aversion Hypothesis and hypothesized that risk aversion of age on the holding of risky assets. Based on U.S.
increases with age. Bakshi and Chen (1994) assumed financial diary panel data, the results showed that for
that risk aversion could be measured by the proportion of individuals younger than 35 years old, the relationship
a household’s assets held in the form of ‘risky’ assets between net worth and risk attitude was not statistically
such as stocks. In this article, a null version of the significant. In contrast, the relationship between net
Bakshi and Chen (1994) hypothesis is tested – the worth and risk attitude was significant for those ages 35
proportion of net wealth held in risky assets does not and over. However, the study did not control for the
vary with age. This article examines the effect of age on effect of inherited wealth, which might account for a
risk tolerance. This article primarily discusses the significant portion of a household’ total wealth.
concept of risk aversion, the term used in the economics

1
Hui Wang, Statistical Analyst with Milliken and Co. in South Carolina. E-mail: [email protected]
2
Sherman Hanna, Professor, Consumer and Texile Sciences Department, The Ohio State University, 1787 Neil Ave., Columbus, OH 43210-1295. Phone:
(614) 292-4584. FAX: (614) 292-7536. E-mail: [email protected].

©1997, Association for Financial Counseling and Planning Education. All rights of reproduction in any form reserved. 27
Financial Counseling and Planning, Volume 8(2), 1997

A study by Riley and Chow (1992) focused on asset Risky assets in this study are defined as assets that
allocation and individual risk aversion in a sample of provide an uncertain nominal cash flow. Thus, the
U.S. households. Riley and Chow derived relative risk market value of all real estate held for investment
aversion indexes from actual asset allocation and found purposes, the total value of business assets, the market
that risk aversion decreased with age until 65, then value of mutual funds, corporate stocks, and precious
increased significantly. The relationship between age metals are included as risky assets. In addition, pension
and risk aversion was also studied by Bakshi and Chen assets in the forms of stocks, bonds, and mutual funds are
(1994) for aggregate U. S. time series data. Bakshi and included as risky assets in the 1989 data.c
Chen (1994) concluded that risk aversion increases as the
population ages. The Effect of Age on Individuals’ Relative Risk Aversion
Following the work of Friend and Blume (1975), Fama
McInish, et al. (1993) and Cohn, Lewellen, Lease, and and Schwert (1977), Morin and Suarez (1983), and
Schlarbaum (1975) found a positive relationship between Schooley & Worden (1996), the effect of age on an
risk tolerance and both net worth and income, with individual’s relative risk aversion is operationalized as
wealthy investors holding a higher proportion of risky the proportion of net wealth invested in risky assets.d
assets. Investors aged 45 to 54 held the highest Human capital is included as part of household net
proportion of risky assets. In contrast, those younger wealth.
than 45 years old held the highest proportion of their
total assets in non-risky assets. Human capital is calculated as the present value of future
earnings and Social Security pensions, plus any defined
Some studies used risk tolerance measures derived from benefit pensions expected. The detailed calculation is
responses to questions. For instance, Sung and Hanna based on the methodology and data described by Wang
(1996) analyzed responses to the Survey of Consumer (1997) and is not discussed here due to space limitation.
Finances risk tolerance question. Age did not have a
significant bivariate relationship with risk tolerance, Previous studies have used an Ordinary Least Squares
although in a multivariate analysis risk tolerance regression with the risky asset proportion of wealth as a
decreased with age (Sung & Hanna, 1996, note e). dependent variable (Friend & Blume, 1975; Morin &
Suarez, 1983; Schooley & Worden, 1996). However,
Data and Methodology two econometric issues arise because of the nature of the
Data and Variables dependent variables. First, because it is a share, the
The major data set used in this study is the 1983-89 panel dependent variable imposes heteroscedasticity of a well
of the Survey of Consumer Finances (SCF). The 1983- known form (Maddala, 1980). Second, many households
89 SCF panel provides the most recent comprehensive have zero shares of various asset categories, suggesting
data about a household’s wealth and its composition and the need for the Tobit model to handle censoring. To take
changes during a period of time.b However, account of both issues, this study employs the
understanding risky asset holding over the life-cycle heteroscedastic Tobit model.
requires information about a household’s present value
of future pension and Social Security wealth and a Results and Discussions
measure of future earnings. Data on present value of The mean value of the ratio of risky assets to net wealth
future pension and Social Security wealth are not in 1989 was 6% (Table 1). The median value of the ratio
included in the panel and were thus imputed from the was 1% and only 25% of households had a value of 6%
1983 cross-sectional SCF data (Avery & Elliehausen, or higher. Note that for most households a collapse of
1990). The 1983 and 1989 Consumer Price Indices stock markets would have a very small impact on total
(CPI) were used to adjust for changes in prices (U. S. wealth. These results are similar to ratios obtained by
Department of Labor, 1992). Life expectancy estimates Lee and Hanna (1995) using financial assets rather than
were obtained from the 1989 Vital Statistics of the risky assets.
United States (U. S. Department of Commerce, 1992).
Projected labor force participation rates were obtained The effects of net wealth, age, education, and other
from the Statistical Abstract of the United States (U.S. socio-demographic variables on proportion of net wealth
Department of Commerce, 1996). Poverty thresholds invested in risky assets are tested using a heteroscedastic
were obtained from the 1989 Current Population Survey Tobit model. Tobit coefficients and the calculated
Report (U. S. Bureau of the Census, 1988; 1989). marginal effects are summarized in the Appendix.

28 ©1997, Association for Financial Counseling and Planning Education. All rights of reproduction in any form reserved.
 Does Risk Tolerance Decrease With Age?

invested in risky assets for both retired and non-retired

Table 1 households. Households with a head who is not retired
Mean and Selected Percentiles of the Ratio of 1989 have a predicted risky asset proportion of 0% at age 30.
Risky Assets to Net Wealth (including human capital.) The proportion increases with age to almost 14% at age
55, 18% at age 65, and 24% at age 80. The predicted
Mean 6.4% proportion does not decrease with age for ages under
th
75 percentile 6.2% 100. The predicted proportion of net wealth invested in
Median 1.2% risky assets for retired households increases with age
th
after age 40, increasing to 8% at age 55, 13% at age 65,
25 percentile 0.3%
and 19% at age 80. Even though retirement decreases the
predicted risky asset proportion, the general pattern is
that eventually the proportion for retired households will
Age is significantly related to the proportion of net exceed the pre-retirement peak proportion. For instance,
wealth invested in risky assets. Figure 1 shows the the predicted risky asset proportion of 18% at age 65 for
relationship between age and the proportion of net wealth someone who is not retired would be exceeded by age 80
for someone who is retired.

Figure 1
Predicted Risky Asset Proportion of Total Wealth, by Age and Retirement Status

Based on tobit results reported in Appendix, assuming mean values of other variables.

©1997, Association for Financial Counseling and Planning Education. All rights of reproduction in any form reserved. 29
Financial Counseling and Planning, Volume 8(2), 1997

Conclusion and Implications b. For more on the dataset and methods, see Wang (1997). In order
to construct nationally representative estimates, the SCF data
Relative risk aversion decreases as people age (i.e., the
contain weight variables. This article does not focus on the
proportion of net wealth invested in risky assets increases estimation of household wealth changes between 1983 and 1989,
as people age) when other variables are held constant. so the weight variable WGT0195 is used.
Therefore, risk tolerance increases with age. Thus, the c. The 1983 SCF dataset does not provide precise information about
constant life-cycle risk aversion hypothesis is not the allocation of pension assets. Therefore, pension assets
invested in stocks, bonds, and mutual funds cannot be identified,
accepted. These results are contrary to Morin and so that pension assets are not counted as investment assets. This
Suarez’s (1983, p. 1201) finding that risk aversion may be a reasonable assumption for the 1980's (Papke, Petersen
increases with age. & Poterba, 1993), although it would not be a reasonable
assumption today.
Human capital accounts for a relatively large portion of d. Net wealth is considered as an exogenous variable ( Friend &
Blume, 1975; Morin & Suarez, 1983).
net wealth for young people, and financial wealth
accounts for a relatively small portion of their net wealth. Appendix
Young people may appear more risk averse since it is Table
hard for them to endure any short-term investment losses Tobit Analysis of Risky Asset Proportion of Net Wealth
with limited financial resources. Future human wealth Constant -.7075***
Age .96E-2**
can not be applied to pay present bills, car loans, Age squared -3.00e-05
mortgage debts, etc. Net wealth in 1989 .20E-4***
Household income in 1988 2.10e-04
Implication for Financial Planning and Education Respondent retired vs. not -.70E-1**
Expect inheritance vs. not .53E-1*
Educators and planners should not assume that risk Married couple vs. not .1160***
tolerance decreases as people age. Overall, the opposite Educational status (vs. less than high school)
seems to be the pattern. Consumers who avoid high High school graduate .1269***
return assets such as stocks should be encouraged to Some college education .2081***
College graduate or more .2383***
allocate part of their investments to broadly diversified Race/Ethnicity (vs. Black)
stock funds in order to maintain household purchasing White .1339***
power. As Hanna and Chen (1997) demonstrate, Hispanic .1792*
objective aspects of risk tolerance, such as the investment Other race .1635***
horizon, may be more important than subjective aspects Respondent & spouse poor health vs. not -.98e-1**
Inadequate retirement income .25e-1*
such as risk aversion. Transitory income during 1983-89 -5.30e-05
Respondent divorced vs. not 7.20e-02
Limitation of this study Respondent changed jobs vs. not -0.80e-1**
The results are based on cross-sectional data, so there Income change during 1983-89 -1.70e-04
Sigma 0.136
may be generational effects. For instance, those who Log-L -274.7
were 87 in 1989 were 27 in 1929 and would have vivid
References
memories of the Stock Market Crash of 1929 and of the Avery, R. B. & Elliehausen, G. E. (1990). 1983 Survey of
Great Depression. It is possible that people who are age Consumer Finances: Technical Manual and Codebook.
87 now may have different attitudes toward risk than Washington D.C.: Board of Governors of Federal Reserve
today’s Baby Boomers will have at age 87. System.
Bakshi, G. S. & Chen, Z. (1994). Baby boom, population
Endnotes aging, and capital markets. Journal of Business, 67(2),
a. Hanna and Chen (1997) assumed that risk tolerance (relative risk 163-202.
aversion) does not change with age. Given their assumption of a Cohn R. A., Lewellen, W. G., Lease, R. C. & Schlarbaum, G.
Constant Relative Risk Aversion utility function, then everyone G. (1975). Individual investor risk aversion and
with the same risk tolerance should have the same level of stocks investment portfolio composition. The Journal of Finance,
as a proportion of wealth, if a static analysis is used.. When the 30 (2), 605-620.
plausible assumption is added that the investment horizon
Fama, E. & Schwert, G. (1977). Human capital and market
decreases as one approaches retirement, the proportion of stocks
in the portfolio should decrease as one ages. However, when a
equilibrium. Journal of Financial Economics, 4, 95-125.
typical lifecycle accumulation of financial assets and decrease in Friend, I. & Blume, M. E. (1975). The demand for risky assets.
human wealth are considered, the normative result using their The American Economic Review, 65 (5), 900-922.
method is that the stock proportion of total wealth increases with Hanna, S. & Chen, P. (1996). Efficient portfolios for saving for
age until retirement, then does not decrease much after college. Financial Counseling and Planning, 7, 115-122.
retirement. Hanna, S. & Chen, P. (1997). Subjective and objective risk

30 ©1997, Association for Financial Counseling and Planning Education. All rights of reproduction in any form reserved.
 Does Risk Tolerance Decrease With Age?

tolerance: Implications for optimal portfolios. Financial

Counseling and Planning, 8(2), 17-25.
Ibbotson Associates (1997). Stocks, Bonds, Bill, and Inflation
1997 Yearbook. Chicago: Ibbotson Associates.
Lee, H. & Hanna, S. (1995). Investment portfolios and human
wealth. Financial Counseling and Planning, 6, 147-152.
Lumsdaine, R. & Wise, D. (1990). Aging and labor force
participation: A review of trends and explanations.
Working paper no. 34320. Cambridge, Mass.: National
Bureau of Economic Research.
Maddala, G. S. (1980). Some extensions of the Nerlove-Press
model. Proceedings of the American Statistical
Association (Business and Economic Statistics Section),
481-85.
Malkiel, B. G. (1996). A Random Walk Down Wall Street. New
York: W. W. Norton & Co.
McInish, T. H., Ramaswami, S. N. & Srivastava, R. K. (1993).
Do more risk-averse investors have lower net worth and
income? The Financial Review, 28 (1), 91-106.
Morin, R. A. & Suarez, A. F. (1983). Risk aversion revisited.
The Journal of Finance, 38(4), 1201-1216.
Papke, L. E., Petersen, M. & Poterba, J. M. (1993). Did 401(k)
plans replace other employer provided pensions? In Wise,
D. A., (Ed.), Studies in the Economics of Aging. Chicago:
The University of Chicago Press. 105-142.
Riley, W. B. & Chow, K. V. (1992). Asset allocation and
individual risk aversion. Financial Analysts Journal, 48
(6), 32-37.
Schooley, D. K. & Worden, D. D. (1996). Risk aversion
measures: comparing attitudes and asset allocation.
Financial Service Review, 5(2), 87-99.
Sung, J. & Hanna, S. (1996). Factors related to risk tolerance.
Financial Counseling and Planning, 7, 11-20.
The Federal Reserve Board. Survey Research Center of
University of Michigan. 1983-89 SCF on-tape
documentation (1989).
Wang, H. (1997). An empirical analysis of household portfolio
allocation based on a rational expectation model.
Unpublished Ph.D dissertation. The Ohio State University.
Wiatrowski, W. J. (1993). Factors affecting retirement income.
Monthly Labor Review, 116 (3), 23-25.
U.S. Bureau of the Census (1988). Current Population
Reports, Consumer Income Series P-60. Washington,
D.C.: Government Printing Office.
U.S. Bureau of the Census (1989). Current Population
Reports, Consumer Income Series P-60. Washington,
D.C.: Government Printing Office.
U.S. Bureau of the Census (1997). Current Population
Reports, Consumer Income Series P-60. Washington,
D.C.: Government Printing Office.
U.S. Department of Commerce (1996). Statistical Abstract of
the United States. Washington, D.C.: Government Printing
Office.
U.S. Department of Commerce (1992). Vital Statistics of the
United States. Washington, D.C.: Government Printing
Office.

©1997, Association for Financial Counseling and Planning Education. All rights of reproduction in any form reserved. 31
 Financial Counseling and Planning, Volume 8(2), 1997

EDITORIAL BOARD
Elizabeth P. Davis, University of Nebraska, Lincoln
Jonathan Fox, The Ohio State University
Suzanne Lindamood, Attorney, Ohio House of Representatives
Jean Lown, Utah State University
Catherine Montalto, The Ohio State University
Kathryn D. Rettig, University of Minnesota
Elizabeth Scannell, University of Vermont
Michael Walden, North Carolina State University
Jing Xiao, The University of Rhode Island

Reviewers for Volume 8

William Bailey Arthur Kennickel
Vickie Bajelsmit Nandinee Kutty
Cathy Bowen Seonglim Lee
Julie Cagle Irene Leech
Regina Chang Jieyu Li
Peng Chen Suzanne Lindamood
Natalie Chieffe Jean Lown
Larry A. Cox Edie Milligan
Sharon Danes Catherine Montalto
Elizabeth P. Davis Kathy Prochaska-Cue
Sharon DeVaney Kathryn Rettig
Musa Essayyad Tim Smaby
Jessie Fan Marlene Stum
Michael Finke Jaimie Sung
Vicki Fitzsimmons Anita Subramaniam
Ray Forgue Jean Turner
Jonathan Fox Katherine Vanderhook
E. Thomas Garman Jill Vihtelic
John Grable Premal Vora
Deborah Godwin Hui Wang
Michael Gutter Michael Walden
Sherman Hanna Mary Winter
Charles Hatcher Walt Woerheide
Celia Hayhoe Robert O. Weagley
Tahira Hira Jing J. Xiao
Jeanne Hogarth Yoon Kyung Yuh
Sandra Huston

32 ©1997, Association for Financial Counseling and Planning Education. All rights of reproduction in any form reserved.

View publication stats