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
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.