``For the equity markets overall, returns of about 5
percent to 7 percent annually above inflation. This range has been
the same for long-term investor for over 100 years and, I suspect,
will be the same for many years to come.'' ---Scott Evans
(Executive Vice President, CREF Investments) in `Stock Investing: Take
the Long-Term View,' TIAA-CREF Participant, May 2003.
Over time, the economy grows. If one were to invest a certain
amount into a reasonably large stock index (and having thus bought a
fraction of the part of the economy represented by publicly traded
companies) and if one were to believe that the system is fair, one
would expect to see real growth that is comparable to the real growth
of the economy (in the long term). However, if the system is not fair
(for example, if stocks are diluted on a yearly basis by grants of
options), one would have to settle for lesser returns.
Inflation Rates 1913-2002
Any money amount is a nominal parameter. For our purposes inflation
is of little significance: we will simply normalize all the data so
that they are presented in dollars from the same year and are,
therefore, directly comparable. There is, however, one aspect of the
U.S. tax system that makes inflation rates important: capital gains
tax is paid on the amount of nominal increase in principal due to
inflation.
In the following figure, CPI for 1913-2002 (1913 is the earliest
year from which CPI data are available) is fitted with an exponential
curve. In 1913-2002, annualized inflation was 4.240%.
Economy Real Growth
The following figure displays the Gross Domestic Product in 1996
chained dollars and a best-fitting exponential curve. The economy has
been growing very steadily at average 3.331%.
Stock Real Returns 1913-2002
In the following figure, Dow Jones Industrial Average adjusted to 2002
dollars is plotted and fitted with an exponential curve. The real
rate of DJIA returns for years 1913-2002 is 2.738%. (This does not
take into account capital gains tax paid on the nominal increase of
principal due to inflation.)
Notice how if one were to invest in DJIA in 1916, one would have
the same real amount in 1982 (0% real returns); that's 66 years
with no growth. (Of course, the inflation was considerable between
these dates; if one pulled out in 1982, one would have to pay
capital gains tax on 89% of one's principal making the investment
a net loss.)
Stock Real Returns 1913-1991
Many believe that what happened in the nineties was an anomaly. Some
ascribe it to speculation, some to productivity growth related to
computers, some to peace dividend, etc. If one were to agree with
these observers it would become interesting to consider the stock
market up to year 1991. A subset of data from the previous figure is
presented here along with a best-fitting exponential curve. The real
rate of DJIA returns for years 1913-1991 is 1.269%.
Real Estate Real Returns 1975-2002
Real estate is a popular portfolio diversification asset. In fact,
many households keep most of their net worth in their house. House
Price Index staring from 1975 (the first year for which it is
available), adjusted to 2002 dollars is plotted below along with a
best-fitting exponential curve. The real rate of real estate returns
1975-2002 is 0.635%.
One can also see that the price is somewhat volatile and is not
approximated very well by an exponential curve in the interval under
consideration. At the same time, the data is continuous and the
best-fitting exponential curve does not match closely the end values.
This is to be expected as the beginning of the interval was during a
bust while the end is during a boom. But even if one were to assume
that it is possible to time the real estate market (buy at the bottom
and sell at the top), the change between 1975 and 2002 HPI is at an
annual rate of 1.082% (note that this is a best-case scenario -- the
0.635% figure better describes average returns).
Population Growth 1913-1991
It is instructive to compare stock real returns in 1913-1991 with
population growth. Best-fitting exponential curve of population
growth is 1.257% annually.
Inflation-linked Bonds and Notes (Real) Returns
These are real guaranteed returns of Treasury Inflation-Protected
Securities (TIPS) auctioned recently:
Issue Date
Maturity Date
Yield
CUSIP
2003-15-01
2012-15-07
2.340%
912828AF7
2002-15-10
2012-15-07
2.260%
912828AF7
2002-15-07
2012-15-07
3.099%
912828AF7
2002-15-01
2012-15-01
3.480%
9128277J5
2001-15-10
2032-15-04
3.465%
912810FQ6
Gold Prices
Gold (Au) is a mostly useless metal that has historically been used as
a money standard. It is no longer used in this way, but its price
remains significantly higher than would be merited by its utility.
The collective folly of investors who buy gold must play a role in
keeping its price up. The price of gold used to be heavily regulated,
but floats freely since 1968 or 1974 (depending on how you count). In
any case, since 1974 regulation has played only a minor role in
forming the price of gold. There is no reason to expect steady
returns from gold, and indeed, gold price does not appear to show a
specific trend.
Comparison of Long-Terms Rates of Real Returns
Real GDP growth in 1929-2002 is 3.331%. (If one were to use this
to estimate stock growth, dilution must be subtracted; it is typically
within 1% to 3% range.)
Real return from DJIA 1913-2002 is 2.738%.
Real return from DJIA 1913-1991 is 1.269% (notably, population
growth is 1.257%).
Real return from HPI 1975-2002 is 0.635%.
Real guaranteed return from TIPS (locked in recently for 9 to 30
years): 2.260% to 3.480%.
TIPS currently offers better long-term return than DJIA. In addition,
TIPS returns are guaranteed, while stock investment might lead to
principal losses even over the long term.
Sources of Data and Methodology
To avoid bias related to selection (conscious or subconscious,
intentional or unintentional) of time periods that are a priori
known to have certain properties, the time periods were selected
simply based on availability of data from a fairly official source.
For example, BLS CPI data goes back to 1913, therefore, nothing
earlier than 1913 is considered.
CPI and GDP data are obtained from Bureau of Labor Statistics.
DJIA data are from Dow Jones and Company. The file of historic DJIA
provided by Dow Jones and Company lacks data for year 1997; a value
had to be added for year 1997 that comes from a different Dow Jones
and Company file. TIPS yields are obtained from the U.S. Treasury.
Population estimates are from the Population Estimates Program of the
Population Division of the U.S. Census Bureau (revised date: June 28,
2000). Gold price data are obtained from `What Was the Price of Gold
Then?', a report published by Economy History Services. House Price
Index (HPI) data are obtained from the Office of Federal Housing
Enterprise Oversight (OFHEO); HPI data is published for quarters only,
so in order to obtain a value for a year the values for its four
quarters were averaged; USA HPI data were used (rather than data for a
specific region).
Data plotted using gnuplot. Curves are exponential best fit
(f(x) = a*exp(b*x));
a and b were normalized to the same order of
magnitude as recommended in gnuplot fitting tips. Percentages
reported are exp(b)-1.
One can also see that the price is somewhat volatile and is not
approximated very well by an exponential curve in the interval under
consideration. At the same time, the data is continuous and the
best-fitting exponential curve does not match closely the end values.
This is to be expected as the beginning of the interval was during a
bust while the end is during a boom. But even if one were to assume
that it is possible to time the real estate market (buy at the bottom
and sell at the top), the change between 1975 and 2002 HPI is at an
annual rate of 1.082% (note that this is a best-case scenario -- the
0.635% figure better describes average returns).
