Return Advantage of Stocks over Bonds Has Shrunk to Lowest in 15 Years But US Endowments’ Equity Beta Likely Highest Ever
A review of the asset allocation history of the largest American endowments shows that the amount of equity risk currently being taken is at the highest level in at least 25 years — and likely ever. This is striking when one considers that the S&P 500 has almost fully recovered last year’s losses and is in the top decile most expensive versus the last ~50 years. And after the massive spike in short and long-term interest rates, the stock-bond risk premium is the smallest in 15 years.
In the table above from a June 15th Goldman Sachs report, the S&P 500 valuation is at the 90th percentile most expensive versus history on eight widely used valuation metrics. Given that the stock market has moved up another ~ 10% since the report was published, absolute valuations look even richer for US stocks today. And after more than a decade of near zero real bond yields (on average), 10 year TIPs are now trading in a range of 1.5-2.0%, a nearly 300 basis point increase in less than 12 months!
As a result, the return advantage of stocks over Treasury bonds has dramatically narrowed. The bottom of the Goldman table above shows that the “yield gap” (i.e., the stock-bond risk premium) is around the 80th percentile smallest versus nominal and real 10 year US Treasury Bonds. The graph below shows that the risk premium calculated as the difference between the S&P 500 earnings yield (trailing 12 months) and 10-year nominal US Treasury bonds – is slightly negative. This is the smallest since the end of the global financial crisis (GFC) 15 years ago.
In other words, the stock-bond risk premium has fallen dramatically yet endowment allocations through the end of March this year show that hardly any change had occurred to mean asset exposures since rates began their surge higher in the spring last year. There’s been multiple reports of pension funds, family offices, and retail investors moving or thinking of moving dollars out of equities but I’ve not seen any about endowments considering doing likewise.
Of course some endowments have surely been making some changes to their strategic asset allocations (SAA) but if the averages haven’t budged, many must be doing very little – or those who are reducing beta are offset by those increasing. It’s also possible that average equity risk exposures were cut after March – after CIOs and ICs had the opportunity to assess and decide.
Some may argue that of course endowments haven’t responded to the 300 basis point jump in real bond yields, the top schools invest with the best managers and those managers generally demand that the capital be locked up for some period of time. But while it’s true that a majority of large endowment portfolio’s are illiquid and can’t be changed quickly, most have the ability to use derivatives to change asset exposures without touching the underlying funds if they believed a change was warranted (though this brings a whole new set of issues).
Strategic doesn’t mean fixed. Either an endowment does asset allocation or it doesn’t. If they say they do but aren’t currently thinking actively about potential asset shifts after such a profound change in relative valuations, when on earth would they?
All of which calls to mind my boyhood idol, that paragon of indifference, MAD Magazine’s Alfred E. Newman. CIOs sitting on portfolios more exposed to equity risk than ever, after a 5% jump in cash rates and almost 3% rise in real bond yields, seem to be saying, What, me worry?
US Endowments and Equity Risk: a 120 Year View
Endowment equity risk has been increasing steadily since the end of the GFC. Last year Cambridge Associates reported mean asset allocation data for a constant universe of 82 large and medium-sized endowments for the 20 years ending in 2021. I’ve taken those figures and for the purpose of this exercise assumed constant equity betas over time for each of the asset classes: 1.0 for public equities, 1.0 for PE, 0.5 for hedge funds, 0.6 for RE, and 0.3 for fixed income. These beta estimates are consistent with the correlations and volatility estimates used by Cambridge, JPMAM, and others for mean/variance analysis. The results show that equity risk, after falling because of the 50% drop in stocks in the GFC, increased by over 15% (.7 to .8) since then. The absolute level is less important that the fact that regardless of whether you use somewhat higher or lower beta estimates, the picture is the same: steadily rising equity risk since 2009 and current levels higher than at any time since 2001. The fact that the post-GFC period was one of exceptionally strong equity returns isn’t coincidental. Rebalancing regularly since 2009 would have substantially reduced returns while “letting it ride” (i.e., allowing equity risk to drift higher as equity investments rose in value) would have been the highest returning strategy of all.
Further evidence that equity risk has been rising and is as high as it’s ever been can be seen in an article by Richard Ennis from May of last year. Ennis took the returns of US endowments with $1 billion or more in assets and calculated the correlations and betas to the Russell 3000 over time periods ranging from the trailing 13 years to the trailing 5 years, all ending 6/30/2021.
The results show that over the 5 years ending in 2021, large US endowments had a beta of 0.97! This compares to a beta of ~ 0.8 over the entire 13 year period. So almost all of the returns of the average large US endowment for the 5 years ended 2021 can be explained by the returns of one asset class: US stocks.
In other words, for all intents and purposes, large endowments are not the slightest bit diversified by asset class. Their portfolios have no significant asset class exposures that diversify equity risk. Assets like private credit, high yield, private real estate, most hedge funds — all do little for diversification because each contributes substantial equity risk.
The most recent asset allocation data I’ve seen, also from Cambridge Associates, is as of March 31st this year. Comparing it to CA data from 2021 (before rates exploded higher) and using the same asset class beta estimates as before, I found that the average portfolio beta is essentially unchanged. I.e., average endowment equity risk was very high in 2021 and through Q1 this year, hasn’t changed since.
Finally, data shown in a fascinating 2020 article in the Financial Analysts Journal, provides evidence that endowment equity risk is likely to be the highest since 1900. In Seventy-Five Years of Investing for Future Generations (Chambers, Dimson & Kaffe), the authors describe the evolution of endowments from their inception some 700 years ago in England to what they are today — as well as provide a history of the returns and asset allocations of 12 of the largest and oldest endowments in the US (the ivy league schools plus Stanford, U of Chicago, MIT, and Johns Hopkins). Compiled by hand from university documents, the results – shown in the four charts nearby) – are illuminating as to large endowment equity beta during the 20th century.
It’s helpful to consider the charts in two distinct periods. From 1900 to about 1970, we see a large fall in fixed income and a large rise in public equities, such that by 1970 (roughly the peak in equity weights), total portfolio beta was likely around 0.6. Nothing ambiguous about this; equity beta risk in 1900 was minimal but by the 1970s endowments had ~ 60% in stocks. As a rough and ready estimate, the equity beta of a portfolio with 60% in US stocks and 40% in US fixed income will be around 0.6.
The picture is less clear during the second period beginning in the 1970s. Bond weights continued to fall (down about 30%) but stocks also fell (~30%) while a whopping ~60% went into alternatives. Given that almost all alternative investments have substantial embedded equity risk, given that fixed income was the only diversifying asset in the portfolio, and given that we know from the Cambridge data cited above that endowment portfolios at the end of the period had a beta of approximately 0.8, it seems very likely that the rise in equity beta continued from the 70s to the 2010s, only via alternatives instead of public equities.
Does it matter?
It’s true that in one sense history isn’t necessarily relevant to how endowment portfolios should be constructed today. The fact that equity beta is likely as high as it’s ever been doesn’t in and of itself mean anything for current exposures. Equity exposures were far lower during much of the past than anyone would consider optimal today. There may be new information which justifies a more optimistic return forecast for stocks and no change to current equity risk exposure.
But in this case, at this point in time, I think the history is absolutely relevant. If a CIO believes the embedded equity risk in their portfolio is acceptable, they should bear in mind that the risk being taken may be as high as it’s ever been for their institution. Maintaining this “highest ever” equity risk portfolio when the excess return of stocks over bonds is the lowest in 15 years, when US stocks are top decile most expensive, and after four decades of disinflation just ended — seems to me to be a very aggressive position. In the lingo of the US Presidential campaigns now underway, there would seem to be an extremely “narrow path to victory” for a portfolio with a equity beta approaching 1.0 with that set of facts.
Related Posts on American Carnage:
Note: this article focuses on asset allocation for US institutions. Non-US institutions face a very different set of circumstances and in the interests of keeping this article less than book length, I will not address them here.
Why Might US Endowments Be Maintaining Equity Risk at All-Time Highs?
Are we going back to zero?
Why might CIOs and investment committees be comfortable with (mostly) unchanged asset weights? What would have to happen for current high equity beta portfolios to outperform more diversified (by asset type) portfolios over the next ten years (the typical forecast period for making SAA decisions)?
In my view, the only potentially compelling reason would be the belief that real bond yields will fall back to something like 0.5% as the disinflationary forces of recent decades eventually return and overwhelm the shorter-term factors pushing inflation higher since Covid. All else being equal, if real yields fall, stocks will do better. Economists appear to be broadly split on this question which tells you that there’s an important debate going on here.
The equilibrium or mid-cycle level of real short and long-term bond yields is a complex topic and beyond the scope of this article. But to summarize, broadly speaking there are two camps. The first, represented by the view given by Olivier Blanchard, former Chief Economist at the IMF, is that real bond yields in the next decade won’t be much higher than they were during ZIRP and QE as the forces of “secular stagnation” continue. This group views the ZIRP years as consistent with a long-term trend of declining real rates and Covid as a temporary deviation from it. They cite papers such as this one from Bank of England staff showing that according to a “global real rate” they constructed, there’s been a steady decline in real interest rates for 7 centuries.
The thinking of the other camp is exemplified in the views expressed by Larry Summers in a recent debate with Blanchard, and by Bill Dudley, former Head of the NY Fed, in a recent Bloomberg column. This group focuses more on secular changes highlighted during Covid such as the reversal of globalization that put downward pressure on inflation for the last 25 years. They view the ZIRP years as the aberration and Covid as a return to a more normal level of real yields.
My view is consistent with that of Larry Summers and Bill Dudley — but I’m not an economist and can’t say I have huge conviction in that view.
To engage with the the debate on this question, it’s necessary to learn a bit of macroeconomic jargon, in particular, “r-star” This is defined as the real short-term interest rate likely to occur when the economy is growing at trend rates and inflation is stable — essentially it’s an equilibrium real cash rate. So if an economist says they think r-star will be 0.5% on average over the next decade, we can add a term premium to get a forecast for real bond yields.
In addition to the perspective provided by economists, I think the following points are relevant for allocators betting on a “back to zero” scenario.
Real yields are still somewhat lower than longer-term average US levels
First, level-setting. Real ten-year US Treasury yields (calculated as the 10 year US Treasury yield minus CPI) have averaged 2.3% over the 130 years before the Fed implemented the first QE and dropped rates close to zero. Since being introduced in the US in 1997, the 10 year TIPs yield has averaged 3.4% but a after an initial period where investors got comfortable with the new securities, yields averaged 2% to 2.5%. So this suggests the period of zero real bond yields from 2010-2020 was an anomaly and that even real bond yields of 1.5-2% as we see currently are somewhat lower than longer term averages.
Ultra-low bond yields have been rarer than black swans for 5000 years
Historically, ultra-low bond yields are rare — whether looking at US history or the history of human civilization. Plato is the father of Western philosophy but even his prodigious intellect simply wasn’t developed enough to conceive of lending for no return. The chart below comes from a book beloved by no one except fixed income geeks, Sydney Homer’s A History of Interest Rates. The lowest recorded interest rate in the West from 3000 B.C. to 500 A.D. according to Homer (not the Iliad-Homer, the bond geek-Homer)
was 4%. Of course we don’t have core CPI for Athens or Rome so on this evidence it’s impossible to know if the ancient world experienced lengthy periods of negative real yields — but most of the time nominal rates were high single digits or greater (the Babylonians appear to have invented the junk bond market). The same can be seen in just about any historical period where there is data. Below you can see that in three centuries, UK bond yields never touched 2% until the GFC.
The effects of QE and ZIRP after the Financial Crisis make a repeat unlikely
Would the Fed turn to QE and ZIRP again if conditions warrant? In my view, it’s questionable whether QE and ZIRP “worked” and hence will be repeated in future recessions. There were obvious benefits to negative rates after the GFC given that the economy was devastated and Congress chose to do as little fiscal stimulus as possible – so the Fed was the only fireman on the scene.
But the negative effects have become equally obvious. To a greater or lesser extent in all developed nations, wealth inequality is the highest since the 1920s while lower income families commonly pay a third to a half of their income for a place to live as ~ zero real bond yields acted as rocket fuel for home prices around the world.
Venture-backed start-ups had billions thrown at them in a frenzy of late-cycle speculation. Pampered cashed-up entrepreneurs replaced real innovation with “blitzscaling” — achieving massive scale quickly based on nothing but astronomical marketing spend. Near-zero yields made PE buyouts insanely attractive so oceans of capital poured in and now PE is a major player in many industries — including healthcare — spurring a backlash from politicians across the political spectrum. For all these reasons, it isn’t obvious that even in a bad recession, there would be the political will to rescue the economy by handing yet another colossal windfall to wealthy investors and Wall Street by keeping rates at/near zero for an extended period of time.
There’s also the small matter of exiting a long period of ultra-accommodative monetary policy. In the US the consequences have been destabilizing to the banking system and are still unfolding. Japan is just beginning to tip toe away from ultra-low rates but with huge government debt levels, no one is relaxed about the risk of a spike in bond yields or other collateral damage after years and years of free money.
But, but, but… AI!
Other justifications for not making changes to equity-dominant portfolios seem far less compelling. One is the idea that there will be a near-term productivity and profits bonanza from new AI products like Large Language Model chatbots. I’m skeptical. Productivity, sure. Meaningful aggregate S&P 500 profit growth, soon enough to matter for a net present value calculation? I’ll take the unders on that one.
Complacency is understandable
After decades of of rising stock valuations and declining interest rates, after ten years of zero real bond yields, the right answer about strategic stock allocations was always, “more, please!” Generally speaking, those who took the most equity risk did the best while those who attempted to manage portfolio risk underperformed. It’s hard to worry about something when not worrying about it has been instrumental in landing your job! Have decades of robust stock returns made it hard to conceive of a world where equity risk doesn’t always pay off in the end?
Feeling tied up?
Or is it the case that since falling rates and higher valuations meant that taking liquidity risk was generally always a good decision in past decades, CIOs have been lulled into a false sense of security about illiquidity? Though the large endowments can use derivatives to adjust major asset exposures, this is far from simple given the complexity of the underlying investments. Have endowments allowed portfolios to become so illiquid and complex that they feel like Gulliver, Jonathon Swift’s giant, so tied down by illiquid investments that moving feels impossible?
The current environment has exacerbated the structural illiquidity of endowments. Distributions from private investments have dropped to the lowest levels in a decade, while selling PE fund interests in the “secondaries” market means taking a 10-25% haircut on NAV. Preqin reports that secondary volume, while up from last year, is well below expectations so it seems CIOs for the most part aren’t selling PE. In addition, endowments over a billion had uncalled capital from private funds worth 14% of the endowment with some having as much as 25% according to a CA report from 2021. Being short a call on your liquidity of that magnitude would make anyone feel constricted.
Worse still, the largest endowments have at least 5% and sometimes 10% or more invested (through VC funds and directly) in Chinese VC-funded companies like ByteDance (owner of TikTok) and Didi Chuxing (taxi company) with no apparent means of listing in Hong Kong or NY, while interest is minimal from potential strategic acquirors given the US/China conflict. This effectively makes up a “frozen block” of equity risk in large endowment portfolios. Adjusting to the profound change in relative valuations that’s taken place might seem impossible given the complex and illiquid make-up of top endowments.
How not to impress your boss
The most important determinant of returns over very long time periods is asset allocation. Everyone knows this but we sometimes act as though we don’t. Remembering was especially difficult during the ZIRP years, when SAA was more or less irrelevant.
As a result, allocators spent more and more of their time searching high and low for individual funds which, even if they invested, mattered less and less to total endowment returns. According to Cambridge Associates, the average $1 billion plus-sized endowment has 140 asset managers with some having as many as 250!
The 141st fund won’t make a bit of difference but it’s still far more interesting for allocators to go deep on a fund or strategy – i.e., on people — than it is discussing dusty old models like the stock-bond risk premium. Perhaps more to the point, getting IC members to think deeply about r-star is unlikely to make them love you.
So it’s not surprising that many allocators haven’t spent much time in recent years thinking seriously about SAA.
A Time for Choosing
Put away your spreadsheets, no regressions required
Readers will notice this article isn’t loaded with numbers and equations and statistical analysis. Sometimes those things are necessary but today they aren’t. Investors are fortunate to be faced with crystal clear signals about the relative value between stocks and bonds. It doesn’t take a mean/variance optimization or Monte Carlo simulation to understand. Short rates have increased 5% while ten year bonds have doubled to 4%: fixed income is much cheaper. US equities, after dropping over 20% last year, have recovered most of what they lost: equities remain very expensive versus history. As a result, the return advantage of stocks over bonds has fallen by ~ 250 basis points.
This is a time for choosing. The fact that endowment equity risk has likely never been higher, though obviously not new, when considered in the context of a top decile most expensive stock market and massively cheaper bonds — should at least spur questioning, debate, and decision among CIOs and investment committees.
If I was on an endowment IC and the CIO was keeping allocations broadly unchanged compared to 2021, the question I would ask is:
- If X was the optimal equity risk when the stock-bond risk premium (using the parameters in the graph above) was 2.5%, why is X still the optimal exposure now that the SB risk premium is zero%? Is your asset allocation fixed?
Whether endowments make changes or stick to their equity guns, in my view the biggest mistake would be to not make an explicit decision and sleep-walk into a far less equity-friendly future. Being a great investor means sometimes doing very little for a long period and then doing a lot in a short period. Which is a big reason why great investors are rare.
Author’s Note: This will be the last substantive investment-related post I write for the American Carnage blog as I’m focused now on writing a book. I’ll continue to post short investment-related thoughts on LinkedIn from time to time – so follow me there if you want to keep tabs on what I’m up to.
Joe I’ve learned a lot following you. I love this passage ,
“The 141st fund won’t make a bit of difference but it’s still far more interesting for allocators to go deep on a fund or strategy – i.e., on people — than it is discussing dusty old models like the stock-bond risk premium. Perhaps more to the point, getting IC members to think deeply about r-star is unlikely to make them love you.”
I guess it’s always been a beauty contest
Comments are closed.