time either the Committee chairman, CEO, or any two Committee
members request a meeting. There may be occasions, in order to com-
plete speciļ¬c Committee business, when the Committee may have to
meet multiple times within a month.
6. Committee members are to make every effort to attend each Commit-
tee meeting. If a member cannot attend in person, he or she should
participate by conference call.
7. Committee members who participate in fewer than 80% of meetings
over a rolling two-year interval are to be terminated from the Commit-
tee, subject to a majority vote for retention by the remaining Commit-
8. Because it is essential to avoid even a perception of conļ¬‚ict of interest,
the Committee should consider preparing a Code of Ethics, to be re-
viewed with legal counsel, which will deal with the appropriate con-
duct for Committee and staff members. The Code should deal with
investment transactions, conļ¬‚icts of interest, and independence issues,
and it should be reviewed and signed by each member of the Commit-
tee and staff annually.
9. The Committee will establish statements of Operating Policies and In-
vestment Policies. The latter is to include a Policy Asset Allocation and
a related Benchmark Portfolio. Draft policy statements are to be sub-
mitted by the CEO [or consultant], who may propose amendments to
these statements at any time. Both policies should be reviewed annually.
10. The CEO [or consultant] is to act at all times within the Committeeā™s
Operating Policies and Investment Policies. If so authorized, the CEO
may deviate from the Committeeā™s Policy Asset Allocation within any
range the Committee may establish for an asset class, but he or she is
Appendix 1: Example of an Investment Committeeā™s Operating Policies
to report promptly to the Committee any deviation from the Policy As-
set Allocation and the reasons for that change.
11. Prior to any Committee meeting, the Committee chairman, upon the
recommendation of the CEO [or consultant], will establish the agenda.
Wherever possible, the CEO [or consultant] will mail presentation ma-
terials to Committee members in time for them to receive the materials
a week before the meeting. Committee members are expected to re-
view these materials in preparation for the meeting.
12. The Committee will appoint a secretary, who will prepare minutes of
all actions decided by the Committee, and retain these minutes, to-
gether with any presentation materials recommending those actions, in
a permanent ļ¬le.
13. Decisions by the Committee are to be made by majority vote, although
Committee members should ļ¬rst endeavor to reach a consensus.
14. The Committee, at the recommendation of the CEO [or consultant],
will appoint a master custodian, and all fund assets are to be held by
the master custodian.
15. The fund may not borrow money except for overnight emergencies, al-
though the Committee may authorize speciļ¬c Investment Managers of
the fund to use leverage.
16. The CEO [or consultant] will submit to Board members a brief quar-
terly report in writing, including
ā“ Recent performance (net of fees) versus benchmarks, in the context
of the long term;
ā“ Current asset allocation versus Policy Asset Allocation;
ā“ Principal actions implemented by the CEO [or consultant] since the
last quarterly report;
ā“ Potential issues or actions for future meetings.
17. The CEO [or consultant] will submit to the Committee a detailed an-
nual report in writing on investment results and follow it with a thor-
ough verbal presentation to the Committee. At this meeting, the CEO
[or consultant] will comment on the continued appropriateness of cur-
rent Operating and Investment Policies and the rationale for continu-
ing to retain each of the fundā™s Investment Managers.
18. The Committee will select an accredited accounting ļ¬rm as the fundā™s
auditor, which will submit an annual audit report to the Committee.
20 THE INVESTMENT COMMITTEE
19. The fund will publish an annual report, cosigned by the Committee
chairman and CEO [if there is one], that will include:
ā“ Investment results,
ā“ Year-end asset allocation,
ā“ Contributions and payouts during the year,
ā“ Key actions during the year,
ā“ A summary of the actuarial reports (if for a pension fund),
ā“ A summary of the audit report,
ā“ Names of Committee members and key staff members [or con-
ā“ Total compensation paid to or accrued by directors and executive
ā“ An appendix that includes statements of the Committeeā™s Operating
Policy and Investment Policy.
If the Committee employs a CEO and investment staff, rather than relying
mainly on a consultant, several more operating policies might be added:
ā“ The Committee is to approve the selection of all Investment Man-
agers and investment funds; or
ā“ if the Committee has authorized staff to appoint any Investment
Manager or investment fund that is to manage less than [X%] of the
fundā™s assets, any decision involving more than [X%] is to be ap-
proved by the Committee. In any case, the Committee must approve
the use of any asset class that is being used for the ļ¬rst time.
21. The Committee is to approve any assets to be managed internallyā”by
the CEO and staff.
22. The CEO will have authority to make all decisions that are not re-
served for the Committee.
23. Each year, the Committee shall approve an operating budget, submit-
ted by the CEO, covering all fund expenses except fees and expenses of
Investment Managers. Fees and expenses of Investment Managers shall
not be a part of the budget but shall be summarized in the CEOā™s an-
nual report to the Committee.
24. The Committee will hire a lawyer, with whom the CEO is to review all
legal documents and consult on all legal issues.
Risk, Return, and Correlation
s the investment committee for an investment fund, what are we trying
A to do? Weā™re trying to earn money; more speciļ¬cally:
To achieve the highest possible net rate of return over the long term
While incurring no more risk than is appropriate for the ļ¬nancial cir-
cumstances of our fundā™s sponsor.
Before we go further, we should deļ¬ne what we mean by return and
risk, since these are critically important concepts to understand.
Whatever game we are learning, whether tennis, bridge, or some other, one
of the ļ¬rst things we should learn is how to keep score. How can we know
how we are doing if we donā™t know how to keep score? This is equally true
of investing, which I view as a āgameā in the classical sense of the term, an
extremely serious game.
How do we keep score in investing? The money we earn (or lose) is
called āinvestment return.ā What constitutes investment return? Invest-
ment return on stocks and bonds includes income (such as dividends and
interest) and capital gains (or losses), net of all fees and expenses. As basic
as that is, we need to keep it in mind. The stock indexes as reported in the
newspaper reļ¬‚ect only priceā”even though dividends have provided in-
vestors with close to half of their total return on stocks over the past 75
years. We must add dividends or interest to a stock index to obtain the total
return on that index.
Our focus should always be on total returnā”the sum of income and
22 RISK, RETURN, AND CORRELATION
capital gains (or losses), whether realized or unrealized.1 Fundamentally,
there is little difference between income and capital gains, in that a com-
pany or an investor can manipulate the composition of income and capital
gains, but one cannot manipulate total return. A company that wants to
shield its investors from taxes can pay very low (or no) dividends and rein-
vest most (or all) of its earnings, either in its business or in the repurchase
of its common shares. As investors, we can easily build a portfolio with
high or low income, depending on whether we invest in securities that pay
high or low dividends and interest. But achieving a high total return re-
mains a difļ¬cult challenge.
The ļ¬nal part of the deļ¬nition of total return is: ānet of all fees and ex-
penses.ā The only return we can count is what we can spend. We must
therefore deduct all costsā”mainly investment management fees, transac-
tion costs, and custodial expense.
Total return is what investing is all about.
Valuing Our Investments
To ļ¬nd the total return on our investments for any time interval, we must
know the value of our investments at the start of the interval and at the end
of the interval. But what value? Book value (the price we paid for an in-
vestment) or market value?
To understand our investments at any time, we must focus on a single
valueā”market valueā”the price at which we could most realistically sell
those investments at that time. Thatā™s what our investments are worth.
Book values are helpful to auditors, and accounting rules require that
book values be taken into consideration. (Book values are also extremely
important to taxable investors.) But for purposes of understanding our
taxfree investments, book values are not helpful.
I often refer to book values as an historical accident. Book value is the
price we happened to have paid for our investment on the day we hap-
pened to have bought it. A comparison of an investmentā™s market and
book values is not enlightening. If the value of our investment is up 50%
since we bought it, is that good? If we bought it only a year ago, thatā™s
We realize a capital gain (or loss) when we sell a security for a price that is dif-
ferent from what we paid for it. We have an unrealized capital gain (or loss) if
the market value of a security we currently own is different from the price we
paid for it.
probably good (unless the market rose even more than 50% in that time).
But if we bought it 10 years ago, a 50% increase is not very exciting.
Book values also can be manipulated. If we want to show a higher
book value for our portfolio, we can sell a security with a large unrealized
appreciation (whose price is much higher than its cost), and the book value
of our portfolio will rise by the gain we have just realized. Or if we want to
show a lower book value, we can sell a security whose price is much lower
than its purchase price, and the book value of our portfolio will decline by
the loss we have just realized.
Market value cannot be manipulated. Always focus on market values.
When making reports about our fund to its board or our sponsorā™s mem-
bership, we should stick with market values. Forget about book values.
Whatā™s a Good Rate of Return?
What does it mean when the newspaper says that Mutual Fund X had an
annual rate of return of 10% for the past three years? Thatā™s simple. It
means that if we put a dollar into the mutual fund three years ago, it would
have grown by 10% per year. We know thatā™s not 3 times 10 equals 30%
for the three years because itā™s a compound rate of growth. The dollar the-
oretically became worth $1.10 after Year 1, plus another 10% was $1.21
after Year 2, and another 10% was $1.33 after Year 3. A return of 33%
over three years is the same as 10% per year.
Fine. But, is 10% per year good? That depends. Based on the way Mu-
tual Fund X generally invests money, what opportunity did it have to make
money? How did the fundā™s total return compare with that of its bench-
markā”usually the most appropriate unmanaged index?
Letā™s say that (a) Mutual Fund X invests mainly in large, well-known
U.S. stocks, sticking pretty close to the kinds of stocks included in Stan-
dard & Poorā™s (S&P) 500 index, and that (b) we should expect Mutual
Fund X to incur about the same level of risk as the index. In that case, the
S&P 500 is a good reļ¬‚ection of the opportunity that the mutual fund
faced. The S&P 500 is a sound benchmark. If the total return on the S&P
500 was 13% per year, then the 10% return on Mutual Fund X was not so
hot. On the other hand, if the S&Pā™s total return was only 7%, then 10%
represents very good performance.
Now wait a moment. Letā™s turn that around. What if Mutual Fund X
returned minus 10% per year, and the S&P was off 13% per year. Are we
saying that Mutual Fund X performed very well?
Absolutely. Investing in marketable securities is a relative game. We
24 RISK, RETURN, AND CORRELATION
know the market can drop precipitouslyā”and will sometimes. And when
it does, Mutual Fund X is just as likely to drop precipitously. We must be
aware of that before we buy into Mutual Fund X. As Harry Truman once
said, āIf you canā™t stand the heat, get out of the kitchen.ā If we have se-
lected a valid benchmark for Mutual Fund X, then the most we can ask of
Mutual Fund X is to do well relative to that benchmark. I view that as a
cardinal rule of investing.
So if the S&P returns 7% and Mutual Fund X returns 10%, then Mu-
tual Fund X is pretty good, right? Itā™s right for that particular year or for
whatever interval is being measured. But virtually every mutual fund that
has underperformed its benchmark over a long interval can select periods
of years when it looked like a hero. And almost every fund with outstand-
ing long-term performance has run into intervals of years when it couldnā™t
meet its benchmark. Hence, evaluating the performance of an investment
manager on the basis of only one or two intervals, such as the past three or
ļ¬ve years, is fraught with danger. What counts is long-term performance,
perhaps 10 years or longer.
As board members, we may be asked to decide whether to hire, retain,
or terminate a given investment manager. Of course we will want to under-
stand historic performance, but what counts is our judgment about the
managerā™s future performance. That involves many additional considera-
tions that we will get into in Chapter 6.
In any case, using the right benchmark to evaluate a manager is criti-
cal. If we have chosen the wrong benchmark, our evaluation might well
motivate us to part company with a strong manager at the wrong time (or
to keep a mediocre manager).
So how do we select an appropriate benchmark? Volumes have been
written about that. But in essence, a benchmark should represent the par-
ticular universe of stocks (or other securities) from which the particular
manager selects his stocks. Weā™ll talk more about selecting benchmarks for
a manager in the next few chapters.
Returns on a Portfolio of Investments
Weā™ve talked about measuring returns on a single investment. What if
weā™ve invested in a whole series of investments? Letā™s say we invested an in-
creasing amount of money over a period of years, some years more, some
years less. In some years, we withdrew some money from our investments.
Also, we invested in not one but a portfolio of funds. How do we keep
score on a portfolio like that?
There are two basic ways: (1) time-weighted returns, and (2) dollar-
weighted returns. Itā™s important to understand the differences between
A time-weighted rate of return measures the rate of return on the ļ¬rst
dollar invested during an interval being measured. Every quarter of
the year is weighted equally regardless of how much money was in
A dollar-weighted rate of return (also called an internal rate of return)
measures the rate of return on every dollar invested during the interval
being measured. For example, a quarter of the year when a lot of new
money was invested is weighted proportionately more than a quarter
when little money was invested.
We can illustrate the differences most easily with a simple example.
If we put $1,000 into Mutual Fund X and it returns 20% the ļ¬rst year,
then at the start of the second year we invest another $5,000 in Mutual
Fund Y, and they both return 10% the second year, what is our rate of re-
turn on our portfolio of the two funds for the two years?
First, how much money do we have (what is our wealth) at the end of
year 2? Our wealth after year 2 is $6,820:
(Contributions Investment Rate of
Date and Withdrawals) Wealth Return Return
1/01/00 $+1,000 $1,000 ā”
12/31/00 ā” 1,200 $200 20%
1/01/01 +5,000 6,200 ā”
12/31/01 ā” 6,820 620 10%
If we weight the results in each year equally (as a mutual fund does),
then the annual rate of return for the two years is about the average of
10% and 20%, or roughly 15%.2 Thatā™s the time-weighted rate of returnā”
the rate we will be principally concerned with.
But be careful of taking simple averages. Itā™s not as simple as it might seem. In this
case, the precise annual time-weighted rate of return for the two years is 14.89%
[(1.10 Ć— 1.20)1/2 ā“ 1 = .1489]
26 RISK, RETURN, AND CORRELATION
But we did not earn 15% on every dollar. We earned 20% on $1,000,
then 10% on a little more than $6,000. What is the dollar-weighted annual
rate of return on every dollar we had invested? The most accepted way to de-
rive the dollar-weighted rate of return is to calculate the internal rate of re-
turnā”which turns out to be 11.5%.3 The 11.5% is what we actually earned
on our money. That may be good or bad compared with our long-term aspi-
rations. But it is very difļ¬cult to compare that with our opportunities (that
is, with any benchmark) to determine whether that is good or bad, because
no benchmark would have invested money with the same timing as we have.
Weighting each year equally gives us a ļ¬gureā”15%ā”that we can
compare with other similar funds or with an appropriate benchmark.
Because time-weighted returns ignore the timing of contributions or
redemptions, time-weighted returns implicitly relieve the investor of the re-
sponsibility for the timing of his investments. That is a critically important
assumption. But is it an appropriate assumption?
Letā™s say we placed $1,000 with Investment Manager A, and after he
achieved a 10% return the ļ¬rst year, we gave him an additional $5,000,
and then the stock market dropped 10% and Manager Aā™s investment did
also. His time-weighted rate of return is about zero (actually ā“0.5% per
year)4 and his dollar-weighted rate of return is minus 7.4% per year.5
Shouldnā™t Manager A have known better than to put our money in the
stock market just before it went down? If we rely on his time-weighted per-
formance, weā™re saying no, he should not have known better. But isnā™t that
why we place our money with a professional investment manager?
Manager A had an opportunity to be a hero by keeping the money in
cash equivalents for the second year, but we are unrealistic if we expect our
manager to be a good market timer. After more than 30 years of investing,
I still donā™t know of a really good market timer. It is realistic to expect a
good manager of stocks or bonds to perform well over the long term rela-
tive to an appropriate benchmark, but not to be clairvoyant enough to
know when to go in or out of the stock market.
Therefore, time-weighted rates of return are best for evaluating a man-
Note that 6,820 = 1,000(1.1152)2 + 5,000(1.1152).
(1.10 Ć— 0.90)1/2 ā“ 1 = .995, or ā“0.5%.
1,000 Ć— 1.10 = 1,100; 1,100 + 5,000 = 6,100; 6,100 Ć— 0.90 = 5,490, the market
value at the end of year 2. Then 5,490 = 1,000(.926)2 + 5,000(.926), and .926 ā“ 1 =
ager of stocks or bonds. They are also the only way to compare the perfor-
mance of our overall fund with other funds that have had different timing
and amounts of cash ļ¬‚ows (contributions or withdrawals), or with an
overall benchmark for our fund.
But ultimately, time-weighted rates of return are not what count. Dol-
lar-weighted rates of returnā”also known as internal rates of returnā”de-
termine our ending wealth. Also, dollar-weighted rates of return are the
only meaningful way to measure returns on private investments, where the
manager controls the timing of when money goes into and out of the fund.
Over a 10-year interval, Fund A earned 12% per year while Fund B earned
only 10% per year. Both funds had negative returns in some years, but in
their negative years Fund A was down 5 percentage points more than Fund