GMO letter for the third quarter ended September 30, 2016.
Rather like a parrot I have been repeating for 10 quarters now my belief that we would not have a traditional bubble burst in the US equity market until we had reached at least 2300 on the S&P, the threshold level of major bubbles in the past, and at least until we had reached the election. Well, we are close on both counts now. My passionate hope was that I would then, perhaps 6 months after the election, recommend a major sidestep of the coming deluge that would conveniently have arrived 6 to 12 months later, allowing us then, after a 50% decline, to leap back into cheap equity markets enthusiastically, more enthusiastically, that is, than we did last time in 2009. Thus we would save many of our clients tons of money as we had (eventually) in the 2000 bust, at least for those clients who stayed with us for the ride, and in 2007. I consider myself a bubble historian and one who is eager to see one form and break: I have often said that they are the only really important events in investing.
I have come to believe, however, very reluctantly, that we bubble historians have, together with much of the market, been a bit brainwashed by our exposure in the last 30 years to 4 of the perhaps 6 or 8 great investment bubbles in history: Japanese land and Japanese equities in 1989, US tech in 2000, and more or less everything in 2007. For bubble historians eager to see pins used on bubbles and spoiled by the prevalence of bubbles in the last 30 years, it is tempting to see them too often. Well, the US market today is not a classic bubble, not even close. The market is unlikely to go “bang” in the way those bubbles did. It is far more likely that the mean reversion will be slow and incomplete. The consequences are dismal for investors: we are likely to limp into the setting sun with very low returns. For bubble historians, though, it is heartbreaking for there will be no histrionics, no chance of being a real hero. Not this time.
Warren Buffett On The Dangers Of Using Complex Math In Investing
When he's trying to figure out if a company is a good investment, Warren Buffett does not rely on complicated formulas, spreadsheets, and higher-level math. He believes higher mathematics may actually be "dangerous and will lead you down pathways that are better left untrod," if used in the investment process. Q2 2020 hedge fund letters, Read More
The 2300 level on the S&P 500, which marks the 2-standard-deviation (2-sigma) point on historical data that has effectively separated real bubbles from mere bull markets, is in this case quite possibly a red herring. It is comparing today’s much higher pricing environment to history’s far lower levels. I have made much of the convenience of 2-sigma in the past as it has brought some apparent precision to the more touchy-feely definition of a true bubble: excellent fundamentals irrationally extrapolated. Now, when this definition conflicts with the 2-sigma measurement – ironically, it was chosen partly because it had never conflicted before – I apparently prefer the less statistical test. But you can imagine the trepidation with which I do this.
Hidden by the great bubbles of 2000 and 2007, another, much slower-burning but perhaps even more powerful force, has been exerting itself: a 35-year downward move in rates (see Exhibit 1), which, with persistent help from the Fed over the last 20 years and a shift in the global economy, has led to a general drop in the discount rate applied to almost all assets. They now all return 2-2.5% less than they did in the 1955 to 1995 era (or, as far as we can tell from incomplete data, from 1900 to 1995).
This broad shift in available returns gives rise to the question of what constitutes fair value in this changed world; will prices regress back toward the more traditional levels? And if they do, will it be fast or slow?
Another contentious question is whether abnormally high US profit margins will also regress, and, if so, by how much and how fast? (This will be discussed in more detail next quarter.)
Counterintuitively, it turns out that the implications for the next 20 years for pension funds and others are oddly similar whether the market crashes in 2 years, falls steadily over 7 years, or whimpers sideways for 20 years. The real difference in these flight paths will be, of course, over the short term. Are we going to have our pain from regression to the mean in an intense 2-year burst, a steady 7-year decline, or a drawn-out 20-year whimper?
The caveat here is that while I am very confident in saying that we are not in a traditional bubble today, all the other arguments below are more in the nature of thought experiments or, less grandly, simply thinking aloud. I am asking you – especially you value managers – to think through with me some of these varied possibilities and their implications. What follows is my attempt to answer these, for me, very uncomfortable questions.
GMO: The Case for a Whimper
1. Classic investment bubbles require abnormally favorable fundamentals in areas such as productivity, technology, employment, and capacity utilization. They usually require a favorable geo-political environment as well. But these very favorable factors alone are not enough.
2. Investment bubbles also require investor euphoria. This euphoria is typically represented by a willingness to extrapolate the abnormally favorable fundamental conditions into the distant future.
3. The euphoric phases of these epic bull markets have tended to rise at an accelerating rate in the final two to three years and to fall even faster. Exhibit 2 shows four of my all-time favorites. True euphoric bubbles have no sound economic underpinning and so are particularly vulnerable to sudden bursting when some unexpected bad news occurs or when selling just starts… “comes in from the country” as they said in 1929.
4. We have been extremely spoiled in the last 30 years by experiencing 4 of perhaps the best 8 classic bubbles known to history. For me, the order of seniority is, from the top: Japanese land, Japanese stocks in 1989, US tech stocks in 2000, and US housing, which peaked in 2006 and shared the stage with both the broadest international equity overpricing (over 1-sigma) ever recorded and a risk/return line for assets that appeared to slope backwards for the first time in history – investors actually paid for the privilege of taking risk.
5. What did these four bubbles have in common? Lots of euphoria and unbelievable things that were widely believed: Yes, the land under the Emperor’s Palace really did equal the real estate value of California. The Japanese market was cheap at 65x said the hit squad from Solomon Bros. Their work proved that with their low bond rates, the P/E should have been 100. The US tech stocks were 65x. Internet stocks sold at many multiples of sales despite a collective loss and Greenspan (hiss) explained how the Internet would usher in a new golden age of growth, not the boom and bust of productivity that we actually experienced. And most institutional investment committees believed it or half believed it! And US house prices, said Bernanke in 2007, “had never declined,” meaning they never would, and everyone believed him. Indeed, the broad public during these four events, two in Japan and two in the US, appeared to believe most or all of it. As did the economic and financial establishments, especially for the two US bubbles. Certainly only mavericks spoke against them.
6. Let me ask you: How does that level of euphoria, of wishful thinking, of general acceptance, compare to today’s stock market in the US? Not very well. The market lacks both the excellent fundamentals and the euphoria required to unreasonably extrapolate it.
7. Current fundamentals are way below optimal – trend line growth and productivity are at such low levels that the usually confident economic establishment is at an obvious loss to explain why. Capacity utilization is well below peak and has been falling. There is plenty of available labor hiding in the current low participation rate (at a price). House building is also far below normal.
8. Classic bubbles have always required that the geopolitical world is at least acceptable, more usually well above average. Today’s, in contrast, you can easily agree is unusually nerve-wracking.
9. Far from euphoric extrapolations, the current market has been for a long while and remains extremely nervous. Investor trepidation is so great that many are willing to tie up money in ultra-safe long-term government bonds that guarantee zero real return rather than buy the marginal share of stock! Cash reserves are high and traditional measures of speculative confidence are low. Most leading commentators are extremely bearish. The net effect of this nervousness is shown in the last two and a half years of the struggling US market shown in Exhibit 3, so utterly unlike the end of the classic bubbles.
10. I just finished a meeting on credit cycles in which yet another difference between today’s conditions and a classic bubble was revealed: They – the bubbles in stocks and houses – all coincided with bubbles in credit. Exhibits 4 and 5 show the classic spike in credit in Japan in 1989, coincident with the high in stocks; the US 2000 tech event and the US housing boom of 2006 also coincided perfectly with booms in credit. Whether stock and house prices rising draws out credit or credit pushes prices, or whether both interact, which seems more likely, is a story for someone else to tell. Credit, needless to say, is complex and there are very often individual credit components that are worrying. For example, financial corporate debt looks fine, but non-financial corporate debt is scary. What is important here is the enormous contrast between the credit conditions that previously have been coincident with investment bubbles and the lack of a similarly consistent and broad-based credit boom today.
11. The current market therefore is closer to an anti-bubble than a bubble. In every sense, that is, except one: Traditional measures of value score this market as extremely overpriced by historical standards.
12. At GMO we have put particular weight for identifying investment bubbles on the statistical measure of a 2-sigma upside move above the long-term trend line, a measure of deviation that uses only long-term prices and volatility around the trend. (A 2-sigma deviation occurs every 44 years in a normally distributed world and every 35 years in our actual fat-tailed stock market world.) Today’s (November 7) price is only 8% away from the 2-sigma level that we calculate for the S&P 500 of 2300.
13. Upside moves of 2-sigma have historically done an excellent job of differentiating between mere bull markets and the real McCoy investment bubbles that are likely to decline a lot – all the way back to trend – often around 50% in equities. And to do so in a hurry, in one to three years.
14. So we have an apparent paradox. None of the usual economic or psychological conditions for an investment bubble are being met, yet the current price is almost on the statistical boundary of a bubble. Can this be reconciled? I believe so.
15. There is a new pressure that has been brought to bear on all asset prices over the last 35 years and especially the last 20 that has observably driven the general discount rate for assets down by 2 to 2.5 percentage points. Tables 1 and 2 compare the approximate yields today of major asset classes with the average returns they had from 1945 to 1995. You can see that available returns to investors are way down. (Let me add here that many of these numbers are provisional. We will try to steadily improve them over the next several months. Any helpful inputs are welcome.) But I do believe that readers will agree with the general proposition that potential investment returns have been lowered on a wide investment front over the last 20 years and that stocks are generally in line with all other assets.
16. This downward shift in the discount rate was probably camouflaged for most of us by its intersection with two of the great genuine bubbles – 2000 tech and 2007 US housing and finance. Those were so much more powerful in the short term and so much faster moving that you could easily lose sight of the slower and very irregular downward drift in the broad discount rate. Some assets would boom for a few years and then bust spectacularly, but all slowly worked their way higher in price and lower in yields.
17. The key point here is that this downward shift in the discount rate has happened and can be measured. The possible reasons for it can, for convenience, be divided into two groups.
18. I believe that the major input has been a sustained policy of the US Fed since just before 1995 to push down short-term rates. Investors initially resisted this effect, assuming it to be very temporary. As the downward pressure on rates continued, though, the 2% drop in T-bill rates worked out along the yield curve until, over a number of years, the 10- to 30-year bonds, both nominal and TIPS, also reflected the 2% drop. This effect, rather like a virus, then moved into high-yield stocks, and eventually into all stocks, real estate, forests, farming, and all investable assets. The US Fed and its growing list of converts to this policy have successfully bullied the entire discount rate structure of assets. They did it to enjoy the economic benefit from the wealth effect, and Yellen, Bernanke, and Greenspan – the three members of this new Fed regime – all overtly bragged about their success in driving asset prices, particularly stocks, higher.
19. The general downward pressure on rates was not continuous. On a cyclical basis, when growth and employment were fine, rates could rise. Indeed, from late 2002 into 2006 the Fed raised rates as much as 430 basis points as the economy recovered, following the time-honored practice. Most asset prices, though, far from stumbling, continued to rise as the markets now believed that the Fed would always act as if it were looking for an excuse to lower rates. Vitally, moral hazard, “the Fed or Greenspan put,” was always in sight: If markets rise you are on your own – Whoopee! But if things go wrong – any bump in the economic or market road – you can count on us to lower rates. This made investment risks asymmetrical and was guaranteed to raise prices. The increasingly miserable rates available to more cautious investors merely rubbed this message in. Exhibit 6 makes it pretty clear to me that around the cyclical moves the Fed’s pressure ratcheted the rates down ? lower highs and lower lows. To roughly neutralize for economic cycles the exhibit looks at real interest rates each time unemployment hits 5.2% since 1982 and shows earlier cycles only when inflation was roughly comparable to today.
20. Exhibit 7 represents the Bank of England’s take on a completely different explanation for the lower discount rate: a series of fundamental factors that they argue have pushed down “normal” rates 4.5%, far more than the 2% or so reflected in broad asset pricing, without a single basis point being allocated to the new policies of central banks. Must be the work of, er… central bankers. Some of their points, though, seem reasonable enough and might have the right sign at least.
21. A third possible reason for low rates might have been added by Schumpeter, of “Creative Destruction” fame: that we are between waves of innovation and that this suppresses growth and the demand for capital. A related school of thought these days is that we have simply run out of low-hanging technological fruit.
22. And, finally, there are those who believe that the reason for lower rates is particularly simple and entirely different: The rapidly aging population of the developed world and China is, cohort by cohort, moving the mix toward middle-aged heavy savers and away from high-consuming young workers. This, they believe, has created excess savings that depress all returns on capital. Clearly, not a crazy argument.
23. I believe the dominant effect is Fed policy, but it seems nearly certain that one or more of these other factors are also contributing. Whatever fraction was caused by Fed actions, the important point here is that we can measure the lower rates and lower imputed returns. They are most definitely there.
24. At GMO our standard assumption for valuing assets has always been that both P/E multiples and profitability would move all the way back to “normal” – the long-term trend – in seven years. The seven-year period was selected because it was close to the historical average.
25. Now this is where it gets interesting and contentious, for I can find no reason why the discount rate this time should return all the way to the old average, nor that it should fully regress in only seven years.
26. The case for partial regression of rates hinges on the possibilities that the broad post war era of 1945 to 2005 will turn out to have been a golden era of global growth that will not be equaled again for at least a couple of decades. We now appear to be in a slower-growing world, the result of a slower-growing and aging population and lower productivity. Quite possibly, lower growth and other reasons mentioned above will lower the demand for capital and thus result in a permanently lower return on all capital, at least to a moderate degree.
27. The current fad for the Fed and central banks trying to influence the economy through lower rates is also unlikely to go away entirely, even over 20 years. Economic dogmas die hard despite evidence of failure. Think of all the unnecessary pain and misdirection from the idea of Rational Expectations and the Efficient Market Hypothesis.
28. History, though, has shown repeatedly that financial and economic ratios have a strong tendency to return to their old, normal levels. These ratios did not spend 100 years or so wandering around a central tendency for no reason. This leads me to believe that the odds still favor some mean reversion: the discount rate moving toward the old normal from today’s extreme. I propose that returning two-thirds of the way to the old normal has a higher probability than either returning all the way or staying indefinitely at current levels, although both are possible.
29. The case for a slower than usual regression rests on my belief that most of the reasons suggested for a lowering of rates are slow-moving: population profiles; Fed policy regimes; temporary (or permanent) lack of a growth push from innovation; income inequality. Much as I would like some of these factors, such as income inequality, to change rapidly, it seems wishful thinking. What do you think? It took 35 years to get from high to low in rates, for example. Bottom line, I suggest a 20-year flight path to get back two-thirds of the way to normal so that T-bills will yield around 1.2-1.5% real again, and developed country equities will return 5.0% instead of our current 5.7% “normal” assumption. Let me point out that all alternate assumptions – whether three years, seven years, or no return at all – are also arbitrary. Granted, each of these is merely a possibility. Our task is to find the least unlikely.
30. And now for another difficult and contentious point: Over roughly the last 20 years there has been a historically unprecedented rise in corporate profit margins. The reasons for this will be considered in part two next quarter. Importantly, many of the reasons are structural and very slow-moving. There is little in the real world data to suggest that there will be a rapid decline back to the old average.
31. In the model I am suggesting here, I assume that corporate returns will slowly, over 20 years, drop back two-thirds of the way to their old normal.
32. This is how my suggested 20-year “evolution” would play out compared to our traditional 7-year full mean reversion model (see Exhibit 8).
33. The 20-year approximate returns likely from “the whimper” flight path from EAFE and emerging equities are shown in Exhibits 9 and 10. In summary, we expect returns of 2.8% a year from US equities, 4% from EAFE, and 5.3% from emerging market equities. (This last number does suggest the only, kamikaze, way to achieve a 5% target return. Anyone for some career risk?)
34. It turns out that all three mean-reverting flight paths end up with remarkably similar and painful outcomes for pension funds and others at a 20-year horizon: 2.4% a year for the standard 7-year slump and 13 years of full returns; 2.6% for a 2-year crash and 18 years of full returns; and 2.8% a year for my 20-year whimper. This narrow range is a significant and perhaps unexpected outcome: Any set of assumptions that includes even a modest reduction in current abnormally high corporate profits and hence some reduction in growth – to be offset by a required increase in yield with a corresponding loss of capital – results in a very disappointing 20-year flight path.
35. Normal bear markets of 15 to 25% can, of course, always occur. They have nothing to do with this analysis though, which is trying to come to grips with a slow-burning shift in the long-term equilibrium. Old-fashioned bubbles breaking causes major declines back to the previously existing equilibriums. A normal bear market is, in comparison, short-term noise. (Although I grant you that at the lows everyone starts to look for arguments justifying even further and more permanent declines. Investing is tough!) Global and domestic political shocks often cause a modest decline. History shows that these are nearly always short-lived and the initial response is almost invariably exaggerated. Domestic economic shocks are the meat and potatoes of market noise as markets over-respond to provisional data that after three changes often ends up with a different sign.
36. The current outlook includes the possibility of a Trump election. This would be an outlier event of its kind because it comes with possibilities of tariff wars and general political unease globally. With the constraints of Congress, however, it would seem unlikely to cause long-lived effects, although this is not impossible. Brexit problems have already caused problems that could easily worsen, as indicated by the weak pound. I believe that the biggest risk is that an extended economic recovery, which I actually expect, lasting for two more years, might start to finally push up wages enough to count. Should oil prices simultaneously bounce back, which I also expect (up to $100 a barrel by 2020), we could have an uptick in inflation, which has a very reliable history of lowering P/Es. All things considered, though, I would say this is market business as usual and not enough to interfere with my suggested 20-year forecast.
37. Consequences for GMO. Like almost all investors, GMO has been effectively bullied by the 20-year Fed Regime. In our Benchmark-Free portfolio, we carry almost 50% of equity or equity-like exposure. If we had been looking at the current data 20 years ago, we would have carried less. Now, we recognize some possibility of the current low-return world continuing forever alongside a substantially larger possibility of our standard 7-year outcome. The possibility of short-term bubbles breaking, which I consider so unlikely, is expressly ignored by our 7-year forecast, so there is no disagreement there. If GMO accepted my current thought experiment of a 20-year whimper as a certainty, what would the consequences be for a benchmark-free portfolio? Perhaps 5-7% more equity exposure. A 60-65% equity position is considered to be what we would hold in a “normal” world with everything priced “fairly,” and, because my whimper flight path is not as attractive as that, we would be under 60% equity. But the gap between the risk-free rate and my whimper forecast for equities, when tilted to emerging and EAFE, is not that far away from normal.
(A shorthand way of viewing this 5-7% more equity exposure is that with a lower chance of a near-term crash or a 7-year slump, the option value of cash would decrease and hence risky assets, including stocks, would become more attractive.)
Over our 28 years of asset allocation, most of our risk reduction for clients was concentrated in the two classic US bubbles of 2000 and 2007, which we sidestepped to a considerable degree. Most of our extra performance, however, came from ranking the different asset classes more correctly than not. We still expect to be able to do this, even in this difficult and novel environment, and are facing an interesting, wide range of potential returns in different global equities and other assets as we write.
GMO does not expect its strategists or any of its analysts to toe a party line. James Montier, for example, four times the highest-rated strategist in Europe,2 is more bearish than our 7-year model. My current analysis is less bearish. Ben Inker, the portfolio manager, is informed by both of us and many others. He weights the alternatives as he sees fit and makes the final decision. It seems a good system, to me.
Brexit: wounds from playing cricket
Yes, London, your financial business will be hurt. Long and hard. And most corporations will lose easy access to the biggest market in the world. As for universities, which receive much EU money, for students and faculty it will be a bone cruncher. Someone, somewhere may benefit? But the people voted. In a sea of misrepresentations by a Brexit leadership who promptly resigned, 51.8% voted to leave. The people voted…in a way that has absolutely no constitutional standing. In contrast, Members of Parliament have an unstated constitutional and ethical duty to do what is in the best long-run interest of the UK. The majority of Members of Parliament know that Brexit is not that. But the people voted; so it would not be cricket (“not kosher” for Americans), to override it. But what a high price for playing cricket!
Happily, my one-third chance of no Brexit still looks good as the Prime Minister does her duty in frightening everyone with a cold and hard Brexit. Just possibly she is combining this hard-Brexit-up-your-nose strategy with a remarkably inclusive program (for a Conservative!) to win over enough previously Brexit voters to have a revote by referendum or Parliament by the spring without a revolution. You never know your luck.
See the full PDF below.