Source: BARRA RogersCasey.
Criteria for Hiring and Retaining Managers
pact is the future. Hence, the future should be the exclusive focus of our in-
How can we judge the predictive value of a managerā™s performance?
Itā™s not easy, and in each case it comes down to a judgment on which rea-
sonable people who have studied all the facts may differ. My judgment has
tended to be inļ¬‚uenced by the following factors:
Decision makers. Who is the individual or individuals responsible for
the performance record (they may not necessarily be the heads of the
ļ¬rm)? Have the same individuals been responsible throughout? If so,
are they still in the saddle? If not, predictive value is essentially nil, be-
cause the past performance reļ¬‚ects somebody elseā™s work. Of all con-
siderations, this is probably the most important.
āInvestors seeking to engage an active manager should focus on
ā˜people, people, people.ā™ Nothing matters more than working with
high-quality partners,ā writes David Swensen of the Yale endow-
Support staff. Material turnover in the research or other support staff
may impair the predictive value of past performance.
Process. In investment approaches where the investment process
is as important as the individual decision makersā”a rarity, in my
judgmentā”I may attribute some predictive value to past perfor-
mance even though there has been turnover in key people. Continu-
ity of methodology is particularly important with quantitative
managersā”where the product of human judgment is the mathemati-
cal algorithm6 rather than individual investment decisions. With
such managers, I am also interested in their commitment to continu-
Size of assets managed. If, adjusted for the growth in market capital-
ization of the overall stock market, a manager is managing a much
larger value of assets today than he did X years ago, his performance
of X years ago may carry very little predictive value. Managing $5 to
$50 million would seem to have little predictive value for a manager
who is now managing over $5 billion.
David F. Swensen, Pioneering Portfolio Management (The Free Press, 2000), p. 252.
These algorithms are mathematical equations that transform raw data about com-
panies and the economy into speciļ¬c buy and sell decisions.
110 SELECTING AND MONITORING INVESTMENT MANAGERS
Number of decisions. Performance that is the result of a thousand
small decisions should have a much higher predictive value than per-
formance dominated by only a handful of decisions, as might be the
case with a manager whose past performance hinges on several key
market-timing calls. The smaller the number of data points, the more
difļ¬cult it is to distinguish skill from luck.
Consistency. Performance that is consistently strong relative to a valid
benchmark would seem to have a lot more predictive value than per-
formance that is all over the place.
Proper benchmark. Is the manager being evaluated against the proper
benchmark? Performance that is compared with a valid, tight-ļ¬tting
benchmark would seem to have higher predictive value than perfor-
mance that is simply compared with the market in general, especially
over intervals as short as three to ļ¬ve years. Comparisons with the
market in general can lead us astray, since the dominant inļ¬‚uence may
be a managerā™s style (which can go in and out of vogue) rather than the
managerā™s skill. It may be useful to review pages 44ā“45 and 48 with re-
spect to benchmarks.
Time. In this case, how many years of a managerā™s past performance
do we think have predictive value? Three years of performance may re-
ļ¬‚ect mainly noise. I am particularly impressed when I see a manager
with 15 years of strong performance that also meet other criteria of
good predictive value.
Some quantitative managers who donā™t have long track records will
show extensive simulations of how they would have performed if they had
been using their quantitative method. Beware! Letā™s understand how the
manager developed his algorithm in order to evaluate how much data-min-
ing7 the manager has done. Itā™s almost impossible to eliminate data-mining
completely, but letā™s make a hard judgment about how academically honest
and objective the manager has been. Then, if he passes both these tests,
letā™s discount his results by several percentage points per year and see if the
manager is still worth considering.
Assessing the predictive value of a managerā™s past performance is not
easy. But itā™s crucial. Assessing the likely impact of the managerā™s volatility
Data-mining is the extent to which 20/20 hindsight has inļ¬‚uenced the managerā™s
and correlation on the rest of our portfolio is likewise not easy, but itā™s also
important. An understanding of the volatility and correlation of the man-
agerā™s particular asset class can be helpful, too.
Managers of Illiquid Assets We need to devote extra care to the selection
of managers of the private, illiquid funds we enter. Because there is such a
wide dispersion of returns between the better managers and the average
managers of private funds, there is extra pressure for us to go only with
The criteria for selecting managers of private investments are the same
as discussed above, but evaluating the predictive value of track records is
more difļ¬cult because (a) track records are often short; (b) track records
are often incomplete, because typically many assets have not yet been sold;
and (c) the track record often comprises a relatively small number of in-
vestments. These factors increase the challenge of selecting managers of
private investments and accentuate the importance of focusing on people,
A common way for a committee to select a manager is for the recom-
mender to bring three or more candidates to meet with the committee, at
the conclusion of which the committee is to select one of them. This is
what I call the ābeauty contest.ā In a 20- to 30-minute presentation, com-
mittee members are able to discern which presenter is the most articulate,
but Iā™ve found little correlation between articulateness and good investing.
Thatā™s why I believe the ābeauty contestā is a poor approach. Those
who have participated in the full, painstaking evaluation of all the man-
agers being considered are the ones best equipped to conclude who should
be hired. Committee members who spend only a relatively few hours per
year on our fundā™s investments canā™t hope to make a meaningful evaluation
on the basis of a 20- or 30-minute presentation.
Should the adviser give the committee a presentation on the three
best managers in a given asset class and then let the committee choose? I
donā™t even favor that approach. The recommender has done the re-
search. He should make a single recommendation and be charged with
112 SELECTING AND MONITORING INVESTMENT MANAGERS
What, then, should I as a committee member do? Be a rubber stamp?
No. I should understand the above criteria, consider whether the recom-
mender has covered them all adequately in his presentation, and ask ques-
tions until I am satisļ¬ed. If I canā™t get comfortable based on the above
criteria, I should try to table the recommendation or vote against it.
But remember: We committee members canā™t do a good job of asset al-
location and manager selection by ourselves. The committee should proba-
bly approve of 95% of the recommendations of its adviser or else
terminate the adviser and hire a new one in whom the committee does have
Caveat about Style
It is helpful initially to categorize managers by style. For example, common
categories of styles of U.S. equity managers are large, medium, or small
cap, and growth or value. But these style categories can only be very gross.
There are great differences of style among large-cap value managers and
among small-cap growth managers, for example. And stocks that might be
considered growth stocks today may, after their bubble has burst, be con-
sidered value stocks tomorrow.
When looking at a large-cap value manager, which benchmark should
we choose? Several value indexes are available. Any particular benchmark
may relate to the managerā™s style in only a very coarse way. Where the
benchmark doesnā™t ļ¬t very well, we should not downgrade the manager
because of benchmark risk.8 Many of the best managers donā™t manage to a
benchmark, and shouldnā™t. That means we just have to work a little harder
to understand and interpret their performance. Still, in the end, we want a
manager that canā”over the long termā”materially outperform his bench-
mark net of fees.
The more valid a benchmark is for a particular manager, and the more he
invests within the universe of that benchmark, the narrower his deviations
from that benchmarkā”and superļ¬cially, at leastā”the easier it is for us to
evaluate his performance. We should never, however, confuse benchmark
āBenchmark riskā is the risk that the managerā™s performance will deviate greatly
(up and down) from his benchmark.
risk with absolute risk, or forget that our objective is to make money. A
manager with a large benchmark risk could possibly have lower volatility
than the benchmark.
Peter L. Bernstein, well-known consultant and ļ¬nancial writer, claims
institutional investors have handcuffed their managers by linking them to
benchmarks whose composition changes every year.9
David Fisher, chairman of the world-class Capital Guardian Trust
Company, has succinctly placed benchmark risk in its proper context with
Risk management is a great deal more than benchmark risk.
Put another wayā”benchmark risk is a small part of risk management.
Organizations should be aware of benchmark risk but not pray at its
Nowhere is it written that criteria that are quantiļ¬able are more im-
portant than those that are not.
Many investment managers are fearful of taking a lot of benchmark
risk because of business riskā”their business risk that if they should ever
underperform their benchmark by a wide margin many clients would leave
them. Thatā™s an understandable concern and needs to be dealt with.
One time while visiting a Canadian investment manager, we noted that
he had some 20% of his portfolio in a single stockā”Nortel. I asked him if
he had that large an allocation because he thought Nortel was by far the
most attractive stock in Canada. He replied no, he was actually far under-
weighted in Nortel, as that stock composed as much as 35% of his bench-
mark, the TSE 300. He was obviously fearful of being any more
underweighted than that because of his own business risk.
We could solve his problem simply by changing his benchmarkā”to a
unique version of the TSE 300, one where the weighting of any single stock
in the index was truncated at X% of the total capitalization of the index.
That unique index would implicitly be rebalanced each quarter, but thatā™s a
small price to pay to relieve the manager from a dysfunctional benchmark.
Some of the best managers are ones for whom there isnā™t a very good
benchmark, and, in fact, they are not much concerned about benchmarks ex-
cept in the very long run. These are managers who will invest our portfolio
Joel Chernoff, Pension & Investments, August 7, 2000, p. 4.
114 SELECTING AND MONITORING INVESTMENT MANAGERS
however they think will make the most money. Their benchmark risk is gi-
gantic. Categorizing such a manager in our asset allocation is fuzzy at best.
Should we include such a manager on our team? By all means, if he is good
enough. We should use benchmarks as tools, not as crutches.
How Much Excess Return to Expect
When our adviser waxes enthusiastic about a prospective investment man-
ager, how much excess return10 above his benchmark, net of fees, might we
realistically expect long-term in the years ahead?
Over intervals of 10 to 20 years, net of fees, few managers of invest-
ment grade bonds can exceed the Lehman Aggregate Bond Index by as
much as 1 percentage point per year, and few managers of large U.S. stocks
can exceed the S&P 500 by 2 points per year. For less well-researched asset
classes, I would be well pleased with 3 percentage points per year in excess
of the relevant index. Considering the fact that we will inevitably choose
some managers who are destined to underperform their benchmarks, I
think we will be doing very well indeed if, in the aggregate over the long
term, all of our active equity managers combined can succeed, net of fees,
in beating their benchmarks by 1 to 11/2 percentage points per year.
Sometimes with a given manager we have a choice between a commingled
fund or a separate account.11 Which route should we take? Some investors
like their own separate account whenever they can get it. My preference,
however, would be for whichever approach is likely to achieve the best rate
of return net of all costs, and that depends on the facts of the matter.
If we want something other than what the commingled fund is offer-
ing, the decision is easy: We can get it only with a separate account. But
what if the manager invests the commingled fund and separate accounts in
a similar manner?
Sometimes, imprecisely, called āalpha.ā
A commingled fund is one in which two or more clients invest. Group trusts
and most limited partnerships are common examples. A mutual fund is an ex-
treme example of a commingled fund. On the other hand, a separate account
(except when the term is used by insurance companies) is an account held for
only a single investor.
A commingled fund can often be preferable to a small separate ac-
count because the commingled fund is more diversiļ¬ed and is usually given
more āshowcaseā management attention.
In short, neither a separate account nor a commingled fund is neces-
sarily more advantageous than the other. It all depends.
How Many Managers?
To gain optimal diversiļ¬cation in common stocks, we have already dis-
cussed the importance of selecting outstanding managers in large stocks
and small stocks, āgrowthā stocks and āvalueā stocks, and managers of
U.S. stocks, stocks from the developed countries abroad, and stocks from
the emerging markets. Also, we should have similar diversiļ¬cation among
managers of the various ļ¬xed-income asset classes. Such diversiļ¬cation is
the way to get the best long-term investment return with the lowest aggre-
A single manager would be most convenient for usā”if it were the best
in each of those specialties. But we have seldom seen managers who are
considered the best in more than one or two of those specialties.
We should aim for outstanding managers in each of these areas. There
is no magic number thatā™s optimal. An extremely large fund can add fur-
ther specialties to the above listā”as long as the managers on its team are
complementary to one another.
Upon ļ¬nding two outstanding managers who ply the same turf, we
may have a hard time deciding which to hire. Should we ease our prob-
lem by hiring both? If itā™s that close a call, ļ¬‚ip a coin. Adding both might
well add more complexity than value. Absolute return managersā”such
as market neutral and other hedge fund managersā”are an exception, as a
portfolio of such managers gives more consistent performance than any
one of them.
There is, however, a reason other than diversiļ¬cation for having
multiple managers. No matter how diligent the selection process, and
how conļ¬dent we are of our ultimate selection, every selection is a prob-
ability. If two-thirds of our selections turn out to be above-average per-
formers, weā™ll be doing well. But letā™s not kid ourselves about our
selections being error free. If we have only one or two managers, the im-
pact of a manager who gives us disappointing performance is greater
than if we have, say, 10 of them. With multiple managers, the probabil-
ity of a home run declines, but (assuming a sound selection process) so
do the odds of striking out.
116 SELECTING AND MONITORING INVESTMENT MANAGERS
What if we have only a $100,000 endowment fund? Or a fund with
All too often, members of the investment committee know a local
banker they trust, and they hire the bankā™s trust department to manage the
endowment fund. The banker is usually highly trustworthy, indeed, but the
approach is often submarginalā”for two reasons:
1. Few bank trust departments have the expertise to invest with real
global diversiļ¬cation, and
2. While the staff members of the bank trust departments work diligently
at their jobs, they are rarely the best investors, for a very simple rea-
son. Hardly any bank trust departments can afford the kind of com-
pensation that will attract, or keep, the best.
Alternatively, many endowment funds place their money with a local
investment management ļ¬rm that has a strong reputation in the commu-
nity. But the same questions should be raised: Does the ļ¬rm have global ex-
pertise, and if so, is it really the best we can get in all areas? Few if any
ļ¬rms meet those criteria, anywhere.
Some brokers who sell mutual funds might tell us they provide con-
sulting services for free, since the commissions cover their compensation.
Their motivations can never be congruent with ours, however, since bro-
kers are compensated on turnoverā”the amount of buying and selling in
our portfolioā”which by itself is irrelevant to us. Consultant remuneration
based solely on a percentage of total assets managed would seem to align
their motivations more closely with ours.
Well, how can a small endowment fund access the best managers?
True, they canā™t be quite as sophisticated in their approach as large
funds, but they can gain most of the beneļ¬ts of diversiļ¬cationā”thanks
to mutual funds.
Among the thousands of mutual funds, there are world class funds in
every category of marketable securities. Even an endowment fund as small
as $50,000 or less can diversify among 10 highly diverse mutual funds.
Exclusive use of mutual funds would entail a marked change in policy
for many investment funds that are either explicitly or implicitly wedded to
the use of local investment management. There is no reason to discriminate
against a local investment manager, but what should lead us to think that
one or more of our local investment managers are among the best in the
world? Our choice of investment managers should be blind as to where a
managerā™s head ofļ¬ce may be located. That blindness in itself is a key ad-
vantage of the exclusive use of mutual funds.
So much is published about mutual funds today in sources like Morn-
ingstar that we might try to select our mutual funds ourselves, without an
adviser. Today, such an approach is more viable than ever, but I still would
not advise it. An adviser should know a lot more about particular mutual
funds than just what is published, and much of his value added is his sub-
jective assessment of the predictive value of a mutual fundā™s track record.
In selecting mutual funds, we should have a sufļ¬ciently broad universe
if we limit our choice to no-load mutual fundsā”funds that do not charge
any brokerage commission for either buying or making withdrawals from
When hiring a manager, we never know what his future performance will