This is the final installment in a series exploring the relationship between stock prices and earnings, particularly with an eye to how much earnings growth will be needed over the long term in order for stocks to continue rising. Broadly speaking, the conclusions have been that there has been a long-term equilibrium relationship between stocks and earnings that is not reflected in historical averages, that the historical averages have departed from that equilibrium level to an unprecedented degree, and that the next 20 years are the likeliest period in which we will see a 'hard reversion' in markets resulting in very low returns despite relatively high earnings growth and inflation.
In this installment, I am going to look at what these findings might say about market behavior over four investment horizons, the Years 2070, 2040, 2030, and 2024. Moreover, I am going to present three different scenarios for each of them, what I will call "Strong", "Weak", and "Bizarro" scenarios. The Strong scenarios will present a world in which things revert to pre-1990s market patterns. Weak scenarios will present a world in which markets revert to historical averages. And, Bizarro scenarios will present a world in which markets break the historical patterns identified in this series. We will also consider investment implications of these scenarios.
First, why 2070? There is really no particular reason for choosing this particular year, except that it would be a roughly 50-year horizon for a young investor today. A few commenters on previous installments asked what the investment implications of my claims were, and as I explained in Part 5, it depends in no small degree on what your horizon is. The more distant the horizon, the less things like PEG and CAPE ratios matter. Assuming a rangebound PE ratio over the very long term, all that matters is earnings growth. Simply plug in whatever your estimated long-term expectation for earnings growth is and that will roughly approximate your price return.
Since the Great Depression, earnings have grown about 6.4% per year. Under the gold standard, it was 0.9% per year (dividend yields were much higher and inflation lower, however). So, assuming a continued 'pro-inflation' monetary system, earnings might be expected to rise 6% or so per year. Under a return to a pre-Fed system, earnings growth might be expected to be barely positive. How economies move from one monetary system to another is beyond the scope of this series.
I have argued before that pro-inflation monetary systems appear to have a tendency to induce "cost disease" which ultimately lowers long-term economic growth (e.g., GDP) in core economies at local, regional, and international scales. But, this does not yet appear to be an idea that is widely accepted, nor has it materially slowed earnings growth yet.
Therefore, in order to raise long-term returns in the face of long-term uncertainty, young investors probably ought to focus on three themes:
1. Avoid heart disease.
2. Build a strong and supportive family.
3. Acquire skills.
Under a Strong scenario, wherein we would see a hard reversion to historical patterns, we might expect the earnings yield to be about 17% (the median level of the three peaks from 1900-1980) in 2040, which is a PE of about 6. Since 1920, the earnings yield has peaked every 30 years, so 2040 would be the next one.
We also saw that very high PEs have been very good at anticipating high subsequent earnings growth (G) over 20-year intervals. 2020 saw one of the highest PEs on record, as can be seen in Chart A below. Let's assume that earnings grow at a near-record pace, 7x, or just over 10% a year (near 2.0 on the righthand side of the logarithmic scale in the chart).
Chart A. (Source: Robert Shiller data)
PE was an arithmetic value of 39 (or an earnings yield of about 2.5%) at the end of 2020.
Using the equation,
the S&P 500's nominal price return would be about 5% over the December 2020 level of 3695. The S&P 500 would be about 3880, roughly -15% relative to today's 4700 level. In other words, if you assume a hard reversion to peak earnings yield levels, it would wipe out any gains produced by even a record level of earnings growth.
Now, if you add 5% per year to the 10% annualized earnings growth we assumed above, the returns will look much better indeed. That would bring the S&P 500 to about 9420, double current levels, or 3.5% per year compound growth.
Under a Weak scenario, let's assume that the earnings yield simply returns to its historical geometric average of 6.8%, or a PE of 14.7. Let's assume average post-Depression earnings growth of 6.4%, which over 20 years amounts to just under 3.5x.
Plugging these values into the equation above produces a value near 1.3x, 30% above the December 2020 level, which would be about 4800 for the S&P 500. That is 0% returns between now and 2040.
Chart B. (Source: Shiller data)
Under the Bizarro scenario, we assume that--contrary to what we have found with markets operating under 'true' equilibrium conditions--yields fall going into 2040, earnings grow at relatively low rates, and earnings growth exceeds the level of the earnings yield. Therefore, let's assume an earnings yield of 2% (a PE of 50) in 2040 and earnings growth at 2.5% per year. That rate of earnings growth amounts to less than 1.7x growth in earnings.
If we do the math, we get a terminal S&P 500 value of 7850, or 3.8% per year. The Bizarro scenario is essentially the one we have been living in since the 1990s but with somewhat lower returns.
Now, which of the scenarios are likeliest? As I have held throughout this series, I know of no theoretical approach to anticipating future changes in growth and yields. I am instead trying to extrapolate from historical behavior. History seems to suggest that high PEs anticipate high growth but that the earnings yield sets limits on ex-post growth. So, if the former (that is, high PEs) are right about growth, that's great. But, high growth is likely to occur under a regime of rising yields (that is, falling PE). History suggests too that since the Fed was established, the Schumpeterian technological cycles that seem to dictate the long-term movement in yields and inflation have increased in frequency, peaking every 30 years rather than every 50-60 years. This puts a peak in inflation and yields in 2040. Thus, in a battle between rising earnings and falling yields, yields will likely capitulate in the end.
Although I think this the likeliest of scenarios, it is only because the infinite number of alternative scenarios seems less likely. Can earnings grow 12% per year for the next 19 years (12% because earnings roughly doubled in the last year, thus reducing the amount of growth necessary to get to a 15% CAGR over the 2020-2040 period)? Perhaps, but I have to point out that this kind of unprecedented earnings growth would almost certainly coincide with white-hot inflation, perhaps pushing the earnings yield beyond its historical highs, thereby lowering returns.
Assuming a 20-year period of low returns, rising yields, and rising inflation, what can investors do? Let's consider commodities. Chart C is from an old article in which I wanted to demonstrate the relationship between real commodity prices and equity yields. The light blue line is the earnings yield, the yellow line the dividend yield. The dark blue line is a historical commodity index called the Grilli-Yang Commodity Price Index. Unfortunately, it only goes back as far as 1900, so I also spliced old wholesale price indexes together with more contemporary producer price indexes to make the pink line.
Chart C. (Please see "Inflation And Yields: The Evolution Of Gibson's Paradox And The Revolution In Prices, Part I" for details on sources)
This chart and charts that I presented in Part 5 suggest that there is a strong long-term correlation between real commodity prices and the earnings yield. If one believed that the earnings yield was likely to be materially higher in the future, it would not be unreasonable to expect real commodity prices to be materially higher. Moreover, because commodity prices, over the long term, have a high "beta" relative to consumer prices, it is likely that nominal commodity prices would rise quite significantly.
But, there are some complicating factors. A glance at the chart above suggests that there are very few 20-year periods, particularly since the establishment of the Fed, where real commodity prices did not experience heavy losses along the way. Take 1929, the peak of a bull market in stocks that drove the earnings yield down to 6.5% on this annual chart. Twenty years later, in 1949, the earnings yield would be over 15%. But real commodity prices rose in that period a little over 0.5% per year; nominal performance was a bit over 2% per year. On the other hand, that beat the negative price return of the S&P 500 over that period while being roughly equivalent to total returns, which we will explore in greater detail in a moment.
First, however, let's briefly talk about the changing role of dividends in total returns, illustrated in Chart D.
Chart D. (Source: Shiller data)
The blue line is 20-year changes in a nominal total return index, the red line the contribution of reinvested dividends to that growth in total return. Where the red line is higher than the blue line, dividends prevented total returns from being negative. Interestingly, as total returns have structurally risen since the Depression of the 1930s (1950 - 20 = 1930), the importance of dividends has fallen. After having seen Chart C, that should not be too surprising. The earnings yield has been falling relative to its historic average, but the dividend yield has been falling relative to the earnings yield on top of that. Stocks have not turned into NFTs quite yet, but they have been gradually moving in that direction. In any case, it is hard to say that dividends will provide the boost in the 2020-2040 period that they did in the 1929-1949 period, but they certainly won't hurt.
Again, we are speculating here on what an investor possessed of a single piece of information about the future-i.e. that there will be a meaningful market reversion between now and 2040-ought to do if they could make a single set of trades today. And, we have found that for most 20-year spans, commodities are not a ready-made solution. (Investing directly in commodities other than precious metals is a difficult proposition on top of that).
So, let's step back and look at the history of 20-year total stock returns in the following two charts.
Chart E. (Sources: Shiller data, World Bank Pink Data, St Louis Fed)
Chart F. (Sources: Shiller data)
Charts E and F tell us that it has been virtually impossible to beat the market with commodities or bonds over 20-year spans. Chart D tells us, however, that dividends might not spare the stock market in the future from negative returns the way it has in the past. Look at Chart G, for example.
Chart G. (Sources: Shiller data, World Bank, St Louis Fed)
Here we have the stock market's performance relative to commodities with and without the impact of dividends. If we supposed the kinds of earnings yield reversions that occurred in the decades of the 1910s, 1940s, 1970s, and (to a lesser degree) in the 2000s, were to occur again in the 2030s, but this time in the absence of dividend relief, it is possible that even total returns could underperform commodities for a very long time.
Therefore, I want to lean towards commodities, yet even here I am hesitant. Chart H is one I presented in Part 5.
Chart H. (Sources: Shiller data, World Bank)
It appears to suggest that commodity prices are already quite above the level that the earnings yield suggests it should be, which could mean that commodity prices are already ahead of the game.
Let me complicate things further. I have argued in previous installments that many of these patterns and relationships are rooted in older economic phenomena that may go back centuries (eg., Schumpeterian "Kondratiev Waves" and Keynes's "Gibson's Paradox"). There has also been an apparent correlation between high inflation and yields and periods of widespread political upheaval, particularly in the shape of armed conflict. From the Napoleonic Wars in the early 19th Century to the localized but bloody civil wars in America and China and the Prussian wars in the 1860s to World Wars I and II in the 1910s and 1940s, respectively, not to mention the Vietnam and Iraq quagmires that the US experienced in the 1970s and 2000s, there appears to be a tendency for breakdowns in the international system that often have an ideological quality to them. If there were to be a significant rise in yields and inflation going into 2040, one might expect a greater likelihood of both serious conflict at both domestic and international levels. This can have implications for certain types of returns, specifically precious metals. First of all, these metals can be confiscated, and second of all, they can lose their value in extreme circumstances (e.g., how much is an ounce of gold worth if you're starving or if it has to be concealed?), as outlined in the late Barton Biggs's Wealth, War, and Wisdom.
In other words, investing in commodities long term is difficult (I would certainly defer to experts in how to structure such an investment). Precious metals may be the simplest way to do so, and my sense is that platinum and silver would see the highest returns, but if you are pouring all of your money into that kind of trade, you are probably not only betting on higher inflation or falling stock prices but a much larger political, social, and economic breakdown that might prevent you from reaping your just reward in case your bet proves right.
So, to wrap up this treatment of 2040, assuming a significant reversion in yields and growth, and assuming that an investor had to decide today how to bet it, a more defensive position than that suggested by 'stocks for the long run' seems reasonable to me. Something like what some call a "Talmud Portfolio" or a "Permanent Portfolio", portfolios which are divided equally between major asset classes (typically equities, bonds, property, and gold).
Of course, there is no reason to have to call 2040 today. So, there are alternatives to equally-weighted portfolios. One, become Elon Musk. Two, become more Buffett than Buffett. Three, time the market.
I have no useful advice on how to achieve the first objective beyond opening a Twitter account. By 'becoming-more-Buffett-than-Buffett', I mean becoming something like a ruthless, fundamentalist, value-investing fanatic precisely at the historical moment that value investing appears to be killed off once and for all. Of course, timing the market is the most foolish and disreputable technique.
First, the Strong Scenario.
Since the establishment of the Fed in 1914, the earnings yield, real commodity prices, and consumer inflation have peaked every 30 years (circa 1920, 1950, 1980, 2010), as we mentioned above. These cycles have typically involved a decade of rising yields and prices (1910-1920, 1940-1950, 1970-1980, 2000-2010) followed by two decades of falling yields and prices (1920-1940, 1950-1970, 1980-2000, 2010-????). Prior to the establishment of the Fed, these cycles were more elongated (50-60 years in length and looking more like a 'V' than a Nike Swoosh (a 20-year gradual drop followed by a 10-year upward surge)). That is, prior to the Fed, yields and prices fell for two or three decades and then rose for two or three decades.
Assuming that this post-gold-standard cyclicality continues into the future, we might expect, therefore, yields and prices to fall or simply remain relatively low going into 2030, before turning higher into 2040. During these double-decade downswings, stocks tend to do very well in the first decade (1920s, 1950s, 1980s, 2010s) and then perform less uniformly in the second decade (1930s, 1960s, 1990s, 2020s?). But, no matter how stocks perform in that second decade, the tendency is for the earnings yield to remain relatively low. In other words, if rising stock prices don't push the yield down (as in the 1990s), then an earnings collapse pulls it down (as in the 1930s).
Thus, the implications of a low/falling yield are not necessarily clear for stocks, but they are certainly not very encouraging for commodity bulls.
What about equities, though? As I have argued in the previous installments, the history of the relationship between earnings growth and the earnings yield suggests that the stock market in 2030 will be roughly equivalent to its 2016 price levels.
I have tried to illustrate this in Charts I and J below.
Chart I. (Sources: Shiller data)
Chart I suggests that the CAPE ratio (which can also be thought of as the ratio of earnings growth to the earnings yield, as demonstrated in previous installments) tends to peak around the time that the correlation between earnings growth and the earnings yield goes negative.
Chart J. (Sources: Shiller data)
Chart J shows that "secular" bear markets tend to last about 14 years. Thus, price returns tend to be very low in the 14 years that follow extremes in the relationship between earnings growth and the earnings yield. With the correlation having first gone negative in 2016 (when the S&P 500 was between 2000 and 2500), that suggests that stock prices in 2030 (that is, 14 years after 2016) could be at similar levels.
Chart K, below, shows that breakdown in the correlation between earnings growth and the earnings yield is typically followed by a collapse in the logarithmic PEG ratio, which has been fairly good at modeling price returns historically.
Chart K. (Sources: Shiller data)
In other words, when we look a couple decades out to 2040, the thing we need to mind is the risk of a reversion of PEs, but when we are looking out 7-14 years, what we need to worry about is the risk of a reversion in growth. Typically, these reversions, as discussed in previous installments, bring the long-term rate of growth below the level of commodity inflation-which, as we explained above, is itself expected to be on the low side this decade.
A Weak Scenario would see, perhaps, a dip in the pace of post-Depression returns rather than the nightmarish negative returns of the Depressionary Strong scenario. Instead of 14 years of negative returns, perhaps 5% annualized returns over the 2016 levels for stocks? That, however, places the S&P 500 at about 4400 in 2030, flat relative to today's level.
For reasons explained in previous installments, I have used the PEG7.5 ratio (the ratio of current trailing PE to subsequent 7.5-year growth rates in earnings (G)) as the primary lens for analyzing the relationship between PE and earnings growth. Seven-and-a-half years does not get quite get us to 2030. I have to use 2021 data and 2021 + 7.5 = 2028.5. Using data from late 2021 gets us to early 2029 at best. So, I am going to use the Year 2029 as a proxy for the Year 2030.
I spent a lot of time in previous installments arguing that the historical equilibrium of the PEG7.5 ratio is higher than the 'true' equilibrium, and that a reversion to the true equilibrium would imply sharply lower returns in the future. But, either way, we do not know what future earnings will be. We have some clues (for example, the correlation in Chart K seems to be a hint; PE itself under circumstances described in earlier installments seems to anticipate earnings growth), but none of it is very direct.
So, we have a cluster of problems in anticipating stock prices at the end of the decade. First is, how much will earnings grow? And, will either the historical or 'true' equilibria apply, and if so, which one? As explained earlier, the chance of a hard reversion is strongest in the 2030-2040 period. Although PE could be meaningfully lower at the end of the decade than it is today, it seems more likely, from the analysis thus far, that it will still be relatively high (that is, the earnings yield will be relatively low). But, let's set that aside for now.
In the table below, I am going to plug in earnings growth scenarios--no-growth (0% CAGR), 150-year historical average growth (3.79% CAGR), average growth since the Depression (6.43% CAGR), and persistent growth (8.14% CAGR) into the average PE ratios of the 1900s, 1930s, 1960s, 1990s, the Year 2000 (that is, a single extreme instance of a very high PE), and of the third quarter of 2021 (for comparison) to produce a sense of the range of possible end-of-decade prices for the S&P 500. Those four decades were each the decades that preceded decade-long upswings in the earnings yield.
I have also included scenarios in which the rate of earnings growth equals the earnings yield ("G = EY") of that particular decade (e.g., the average earnings yield in the 1930s was 6.14%, so we assume a scenario where G = 6.14% to calculate a Year 2030 S&P value of 4900). As I have tried to argue in this series, there appears to be a tendency for the earnings yield to act as a cap for long-term rates of earnings growth (especially in the 7- to 8-year range) in a manner reminiscent of what we have found with "Gibson's Paradox".
Table 1. (Sources: Shiller data, own calculations)
The red numbers are scenarios in which the rate of growth approaches but does not exceed the earnings yield. These, especially those in the "Average" column, would be plausible Weak scenarios.
As you can see, a rising PE can make up for a multitude of sins. Earnings growth above the level of a rising yield is perhaps better but had been relatively rare until the last three decades.
Chart L. (Sources: Shiller data)
We can use these Weak scenarios to imagine what Bizarro scenarios might look like, and they break into essentially two elements: growth in excess of the earnings yield and/or an even higher PE.
The question we have been asking is, how long can either--and especially both--continue? Without a theoretical explanation of the relationship between the two, there is no way to know, but as I have tried to explain in this series, the brief history that we have of global equity markets (only 150 years) suggests that it cannot continue for much longer. I tend to lean, therefore, towards the Strong scenario: low growth with little to no help from PE resulting in negative price returns.
Chart M. (Sources: Shiller data)
For those who have had the patience to read this series, this may not seem to square with what we found out about supercycles in earnings, yields, and employment. Recall that we found that about every 12.5 years since the establishment of the Fed and beginning in 1921 (although there is a hint of such a pattern pre-Fed, as well), there has been a crisis that would then kick off a new earnings boom that typically lasted 7.5 years. This appears to have been especially true when the PE ratio was also very high. Like clockwork, the last such crisis appears to have been the 2020 pandemic shock which turned growth negative, raised the PE ratio, and decimated employment. Thus, we might expect the long-term earnings growth rate to remain strong well into 2028 and perhaps beyond.
The trick is that markets have packed in a supercycle's worth of growth in a half dozen quarters already. In other words, if earnings remained flat until the end of the decade, the 7.5-year rate of growth would peak right on time in 2028 regardless at about 10% per year. Thus, there is nothing in the historical patterns we have discussed, so far as I can tell, that says earnings must continue to rise from their 2021 levels.
But, this raises the problem-as I see it-of violating yet again the notion that the earnings yield acts as a cap on earnings growth. There is no historical scenario (as illustrated in Table 1) that suggests a major upward shift in the earnings yield to match the earnings growth rate at this stage of (what I have deemed as) Schumpeterian Kondratiev cycles.
The history of the relationship between PE and subsequent earnings growth (G) strongly suggests that the rate of earnings growth will be high in 2028. The history of the earnings yield and inflation since 1914 suggests that the yield will be relatively low throughout the 2020s. The history of the relationship between earnings growth and the yield--whether we combine them to form Shiller's CAPE or, in the idiosyncratic fashion employed here (looking at the correlation of the two)-suggest that earnings will be severely hit in the immediate future (2023 and 2024) with a direct impact on stock prices.
I have already showed how nearly every one of the laws I have outlined has been twisted and even broken at some point in the history of the market, so we cannot discount that happening again. Indeed, it appears that that will happen again in the coming decade when earnings growth again exceeds the yield. But, I think we can also conceive a scenario that squares these circles, a scenario in which we can have both a high rate of earnings growth (as anticipated by PE), a low earnings yield, and a rate of earnings growth that 'treats' the earnings yield as an upside cap. Let me explain how.
A sharp cyclical decline in earnings that ripples through stock prices which was then followed by an equally strong cyclical rebound in earnings (but somewhat less by stock prices) could check these boxes. This was more or less the scenario that unfolded in the 1930s. In Table 1, we used the average earnings yield for the 1930s, but in actuality, the yield was quite volatile, oscillating between 4% and 10% (that is, a PE ratio that oscillated between 10 and 25). Although the earnings yield did not cease to observe the Kondratiev waves, it also became highly cyclical within that constraint.
Chart N shows how volatile the earnings yield became in the 1930s.
Chart N. (Sources: Shiller data)
That volatility-in yields, earnings, and prices-makes it possible for history to thread its needle.
Chart O illustrates a scenario that essentially fits all of the constraints that we have described thus far:
1. Flat returns from 2016-2024 and 2016-2030.
2. A low earnings yield.
3. An earnings shock circa 2024.
4. A high 7.5-year earnings growth rate in 2028.
5. A long-term earnings growth rate that is reluctant to breach the level of the earnings yield.
6. Stock price returns that revert to the 'true' PEG equilibrium.
Chart O. (Sources: extrapolations)
This would constitute an immediate reconvergence with the 'true' equilibrium of the logarithmic PEG7.5 ratio.
Chart P. (Sources: Shiller data & own extrapolations)
The red line here shows 7.5-year forward price returns for stocks using the extrapolated data from Chart O.
The average earnings yield for the 2020s would be about 5%, roughly the level it was in the 1990s.
There is a temporary breach of the earnings yield by the long-term rate of earnings growth in 2028, partly accommodated by a 1930s' style surge in the earnings yield, as illustrated in the following chart.
Chart Q. (Sources: Shiller data & own extrapolations)
By adhering to the six conditions mentioned above, the resultant scenario largely resembles, it appears to me, the behavior of price and earnings in the 1930s: a uniform descent followed by an uneven surge followed by an uneven decline.
Chart R. (Source: own extrapolations)
Chart S. (Sources: Shiller data)
In a Depressionary context, there is little reason to expect a positive performance in commodities, especially since commodities already appear to be overpriced relative to the earnings yield, as mentioned in the discussion of 2040 and illustrated in Chart H.
But, commodities held up pretty well against stocks during the 1930s, especially if we exclude dividends.
Chart T. (Sources: Shiller data, World Bank, St Louis Fed)
Curiously, since the establishment of the Fed, stocks have become riskier relative to rocks and crops, even if the long-term trajectory has remained the same.
Chart U. (Sources: Shiller data, World Bank, St Louis Fed)
And, what of Treasuries?
Chart V. (Sources: Shiller data)
In the 1930s, bonds outperformed stock price returns by far, with Treasury yields falling even as the earnings yield remained rangebound. We find much the same even when we include dividends. But, there are two complicating factors. First, as we have noted, dividends play less of a role than they used to, and Treasuries, like commodities, are near long-term historic highs.
Bonds outperformed commodities during the Depression, but after the initial collapse in commodities, the ratio between the two was essentially flat.
Chart W. (Sources: Shiller data, World Bank, St Louis Fed)
So, if we had to make a bet today on where markets are going to be in 2030, how ought we position ourselves?
There is some overlap between 2030 and 2040 strategies. Setting aside the Elon Musk strategy, there is a strong case to be made for turning to a strict, fundamentalist value-investing strategy. And, although the argument here has been that the 2020s will tend towards deflation while the 2030s tend towards very high inflation, both scenarios appear to be relatively negative for stock returns. This suggests, again, something like a portfolio evenly split between asset classes.
But, the emphasis within respective asset classes in a 2022-2030 portfolio might look different from that in a 2022-2040 portfolio. For example, where silver and platinum might look better in a more inflation-oriented 2022-2040 portfolio, gold might look better in a deflation-oriented 2022-2030 one, keeping in mind that gold could be confiscated in case of a repeat of the 1930s. In equities, a stronger case might be made for some sectors over others in the case of the 2022-2030 portfolio. As I argued in an essay on long-term sector rotations last year, the energy sector-and, to a lesser degree, industrials, materials, and mining-is likely to outperform tech- and service-oriented sectors such as the FANGs, semiconductors, the health sector, retailers, consumer discretionary, and car manufacturers (particularly those rhyming with "Tesla") over the long term. But, I also pointed out that although energy (the key sector) outperformed the broader market during the 1930s and '40s, it fell just as dramatically as others in the initial 1929-1932 crash.
If 2030-2040 is to be, as anticipated, an inflationary decade, the energy sector and its cousins could be expected to fare quite well then, as well, relative to other stock market sectors. And, in the 2030s, energy et al would almost certainly crush bonds, if the decade is, in fact, to see rising yields and inflation.
So, for this decade: in commodities, gold; in equities, energy (among some others); and Treasuries. I have done very little analysis of the history of property markets, so real estate (farms?) might be another way to diversify, particularly for a 2022-2040 portfolio.
Finally, there is timing the markets. If this 'hard reversion' scenario, the Strong scenario, pans out, getting out and in at key moments over the coming decade could potentially provide exceptional returns in a deflationary environment. I have shown in previous articles that in periods of persistently low equity returns ('secular' bear markets), equities are quite cyclical. That is, they rise and fall with the cyclical elements of markets-earnings, industrial metals, bond yields. The moves can be quite dramatic. In the 1930s, stocks dropped something like 80%, then quadrupled over the next few years, then dropped 50% again. And, as mentioned above, cyclical sectors tend to outperform less cyclical sectors on a long-term basis during long-term bear markets.
But needless to say, picking market tops and bottoms is not easy. A book like Anatomy of the Bear is a great place to bone up on how to pick major bottoms in the stock market, one of which was the Depression low. One such low could appear in 2024.
We have already seen how the long-term behavior of earnings growth compared to the earnings yield points to a dramatic decline in stock prices over the next two years, to as low as 2500 points for the S&P 500 by 2024, and perhaps lower.
Chart X. (Sources: Shiller data & own extrapolations)
Having covered that ad nauseum in previous installments and sections, I am not going to continue belaboring the point here. Rather, I want to focus on how and when investors might position themselves in anticipation of a significant, cyclical bear market in equities.
The most reliable place to hide during cyclical downturns is long-term Treasury bonds, gold, and defensive sectors like staples, utilities, and healthcare. The worst places have typically been cyclical sectors, although it depends to a certain degree where you are in the cycle, with energy and financials typically stronger late in the cycle.
There have been two fairly reliable indicators of cyclical peaks: energy prices, especially crude oil "shocks", and the yield curve spread. Stephen Leeb argued in his book The Oil Factor that a year-over-year near-doubling in crude oil prices is typically followed by periods in which bonds, especially zero-coupon Treasuries, outperform major stock indices. With the sudden drop in oil prices in Spring 2020, the surge that followed triggered Leeb's signal in Spring 2021. Since then, zero-coupon Treasuries have been more or less neck-and-neck with equities. I found, as well, in my article on long-term sector rotations, mentioned above, that both short-term and long-term changes in energy prices play decisive roles in signaling, if not provoking, a) short-term market crashes as well as b) long-term bear and bull markets in equities, and c) in long-term sector rotations.
I will not attempt to recapitulate that argument here, but the combination of sectoral divergences (especially tech vs energy), a transition in the long-term rate of change of energy prices, and the 'Leeb oil shock', which has yet to subside, point to a likely inversion of the 'sectoral hierarchy' over the course of the 2020s, low returns for the market as a whole throughout the decade, and a looming market crash. In other words, the history of sectors and energy prices appear to point to a scenario much as we described with respect to the 2030 horizon.
Beyond that, I have written in previous articles that the market cycle is based fundamentally on the earnings cycle and that commodity prices and Treasury yields oscillate around that cycle, giving us, in effect, live updates on the state of the market cycle generally and the earnings cycle in particular. Precious metals, industrial metals, and energy prices each have their own relationships with the earnings cycle, with precious metals leading, industrial metals shadowing, and energy prices lagging the earnings cycle.
Typically, I measure the market cycle as changes relative to a 36-month moving average. If energy prices are spiking, as indicated by something like Leeb's indicator, that usually suggests that the cycle has effectively run its course. By that point, precious metals should be in a clear state of deceleration, if not decline, and industrial metals should be decelerating, as well. You can also observe decelerations in key cyclical and/or high-beta stock sectors (e.g., industrials and materials) and industries (e.g., transportation, infrastructure, mining). What holds for industrial metals holds for long-term Treasury yields, as well. Short-term yields typically rise late in the cycle, much as energy prices do, thus producing the late-cycle compression of the yield curve spread.
Most of these conditions are in place. Precious metals have been soft since mid-2020, industrial metals are still high but their momentum has decelerated sharply, and the same can be said for cyclical stocks and long-term bond yields. This does not mean they cannot go higher on an absolute basis, but their deceleration points to a likely absolute decline in the coming months and years.
Chart Y. (Sources: Stockcharts.com)
An inverted curve has not occurred yet, however. There are still 120 basis points separating 20-year yields and 2-year yields, a gap that has closed only 100 basis points since last spring.
Not every stock market crash has been preceded by an inverted yield curve. The 1987 crash saw a 30% drop in the S&P 500 that took nearly two years to recover but also marked the beginning of a period that saw lower returns for the next seven years. That crash was preceded by a 'Leeb shock'. The 2020 crash, on the other hand, was preceded by an inverted yield curve but no 'Leeb shock'. In my opinion, history suggests that energy is the more consistent 'predictor' of stock market crashes going back a century or more while the yield curve is the more consistent 'predictor' of post-1960 recessions in GDP.
The type of scenario I outlined above in Chart R showing a collapse in earnings forcing the S&P 500 down to the 2500 level in 2024 suggests a crisis of a magnitude sufficient enough to be reflected in GDP data. In other words, if that scenario proves correct, it is likely to coincide with a recession. That being the case, we ought to see an inversion of the yield curve as that moment approaches.
But, that does not necessarily mean it is too early to start shifting to more defensive positions like Treasuries, gold (perhaps), and defensive sectors. Sometimes the yield curve inverts early in the cycle, sometimes late. With 10-year yields at cyclical highs and inflation at 7%, it is difficult to call Treasuries a screaming buy, but history suggests that there is more downside potential than upside potential in bond yields at this point in the cycle.
Over the three closest horizons discussed above (2024, 2030, 2040), the broad stock market indices do not look promising. Tech looks least promising of all. Energy, the yang to tech's yin, looks much more promising, although it could suffer significantly in any short-term or cyclical sell-off. Commodities, although largely unimpressive and likely ripe for a fall in the coming two years, have a fair chance of outperforming equities over the longer term, as do Treasuries.
Based on the historical patterns examined in this series, I would imagine an all-weather portfolio designed for long-term performance would look something like the Talmud or Permanent Portfolios described above. The equity portion would be energy-heavy and tech-light. Commodity allocations would probably be some mix of gold, silver, and platinum. Bonds would be nearly all long-term US Treasuries. Some portion might be set aside for real estate.
If one wanted to maintain a more equity-oriented portfolio, it might be best to emphasize quality above all. I am personally more interested in the macro level and historical processes, but my impression is that if traditional value metrics were ruthlessly applied to current markets, very little would be worth buying at the moment (I am happy to be corrected on this), but there are certainly some better than others.
What about markets outside of the US? Based on my previous research, emerging markets tend to move in seven-year cycles (largely unrelated to those described above) that follow the strength of the US dollar. Thus, the back and forth of developed markets and emerging markets is likely to be something of a wash. Moreover, with these major upswings in yields often coinciding with major military conflicts and increased political violence more generally, plus the growing number of major potential conflicts (between Western-style democracies and China and/or Russia, most obviously, not to mention Iran and North Korea), there could be considerable political risk in employing capital abroad, even to 'friendly' countries in the Asia-Pacific region or Europe.
Of course, the risk is not only from 'abroad'. Americans have to consider the possibility of domestic political risk, as well. On that basis, there may be some justification for a modicum of diversification into stable developed markets.
How much growth will be necessary to maintain current prices? How likely are we to get that growth? History seems to suggest that the answers depend on when you ask and how far out into the future you are looking. We are about 30 years into a marked departure from previous equilibrium levels of the PEG ratio. This departure has been driven by persistently "high" PE ratios. But who is to say what constitutes a "high" PE ratio or a "low" earnings yield? It appears to me that the earnings yield for the S&P 500, which is probably a proxy for the global earnings yield, is both correlated with commodity prices and consumer inflation and acts as a limit to the various long-term rates of inflation, including the inflation rate for corporate earnings. When this limit has been persistently violated in the past, there have been long-term negative consequences for stock prices. When the earnings frenzy ends, stocks collapse and then stay relatively subdued even as earnings growth resuscitates.
Stocks in the 1990s rose to unprecedented heights. So, even with the spectacular dot.com burst at the end of the 1990s and the this-might-be-the-end-of-capitalism crisis of 2008, the earnings yield remains at historically low levels. This acts as a kind of check on the level of growth that is sustainable. Growth can rise, but not without resulting in crisis.
Since the establishment of the Federal Reserve in 1914, there has been a persistent convergence of yields and inflation. Yields, albeit not in a straight line, have fallen to lower and lower levels while inflation-consumer, commodity, and earnings-has risen. That has increased the likelihood of inflation and growth exceeding the level of the yield.
Chart Z1. (Sources: Shiller data)
Chart Z2. (Sources: Shiller data, World Bank, St Louis Fed)
Perhaps, as time progresses, we will find that inflation and growth will continue to rise while yields continue to fall quarter-century after quarter-century. Perhaps the average earnings yield level in the 2050s will be 2.5% (an average PE of 40) while growth is 5%, just above the historical average. Five percent growth from now until 2060 paired with a 1.6x increase in the PE multiple puts the S&P 500 at about 50000 in 2060. This is the path markets have been on for the last century despite all of the horrors that we read about in our history books and in our newspapers (or their modern equivalents). You just need to tap into that river of earnings that churns beneath the pages of the history books and then minimize exposure to the risks that periodically crop up in those books. Be Switzerland.
My sense is that the bulk of this process has been driven by Fed-financed, US-centered technological development. Whether that river continues to gush depends largely on whether or not the technological development (which has both created and thrived upon the creation of mass consumerism) ultimately washes away its sociopolitical foundations in the process. This is an argument for another time, but what I want to say is that however bleak a view one may have of the process, you would have to be rather daring to simply pick a point along the time continuum, declare it over, and bet against it.
This series argues rather that it is likely that this process, whatever its underlying mechanism, has run ahead of itself and is "due" for a severe, multi-decade correction-a "hard reversion"-and that the next 20 years are the likeliest time for that to occur. We have been living in Bizarro world for the last three decades, a period of "irrational exuberance". History suggests that this period is drawing to a close.
This article was written by
Disclosure: I/we have no stock, option or similar derivative position in any of the companies mentioned, and no plans to initiate any such positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article.
Additional disclosure: I am short tech stocks, long Treasuries, short commodities.