izationā”its number of shares outstanding (or available for trading) times the
price of its stock. Almost all stock indexes are cap-weighted. That is, a stock
like GE, which may have a cap of $300 billion, is weighted 3,000 times as
heavily as a micro-cap stock with a capitalization of only $100 million. There
is an elegance to cap-weighted indexes:
They reļ¬‚ect the total market value of all stocks in the index.
We can create a portfolio that contains the same stocks in the same
proportions as the index (an index fund), and if properly assembled
the portfolioā™s performance should very precisely mirror that of the in-
dex. Weā™ll talk more about index funds in Chapter 6 in the context of
selecting investment managers.
Cap-weighting is appropriate because all investors together must hold
the same aggregate value in each stock as its capitalization, and they
can hold no more aggregate value in each small stock than its capital-
ization. Although any individual investor can be different, all in-
vestors together cannot.
We havenā™t even mentioned the best-known stock index of allā”the Dow
Jones Index of 30 Industrial Stocks. Why not? After all, itā™s the market
barometer most often referred to in the press. The Dow is not a cap-weighted
index. Each stock is weighted by a factor relating to its price, unrelated to
how many shares are outstanding. It is difļ¬cult to manage an index fund of
the Dow Jones Industrials. For this reason, while the Dow serves well as a
rough measure of the performance of very large U.S. stocks, it is not a very
useful analytical tool.
The Russell 1000 includes the 1,000 largest U.S. stocks, and the Russell
2000 the 2,000 next largest.
46 SETTING INVESTMENT POLICIES
Selecting the Benchmarks
Well, given a Policy Asset Allocation composed of a range of different asset
classes, how do we go about selecting the index that will serve as our
benchmark for each asset class? The boxes on the two previous pages sug-
gest the kind of benchmarks that would be appropriate for the most com-
mon marketable securities.
Many investment funds ļ¬nesse their investment objectives by using as
a benchmark whatever rate of return is earned by their peers, with a target
standard deviation of annual investment returns no higher than the aver-
age of their peers. I understand the motivations for this, because most insti-
tutions are always looking over their shoulder to see if they are doing as
well as their peers. But I think this is not an advisable investment objective.
Framing our objectives as a function of what our peers are doing
makes us a prisoner of their investment objectives and constraints,
whether or not their objectives and constraints are optimal for us. Fac-
tors that inļ¬‚uence their investment policies may be quite different from
those that should inļ¬‚uence ours. We should set our investment objec-
tives based on our own independent thinking about the reasons for each
part of those objectives. Thatā™s because our peers are not always right,
especially for our situation. Moreover, they are often inļ¬‚uenced by con-
To illustrate differences in conventional wisdom, U.K. pension funds
historically invested 80% or more of their portfolios in common stocks,
U.S. pension funds 60% to 70%, Canadian pension funds were at one
time more like 40%, and Swiss pension funds closer to zero. (Many
Swiss pension plans preferred simply to buy annuities.) Such asset allo-
cations are inļ¬‚uenced partly by local laws, but in most cases, Company
A follows an approach because itā™s conventional wisdomā”itā™s the ap-
proach followed by peers in its country. Arenā™t all pension funds world-
wide trying to do the same thing? An optimal approach for investing
a pension fund in one country is probably pretty close to an optimal
approach in another country. Except for conventional wisdomā”herd
One who has done his own independent thinking is David Swensen,
who has led the Yale University endowment fund to achieve one of the best
long-term results among all institutional funds. Swensen āpioneered the
move away from heavy reliance on domestic marketable securities, empha-
sizing instead a collection of asset classes expected to provide equity-like
returns driven by fundamentally different underlying factors. Aside from
Preparing a Statement of Investment Policies
reducing dependence on the common factor of U.S. corporate proļ¬tability,
the asset allocation changes ultimately exposed the portfolio to a range of
less efļ¬ciently priced investment alternatives, creating a rich set of active
Hence . . . letā™s do our own independent thinking, set our own invest-
PREPARING A STATEMENT OF INVESTMENT POLICIES
Ultimately, every investment fund should prepare a written statement of in-
vestment policies. The following is an example of such a statement.
Investment Policies of XYZ Fund
Overall Objectives. Investment policies and individual decisions are to be made
for the exclusive beneļ¬t of the endowment fundā™s sponsor [or for the Pension
Planā™s participants], and any perception of conļ¬‚ict of interest is to be avoided.
Within the relevant laws, investment objectives are:
Payments. Without fail, to make every scheduled payment to the fund spon-
sor [or to Plan participants] on the date it is due.
Liquidity. To maintain enough liquid assets and other assured sources of
cash to cover projected payouts for at least the next ļ¬ve years.
Policy Asset Allocation/Benchmark Portfolio. The Fundā™s Policy Asset Alloca-
tion also serves as its Benchmark Portfolio, with each liquid asset class to be
benchmarked against a speciļ¬ed index. Over intervals of ļ¬ve years or longer, the
portfolioā™s net total return is intended to exceed that of the Benchmark Portfolio by
as much as possible without the portfolioā™s overall volatility materially exceeding
that of the Benchmark Portfolio.
Quarterly returns on the Benchmark Portfolio are to be calculated as follows:
a. The actual return on illiquid investments (targeted at __% of the portfolio)
times their actual percentage of the portfolio at the start of the quarter.
David F. Swensen, Pioneering Portfolio Management (The Free Press, 2000), p. 1.
48 SETTING INVESTMENT POLICIES
b. Index returns on liquid assets (the balance of the portfolio) weighted as
shown in Table 3.1.
The Fund should periodically review its Policy Asset Allocation to ensure that
it remains appropriate for the needs of the Fund, although it is not expected that
changes will need to be made frequently.
Because short-term ļ¬xed income securities are the lowest-return asset class
over any long-term interval, the Fund should target its holdings of these securities
at the lowest possible level commensurate with its immediate cash needs. This
generally means selling stocks and bonds ājust in timeā to meet cash needs.
New contributions to the Fund should be applied to, and payments by the
Fund withdrawn from, asset classes in such a way as to bring the Fundā™s asset al-
location toward its Policy Allocation. If these actions are insufļ¬cient to return the
portfolio to its Policy Allocation, then the Fund should make additional transac-
tions to rebalance the portfolio at least once a year.
Liquid Assets. The term āliquid assetsā includes all investments that the Fund
can convert to cash within a year, such as marketable securities, both equity and
The Fund should consider investing in all liquid asset classes in which it can
gain competency to invest, and it should base its portfolio weight in each class on
whatever combination it expects will provide optimal risk/return characteristics for
the aggregate portfolio. Where feasible, the Fund should also seek diversiļ¬cation
within asset classes. For example, in common stocks, the Fund should normally
TABLE 3.1 Benchmark Portfolio
Asset Class Benchmark Index
___% Large U.S. stocks Russell 1000 Index
___% Small U.S. stocks Russell 2000 Index
___% Real estate investment trusts NAREIT Equity Index
___% Large non-U.S. stocks MSCI World Index, ex U.S.
___% Small non-U.S. stock MSCI EAFE Small Cap Index
___% Emerging markets stocks MSCI Emerging Markets Free Index
___% Investment grade bonds, Lehman Government Corp.
10-year duration Long-Term Bond Index
___% Inļ¬‚ation-linked bonds Index Barclayā™s U.S. Inļ¬‚ation Linked Bond Index
___% High-yield bonds Chase High Yield Developed Market Index
___% Emerging markets debt J. P. Morgan Emerging Markets Bond Plus
___% Market neutral programs Treasury bills plus 4%
Preparing a Statement of Investment Policies
seek to have managers with different styles. The Fund may therefore hire multiple
specialist managers in a single asset class.
Illiquid Assets. 5 The term āilliquid assetsā includes any investment that the Fund
cannot readily convert to cash at fair market value within a year, such as partner-
ships invested in real estate, venture capital, oil & gas, and timberland.
Each new illiquid investment should be selected on an opportunistic basis
so as to improve the overall portfolio diversiļ¬cation and to enhance its return.
To subscribe to a new illiquid investment, the Fund should have a realistic
expectation that it will provide a materially higher net rate of return than a com-
parable marketable investment in order to compensate for the risk and incon-
venience inherent in illiquidity. A still higher expected return should be required
of an illiquid investment to the extent that its underlying risk is greater than that
of common stocks.
The attractiveness of a prospective illiquid investment will be enhanced by its
expected diversiļ¬cation beneļ¬t to the Fundā™s overall portfolio, that is, the extent to
which the key factors affecting its investment returns differ from those that affect
the Fundā™s other investments.
No single commitment to an illiquid investment should normally exceed
__%6 of the Fundā™s total assets. Such commitments are much smaller than
commitments the Fund typically makes to managers of liquid assets. This is
because a great diversity of private asset classes and managers, including
time diversiļ¬cation, is desirable due to the illiquid and often specialized nature
of private investments.
The success of the Fundā™s illiquid investments will depend on the extent that,
over time, its portfolio of illiquid investments either (a) exceeds the return on the
Russell 3000 Index by 3% per year or (b) achieves a net IRR of 15% per year.
Manager Selection and Retention. In every asset class, the Fundā™s goal is to
have its investments managed by the best possible investors that the Fund can
access in that asset class. Until such time that the Fundā™s investment staff can be
realistically expected to achieve world-class results in managing any particular
asset class (at least equal, net of all costs, to the best the Fund can obtain out-
side), the day-to-day portfolio management of all Fund investments shall be per-
formed by outside managers.
To achieve superior investment returns within the Fundā™s volatility con-
straints, the Fund should continuously seek investment managers and
Small endowment funds probably shouldnā™t consider illiquid assets.
Iā™d suggest inserting perhaps 0.5% for a very large fund and 2% for a relatively small fund.
50 SETTING INVESTMENT POLICIES
investment opportunities that have expected rates of return higher than those
expected from its existing managers, especially if this would reduce the Fundā™s
aggregate volatility or improve the manageability of the Fundā™s overall portfolio.
All managersā”both prospective and existingā”should be evaluated under
the following criteria:
Character. Integrity and reliability.
Investment approach. Do the assumptions and principles underlying the
managerā™s investment approach make sense to us?
Expected return. The managerā™s historic return, net of fees, overlaid by an
evaluation of the predictive value of that historic return, as well as other fac-
tors that may seem relevant in that instance and may have predictive value.
Expected impact on the Fundā™s overall volatility. Two facets:
a. Expected volatilityā”the historic volatility of the managerā™s investments
overlaid by an evaluation of the predictive value of that historic volatility,
and recognition of the historic volatility of that managerā™s asset class.
b. Expected correlation of the managerā™s volatility with the rest of the
Liquidity. How readily in the future can the account be converted to cash,
and how satisfactory is that in relation to the Fundā™s projected needs for
Control. Can our organization, with the help of its adviser, adequately moni-
tor this investment manager and its investment program?
Legal. Have all legal concerns been dealt with satisfactorily?
Managers should be selected without regard to the geographic location of
their ofļ¬ces and without regard to the nature of their ownership except as those
factors may impact the above considerations.
In each asset class, unless it is viable for the Fund to select active managers
whom it expects, with high conļ¬dence, to add meaningful excess long-term value
(net of fees and expenses) above their benchmark, then the Fund should invest
its assets in an index fund.
Criteria applicable to the selection of an index fund manager include that of
character and integrity and the managerā™s historic performance (net of fees,
taxes, and transaction costs) relative to the respective index.
This sample statement of investment policies contains a number of
concepts we may not have discussed adequately. Letā™s address them now.
The opening statement about āexclusive beneļ¬tā and the nextā”of
Preparing a Statement of Investment Policies
making all scheduled payments to the fund sponsor without failā”are a bit
of motherhood and apple pie, but I think they are so basic that any policy
statement should begin with something like them.
Liquidity requirements are necessary to meet the ļ¬rst objective. That
statement, however, serves best to remind us that we have great ļ¬‚exibility.
The minimum-liquidity requirement allows vastly more illiquid invest-
ments than are held by any investment fund I am aware of.
The goal of broad diversiļ¬cation is embodied in the range of asset
classes that are included in the policy statement. The particular asset
classes our plan selects may differ from those in the sample statement, but
they should in any case be broadly diversiļ¬ed.
In the calculation of quarterly returns on the Benchmark Portfolio,
why would we benchmark illiquid investments against themselves?! Rarely
is there a good way to measure quarterly returns on an illiquid asset. The
quarterly valuations on these assets often tend to show little movement
over time and then incur a sudden change that can obfuscate relative re-
turns on our liquid assets. Returns on illiquid assets certainly need to be
benchmarked! But we will proļ¬t most by benchmarking their aggregate in-
ternal rates of return, over multiyear intervals, against the hurdle rate of
return below which, on an expected basis, we wouldnā™t have chosen to in-
vest in those asset classes.
It is important to establish return and diversiļ¬cation criteria for Illiquid
Assets. We should certainly require higher returns from private investments
than from liquid assets. And when we ļ¬nd a private opportunity that strikes
us as the greatest thing weā™ve ever seen, we need, in order to control our
own enthusiasm, a constraint on the maximum percentage of assets we
should commit to that opportunity, because we canā™t change our mind.
If our investment fund is small, much less than $100 million, we may
want to avoid illiquid investments altogether and omit any reference to
them in our policy statement.
In evaluating our returns on Liquid Assets, we should remember we
are long-term investors. We must measure quarterly results, but letā™s not
get hung up on them. Our focus should be on results relative to benchmark
over the last 5 or 10 years.
The target of minimizing short-term ļ¬xed income assets is worth artic-
ulating, as most pension and endowment funds tend to retain more cash
equivalents than they should.
The paragraph about using contributions and withdrawals to rebal-
ance toward the Policy Asset Allocation codiļ¬es our commitment to rebal-
ance our portfolio back to its Policy Allocation as market action drives it
52 SETTING INVESTMENT POLICIES
away from that allocation. Rebalancing is a critical concept we will cover
in detail in Chapter 6.
Our policy statement of using contributions and withdrawals to rebal-
ance implies that this might be done more or less mechanically, without
judging what asset classes are attractive or unattractive at the time. Thatā™s
exactly what I mean to imply! There are few if any mortals who can time
asset classes. The sentence takes judgment out of a decision where judg-
ment isnā™t likely to add value. Also, as mentioned before, it is the lowest-
transaction-cost method of rebalancing.
As part of these criteria, we should aim to use the best possible man-
agers that we can access in each asset class. That is obviously an unreach-
able target, but thatā™s the direction in which we should always be striving. I
also think we should include a statement about in-house management and
its rationale, and also about the role of index funds.
Criteria for manager selection and retention are helpful as guideposts
against which future manager recommendations should be evaluated. Weā™ll
discuss more about this in Chapter 5 on selecting investment managers.
Every investment fund should have a written statement of investment
objectives. The statement sets directions and criteria that will help to
focus everyone who will be involved in subsequent decisions.
We should ļ¬rst establish the time horizon for our portfolio (usually
very long term), then how much risk, or volatility, we can afford. Our
target rate of return should then be āthe highest we can achieve with-
out exceeding our target level of volatility.ā
Our statement of investment objectives should include our Policy Asset
Allocation as well as benchmarks against which we can measure the
success of our investment program.
y far, the most important single investment decision that our investment
B fund makes is not the particular investment managers we select, but our
asset allocation. Thatā™s the proportion of our total assets we put into each
asset class, such as large U.S. stocks, long-term bonds, or real estate equity.
Asset allocation determines up to 90% of our future investment returns.
If we think ļ¬rst of manager selection, we are implicitly making allo-
cations to asset classes. Why? Because investment results within an asset
class are so dominated by the wind behind that asset class,1 any man-
agerā™s results will be highly impacted by that wind. If large U.S. stocks
achieve high returns, so will nearly all managers of large U.S. stocks; and
those managers will not be able to escape the slaughter if large U.S.
stocks should crash.
Historically, many endowment funds have drifted toward an asset allo-
cation something like 60/30/10, that is, 60% in stocks, 30% in bonds, and
10% in cash equivalents, all U.S. based. Is there something inherently ideal
in that asset mix? If there is, it isnā™t apparent in other countries, where typ-
ical asset allocation differs greatly from country to country.
Which nationā™s conventional investment wisdom is best? The over-
whelming reason for each fundā™s asset allocation is . . . thatā™s how itā™s al-
ways done here. Fund sponsors feel safety in numbers, and many are timid
about investing their money differently.
Letā™s obey all laws diligently. But I suggest that an investment fund does
well to ignore how its peers are investing their money. Instead, letā™s set our
asset allocation on the basis of (1) our objectives, as discussed in Chapter 3,
The āwind,ā as I call it, means the things that tend to affect the returns of all in-
vestments in a particular asset class at any given time.
54 ASSET ALLOCATION