Investment Portfolios and Human Wealth


Investment Portfolios and Human Wealth


Financial Counseling and Planning, 1995, Volume 6, pp. 147-152.


Hye Kyung Lee
,

Sherman Hanna
,
Professor, Consumer and Textile Science Department, The Ohio State University,
1787 Neil Ave., Columbus, OH 43210. Phone: (614) 292-4584.  Email:
hanna.1@osu.edu
.

The optimal proportion of a household’s investment
portfolio that should be in risky assets such as stocks depends on what proportion
of total wealth, including human wealth, the investment portfolio represents.
This article estimates the total wealth of households in the U.S. Survey
of Consumer Finances, and finds that financial assets represent less than
2% of the total wealth of most households. Only the elderly are likely to
have investment portfolios representing a high proportion of total wealth.

KEY WORDS: household portfolios, investment, risk,
wealth, stocks, Survey of Consumer Finance

It is well known that higher rates of return may be obtained
on investments by accepting greater risk. The best normative approach to
analyzing risky choices is to view utility as a function of wealth. In evaluating
choices under uncertainty, utility can be modeled as a function of wealth
(Hanna, 1988). Wealth represents potential consumption for the rest of a
consumer’s life. Therefore, wealth should be defined to include both net
worth and the present value of non-investment income, as both can be used
for future consumption. This is consistent with Malkiel’s (1990) suggestion
that the portion of the portfolio for stocks should decrease as a person
ages. As a person ages, human wealth (the present value of non-investment
income) will decrease, and typically, net worth will increase. Therefore,
the relationship between a household’s investment portfolio should be considered.

Even a risk averse investor can increase expected utility
by taking chances with a portion of his or her assets (Hanna, 1988). Hanna
and Chen (1995) showed that all households should hold stocks if the time
horizon is five years or more, and all of the investment portfolio should
be in stocks if the portfolio represents less than 10% of total wealth, including
human wealth. Their result is similar to Arrow’s statement (1971, p. 100)
that “… for small amounts at risk, the utility function is approximately
linear, and risk aversion disappears.” If the investment portfolio represents
10% of a household’s wealth, then a 20% loss in the portfolio represents
only a 2% loss in the household’s total wealth.

Friend and Blume (1975) incorporated nonmarketable assets
(human wealth) in the form of capitalized labor income until age 65 into
a second measure of wealth, total resources, which is the sum of net worth
and nonmarketable (human wealth) assets. In addition to the capitalized
value of labor income to the age at which the respondent expected to retire,
the capitalized values of social security income, pension income, and transfer
payments were included in the measure of nonmarketable wealth (Friend &
Blume, 1975).

Graham and Webb (1979) measured human wealth by calculating
the present value of future earnings for the male population for 1969 using
summary census data for 1950 and 1970 and detailed cross-sectional census
data available in the Public Use Sample of the 1970 Census. They implemented
this approach by equating expected returns with market earnings derived from
cross-sectional earnings data for out-of-school males adjusted by growth rates
that vary with levels of education. Both cross-sectional and time-series
wealth profiles confirmed the notion that education is positively associated
with wealth at all ages (Graham & Webb, 1979).

How important are investment portfolios of households
in relation to net worth and human wealth? This paper presents the distribution
of the ratio of financial assets to total wealth using an estimate of the
human wealth of a national sample of households in the United States. The
results show that financial assets, and therefore investment portfolios represent
a very small portion of total wealth for most U.S. households. Therefore,
based on Hanna and Chen’s (1995) results, most households in the U.S. should
have investment portfolios consisting entirely of stocks, especially in retirement
funds.


Methods

Description of Data and Sample

The data used for the analysis are from a public use
tape of the 1983 and 1986 Survey of Consumer Finances (SCF). The 1986 data
are used only to estimate the average of 1982, 1983, 1984, and 1985 total
household income. The Survey Research Center (SRC) of the University of
Michigan conducted interviews for the 1983 Survey of Consumer Finances between
February and August of 1983. The survey sample consists of 3,824 randomly
selected U.S. households and a supplemental sample of 438 high-income households
drawn from federal income tax files. The supplemental high-income sample
provides better representation of the upper tail of the wealth distribution
than that provided by most other surveys. In the summer of 1986, a limited
telephone reinterview was conducted for 2,822 of the 1983 SCF respondents
(Avery & Elliehausen, 1988).

Imputation

The problems of missing or inconsistent information make
analysis of the raw data difficult and, depending on the pattern of errors,
may bias conclusions. In order to eliminate these kinds of problems, a series
of consistency checks and imputation procedures were developed at the Federal
Reserve Board to clean the raw data and to estimate values for the missing
data (Avery et al., 1984b). From the high-income sample, missing values
for all observations were imputed. From the area probability sample, only
159 of the original 3,824 observations were discarded due to missing dollar
amounts for all income and assets. Finally, all missing values for the remaining
3,665 observations were imputed (Avery & Elliehausen, 1986).

Sample Weight

The construction of weighting variables was necessary
because of nonrandomness from inclusion of the high-income supplement drawn
from Federal Income Tax files. In this study, the recommended full sample
weight (B3016) is used. (Avery & Elliehausen, 1988).

The income adjustment is very slight for those area probability
sample observations with incomes below $50,000. The observations from the
area probability sample in the higher income strata have a more significant
reduction in their weight. The high-income sample weight is given only
for the high-income sample and gives relative sampling weights within that
sample as computed by the IRS and the Office of Tax Analysis. The weight
used in this study applies to the cleaned area probability sample and uses
the 1983 post-stratification weight. It was constructed by post-stratification
to the 1982 IRS tables using extended income (Avery & Elliehausen, 1988).

Sample

In this study, both the high-income sample (438 cases)
and the area probability “cleaned” sample (3665 cases) are included. Native
Americans were deleted from the sample because the sample size is too small
(9 cases) to use for meaningful analysis. After deleting missing and invalid
values, total sample size for this study is 2,691 households.


Measurement of Variables

Financial Assets: The total dollar amount of
financial assets is the sum of checking accounts, money market accounts,
saving accounts, IRAs, Keoghs, CDs, saving bonds, bonds, stock and mutual
fund holdings and trust accounts owned by household.

Income: Average value of total household income
of 1982, 1983, 1984, and 1985. All income values are adjusted using the
Consumer Price Index (CPI) and are expressed in 1986 dollars before computing
the average.

Composition of Total Income:

Total income was composed of income in wages and salary;
income from a professional practice, business, or farm; income from non-taxable
investments such as IRAs or municipal bonds; taxable interest income;

dividend income; net gains from the sale of stocks/bonds
or real estate; rent, trust income, or royalties from another investment;
workers or unemployment compensation income; child support, alimony, inheritance,
gifts, financial support; ADC, AFDC, food stamps, SSI, welfare, other public
assistance; retirement, annuity, pension, disability, survivor benefits;
and other income.

Wealth: In this study, wealth is defined as
the sum of net worth and human wealth. Net worth is the value of assets minus
liabilities. Human wealth is calculated as a present value of cumulative
life time non-investment income (formula: Cissel, Cissel & Flaspohler,
1990).

Wealth = Net Worth + Human Wealth (1)

Net Worth = Assets – Liabilities (2)

Liabilities: Total real estate debt (house mortgage
plus other property mortgages) + total other debt (consumer debt plus other
debt.)

Assets: Total paper assets (sum of stocks and
mutual funds, bonds, checking and savings accounts, IRA and Keogh accounts,
money market accounts and CDs, profit sharing and thrift accounts, cash value
of life insurance, and other financial assets, plus total real assets (sum
of the current market value of the home, other properties, businesses, and
vehicles.)

Human Wealth

For people who were not retired yet or whose expected
retirement age is accurately reported or who have positive value of gross
present value of social security and pensions,

HW =[NI*{1-(1+r)(-a)}/r]+PP+PS (3)

where

NI: non-investment income until retirement

a: period between current age and expected retirement
age (working years)

PP: gross present value of pensions (estimated by SCF)

PS: gross present value of Social security (estimated
by SCF)

r: real interest rate

For people who were already retired or whose expected
retirement age is uncertain,

HW = PP+PS (4)

For people who were not retired and who have no gross
present value of Social Security and pensions,

HW = NI*{1-(1+r)-b}/r (5)

where b: life expectancy

The real interest was estimated using the nominal rate
of 10.85% (the rate on 1983 long-term U.S. government bond rate) and inflation
rate of 4%, which are the rates used by SCF. Therefore, the real interest
rate for the analysis is 6.59% which is ((1.1085/1.04)-1). No information
is reported as to whether the SCF used different life expectancy by race
or sex (Avery & Elliehausen, 1988).

Non-Investment Income: Sum of income in wages
and salary; income from a professional practice, business, or farm; workers
or unemployment compensation income; ADC, AFDC, food stamps, SSI, welfare,
other public assistance; disability, survivor benefits; and other income.

Non-investment income was measured by the average of
1982, 1983, 1984, and 1985 of non-investment incomes. Since non-investment
incomes of 1983, 1984 and 1985 were not available, each year’s non-investment
income was estimated by each year’s total income multiplied by proportion
of non-investment income to total income of 1982. It was assumed that the
household’s real non-investment income until retirement would be the same
as the average annual income from 1982 to 1985.

Life Expectancy

Life expectancy was determined by age, gender, marital
status of the householder and race. Asians and non-black Hispanics were
assumed to have the same life expectancy as Whites (American Council of Life
Insurance, 1986). For unmarried householders, estimates of individual
life expectancy by gender, age and race were used (Statistical Abstract of
the United States, 1992). For married couples, the approximate joint life
expectancy was calculated by adding 5 years to the life expectancy of the
household head This method was based on the fact that the State Teachers
Retirement System of Ohio sets the pension of a joint annuity (with 100%
pension going to a surviving spouse) at approximately the same level as a
single pension for a recipient 5 years younger..

Investment Assets: Rather than identifying specific
assets as investment assets, it was assumed that the portion of financial
assets in excess of three months of average income (1982-1985) represented
investment assets, or the investment portfolio.


Results

Sample Distribution

Table 1 shows the sample distribution of variables that
were involved in estimating wealth. Except for age, life expectancy, and
age of expected death, the mean values are higher than median values. The
mean of life expectancy, 35 years, is a reasonable value according to the
mean value of expected death (81 years) and mean of age (46 years). A person
at the mean value of age, 46 years old at the time of interview, would expect
to die at the age of 81 years, so the value of life expectancy would equal
35 years (Table 1).

The median level of financial assets (Table 1), $6,500,
is relatively low, especially considering it include retirement savings.
The 75th percentile level for financial assets, $27,370, represents the
dollar level for which 75% of the households fall below. Thus, 75% of the
households had less than $27,370 in financial assets. The median level of
net worth was $48,100, but 25% of the households had a level of net worth
of $11,840 or less. The median level of human wealth was $177,150. The
median level of total wealth (net worth plus human wealth) was $276,100.
(It is in general not the case that the sum of medians equal the
median of the sum, as the median will represent different households, so
no addition or division should be attempted for different variables in Table
1).

The median level of the ratio of financial assets to
total wealth was 1.3%, thus, for half of the households, financial assets
represented less than 1.3% of total wealth (Table 1). The 75th percentile
of this ratio was 5.7%, thus, for 75% of the households, financial assets
represented less than 5.7% of total wealth. The 90th percentile of this
ratio was 17.4%. Therefore, only for a small proportion of U.S. households
did financial assets represent a high percent of total wealth.

This conclusion is even stronger if investment portfolios
as a percent of total wealth are considered. Investment assets were defined
as the amount by which financial assets exceeded three months of income.
The ratio of investment assets to total wealth was 3% at the 75th percentile.
Thus, for a large majority of households, what might be considered the investment
portfolio represents a tiny percent of total wealth.

The elderly are more likely than the general population
to have financial assets representing a high proportion of total wealth.
Table 2 shows the distribution of total wealth and the ratios of financial
assets and investments assets to total wealth for households headed by someone
age 65 or older. The median level of wealth was $183,790. The median level
of the ratio of financial assets to total wealth was 7.0%, compared to 1.3%
for all households. The 75th percentile for the ratio was 19.8% for the
elderly, compared to 5.7% for all households. The 90th percentile for the
ratio was 40.2% for the elderly, compared to 17.4% for all households.

The median level of the ratio of investment assets to
total wealth was 5.1%, compared to 0.0% for all households. The 75th percentile
of this ratio was 17.2%, compared to 3.0% for all households. The 90th percentile
for the ratio was 37.4% for the elderly, compared to 13.8% for all households.


Conclusions

Summary

A conservative estimate of human wealth was used to calculate
total wealth of households. The ratio of financial assets to total wealth
is small for most U.S. households. Fluctuations in financial wealth represent
very small differences in the total wealth of most households. Therefore,
investments in stocks may be rational for almost all non-elderly households,
to the extent they can commit money to retirement and other long term goals
(Hanna & Chen, 1995).

Relatively few households have enough financial assets
to merit active management of investments. Only 10% of all households have
$73,700 or more of financial assets. Therefore, the need for financial education
is obvious.

If the stocks and other financial assets fluctuate substantially,
there may be little impact on the total wealth of most U.S. households.
A stock market decrease of 20% may represent a loss of only less than 1%
for 75% of households, and less than 0.2% for most households.


Table 1

Distribution of Wealth, Financial Assets, Ratio of Financial
Assets to Wealth, and Related Variables














Percentiles
Variables Mean 10% 25% 50% 75% 90%
Average Annual Income, 1982-1985 33,716 7,975 15,060 25,850 40,000 57,900
Non-Investment Income, 1982-1985 26,285 5,905 11,395 20,640 32,550 46,700
Financial Assets 38,171 125 975 6,500 23,370 73,700
Net Worth 147,875 1,495 11,840 48,100 117,200 279,500
Human Wealth 251,664 38,100 75,850 177,150 358,400 538,500
Total Wealth (including Human Wealth) 399,395 68,600 143,650 276,100 481,500 725,500
Ratio of Financial Assets to Total Wealth 0.0644 0.0004 0.0024 0.0133 0.0569 0.1740
Ratio of Investment Assets to Total Wealth 0.0489 0.0000 0.0000 0.0000 0.0301 0.1375
Age (years) 46 26 34 46 61 70
Life Expectancy (years) 35 16 23 36 46 52
Age of Expected Death (years) 81 76 80 81 83 86

*Investment assets defined as amount of financial assets
beyond 3 months income (average of 4 years)

All figures reported are weighted; n=2,691



Table 2

Distribution of Wealth, Financial Assets and Ratio of Financial Assets
to Wealth, for Households with Reference Person Age 65 and Over






Percentiles
Variables Mean 10% 25% 50% 75% 90%
Total Wealth (including Human Wealth) 366,892 49,325 98,900 183,790 336,800 632,400
Ratio of Financial Assets to Total Wealth 0.1401 0.0005 0.0104 0.0701 0.1975 0.4016
Ratio of Investment Assets to Total Wealth 0.1224 0.0000 0.0000 0.0512 0.1715 0.3735


*Investment assets defined as amount of financial assets beyond 3 months
income (average of 4 years)


Implications for Future Research

This analysis should be repeated for more recent Surveys
of Consumer Finance. Estimates of human wealth based on only one year of
household income data should be made cautiously, however, as some households
may have abnormally low income in a particular year. Some method of calculating
future household income would improve the estimate of human wealth. In this
article, household income was assumed to remain at the real level of the
average annual income of 1982-1985, until the planned retirement age.

The appropriate interest rate to use to estimate human
wealth is the real cost of capital for each household. For some households,
the after-tax real rate of return on financial assets may be very low (e.g.,
2%), and thus the real interest rate (6.59%) used to estimate human wealth
may have produced an underestimate, perhaps by 50% for some young households.
For households with low levels of financial assets, the real interest rate
on credit may be the appropriate interest rate to use. For some households
with poor credit ratings, an interest rate of 30% or more might be appropriate
for estimating human wealth, which would imply that the 6.59% rate used in
this study would produce a substantial overestimate of human wealth.

It might be appropriate to use different interest rates
for calculating the present value of non-investment income. In particular,
using a different rate depending on the portfolio composition of the household
might be appropriate. For instance, for a household with financial assets
amounting to a month’s income or more, the estimated after-tax real rate
of return on the assets might be the best rate to use. For a household with
very low or negative levels of net worth, the real interest rate on a credit
card might the correct rate to use. If this method were used, the estimates
of human wealth would tend to be lower for young households and higher for
older households.

The estimates of human wealth and total wealth presented
in this article are conservative, because of the high interest rate used
to calculate present value, and also because future income was assumed to
remain constant in real terms until retirement. Therefore, actual human wealth
of households is higher than the estimates presented, and the bias is higher
for younger households than for older households. This bias strengthens
the major conclusion: because financial assets represent such a small portion
of total wealth for most non-elderly households, investments for long-term
goals such as retirement should be primarily in stocks.

Implications for Investing

For households who rent and desire to buy a home, investing
in stocks would probably be inappropriate for that saving goal, because
the time horizon may be short. For households who have low levels of liquid
assets, and therefore, are not prepared for emergencies (e.g., Chang &
Huston, 1995), even investing in stocks for a retirement account may present
problems. However, for households who can be confident that they will not
need the funds in a retirement account for at least 5 years, investing all
or most of their retirement fund contributions in stocks may be rational.


References

American Council of Life Insurance. (1986). Life Insurance Fact
Book
.

Arrow, K. J. (1971). Essays in the theory of risk-bearing. Chicago:
Markham Publishing Co.

Avery, R. B. & Elliehausen, G. E. (1988). 1983 Survey of Consumer
Finances: Technical manual and codebook
(Rev. ed.). Washington, DC:
Board of Governors of the Federal Reserve System.

Avery, R. B. & Elliehausen, G. E. (1986, March). Financial characteristics
of high-income families. Federal Reserve Bulletin, 163-177.

Avery, R. B., Elliehausen, G. E., Canner, G. B. & Gustafson, T.
A. (1984a). Survey of Consumer Finances, 1983. Federal Reserve Bulletin,
70
, 679-692.

Avery, R. B., Elliehausen, G. E., Canner, G. B. & Gustafson, T.
A. (1984b). Survey of Consumer Finances, 1983: A second report. Federal
Reserve Bulletin, 70
, 857-868.

Chang, Y. R. & Huston, S. J. (1995). Patterns of adequate household
emergency fund holdings: A comparison of households in 1983 and 1986.
Financial Counseling and Planning
. 6, 119-128.

Cissel, R., Cissel, H. & Flaspohler, D. (1990). Mathematics
of finance
(8th ed.). Boston: Houghton Mifflin.

Friend, I. & Blume, M. E. (1975). The asset structure of individual
portfolios and some implications for utility functions. The Journal of
Finance
, 30(2), 585-603.

Graham, J. W. & Webb, R. H. (1979). Stocks and depreciation of
human capital: New evidence from a present-value perspective. The Review
of Income and Wealth, 25
(2), 209-224.

Hanna, S. & Chen, P. (1995). Optimal portfolios: An expected
utility/simulation approach. Proceedings of the Academy of Financial
Services
.

Hanna, S. (1988). Risk versus return for risk averse investors:
CREF versus TIAA. Proceedings of the Association for Financial Counseling
and Planning Education.

Malkiel, B. G. (1990). A random walk down Wall Street. New York:
W.W. Norton & Co.

Note — as published, this article was on pages 147-152.