Professor David Swensen:
Let me start out by putting what I think is a
relatively controversial proposition on the table and
that's that this investment management business,
when stripped down to its bare essentials, is really quite
simple. Now, why do I say that?
Well, I think if we took the group here today and divided you
up into smaller groups of four, or five, or six and asked you
to talk about what's really important in managing a
portfolio that has a very long time horizon,
I think that almost all the groups would come to very
similar conclusions. If you're investing with a long
time horizon, having an equity bias makes
sense; stocks go up in the long run.
Bob Shiller's friend, Jeremy Siegel,
wrote a book that has the very simple title,
Stocks For The Long Run. Well, the book is assigned;
you all know it. The other thing that I think
would come out of the discussions is that
diversification is important. Anybody whose read a basic
finance text, as a matter of fact,
I think anybody who thinks about investments in a common
sense fashion knows that diversification is an important
fundamental tenet of portfolio management.
As a matter of fact, Harry Markowitz called
diversification a "free lunch." We spend all our time in intro.
econ. figuring out there is no such
thing as a free lunch but Markowitz tells us that
diversification is a free lunch. For any given level of return,
you can reduce--For any given level of risk,
you can increase the return; sounds pretty good.
That's pretty simple, right? Two tenets, an equity bias for
portfolios with a long time horizon and diversification.
Bob mentioned in his introduction that I showed up at
Yale in 1985, after having spent six years on
Wall Street, and I was totally unencumbered
by any portfolio management experience.
I thought that was pretty neat. Here I was, back at Yale,
with a billion dollar portfolio--it seemed like a lot
of money at the time--no portfolio management experience.
What do I do? Well, one of the things I think
is a sensible thing to do in life is look around at what
others are doing, so I looked at what colleges
and universities had done in terms of asset allocation.
Turns out that 50% of endowment assets in the mid-1980s were
invested in common stocks, 40% of endowment assets were in
U.S. bonds and U.S.
cash, and 10% in a smattering of alternatives.
Well, I looked at that and I thought, this doesn't really
make a lot of sense. You have half of your assets in
one single asset class: U.S.
common stocks. You've got another 40% of your
assets in U.S. bonds and cash.
So 90% of your portfolio is in domestic marketable securities
and only 10% is invested in things like real estate or
venture capital or private equity--hardly enough to make a
difference in terms of the portfolios returns.
Unencumbered by, I guess, the conventional
wisdom, we started out at Yale on a path that I think
is--fundamentally that changed the way that institutions manage
portfolios. A few years ago,
I wrote a book called Pioneering Portfolio
Management. The reason you could put an
audacious title like Pioneering Portfolio
Management on the cover of the book was that we moved away
from this traditional model with 50% in stocks and 40% in bonds
and cash to something that was much more equity-oriented and
much more diversified. What I'd like to do today is
talk to you about how it is that we moved from this old model to
what it is that today many institutions call the Yale
model. The way that I would like to
talk about this journey that we took is by looking at the tools
that we have available to us as investors--these tools are the
same tools that we have whether we're operating as individual
investors or institutional investors--and describe how we
employ those tools at Yale and how they led us to the portfolio
that we have today. Those three tools are asset
allocation, market timing, and security selection.
The first, asset allocation, basically deals with which
assets you have in your portfolio and in which
proportion you hold each of those assets.
The second, market timing, deals with short-term
deviations from the long-term asset allocations that you
establish. And the third,
securities selection, speaks to how it is you manage
each of your individual asset classes.
Are you going to hold the market portfolio,
index your assets, match the markets results?
Or are you going to manage each individual asset class actively,
trying to beat the market and generate risk-adjusted excess
returns? Let's start out with the first:
asset allocation. I think it's pretty widely
known that asset allocation is far and away the most important
tool that we have available to us as investors.
As a matter of fact, it's so widely believed that
asset allocation is the most important tool that I think some
people have come to the conclusion that it's some sort
of law of finance that asset allocation is the most important
tool. It turns out that it's not a
financial law that asset allocation takes center stage;
it really is more a description of how it is that we behave.
Yale actually has a lot more than the billion dollars that we
started with in 1985. I think the estimate sheet that
I got yesterday morning said that we've got about $22.5
billion dollars; so that's been a nice run.
If I went back to my office after speaking with you this
morning and took Yale's $22.5 billion dollars and put all of
it into Google stock, asset allocation would have
very little to say about what Yale's returns would be.
As a matter of fact, security selection would
absolutely dominate the results. The idiosyncratic behavior of
Google stock from the time that we purchase it to the time that
we sell it would define Yale's returns.
Alternatively, if I went back to the office
and took Yale's $22.5 billion dollars and decided that I was
going to day trade bond futures, security selection wouldn't
have anything to say about the returns;
asset allocation wouldn't have anything to say about the
returns. The returns would be
attributable solely to my ability to market time the bond
futures market. Now, I'm not going to do either
one of those things. I'm not going to put Yale's
entire portfolio in Google stock, I'm not going to go back
and take Yale's entire portfolio to day-trade bond futures;
in part, because it would be bad for me personally.
I think I would be fired as soon as people found out what it
was that I was doing with the portfolio and,
overwhelmingly more important, it would be bad for the
University. It's not a rational thing to do.
What will happen is that Yale will continue to hold a
relatively well-diversified portfolio as defined by the
range of asset classes in which it invests.
When you look at each of those individual asset
classes--domestic equities, foreign equities,
bonds, real assets, absolute return and private
equity--each of those individual asset classes is going to be
relatively well-diversified in terms of exposures to individual
positions or individual securities.
Because that's true, then asset allocation ends up
being the overwhelmingly important determinant of the
University's results. Because we hold relatively
stable, relatively well-diversified portfolios,
security selection turns out not to be an important
determinant of returns for most investors and market timing
turns out not to be an important determinant of returns.
The last man standing is asset allocation and that tends to
drive both institutional returns and individual returns.
Roger Ibbotson, who is a colleague of Bob
Shiller's and mine at the School of Management,
has done a fair amount of work, studying the relative
importance of these sources of returns.
He's come to the conclusion that over 90% of the variability
of returns in institutional portfolios is attributable to
asset allocation and that's the number that I think most people
hear cited when they are looking at Roger Ibbotson's work.
I think one of the more interesting and even simpler
concepts that comes out of his study is that more than 100% of
returns are defined by asset allocation.
Now, how can that be true? How can asset allocation be
responsible for more than 100% of investment returns?
Well, it can only be true if security selection and market
timing detract from institutional returns or
individual returns in the aggregate.
Of course, if think about it, as a community,
the investment community is going to lose from security
selection decisions. If security selection is a
zero-sum game, the amount by which the winner
wins equals the amount by which the loser loses--winners and
losers being defined by performance after a security
selection that has been made--well,
that sounds like a zero-sum game.
But then, if you take into account that you create market
impact when you trade, that you pay commissions when
you trade and you frequently pay advisors substantial amounts of
money--whether they're mutual fund managers or institutional
fund managers--there's this leakage from the system that
causes the active results for the community as a whole to be
negative. Absolutely the same thing is
true on the market timing front. I mean, to the extent that
you're making these short-term bets against your long-term
policy, it requires trading and trading is expensive.
It's very expensive when you take into account not only the
direct costs, but also the costs that you pay
advisors to help you make these decisions.
So, it's not surprising that asset allocation explains more
than 100% of returns and that, for the community as a whole,
market timing and security selection are costly and lower
the community's aggregate investment returns.
It's a little bit of a digression, but one of the
things that I've witnessed over the past twenty years is that
the leakage of the--the leakage from the system in terms of the
returns that go to the owners of capital--leakage has increased
enormously. Think about the advent of hedge
funds--twenty or twenty-five years ago, hedge funds were a
blip on the radar screen. Today, they're a very important
part of the fund's management framework.
Well, those hedge funds charge enormously more than what a
standard manage or marketable securities firm charges.
Well, that leakage--that 1.5% or 2% that you pay your hedge
fund manager--plus the 20% of profits really reduces the
amount of return that's available for the owners of
capital. This idea that the difference
between the returns that you would get if you took your asset
allocation, implemented passively,
and the actual results that the active investors get--the gap
between those two numbers--is becoming larger and larger over
time, generating more and more
returns for the provider of investment management services
and lower and lower returns for those that are hiring those
external advisors. To get back on track,
let's look at the basic underpinnings to this notion
that asset allocation is at the center of the investor's
decision-making process. There are two points that we
talked about--the hypothetical points that came out of the
small group discussions that I suggested we might think about
at the beginning of this talk. First, in terms of equity bias.
Now, we're going to go back to Roger Ibbotson at the School of
Management. He did some path breaking work
in terms of describing capital markets returns over reasonably
long periods of time. I guess you've already looked
at Stocks for the Long Run;
you've seen 200 years worth of data.
Roger Ibbotson's data goes back to 1925 and these are the actual
numbers we used when we first started doing our mean-variance
optimization in our simulations, trying to come to conclusions
about what the appropriate allocations would be for Yale's
portfolio. I'm sure you're familiar with
the drill--you put a dollar into various asset classes,
in this case, at the end of 1925 and hold
those asset classes for, in this case,
eighty-one years; the numbers go through the end
of 2006. As you put a dollar in treasury
bills, you end up with a nineteen multiple;
that sounds pretty good. You get nineteen times your
money over eighty-one years, but then if you take into
account the inflation consumes a multiple of eleven and you're an
institution like Yale that consumes only,
after inflation returns, putting your money into
treasury bills really didn't get you very much.
Suppose you step out in the risk spectrum and put a dollar
into the bond market. Over that eighty-one year
period you would have gotten a multiple of seventy-two.
Well, now we're talking some real after inflation returns
that can be umed. But, when you move from lending
money to the government--either short-term with bills or longer
term with bonds--to investing in the equity market,
there's a stunning difference in terms of the returns.
Just by putting money into a broadly diversified portfolio of
stocks you would have gotten 3,077 times your money.
If you would have stepped further out of the risk spectrum
and put your money into a portfolio of small stocks you
would have gotten 15,922 times your money.
So, ownership of stocks absolutely crushes buying
bonds--almost 16,000 times your money or more than 3,000 times
your money in the stock market as opposed to 72 times your
money or 19 times your money in the bond market or the bill
market. It almost makes you wonder
whether this diversification thing makes any sense.
I mean, why would you do that? Why would you put any of your
assets in bonds if stocks are going to give you 16,000 times
your money? That bond multiple of 72 is
just a drag on returns--what's the point?
This question, particularly in the late 1980s,
was very important to me personally because we were
trying to put together a sensible portfolio for Yale and
if that sensible portfolio just involved identifying the
high-risk asset class and putting all your assets into,
let's say, small stocks, it wouldn't take the investment
committee very long to figure out that they didn't need to pay
me to do that; they could do that on their own.
And if they didn't need to pay me, then I wouldn't have any
income to put food on the table for my wife and children.
So, there had to be more to it than just identifying the
high-risk asset class and putting your assets there and
letting it rip. I went back and took a closer
look at Roger Ibbotson's data and there are lots of examples
that will illustrate this point, but the most dramatic occurs
around the crash in October 1929.
For every dollar that you had in small stocks at the peak of
the market, by the end of 1929, you lost 54% of your money.
By the end of 1930, you lost another 38% of your
money; by the end of 1931,
you lost another 50%; and by the end of--by June of
1932, you lost another 32%. So, for every dollar that you
had at the peak, at the trough you had $.10
left. At some point,
when your dollars were turning into dimes, you'd say,
forget this, this is ridiculous,
it doesn't make any sense for me to own these risky small-cap
stocks. And you would sell your small
stocks and put your money where? Either in treasury bonds or
treasury bills. And of course,
that's what the overwhelming portion of the investment
community did in the 1930s, and in the 1940s,
and in the 1950s. As long as there was a memory
of the searing experience that people had in the equity markets
around the time of the great crash,
people reacted to it by saying, avoid this risky asset,
it doesn't make any sense for a fiduciary or for an individual
to own these risky things called stocks.
As a matter of fact, I was looking at some of the
contemporary literature, the popular literature,
and there was an article in the Saturday Evening Post that
basically said, you shouldn't call stocks
securities--that was a ridiculous thing to call them;
they should be called insecurities because they were
so risky. Of course, this attitude came
at exactly the wrong time. If you put a dollar into small
stocks in June of 1932, by the end of 2006,
you would have had 159,000 times your money.
Just at the point of maximum opportunity people were at the
point of maximum bearishness about the equity markets.
The take-aways are that an equity bias is an absolutely
sensible underpinning for investors with long time
horizons but that diversification is important.
You have to limit your exposure to risky asset classes to a
level that allows you to sustain those positions even in the face
of terribly adverse market conditions.
Let's move to the second point: market timing.
I actually have a quotation here.
A few months ago, some former students of
mine--former colleagues of mine--gave this very nice party
at the Yale Club. I used to teach a big lecture
class when I first got to Yale in the late 1980s and my last
lecture always involved taking Keynes's General Theory,
and quoting from what I think is Keynes--is one of the most
wonderful writers about issues surrounding investment
management. This particular copy was pretty
dog-eared; as a matter of fact,
it was a paperback copy and I think it was in about eight or
ten different pieces and the people that threw this party
remembered that, so they gave me it at this
celebration. It made me wonder if they were
trying to tell that I should retire;
it felt like a retirement party. I feel like I'm way too young
to retire. But as a gift,
they gave me a first edition of Keynes's General Theory.
I was coming back to New Haven on the train afterwards and I
came across this quote. Keynes wrote that,
"The idea of wholesale shifts is for various reasons
impracticable and indeed undesirable.
Most of those who attempt to sell too late and buy too late
and do both too often, incurring heavy expenses and
developing too unsettled and speculative state of mind."
He's absolutely right. I wrote my first book--I
already talked about that, Pioneering Portfolio
Management--that deals with the challenges that face
institutional investors. Subsequently,
I wrote a book called Unconventional Success
that deals with individual investors.
In Unconventional Success, I did a study of
individual behavior in their mutual fund purchases and sales
around the collapse of the Internet bubble in March of
2000. What I did was I took the ten
best-performing Internet funds and looked at the returns from
1997 to 2002. Now this is,
I think, a surprising starting point.
If you look at the ten best-performing Internet funds
from 1997 to 2002, the time-weighted return is
1.5% per year positive, so the funds went way up and
then they went way down. But it's positive 1.5% per
year, time-weighted--that's the number that you see in the
prospectus or the number that you see in the
advertisements--so you say, what's the big deal,
no harm no foul. Well, there's another way to
look at returns--those are the dollar-weighted returns--and the
dollar-weighted returns actually do a better job of describing
the experience of the group of investors that participated in
these funds. Dollar-weighted obviously takes
into account when the cash flows come in and when they go out.
When you do the dollar-weighted returns, you find out that there
was $13.7 billion invested in these funds and the investors
lost $9.9 billion out of the 13.7 that they committed;
so, 72% of the money that was invested in these funds was
lost. Because of the way that we deal
with taxes and mutual funds, you can get a tax bill for
gains that were realized by the investment manager turning over
the portfolio even though you might not have held the shares
during the period when the gains were realized.
So, in addition to losing $9.9 billion, there were capital
gains' distributions of $3.3 billion dollars representing
about 24% of the money that was invested.
So, adding insult to injury, you lost 72% of the money and
then you got a tax bill for 24% of the amount that had been put
in; not a very happy experience.
After I wrote the book, Morningstar did a much more
comprehensive study of every single one of the equity
categories that they follow. There were seventeen categories
of equity mutual funds and they compared the dollar-weighted to
the time-weighted returns. In every one of those seventeen
categories, the dollar-weighted returns were less than the
time-weighted returns. Well, how does that happen?
The same way that these investors and the Internet tech
funds lost their money. They bought after the funds had
gone up and they sold after they had gone down.
When you buy high and sell low it's really hard to generate
returns, even if you do it with great enthusiasm and great
volume. The Morningstar study is
incredibly damning in terms of the market timing abilities of
individual investors. Systematically,
investors are buying after things have gone up,
selling after they've gone down,
and the problem is most severe in those funds that show the
greatest volatility. The gap in what Morningstar
calls the "conservative allocation fund" is .3% per
year. Now, that's not a huge number
but, obviously, when you're hoping to beat the
market by a point or two, losing by .3% per year because
of your market timing inability is a bad thing.
But if you look at the tech fund category,
the difference between the dollar-weighted and the
time-weighted returns--this is over a ten-year period--is 13.4%
per annum; that's stunning.
Compound that 13.4% over ten years and there's just an
enormous gap between those mutual fund numbers that are in
the prospectus and in the advertisement--the time-weighted
returns and the dollar-weighted returns that talk about the
actual experience of the investment community.
I'm not just going to pick on individual investors,
I'm going to pick on institutional investors too.
One of the studies that I did for my first book,
Pioneering Portfolio Management,
looked at the behavior of endowments and foundations
around the crash in October 1987.
I used to talk about the crash in October 1987 without
explaining what it was and I do still teach a seminar in the
economics department in the Fall.
I started talking about what happened in October 1987 and I
looked around the room and I realized that I think the
students were three or four years old in 1987 and weren't
yet reading The Wall Street Journal.
So, just to give you a little bit of context,
the crash was really an extraordinary event.
According to my calculations it was a twenty-five standard
deviation event. One standard deviation happens
one draw out of three, two standard deviations one out
of twenty, three standard deviations is one out of one
hundred. An eight standard deviation
event happens once out of every six trillion trials.
You can't come up with a number to describe the twenty-five
standard deviation event; it's just too large a number,
I think, for any of us to really comprehend.
In essence, this collapse in stock prices--the one-day
collapse in stock prices--I think in the U.S.
the price was, depending on which index you
were looking at, were down 21-22% in a single
day. Interestingly,
most major markets around the world were off by a similar
magnitude. This one-day collapse in stock
prices was a virtual impossibility.
Of course, this was just a change in stock prices;
it wasn't related to any fundamental change in the
economy or any fundamental change in corporate prospects.
It was just a financial event. If stock prices went down--by
the way, bond prices went up. When people were selling
stocks, money had to go somewhere.
Well, it went into the bond market.
There was a huge rally in treasury bonds on October
19,1987. So, stocks were cheaper and
bonds were more expensive. Well, what do you do?
You buy what's cheap and sell what's expensive.
But what did endowments and foundations do?
Well, if you look at the annual reports of their asset
allocation, in June of 1987, their equity allocation was
higher than it had been for fifteen years.
The '70s were a terrible time to invest in stocks,
a bull market had started in 1982.
We were five years into this bull market and people were
getting excited about the fact that stocks were going up and
equity allocations were at a fifteen-year high.
Of course, the money had to come from somewhere,
so bond allocations were at a fifteen-year low.
Fast forward to June 30,1988 and stock allocations had
dropped and, not only had they dropped,
they dropped by more than the decline in stock prices
associated with this collapse in October 19,1987.
Bond allocations had increased by more than could be explained
by the increase in bond prices over the course of the year.
The only conclusion that you could draw is these supposedly
sophisticated institutional investors sold stocks in
November and December and January because they were
fearful and they bought bonds in October,
November, and December--maybe because they were fearful or
maybe because they were greedy. Emotion ruled the decisions,
not rational economic calculus. The costs were huge--not just
the immediate costs in terms of the move from stocks to bonds.
It took these institutions until 1993--a full six years--to
get their bond allocation back down to where it had been prior
to the crash in October 1987. And this is in the context of
one of the greatest bull markets ever.
You certainly have to measure the bull market,
from 1982 to 2000 and some people would say that 2000 was
just a blip and we're still in this bull market.
But regardless of how you measure it, for a full
half-dozen years, in the midst of this bull
market, colleges and universities were
over-allocated to fixed income relative to where they had been
in June of 1987. The take-away is to avoid
market timing. The underlying driving force
behind market timing decisions seems to be emotional--fear,
greed, chasing performance--buying something
after it has gone up, disappointment,
and sales after something has declined.
As opposed to rationally stepping up when something
appears relatively attractive and overweighting and then
leaning against the wind by selling something that's
performed well. Final source of
returns--security selection. We've already talked about how
security selection is a zero-sum game.
The only way that somebody can overweight Ford Motor Company in
the market is to have somebody have a counter position where
they underweight Ford Motor Company;
only one of those is going to be right.
It's measured by subsequent performance in the amount by
which the winner wins equals the amount by which the loser loses,
but it costs a lot to play the game.
As a matter of fact, it costs an increasing amount
to play the game when you look at the fees that are paid to
investment managers and hedge funds.
So, after taking into account the market impact,
and the commissions, and the fees,
this zero-sum game becomes a negative-sum game.
When you look at the returns for institutions,
you see exactly what it is that you'd expect.
Here's ten years worth of data from the Frank Russell
Corporation, the benchmark Wilshire 5000.
For the ten years ended June 30,2005, it returned 9.9% per
year and then the average return for the actively managed equity
fund was 9.6% per year, so we're back to that thirty
basis points. Maybe on average institutions
lose thirty basis points, but it's kind of Lake Wobegon,
where we all believe that we're better than average,
so we're going to overcome that thirty basis points--that's not
such a big hurdle. There's a very important
phenomenon that you need to take into account when you look at
these histories of returns that are generated by active
managers. This is true whether you look
at the universe of the mutual fund managers that we might have
available to us as individuals or whether it's institutional
data, such as those that I just cited;
that concept is survivorship bias.
The only numbers that appear for the trailing ten years are
numbers that are associated with firms that are still in
business. There were probably a number of
firms that, over that ten-year period, went out of business.
Now, which firms do you think went out of business?
Not the ones that are producing great results.
The problem is even more severe when you're looking at mutual
funds because there's kind of a cynical game that mutual fund
management companies play. If they have an underperforming
fund, sometimes they allow it to die a dignified death;
although, that doesn't happen very often.
What they usually do is they take the underperforming fund
and they merge it with one that has a better track record.
All of a sudden the underperforming fund's record
disappears and the assets are in a fund that has a better
record--a record that you can actually market.
Then when we look at the statistics, all we see are a lot
of assets in the fund that performed well and the
underperforming fund that was merged out of existence isn't
there anymore. How important is this
survivorship bias? If you look at the Frank
Russell data--and I just cited ten-year returns ending June
30,2005, so that period started in 1996--well,
in 1996 there were 307 managers that reported returns.
By the time 2005 rolled around, there were only 177 managers
that reported returns, so 130 managers disappeared.
Now, more than 130 managers failed because,
in addition to survivorship bias, there's something called
backfill bias. That's when a new manager
appears subsequent to the beginning of the ten-year
period; they'll put not only the new
numbers in, but they'll take the history of the new manager and
put that history into the database.
Which direction is that going to move the numbers?
Well, that's going to inflate the numbers too because the only
managers that kind of raise their hand and say,
hey I've got this interesting new approach to managing
domestic equities--or whatever the asset class is--are the ones
that have succeeded. You've got survivorship bias
taking out bad records and then you've got backfill bias adding
good records. They both cause the universe of
active management returns to appear to be better than the
reality because there's a lot in there that doesn't have anything
to do with the average experience of,
in this case, an institutional investor.
Sometimes the numbers can be pretty dramatic;
I mean, 2000 was a year of great flux in the markets
because that's when the Internet bubble burst.
If you looked at the domestic equity return--the average
return that was posted in 2000--it was -3.1%.
Then if you fast forward to 2005 and look at the average
return that was posted for 2000, it was +1.2%.
So, the combination of survivorship bias and backfill
bias for that one year made 4.3 percentage points difference.
As reported contemporaneously in 2000, the number was -3.1%
but if you look at the number reported for 2005,
because bad records had disappeared and good records had
been added, all of a sudden the average experience for that year
went up to +1.2%. This is incredibly important
because, when you look at this number that we started out with,
saying the benchmark was 9.9 but net of fees the managers on
average only lost thirty basis points--or .3%--you'd say,
well that's a game I don't mind playing.
Then if you adjust for survivorship bias,
you end up concluding that the deficit wasn't .3% but the
deficit was actually 2%. In a world where,
if you could win by a percentage point or two relative
to the market, to have the average be minus
two full percentage points is pretty daunting.
That's the kind of issue with survivorship bias and backfill
bias in the relatively established asset class of
domestic equities. The problem is even more severe
when you look at something that's relatively new,
like the hedge fund world. Now, why is that?
Well, if hedge funds first became mainstream maybe fifteen
years ago, then what are you looking at in terms of history?
The only history that you would have had fifteen years ago would
have been those funds that produced great returns,
so it's all identified after the fact.
At least in the domestic equity world you've got a pretty stable
base that you were looking at ten years ago,
so the survivorship bias and the backfill bias would be much,
much more of a problem in the hedge fund world.
Burt Malkiel who wrote a book called A Random Walk Down
Wall Street, which if it's not on your
reading list you ought to pick up and take a look at because
it's really fun to read but it's also extremely insightful,
took a look at survivorship bias and backfill bias in the
hedge fund world. He looked at a group of hedge
funds that numbered 331 in 1996 and by 2004, eight years later,
75% of them had disappeared. Looking at this particular
group, he estimated survivorship bias to be 4.4% per year and
backfill bias to be 7.3% per year.
So, we're talking about a group of funds that in aggregate
probably produced somewhere in the low teens returns and he's
got 11.7% per year combined survivorship bias and backfill
bias. Roger Ibbotson took a look at a
larger group of funds--3,500--funds over a
ten-year period and found survivorship bias at 2.9% per
year and backfill bias at 4.6% per year.
So, huge amounts of institutional funds and
individual funds are going into this hedge fund world.
You look at the returns that are reported for hedge funds in
aggregate--they're generally 12%, 13%, 14% per year for the
last five or ten years. In the case of Burt Malkiel's
data, more than 11% per year and in the case of Roger Ibbotson's
data, between 7% and 8% per year of
those returns can be explained either by backfill bias or
survivorship bias. If you subtract those numbers
from the reported numbers, the returns that the investors
that were actually investing in the funds that are defined as
part of the universe at the time are low,
maybe mid-single digits--far less than people would expect
for the amount of risk that they're taking to be exposed to
this particular group of active managers.
The final point that I want to make with respect to security
selection actually is a little bit different.
It has to do with the degree of opportunity.
This is once you've decided that you're going to be an
active manager and try and pursue market beating
strategies, how do you decide where it is
that you want to spend your time and energy?
Now, I think it's logical that if you're going to try and beat
the markets, you'd want to beat the markets where the
opportunity was greatest. Where's the opportunity
greatest? The opportunity's greatest
where assets are least efficiently priced.
How do you figure out where things are least efficiently
priced? Well unfortunately,
financial economists don't have any direct measures of market
efficiency, but I think there's a story
that you can tell about groups of active manager returns that
will help point you toward those asset classes that are least
efficiently priced. If an asset class has
constituents that are efficiently priced,
then it's very hard to generate excess returns.
As a matter of fact, if things were perfectly
efficiently priced, there wouldn't be any
opportunity to generate excess returns and if you make active
bets--if you make bets against the market--then whether you win
or lose has to do with luck. How are managers going to
behave in an asset class where things are efficiently priced?
Well, they're not going to make big bets, right?
If they do make big bets maybe they get lucky once,
or twice, or three times, but ultimately their luck is
going to run out. And when their luck runs out,
they'll post bad results and get fired.
How do you stay in business? You stay in business by looking
a lot like the market. What market might be
efficiently priced? The bond markets,
in general, and the high-quality bonds in particular
are probably easiest to value. It's all about math.
The government bond, you don't have to worry about
default. Generally, you don't have to
worry about optionality or call provisions and so it's math.
You're given coupon payments every six months and then,
when the bond matures, you get your money back.
So there's not a lot of room in the government bond market or
other high-quality bond markets to generate excess returns.
How about the other end of the spectrum?
The other end of the spectrum is a market that is very hard to
define. As a matter of fact,
there might not even be a benchmark against which you can
measure results and you'd think about the venture capital world.
How do you hug the market in the venture capital world?
You can't; it's very idiosyncratic.
If you're doing early-stage venture investing,
you're backing entrepreneurs and ideas and they're operating
out of their garage. I mean, this romantic notion of
what goes on in Silicon Valley actually still holds true in a
lot of cases but there's absolutely no way,
as a venture capital investor, you could index the venture
capital market. If you look at the behavior of
groups of active managers and the dispersion of returns,
I think it gives you some idea of what the efficiency is with
which assets in these individual assets classes are priced.
Just as I foreshadowed, if you look at the difference
between the first and third quartile in the bond
market--these are active returns over a ten-year period,
again ending June 30,2005--and the fixed income market,
the difference between first and third quartile is a half a
percent per annum. That's an incredibly tight
distribution of returns. Half of the returns are within
a spread of a half-percent. Then as you move out to the
equity markets where it's harder to price things as
efficiently--large-cap stocks--there are two-fold
percentage points, first to third quartile.
Small-cap stocks are tougher to price than large-cap stocks,
so there's a 4.7% differential, first to third quartile.
The hedge fund world is 7.1% first to third quartile,
real estate 9.3% per annum, leveraged buyouts 13.7% per
annum--this is over a ten-year period, so now we're starting to
talk about some pretty significant dispersion.
Of course, in the venture capital world,
the least efficiently priced of all,
there's a 43.2% differential between the top quartile and the
bottom quartile. If I'm going to be active in
terms of managing my portfolio, should I spend my time and
energy trying to beat the bond market?
Where even if you can find somebody who's going to be a
first quartile manager, there's almost no difference
between the first quartile return and the third quartile
return. Or should I spend my time and
energy trying to find the top quartile bond,
top quartile real estate manager,
or buyout manager, or venture capital manager?
I think the answer is pretty obvious.
You want to spend your time and energy pursuing the most
inefficiently priced asset classes because there's an
enormous reward for identifying the top quartile venture
capitalist and almost no reward for being in the top quartile of
the high-quality bond universe. The overall conclusions are
that, with respect to asset allocation, you want to create
an equity-oriented diversified portfolio.
With regard to market timing, you don't want to do it.
And with respect to securities selection, you want to consider
your skills and you want to consider the efficiency of
markets when you're making your decisions as to whether or not
to pursue passive management or active management.
Where did this lead us in terms of Yale's portfolio?
Our current portfolio has 11% allocated to domestic equities,
15% to foreign equities, and 4% to bonds,
so traditional marketable securities account for 30% of
assets. The absolute return portfolio,
which is a group of hedge funds that strive to produce
fundamentally uncorrelated returns, accounts for 23% of
assets; our real assets portfolio,
which includes timber, oil and gas,
and real estate, amounts to 28% of the
portfolio; and private equity,
which includes venture capital and leveraged buyouts,
is 19% of assets. So, 70% of the portfolio is in
absolute return, real assets,
private equity, alternatives--broadly defined.
If you take this portfolio and apply the tests that we
articulated at the outset of the lecture today--equity
orientation and diversification--the portfolio
is clearly equity-oriented; 96% of assets are invested in
some type of vehicle that we would expect to generate
equity-like returns over reasonably long periods of time.
In terms of diversification, there are half a dozen asset
classes with weights that range between 4% and 28%.
So, if you just came down and took a look at that and compared
it to 50% in domestic stocks, 40% in domestic bonds and cash,
and 10% in a smattering of alternatives,
you'd say that this is really a much,
much better diversified portfolio than the one with
which we started. The results have been okay.
Over the past twenty years, we've generated 15.6% per annum
return, but that headline number obviously has a lot to do with
the equity orientation of the portfolio but doesn't describe
the importance of the diversification.
We've had no down years since 1987--1987 that was the crash in
October that I talked about earlier.
In that year, we were early on in terms of
diversifying the portfolio--we'd only been working on that
program for two years--and even so,
the negative return was less than 1%, so it was a modest
negative return. Probably a more important test
of the portfolio was what happened around the collapse of
the Internet bubble in 2000. In the year ending June 30,2001
and 2002, returns for institutional investors were on
average negative in both of those years and actually in
every year since 1987 Yale has had positive returns.
The equity orientation drove the returns but the
diversification allowed us to deliver those returns in a
stable fashion, which is incredibly important
for an institution like Yale that requires a steady supply of
funds to finance its operations. When I started in 1985,
the distribution to the operating budget was $45
million. That represented 10% of
revenues and that was the lowest level for the entire
century--the entire twentieth century--10% of revenues.
The amount that we're spending for the year ending June 30,2008
is $843 million--that represents 37% of revenues--and we're
projecting expenditures for the following year of $1.15 billion.
The results have been really quite extraordinary.
My favorite way to measure the results is actually to compare
what Yale achieved with what we would have had if we would have
just experienced average returns over the past twenty years.
The difference between the average return for colleges and
universities and Yale's returns has added $14.4 billion dollars
to the University's coffers. Whether you measure it in terms
of dollars of value added or in terms of returns,
Yale has the best record among colleges and universities for
the past two decades. So with that,
I'd be happy to take any questions that you might have.
Student: [inaudible] Professor David Swensen:
The question is, if a group of Yalies started a
hedge fund, what would they have to do to convince me to invest
in them? One of the things that we've
done over the years has been open-minded about backing groups
that don't have traditional investment credentials.
If you went to a corporate pension plan or a state pension
manager, they'd have a very bureaucratic process--probably a
fifty or hundred page questionnaire that you had to
fill out, you'd have to deal with
consultants, and you'd have to have ten years or five years
worth of audited performance statistics.
We tend to think that that's not the richest pond within
which we should fish. We think that the more
interesting investment opportunities are kind of
outside of the mainstream with more entrepreneurial firms and
ones that might have less traditional backgrounds.
That said, we just don't take flyers on people that we think
have interesting resumes; we want to have a demonstrated
ability to operate in the markets that the investment
management firm is suggesting that we back.
I would say, part of what we look at are
hard quantitative factors, but probably more important
than the numbers are the soft qualitative attributes.
It's almost like what you looked for in a Boy Scout or a
Girl Scout. You want people of high
integrity. You want people of
unimpeachable character. You want people that are smart,
incredibly hard-working. And in the investment world,
you want somebody who's really obsessed with the
markets--somebody who doesn't define winning by getting as
rich as they possibly can because,
if that's their goal, there are all sorts of things
that they can do to get rich that don't have anything to do
with generating investment returns.
We want people who are maniacally focused on beating
the markets, generating superior investment returns.
That's an incredibly important distinction because,
think about it, if what you want to do is get
rich, you can put together a
reasonable investment record and then raise staggering amounts of
money. Size is the enemy of
performance. So that staggering amount of
money then impairs the fund managers' ability to continue
generating excellent returns, but they can stay in business
and collect the fees that they get for having this huge pile of
money. The type of manager we're
looking for is somebody who strives to generate excellent
returns and they'll frequently raise modest amounts of money
and close to new investors, measuring their success by
beating the market not by generating huge flows of fees
for themselves. It's a combination of looking
at kind of objective attributes and subjective characteristics
and finding people who ultimately will be good partners
for the University. Student: How has Yale's
endowment dealt with the falling house prices?
You said, if we invest in real estate [inaudible]
Professor David Swensen: The question is how we've dealt
with decline in housing prices. We don't have really much of
any direct exposure to homebuilders or to the housing
industry. Most of our real estate
exposure is institutional--acquisitions of
office buildings--largely in major markets--central business
districts. So, you'd find Yale with
interests in office buildings in New York, Washington D.C.,
Chicago, San Francisco, Los Angeles,
some in secondary markets as well, but predominantly in large
metropolitan downtown areas. There are also some hotel
investments, retail properties, smattering of industrial
properties--not a lot of exposure to individual houses.
The only way that we would get that occasionally would be
through some sort of lot-financing activities,
but that's not something that I've generally liked.
I don't think the housing industry, in general,
is a good place to be because of its, sometimes,
violent cyclicality. We did have a large,
short position in subprime mortgage-backed securities,
which has paid off enormously for the University and really
helped protect assets in the past nine months or a year.
I think that, generally speaking--and Bob
Shiller can speak to this with a lot more authority than I
can--this bubble was not something that should have
surprised people. I thought the University
positioned itself well to take advantage of this really not
surprising collapse in housing prices.
Isn't that market timing? I mean, it all depends on your
perspective. I think market timing,
as I've defined it, has to do with short-term
deviations from your long-term policy targets.
I mentioned that our domestic equity target was 11%.
If I came to the office next week and decided domestic stocks
were too high--I want to move that target down to 8%--in the
way that I've described market timing,
that would be a market timing move and we're very careful not
to do that. We establish these targets,
we review them once a year, we don't make changes in many
years, they're quite stable,
and when we do move them we don't move them by a lot.
That doesn't mean that we don't manage the portfolio actively.
So, if we see areas that are particularly interesting,
we're more than happy to deploy capital to take advantage of
what we think are cheap assets or expensive assets.
We made a big bet against Internet stocks in 1999 and 2000
that was very profitable for the University.
As I mentioned, there was a big bet that credit
spreads, both in mortgages and in corporates,
were way too narrow in the past couple of years and that--we
thought that if they were priced rationally those spreads would
widen and we put ourselves in a position to profit from that.
Today, we're looking at opportunities in distressed
securities. A lot of these loans that were
made in 2005 and 2006 and early in 2007 were made at very,
very narrow spreads and there are opportunities out there to
buy bank loans, which are at the very top of
the capital structure, that we believe will be money
good for prices in the '80s. If it turns out that they're
money good, you get your interest and you get $1 for
every $.85 that you invested in a few years.
If markets offer us opportunities,
we're more than happy to take advantage of them. So, we will make valuation bets.
We'll look at things--sectors--say they're
cheap or expensive and exploit the opportunity;
but at least in terms of how I define market timing,
it wouldn't be included in that--it wouldn't be included in
that definition. Student: [inaudible]
Professor David Swensen: The first question is,
what's the beta of the Yale portfolio?
That's not a way that we really think about it,
but I do believe that the risk level of the University's
portfolio is really quite low in statistical terms--much lower
than the risk level that you'd have if you had a traditional
portfolio dominated by marketable securities.
The reason it's low is that we do have, what I think is,
superior diversification and that really lowers the
University's risk. A lot of people look at Yale's
portfolio and say, oh it's risky because you've
got venture capital and you've got timber--we have all these
things that you might believe are individually risky,
but part of the magic of diversification is if you've got
things that are individually risky but they're not well
correlated one to another, the overall portfolio risk
level is quite low. I believe that we have quite a
low risk portfolio. The second part of the question
dealt with the changes in our exposure to foreign assets and
that's an area that we've been very interested in.
Our foreign exposure is not limited to the marketable
security exposure, which I cited as being 15% of
the fund, but there's foreign exposure in
real estate, there's foreign exposure in leverage buyouts,
there's foreign exposure in venture capital.
It's something that permeates the portfolio and,
I think, provides really interesting investment
opportunities because a lot of the foreign markets are less
efficiently priced than those that you find in the U.S.
And I think the fact that our foreign investments are
generally denominated in currencies other than the dollar
is also attractive--a good diversifying tool for the
university. Student: [inaudible]
Professor David Swensen: The question was whether we were
looking to take more short positions as the economy appears
to be moving into recession and I guess the second part of the
question was how do you remain bullish in this kind of
environment. I think the best answer to that
is a quote from one of my contemporaries,
who I think is one of the best investment managers out there.
A guy named Seth Klarman, who works at a fund in Boston
called Baupost, said that what he does is
worries top-down and invests bottom-up.
I read The Wall Street Journal every morning and I
worry about the credit crisis, and I worry about credit cards,
and I worry about auto loans, and I worry about corporate
loans, and I worry about the solvency
of the banking system, and then I go to work and I try
and find the best opportunities that I possibly can.
So, the worrying top-down helps because you don't want to put
yourself in a position where you're going to get hurt by some
adverse macro, sectoral circumstance,
but there's no way that you can take $22.5 billion dollars and
be in the markets when they're attractive and out of the
markets when they're not attractive.
So you just say, okay fine, this is the macro
circumstance that we're dealing with and we're going to do
absolutely the best job we can identifying individual,
specific, bottom-up opportunities to deploy the
funds. Student: [inaudible]
Professor David Swensen: Well, I think one of the
questions---the question is how can you successfully invest in a
market where, I guess, people say you might
catch a falling knife. You buy something that's down
30% but it's got another 50% to go and I think it just has to do
with time horizon. Particularly if you have a
value orientation, you tend to buy things early.
If you bought them with a good, sound, fundamental investment
case and prices are down from where you made your purchase,
have enough dry powder so that you can purchase some more at
the now lower price but have enough confidence in your thesis
to be able to hold the position through the decline and wait for
the markets to recognize the value that you identified.
I think one of the most pervasive problems in the
financial markets is investment with too short a time horizon.
The fact that people look at quarterly returns of mutual
funds is incredibly dysfunctional.
I mean, there's no way that you can expect somebody quarter in
and quarter out or month in and month out to produce superior
returns. There just aren't pricing
anomalies that are significant that are going to resolve
themselves in a matter of months or weeks and so it's a silly
game to play. By extending your time horizon
to three years, or four years,
or five years, it opens up a whole host of
investment opportunities that aren't available to people that
are playing this silly, short-term game.
So, it's not a big deal to buy something at a price that you
think is attractive, have it go down 20,
or 30, or 40%; that ought to be almost a
positive thing because you get a chance to add to the position of
even lower prices, as long as you're ultimately
right that sometime in the three-, or four-,
or five-year time horizon you have your investment thesis
proves out and you're ultimately able to exit the position at a
profit. Student: [inaudible]
Professor David Swensen: The question is about housing
indexes. I'll defer those to Bob
Shiller-- I couldn't answer a question like that in front of
him. Great, thank you very much.
