SPECIAL REPORT ON BIG ENDOWMENT PERFORMANCE
Below, we present a special report ranking twelve of the most-watched U.S. endowments: the Ivys and Alt-Ivys, with the latest data on investment returns and leadership compensation through FY 2013.
We recruit chief investment officers and senior asset managers, so we need to track the performance of the best in the business. The boards and committees who retain us want hard evidence to support their hiring decisions and benchmark their own organizations. This is one of the ways we try to help.
We also analyze the longer-term performance of these important funds, popping open the hood to get a better understanding of their investment process. Who really drives returns? Is it the investment committees making allocation decisions, or active management by the investment office and external managers? What’s the division of labor and who should get the credit (or blame)?
We won’t keep you in suspense. After looking at some academic work and deconstructing some endowment returns, our conclusion is: What the CIO and his/her team do all year, even after the policy portfolio has been set, matters a great deal. Allocation is not destiny.
In our next letter we’ll be recapping and critiquing recent comings-and-goings among the country’s top investment managers and other hot industry topics.
And, we’ll soon be offering a revamped and updated version of our list of outsourced CIO (OCIO) vendors. So, to those of you amongst the ever-growing ranks of OCIO firms, please send me your latest AUM under full discretion and contact information for the list. If potential customers can’t find your number, they won’t call you.
Remember: If you like your newsletter, you can keep it! And that’s a promise!
If you currently enjoy our platinum-level coverage, we’ll keep it coming. And, we invite you to forward our letter to friends and associates; or give us their email address, and we’ll add them to the list.
Skorina in the news:
midsize endowments revisited:
Inspired by our own previous report on the subject, and using some of our stats, our friend Fran Denmark, writing in the currentInstitutional Investor, has a feature-length story about some high-performing but overlooked midsize endowments.
For our take, see Skorina Letter No. 50 archived on our website, here:
For Fran’s story, see:
Using her real-journalism skills, Fran scored in-depth interviews with a lot of the key players: board members, investment committee chairs, chief investment officers, and even a couple of outsourcing firms who serve this market.
If you’re a leader at a midsize-endowment school (or one that’s aspiring to that bracket), or hope to do business with them; you’ll want to give this piece a careful read.
Ivy and Alt-Ivy Endowment Performance: FY2013 Edition:
I’ve been back and forth to New York in recent weeks and I’ve watched a beautiful autumn unfolding in the Great Northeast.
In Ivy League football, if anyone cares, Harvard has just crushed Yale in The Game, which leaves them tied with Princeton for the league championship. The Columbia Lions, alas, are winless at zero-for-seven. But, win or lose, the leafy campuses all look glorious.
Among our Alt-Ivy schools: Notre Dame has won seven out of ten. And here in northern California, the Stanford Cardinal has won eight out of ten. The big Stanford-Notre Dame shootout looms next week in Palo Alto.
In the Ivy League, where football is still more of a pastime than a profession, chief investment officers are far better compensated than the football coaches, and they are locked into their own fierce competition for dominance in the investment game. We’re here to tell you how they’re doing.
We’re going to look at an even dozen of the most closely-watched endowment funds: the eight traditional Ivy schools, plus MIT, Stanford University, University of Chicago, and University of Notre Dame: all top-tier schools with over a billion in endowment assets and run by redoubtable investment pros.
The last of the Ivys, Cornell, reported their results just last month, so now we can finally crunch the numbers.
Here’s our Gang of Twelve, ranked by self-reported 1-year investment return. All use the same fiscal year ending June 30, 2013. The benchmark is the preliminary average returns for endowments over $1 billion reported by the NACUBO-Commonfund NCSE study. We also include variance with respect to a traditional 60-percent stock/40-percent bond portfolio.
Rank by 1-Year Return: FY2013
Longer-run returns are much more important than a one-year snapshot, and we’ll look at those below. But even long-run returns are achieved one year at a time, and a close look at these numbers will give us a sense of how each fund rode out the specific market tides of FY2013.
We first note that MIT and Chicago, who have ridden high in recent years, are ranked 11 and 12 this year. The returns for most schools are clustered between 11.2 and 12.6. The outliers are Penn at the top and Chicago at the bottom, with Chicago missing the NCSE average by 5.2 percentage points.
Closely-watched Harvard, who finished 12th last year, earned a not-bad 11.3 percent, hoisting themselves up into ninth place. And Penn, in the middle of the pack last year, finished first with a strong 14.4 percent return.
The short explanation for this ranking is that stocks had a very strong year. If you were overweight equities you did well; if not, not. But there’s more to it than that. The Russell 3000 was up more than 21 percent; but even a stock-heavy 60/40 portfolio would have brought you only about 10.4 percent, since bond returns were close to zero. That leaves only alternatives to take up the slack, and obviously they had to have pulled their weight in most of these funds to give them overall returns in the 11-to-12 percent range.
In any case, allocation isn’t destiny; good active management by the investment office and external managers is also crucial. We’ll delve more deeply into the allocations vs. execution question as we go along.
The Half-decade numbers:
Now, here are the 5-year annualized returns. We’re recruiters, so we’re more interested in the finite careers of individual managers than the theoretically-infinite lives of institutions. A 5-year return can usually be used to gauge the success of a specific management team versus their peers.
Some of these figures are self-reported. In other cases we computed our own 5-year numbers from public annual returns, and they may be slightly different from the fund’s internal numbers.
Rank by 5-Year Annualized Return: FY09 – FY13
MIT and Chicago, who had off-years in 2013, still stand high in the more important 5-year ranking. Their investment teams have delivered consistently good performance. And Harvard, even with an OK return this year, still stands at the bottom among its peers over the longer period.
More important even than 5-year absolute returns are risk-adjusted returns. This is the gold standard for investment professionals, and the most popular metric is Sharpe ratio. It tells us how many units of return were achieved per unit of risk.
And here we name names, attributing risk-adjusted returns to the responsible chief investment officers. Most of them have been on duty for five years, or at least the last three. The two newbies -Pamela Peedin at Dartmouth and A.J. Edwards at Cornell – are noted below. They haven’t been aboard long enough to be blamed or credited for their school’s 5-year performance.
Rank by 5-Year Sharpe Ratio: FY09 – FY13
The sharpe ratio ranking is very similar to the absolute return ranking. Columbia, MIT and Penn, all of whom had the same investment team over five years, are closely clustered at the top with SRs around 0.4.
Harvard finishes last on this metric, just as Ms. Mendillo finishes her fifth year in office. In fact, this is the second consecutive year in which Harvard finished last for both 5-year absolute return andSharpe ratio.
As recruiters we pay close attention to executive compensation. To put it bluntly: are these schools getting the performance they pay for from their investment offices?
To get some perspective on performance-versus-pay, let’s look at one more chart:
Rank by Leadership Compensation
1. Less than three years in office: Pamela Peedin at Dartmouth and A.J. Edwardsat Cornell.
2. Ms. Kristin Gilbertson resigned her position at the University of Pennsylvania in Oct, 2012. Peter H. Amman became CIO in April, 2013.
3. Ms. Cynthia E. Frost departed Brown University at the end of calendar year 2012. She was succeeded by Joseph E. Dowling in April, 2013.
The W2 compensation figures are from federal tax filings and are for calendar year 2011, the latest available.
The bonus formulas are complicated, confidential, and vary from school to school. Typically, each year’s bonus is based on rolling-average performance versus investment benchmarks over the preceding three or four years.
Suppose we create a little ratio like this: divide 5-year annualized return (in basis points) by comp dollars (in hundreds of thousands). This is, in fact, our notorious Skorina Performance-for-Pay Ratio.
For Ms. Mendillo for instance, we get: 170/52.8 = 3.2. That is: she has generated 3.2 basis points of 5-year return per $100K of her annual compensation.
Here’s the P-for-P ratio for the ten schools with long-serving leaders:
Chief Investment Officer Performance-for-Pay Ratio:
5-Year Return Basis Points per $100K Compensation:
N.B.: Leaders with < 3 Years in office are omitted
Mr. Alexander at MIT gets bragging rights here. His board should be pleased to see that he’s more than earning his keep per this measurement.
We’re not suggesting that 5-year return is the only relevant factor in judging compensation. Ms. Mendillo is the chief executive of a large organization, with years of seniority and a much higher AUM than her peers. In any case, Harvard is entitled to make their own judgment about what their people are worth. It’s their money, after all.
But we think the downward slope in performance-for-pay ratio, from Mr. Alexander’s 47.0 to Ms. Mendillo’s 3.2 isn’t just a statistical trick, and is worth pondering.
Allocation is not destiny:
The 90-percent non-solution:
Once upon a time (1986), an article in the Financial Analyst’s Journal (“Determinants of Portfolio Performance” by Brinson, Hood, and Beebower) argued that 94 percent of the average portfolio’s return variation over time was explained by its policy portfolio.
Many investors concluded that allocation decisions are overwhelmingly the driver of returns. But, according to Yale professor Roger Ibbotson, the inferences people drew from the BHB paper were mostly wrong. He calls it “accepted folklore.”
Most of our sophisticated readers know all this, but it’s taken years to thoroughly discredit that old story. What the BHB paper explained was volatility as a function of asset mix, not relative performance, which is a whole different thing.
(Prof. Ibbotson earned his PhD at University of Chicago, where he taught for a time, and is now on the faculty of the Yale School of Management. He’s a practitioner as well as an academic, as founder and CIO of hedge-fund manager Zebra Capital.)
Writing with colleagues in 2010, he said: “The time has come for folklore to be replaced with reality. Asset allocation is very important, but the variation in returns caused by the specific asset allocation mix is typically nowhere near 90 percent.”
“The question that an investor really needs to answer is: Why does your return differ from mine?”
We will skip over the math, but Prof. Ibbotson, et al concluded that “active management has about the same impact on performance as a fund’s specific asset allocation policy.”
In plain language: among actively-managed portfolios, allocation and execution typically contribute to relative performance on roughly a 50/50 basis. That’s a statistical construct for a large universe, of course; each portfolio’s mileage will vary.
And, more active management is not necessarily better management. That door swings both ways. A great active manager boosts performance; a poor one can crush it. If you don’t have a great active manager handy, then a prudent allocation earning index returns may not be a bad thing.
Speaking of great managers, Yale’s David Swensen provided one more nail in the 90-percent coffin last year, if one was needed, when he wrote: “…for the twenty years ending June 30, 2012, nearly 80 percent of Yale’s outperformance relative to the Cambridge Associates endowment was attributable to the value added by active managers, while only 20 percent was the result of Yale’s asset allocation.” Yale’s returns, that is, were driven by a 20/80 mix of allocation to execution as compared to the 50/50 ratio in Prof. Ibbotson’s larger universe. And, Dr. Swensen was too modest to add that those twenty years were entirely under his aegis.
In our endowment rankings above, how much of the relative performance of the 12 funds is explained by their respective allocations (or policy portfolios) at the beginning of the fiscal year, versus the impact of active management during the year?
It would be nice if we could answer that question for specific funds in specific periods, as Dr. Swensen did for Yale, because it goes directly to the division of labor between the investment committee and investment office, and what the CIO’s work is actually worth. That’s something about which we, as recruiters, are professionally curious.
Any readers who want to see the Ibbotson papers can follow links in an appendix at the end of this report.
Performance forecasting for dummies:
Back in July, Aaron Cunningham, who is the numbers guy at Pensions & Investments (i.e., Director of Research & Analytics), posted an article which estimated the returns at big endowments for the just-ended FY2013, even though the funds wouldn’t be publicly announcing their results until September or October. How did he do that?
Well, if you know the average portfolio allocation for each asset class at the beginning of the FY, which we do (courtesy of NACUBO-Commonfund); and the performance of the major indexes through the end of the FY in June, which we also do; then you just extend the former by the latter, blend them together and voilà: there’s your estimated annual return for the big endowments as a group.
As it happens, we had been fooling around with the same idea. We did our modeling independently, using different indexes in some cases, but we were interested to see that our forecasts were about the same.
Mr. Cunningham concluded that the big endowments would return “a little less than 12 percent.” We came up with 11.6 percent, which led us to think our little model was about right.
Then we awaited the preliminary results from the NACUBO-Commonfund survey for FY2013. On November 6 they announced that the average return for all endowments was about 11.7 percent. They also implied (not quite saying it explicitly) that the over-$1 billion endowments had returned about 11.8 percent.
Well alright. Yay, us (and yay, Mr. Cunningham).
But what difference does it make to our busy readers? We now have the actual, post-facto returns, so who cares about our retro-predictions at this point?
Let’s think about this. Those predictions we made were pure beta. They assumed that the big endowments would earn index returns; no more, no less. In fact, as we saw above, actual returns varied from 6.6 to 14.4.
We also used the (equal-weighted) average allocation for big endowments as calculated by NACUBO, whereas we know (from their annual reports) that the actual (or policy) allocations of the various endowments were very different from each other.
Thinking about Prof. Ibbotson’s argument, it occurred to us that we had enough information to do a crude “attribution analysis” for these endowments, decomposing their total returns to see how much of the peer-to-peer variation was driven by allocation and how much by execution (i.e., active management).
We can’t get as fancy as the consultants (like Cambridge) who provide such analyses to their clients using richer, internal data; but maybe we could see how Ibbotson’s thesis holds up in some particular cases.
More important, we might get some insight into how the division of labor works in individual cases. Allocation is primarily the responsibility of the investment committee (although the CIO usually gets to put his or her oar in), whereas execution falls squarely on the investment office.
In some cases, allocation might explain everything in a given year (which implies that the fund earned overall beta returns). Or, in other cases, returns might be greatly augmented (or undercut) by the work of the CIO.
Here’s how we calculated the contribution of each (of NCSE asset classes to the NCSE total benchmark return of 11.6 percent in FY2013
(N.B.: In some cases it was convenient to use the stats provided by ETF vendors, such as iShares, but the returns are for the underlying index, not the tracking ETF):
Contribution of each (of NCSE asset classes to the NCSE large endowment benchmark return of 11.6 percent in FY2013:
|This eight-bucket breakdown is based on NCSE stats provided to members. It's more detailed than the 5-bucket version in their public tables, which squashes private capital, hedge funds, real estate, and natural resources together into just "alternatives."
We picked what seemed to us to be widely-used, industry-standard indexes, although each endowment gets to pick its preferred benchmarks, or make up its own.
The next and hardest step was to map the stated allocations of our Gang of 12 into the eight NCSE buckets. They all have their own idiosyncratic ways of labeling their allocations and we had to do a certain amount of estimating (i.e., guessing) to re-state them all on an apples-to-apples basis. We won't apologize for this. Clearly, NCSE does the same thing to arrive at their average allocations, although they aren't telling how their sausage was made.
As we puzzled through this exercise we learned to appreciate this bit of wisdom from Dr. Swensen:
"The deﬁnition of an asset class is quite subjective, requiring precise distinctions where none exist."
With allocations all normalized to the eight NCSE buckets, and with indexes and actual returns in hand, we can try to estimate the relative contributions of allocation and execution for the various endowments. (We used actual beginning-of-the year allocations when available; otherwise, we used the policy portfolio as of June, 2012)
We won't impose on our readers by posting all our work, but here's Harvard as an example:
Harvard Endowment: Contributions of Allocation vs. Active Management to FY13 Returns:
|Note: Decomposing Harvard-NCSE variance: -0.1 percent due to allocation + (-0.4) percent due to alpha = -0.5 percent total variance (11.3 - 11.8 = -0.5)
Conclusion: about 20 percent of variance from NCSE is due to allocation; about 80 percent due to execution.
If we did our sums right, it appears that only about 20 percent of Harvard's return can be attributed to allocation, and about 80 percent to execution. Coincidentally, that's the same ratio claimed by Yale over 20 years.
If Harvard had used our indexes (which they didn't) we think their policy portfolio would have earned about 11.5 percent. Their actual return was slightly below that, at 11.3 percent. Per our model, that negative 20 basis point difference must be blamed mostly on execution. In other words, they had a slightly negative alpha.
How about Penn? They're number one for the year, beating the big-endowment benchmark by a handsome 260 basis points for FY2013: 14.4 versus 11.8 percent. How did they do that?
Per our model, 1.6 percent of that variance can be attributed to differences from the average NCSE allocation. Specifically, they had about 45 percent in equities versus only 31 percent for the NCSE average. That was a good place to be in FY2013.
They were also underweight in hedge funds (9 versus 20 percent) in a year when HF returns were mediocre relative to public markets (they returned just 7.3 percent per our HFRI index). But, overall, Penn beat our indexes, contributing 1.2 percent of execution alpha to their total return. So their return was driven about 43 percent by allocation and about 57 percent by execution.
I recently had a pleasant conversation with Ms. Gilbertson who left Penn late last year. I don't think she wants to be quoted directly, but I shall paraphrase. Although her allocations were better than most when the storms came in 2008/2009 -- for which Penn was rightly praised at the time -- she says there was a lot more to Penn's success. While coloring inside the allocation-lines, her team did a whole lot of active managing, fighting for every basis point.
Our allocation-versus-execution analysis pertains only to Ms. Gillbertson's last year in office, but it's certainly consistent with her account.
She gives particular credit to Penn's public-market chief, David Harkins, who also acted as interim CIO for a while after Ms. Gilbertson's departure.
I would add that I think Howard Marks, founder of Oaktree Capital, who chaired the investment committee from 2000 to 2010, had a lot to do with creating a high-performance endowment at Penn.
Just one more. Let's look at Chicago, whose 5-year performance has been good, but who had an off-year in FY2013 and finished 12th out of 12 relative to its peers.
Mr. Schmid's team had a negative 5.1 variance against the big-endowment average: 6.6 versus 11.8 percent. We think about 1.9 percentage points of that total negative variance can be blamed on allocation. They were sitting on only 13 percent equities at the beginning of the FY, versus 31 percent average for the big endowments. That generated a negative 3.1 percent toward total return. On the other hand, they were overweight real estate, hedge funds and natural resources, which made a 1.2 percent positive contribution. So the effect of allocations nets out to negative 1.9.
Our model attributes the other negative 3.0 percent of return variance to execution. As we interpret it, they fell short of our index returns overall, although the limitations of our data preclude pinpointing the asset classes which underperformed.
And, again, we have to emphasize that no fund is compelled to use our indexes. Mr. Schmid may or may not judge his own returns as satisfactory relative to his own internal benchmarks and the specific needs of his school. And, even including FY2013, Chicago still stands 4th out of 12 for returns over five years among its peers, which is quite good.
About 37 percent of Chicago's return variance seems to be allocation-driven, and about 63 percent execution-driven for FY2013.
Overall, our crude analysis seems to bear out Prof. Ibbotson's thesis. While we're looking at only a few endowments in a single year, we haven't seen any case where allocation explains anything close to 90 percent of variance against the NCSE benchmark returns. In fact, the allocation/execution ratio in our small sample ran closer to 40/60.
What the CIO and his/her team do all year, even after the policy portfolio has been set, matters a great deal. Allocation is not destiny.
One more look at Harvard:
Harvard is by far the world's biggest endowment, which guarantees a certain level of not-always-sympathetic scrutiny. To the extent we've been critical of the good people at HMC, we think we've also been civil and supported our comments with hard data.
Some other commentators have not been so restrained.
A few weeks ago Dan Primack, a senior editor at Fortune looked at HMC's 5-year performance and had this to say:
...relative to other large schools (or to broader market indices)...Harvard has been a virtual trainwreck.
Mr. Primack continues:
Those close to the situation offer three defenses for Harvard's sub-par performance:
1. The larger you are, the harder it is to outperform.
2. Harvard has much less exposure to domestic public equities than do many other endowments, which have been able to ride the recent bull run.
3. Harvard has a particularly large slug of illiquid assets, some of which it felt compelled to sell at deep discounts during the financial crisis. Those sales continue to be a drag on five-year returns.
...each of those arguments are [sic] easily countered.
First, Harvard has always been the nation’s largest endowment — but it had little trouble outperforming in the past. For example, through June 30, 2012 it ranked #5 over a 20-year period.
Second, it certainly is true that Harvard’s rank has slipped due to its relative dearth of public equities. But other schools with even smaller public equity allocations — such as Yale — have performed much better.
Third, Harvard has actually increased its target exposure to illiquid assets over the past five years, including in private equity and real estate. So not only did current management approve the secondary sale (arguably too soon), but it also may be setting itself up for a similar liquidity crunch were the economy to tank again.
Regarding HMC’s private equity portfolio, we used 15.4 percent as our index return for “private capital” (i.e., private equity). That’s theCambridge Associates Private Equity Index, which looked pretty respectable to us. They even calculate it right there in Boston.
Harvard’s own PE benchmark was 10.6. So, their 11.0 percent PE return managed to beat their own benchmark by a scant 0.4 percent, but it missed the CA index by a fat 440 basis points.
This is one of several instances where Harvard has beaten its own benchmarks while still, somehow, underperforming against its peers.
Ms. Mendillo wrote in the 2013 annual report:
I would characterize our Private Equity performance this year as fair…, a strong nominal return, but well below the return on public market equity, and only slightly above our benchmark…While this asset class still presents unique opportunities for attractive returns, it has gotten much more crowded and there is less of an illiquidity premium. As a result, we are actively focused on honing our private equity strategy.
That strikes us as pretty faint praise. In fact, when you unpack it, it sounds more like chagrin.
We note that HMC recently hired a new head of PE to succeed departed Peter Dolan. This may be some of the “honing” Ms. Mendillo promised.
Mr. Dolan left back in May, after 20 years on the job, and neither he nor Harvard had much to say about his departure.
The new guy is Lane MacDonald, whom they’ve moved over from the public equity group (although he also has deep prior experience in private equity). Mr. MacDonald, in addition to his investment smarts, is a local hero for having led Harvard to its only NCAA hockey championship in 1989.
Interim PE head John Shue was a plausible candidate for the top job, but was passed over. The well-informed Mr. Primack has also revealed that Mr. Shue is the brother of Hollywood beauty Elisabeth Shue, herself a Harvard alumna (BA, Political Science, 2000).
And, as for PE becoming an over-crowded space, we note that Yale, which has a much bigger PE commitment, has announced a cutback in its allocation for the first time anyone can remember, which may be a harbinger for other top institutions.
Public pensions, on the other hand, hungry for returns, seem to be doubling down on the category. Of course, that would be just fine with the general partners, since the pensions are a much bigger market.
If this is so, it wouldn’t be the first time less sophisticated investors jumped into a sector just as the smart guys were backing out. We’ll see how that works out.
Papers by Roger Ibbotson, et al, on the role of allocations in portfolio returns:
01 Ibbotson and Kaplan: Does Allocation Policy Explain 40, 90, or 100 Percent of Performance?
Financial Analyst’s Journal, Jan/Feb 2000
02 Xiong, Ibbotson, Idzorek, and Peng: The Equal Importance of Asset Allocation and Active Management
Financial Analyst’s Journal, March/April 2010.