Large U.S. Stocks Large U.S. stocks have been the least volatile stocks in
the world. Weā™ve talked a bit about large U.S. stocks earlier in this chapter,
but now letā™s compare their returns with those of bonds.
There hasnā™t been a 20-year interval in the last 70 years when bonds
provided a higher rate of return than stocks. Such historical results also
square with good old common sense. After all, unless we had a rational ex-
pectation that stocks would give us a materially higher return, why would
we buy a stock whose future price could be anything, high or low, when we
could buy a bond that we can redeem at par value (usually $1,000) X years
from now? For the long term, unless investors en masse are irrational, we
have to expect a materially higher return from stocks than from bonds.
The key question isā”how much higher?
Over the 79 years through 2004, the S&P 500ā”a measure mainly of
the largest stocksā”returned 41/2 percentage points per year more than
bonds and over 7 points more than inļ¬‚ation.4 My guess is that in the years
ahead, both the real return on stocks and the return differential between
stocks and bonds will be distinctly smaller than they have been heretofore.
Even so, the return differential over the long term should still be material.
Stocks of any size are often arbitrarily divided between categories of
growth and value stocks. No one quite agrees on the precise quantitative
deļ¬nitions of āgrowthā and āvalue,ā but in general, stocks with higher
earnings growth rates are categorized as growth and those with low price-
to-book-value ratios are categorized as value.
There are multiple indexes of growth and value stocks, and while each
is a little different, all show that growth and value tend to move in some-
what different cycles. Unless we recognize this, we might regard all man-
agers of growth stocks as brilliant during some intervals, and as dunces
during other intervals (and vice versa for value managers). Obviously, we
must understand a managerā™s style to evaluate him properly.
For diversiļ¬cation in our asset allocation, we should probably have
both growth and value managers.
Small U.S. Stocks Step one is to deļ¬ne small stocks. The Russell 2000 in-
dex deļ¬nes them by market capitalization, as the 2,000 largest U.S. stocks
after eliminating the largest 1,000 stocks, rebalanced annually.5 The largest
1,000 U.S. stocks (measured by the Russell 1000) account for some 90%
of the total market capitalization of U.S. stocks; the Russell 2000 for an-
other 81/2%; and some 8,000 tinier stocks (which we might refer to as mi-
crocaps) account for the ļ¬nal 11/2%.
Small U.S. stocks are often treated as a separate asset class, because
over the years they have at times had quite different returns from large
These ļ¬gures overstate the advantage of stocks to some extent. In 1926, at the be-
ginning of the 78-year interval, stocks sold at prices that provided an average divi-
dend yield of more than 5%. Today, prices have risen so high that the average
dividend yield is scarcely 2%. That decline couldnā™t happen again from todayā™s divi-
dend yield. If the marketā™s dividend yield and price/earnings ratio had remained un-
changed over the years (and that might be the best we can expect going forward from
today), then the 78-year annual return on the S&P 500 would have been well below
10%ā”barely 3% more than the return on bonds and about 51/2% above inļ¬‚ation.
As of May, 2004, the Russell organization reconstituted the Russell 2000 index (as it
does annually) to include companies with market caps between $176 and $1,600
million. But, of course, by the time the reconstituted index was put in place on June
25, market price changes had materially widened the range of market caps.
64 ASSET ALLOCATION
stocks. From 1979, when the Russell 2000 index was created, through
2004, its annual rate of return lagged that of the S&P 500 by 0.4%/year.
Over the 78 years 1926 through 2003, Ibbotson data shows that small
stocks (deļ¬ned differently from and materially smaller than the Russell
2000) returned more than 2 percentage points per year more than the S&P
500. But much of this excess return was earned in a single 10-year interval,
under conditions unlikely to be repeated. Based on the Ibbotson data, wit-
ness the cycles shown in Table 4.1.
What expectation is most rational for us to make going forward?
Individual small stocks have been a lot more volatile than large stocks,
and even a broad portfolio of small stocks like the Russell 2000 has aver-
aged several percentage points more in annual volatility than large stocks.
The correlation has been low enough, however, that a mixture of, say, 20%
small stocks with the balance in large stocks would have had a slightly
lower volatility than a portfolio of large stocks alone.
Because investment analysts donā™t follow small stocks as widely as
larger stocks, a good manager of small stocks should be able to add more
value to an index of small stocks than a good manager of large stocks can
add to an index of large stocks. The ļ¬‚ip side, of course, is that a below-av-
erage manager of small stocks is more likely to get bagged!
As with large stocks, the use of both growth and value small-stock
managers can add useful diversiļ¬cation.
Possibly a separate category is microcap stocksā”stocks smaller than
those in the Russell 2000 index. It is hard to get much money into micro-
cap stocks because they are simply too small. To the extent it is possible,
however, microcap stocks act as a further diversifying element, since they
behave somewhat differently from Russell 2000 stocks. They have much
higher volatility and transaction costs, but if we can stand the volatility, a
strong manager can earn good returns from them.
TABLE 4.1 Annual Rates of Return for Stocks over Different Time Periods
No. of Advantage of
Years Interval S&P 500 Small Stocks Small Stocks
34 1926ā“59 10.3% 10.5% + 0.2 points
8 1960ā“67 9.6 19.5 + 9.9
6 1968ā“73 3.5 ā“5.6 ā“ 9.1
10 1974ā“83 10.6 28.4 +17.8
15 1984ā“98 17.9 11.0 ā“ 6.9
5 1999ā“2003 ā“0.6 16.3 +16.9
Mid-Cap U.S. Stocks Some investment funds include midcap stocks as an
additional asset category. There are multiple deļ¬nitions of mid-cap stocks,
but basically they are ones larger than those in the Russell 2000 index, per-
haps as large as $10 to $15 billion in market cap. They do have somewhat
different characteristics from large or small stocks, and theyā™re not quite as
volatile as small stocks. But, relative to large stocks, they donā™t provide as
much diversiļ¬cation beneļ¬t as small stocks.
Real Estate Investment Trusts (REITs) REITs are a form of common stock.
They trade on stock exchanges but, unlike regular corporations, they pay
no income tax. They pass their income tax liability on to their sharehold-
ers, which is just ļ¬ne for a taxfree investment fund. To qualify for such tax
treatment, an REIT must meet speciļ¬c legal criteria, such as earning 75%
of its gross income from rents or mortgage interest, and distributing 90%
of each yearā™s taxable income to shareholders.
REITs have been around for more than 30 years, but for a long time
they were small in number, with a sizable proportion devoted to investing
in high-risk construction lending. The number of REITs devoted to owning
properties has mushroomed since the early 1990s, and their aggregate
value has gone up more than 20 times. Yet today, they may own little more
than 5% of total commercial real estate in the United States. Some industry
observers expect that eventually REITs will own the majority of commer-
cial properties in the country. Some REITs are private, but most are pub-
Why distinguish REITs from any other U.S. common stock? Although
to some extent they share in stock market cycles, their returns are driven
by commercial real estate values, whose cycles have a low correlation with
those of the stock market.
Should we rely on our regular common stock managers to invest in RE-
ITs, or should we hire a specialist in REITs? Because investing in marketable
REITs requires a lot more real estate savvy than most investment managers
have, and requires lots more research about the particular properties owned
by each individual REIT, I favor an REIT specialist manager. And because
REITs may have only a .4 correlation with the S&P 500, I favor treatment
of REITs as a separate asset class in our Policy Asset Allocation.
Nonā“U.S. Stocks Unlike Scottish investors, who have been global investors
for nearly 200 years, U.S. investors have until relatively recent years been
among the more provincial. They implicitly assumed that appropriate in-
vestment opportunities began and ended in the United States even though
the total value of U.S. stocks has for much of the last 20 years been below
66 ASSET ALLOCATION
half of the total value of all stocks in the world (although by year-end
2003 U.S. stocks again comprised about half of the worldā™s market cap).
Some U.S. investors have now moved 20% or more of their equity
portfolios outside the United Statesā”and for good long-term reasons. It is
hard to argue that expected returns from stocks in the developed countries
of the world should be materially different from those in the United States.
For the full 34-year interval 1971ā“2004 since the MSCI EAFE index was
started, their returns were about the same. But differences over shorter in-
tervals have been dramatic!
For example, during the four years from 1985 to 1988 stocks of the
developed countries outside the U.S. outperformed U.S. stocks by 26 per-
centage points per year. Then for the next 13 years (1989ā“2001) U.S.
stocks outperformed by 11 points per year. We donā™t need to calculate a
correlation coefļ¬cient to see that U.S. and nonā“U.S. stocks have provided
real diversiļ¬cation for one another.
U.S. investors often worry that foreign currencies will lose value rela-
tive to the dollar. On the other hand, foreign currencies can be an opportu-
nity as well as a risk. Overall, from the beginning to the end of the 33 years
from 1970 to 2003, changes in foreign exchange values had very little im-
pact on investment returnsā”despite substantial impact during shorter in-
If investors are unduly worried about foreign exchange risk, they can
always hedge that risk through the purchase of foreign exchange futures. I
am not much inclined, however, to spend my money on such an āinsurance
policyā unless a very large percentage of our portfolio is at foreign-ex-
change risk. In any case, foreign exchange risk is not a reason to avoid con-
sidering nonā“U.S. investments.
David Swensen of Yale refers to diversiļ¬cation as a āfree lunch.ā6 In a
simplistic way, nonā“U.S. stocks can be used to illustrate the point. Even
though the nonā“U.S. EAFE stock index had a slightly lower return and was
more volatile than the S&P 500 during the 30-year interval from 1970 to
1999, a portfolio consisting of 40% EAFE and 60% S&P would have pro-
vided a slightly higher return than the S&P 500 alone, and at a volatility
nearly 2 percentage points per year lower.7 A modest allocation to highly
volatile emerging markets stocks would have made this simple portfolio
more efļ¬cient yet.
Swensen, Pioneering Portfolio Management (The Free Press, 2000), p. 67.
Source: Ibbotson Associates.
Small Nonā“U.S. Stocks Small stocks outside the U.S. offer further diversiļ¬-
cation value. Just as in the United States, their returns have had materially
different patterns from large stocks. Likewise, small stocks outside the U.S.
have had lengthy intervals of materially outperforming and underperform-
ing large stocks.
From country to country, the correlations among small-stock returns
are substantially lower than the correlations among large-stock returns.
Within each country, of course, small stocks are considerably more volatile
than large stocks. But because of the low country correlations, the MSCI
EAFE Small Stock index is only marginally more volatile than the large
stock MSCI EAFE index.
The comments earlier in this chapter about growth and value relative
to U.S. stocks apply equally to nonā“U.S. stocks, both large and small.
Emerging Markets Stocks With the rapid spread of private enterprise
among the less-developed countries of the world, especially since the end of
the Cold War, a new asset class has come about. Stocks of Singapore, Hong
Kong, and the less-developed countries now account for some 10% of the
value of the worldā™s common stocks.
Over the last dozen years or more, the GNP of a number of those
countries has been growing at a rate of 5% to 10% per year, compared
with 2% to 4% for the developed economies. Hence, there is reason to
expect companies in the emerging markets to grow faster and their
stocks to provide a greater return than in the developed worldā”espe-
cially if the accounting and shareholder orientation of those companies
continue to improve.
But what about their volatility? It is not at all unusual to see the aggre-
gate return on stocks in a particular developing country go up by 100% in
a year or down by 50%ā”or more. If we could invest only in a single devel-
oping country, the risk would be tremendous. But today we can invest in
some 60 developing countries, and over time their returns have had a rela-
tively low correlation with one another. Stocks in one country may be way
up when those in another country go into a tailspin. Indexes of emerging
markets stocks are composed of 25 to 30 different countries, and these di-
versiļ¬ed indexesā”while still a lot more volatile than those of the devel-
oped worldā”still have low enough volatility to be fruitfully considered by
Emerging markets stock indexes provide a good example of the advan-
tage of low correlations. On average, the volatility of the stock market of
individual developing countries may be well over 40 percentage points per
68 ASSET ALLOCATION
year, but taking all countries together, the volatility of the emerging mar-
kets has been in the range of 25 to 30 percentage points.
Tactical Asset Allocation (TAA)
In the mid-1980s, a number of managers developed complex computer
programs that moved assets unemotionally back and forth between stock
and bond index funds, depending on which seemed to their quantitative
models the most attractively valued at the time. These Tactical Asset Allo-
cation programs are now sometimes invested entirely through index fu-
tures, because futures are most cost-efļ¬cient. And the models have now
become increasingly sophisticated, using futures for different sizes of U.S.
stocks, and futures for stock and bond markets of more than 15 countries
outside the United States.
Because of their quantitative models, TAA managers can readily tailor
their products to whatever mandateā”or benchmarkā”a client might prefer,
such as the MSCI World stock index, or 50% S&P 500/50% Lehman Ag-
gregate, or any other index or combination of indexes. What counts, of
course, is their risk-adjusted returns relative to their benchmarks.
So how have they done? Well, there are differences among TAA man-
agers, of course, but on average they have not tended to outperform their
benchmarks nor to keep their volatility measurably below their benchmarks.
If we want at least one of our accounts to vary its asset allocation tac-
tically, a TAA account may be our best choice, if we are able to select a
TAA manager who in the future can achieve above-average returns. Use of
an experienced TAA manager may at least be a better way to vary our asset
allocation tactically than doing it intuitively based on our own predilec-
tions or those of our adviser.
Alternative Asset Classes
In the minds of many investors, the concept of asset allocation ends with
traditional marketable securities. Perhaps they might identify real estate as
another viable asset class. But we can strengthen our portfolio materially
with additional asset classes, such as:
Start-up venture capital funds
Leveraged buyout funds
Corporate buy-in funds
Oil and gas properties
Market neutral funds, such merger and acquisition arbitrage and con-
Hedge funds (funds that may be short some common stocks while also
holding a long equity portfolio)
Commodity Futures (Including Foreign Exchange)
We shall discuss these asset classes in some detail in Chapter 5. But for
now, letā™s just consider how we go about estimating future returns, volatil-
ity, and correlations.
With arbitrage programs or hedge funds, where the skill of the man-
ager is more important than the asset class itself, the particular managerā™s
historic returns may be useful indicators. Asset classes that add great diver-
siļ¬cation beneļ¬t to a portfolio are those that are market neutralā”whose
correlation with the stock market is close to zero. Many arbitrage strate-
gies get down to correlations of 0.3 or less. Also close to zero correlation
are many long/short common stock funds whose short positions are equal
in value to their long positions.
When it comes to illiquid investments, such as real estate or venture
capital, estimating their volatility and correlation with other asset classes is
harder yet. The āmarketā values at which we carry these assets on our
books are much less meaningful because each asset is unique. No identical
asset is being bought and sold every day in the marketplace. Valuations are
established by (1) judgmental appraisals, as with real estate, or (2) the
price at which the last shares of a stock were sold, even if that was two
years ago, or (3) the book value of the investment, which is the usual valu-
ation of a private investment in which there have been no transactions,
perhaps for years, or (4) a written-down value if the manager has strong
evidence that an investmentā™s value has been impaired.
Given these approaches to valuation, illiquid investments often appear
to have materially less volatility than common stocks. But note the empha-
sis on the word āappear.ā The price at which a particular investment could
be sold certainly goes up and down each quarterā”undoubtedly with great
volatility for a start-up venture, for exampleā”even though its reported
value is kept unchanged quarter after quarter.
Which volatility of an illiquid investment should we assessā”the
70 ASSET ALLOCATION
volatility of its reported returns or the estimated volatility of its underly-
In the reports we make on our investment fund, we have to base our
returns on reported valuations. But letā™s stop and consider two invest-
ments: a marketable stock and a start-up venture capital stock. Letā™s say
each is sold after seven years and each returned 16% per year over that in-
terval. Which was the more volatile?
The marketable stock had lots of ups and downs, whereas the venture
capital stock was kept at book value much of the time. Was the marketable
stock more volatile? If our time horizon for measuring volatility is seven
years, we could say they had the same volatility. But there was obviously a
lot greater uncertainty as to the seven-year return on the venture capital
stock than on the marketable stock. The underlying annual volatility of the
venture capital stock had to have been a lot higher.
We could make a good case that our expected volatility of an illiquid
investment should reļ¬‚ect the innate uncertainty in its returnā”that is, its
Next, how does one assess correlations between liquid and illiquid in-
vestments? We should study whatever data we have, but ultimately weā™ll
have to go with an educated guess.
PUTTING IT ALL TOGETHER
Perhaps with the help of our adviser, we have developed a diverse range of
assumptions for the return, volatility, and correlation for each of the asset
classes we are going to consider. What do we do now?
Why Not 100 Percent Equities
A question that has long bugged me is: Why not 100% equities? If we
agree that we should be very long-term oriented, and if we are convinced
that over any 20-year interval stocks should outperform bonds, then why
not target 100% stocks?
Well, the roller coaster ride of the stock market could be very upset-
ting. The worst eventuality would be if, at the bottom of a bear market, the
stomach of some future investment committee weakened and the commit-
tee reduced the allocation to common stocks at that time. So how can we
ease the roller coaster ride a little but not impair expected returns unduly?
At this point, I would like to introduce my personal deļ¬nition of the
Putting It All Together
term āequities.ā By equities I mean all investments whose expected returns
are generally as high as, or higher than, common stocks. I am big on diver-
siļ¬cation and believe in reducing the volatility of our aggregate portfolio
through diversiļ¬cation. But I am convinced that strong diversiļ¬cation can
be achieved without resorting to large allocations to assets whose expected
returns we believe are materially below that of equitiesā”such as tradi-
tional ļ¬xed income.
Judicious use of ļ¬xed income might let us boost our aggregate return
per unit of risk, but unless we can leverage our overall portfolio (and thatā™s
tough to do), we canā™t spend risk-adjusted returns. If we are truly long-
term oriented, why not accept a little higher volatility in exchange for
Before relegating ļ¬xed income to oblivion, letā™s ask what purpose ļ¬xed
income should serve in a portfolio, and how best we can fulļ¬ll that pur-
pose. For an endowment fund or foundation, traditional investment-grade
ļ¬xed income serves two key purposes:
1. Traditional ļ¬xed income lowers the expected volatility of the portfo-
lio. This is the most common purpose of ļ¬xed income, and the purpose
I would hope to achieve instead through the use of diverse asset classes
that have materially higher expected returns than ļ¬xed income.
2. Fixed income gives the portfolio needed strength whenever interest
rates decline and stock prices decline at the same time, as in a reces-
sionā”or heaven forbid, in a depression. No asset class serves this func-
tion as well as ļ¬xed income.
So maybe there is a bona ļ¬de rationale for ļ¬xed income, after all. If we
must use ļ¬xed income with a lower expected return to fulļ¬ll purpose 2
above, how can we do it most efļ¬ciently? The answer, it seems to me, is in
long-duration high-quality bonds. This approach will (1) give us the maxi-