The Great Earnings Yield Spread Divergence: The Bullish Case

by: James A. Kostohryz

In the first article in my series addressing the "great earnings yield spread divergence," as Felix Salmon put it, I argued that this was potentially a very significant financial event of profound significance.

There are essentially two main ways to interpret this divergence:

  • On the one hand, the "great divergence" in the earnings yield spread could simply reflect a mispricing of stocks and/or bonds in the context of a financial panic, thereby providing a major opportunity for investors to exploit anomalies.
  • On the other hand, the very large divergence in the earnings yield spread could be presaging a "great divergence," in the financial and economic future of the United States.

In the next two articles in this series, I will summarize various scenarios and possibilities associated with these points of view. My goal is not to say anything definitive, but to simply to provide a framework for analysis, debate and discussion.

This article will summarize the bull case.

The bull case essentially rests on the hypothesis that due to excessive risk aversion and panic in financial markets, stocks and/or bonds are being mispriced. There are three main perspectives from which to advance this argument.

  1. Treasury Bonds are extremely overvalued.
  2. Stocks are extremely undervalued.
  3. All of the above.

Historical Context

Prior to analyzing the various arguments in favor of the bullish view, it is important to place the discussion in a proper historical context. The Great Earnings Yield Spread Divergence is only "great" when the historical framework is limited to data subsequent to 1985.

Earnings Yield Spread 1985-2011

However, if we look at data going back to 1871, the current earnings yield spread is not nearly so "great." As can be appreciated below, the current earnings yield spread of roughly 6.25%, or 625 basis points, is high by historical standards going back to 1871, but not extremely so.

Earnings Yield Spread 1871-2011

In the 140 years since 1871, the earnings yield spread has had an average value of 2.8%. This is substantially above the zero% level hypothesized by the so-called Fed Model. And it is substantially above the level (around that same zero%) around which the earnings yield spread seemed to mean revert around fairly consistently between 1980 and 2000.

The upshot is that the current earnings yield spread does indeed appear high. Perhaps not as high as one would be led to believe looking at data after 1980, but still high compared to the historical average of 2.8% since 1871.

However, there is an important caveat. If we only focus on data from 1871 through 1960, the 6.00% level was roughly the average for the earnings yield spread during this period. Is it possible that the period after 1960 has been anomalous and that the earnings yield spread will revert to its pre-1960 mean?

Any bullish case based premised on the hypothesis that stocks are mispriced relative to bonds, needs to take the longer historical view into account. Specifically, the bullish case must address why the earnings yield spread will revert to the 1980-2000 mean, rather than the longer 140-year historical mean. Furthermore, the bullish case should address the concern about whether the post-1960 period represents an economic and/or financial anomaly and whether a reversion to the mean prevalent during the 1871-1960 period might be in the cards.

Hypothesis: Treasury Bonds Are Extremely Overvalued

It can plausibly be argued that recent financial panics and the ensuing "flight to quality" have driven yields on the 10Y Treasury Bond to unsustainably low levels.

This hypothesis has much to recommend it. Since 1871, the real (inflation-adjusted) rate of return on 10Y Treasury bonds has averaged around 2.6%. This rate of return demanded by holders of U.S. Treasury Bonds has remained remarkably constant over reasonably long stretches of time.

If we assume that in the long run investors will continue to demand this real rate of return for holding long-dated Treasuries, then with 10Y yields at 2.2%, it would appear that bond yields are discounting long-term deflation, of about 0.4% per annum during the next 10 years.

Is such long-term deflation a real possibility? Given the Fed's legal mandate, current academic and policy consensus and the current Chairman's publicly expressed views, the prospects for long term deflation are close to nil, in my view.

The U.S. has a fiat monetary system. To the extent that one considers a long-term deflationary outcome unlikely in the context of such a system, then nominal bond yields will tend to rise in order to more realistically reflect reasonable inflation expectations and meet the real rate of return that investors have historically demanded.

For example, if we assume that inflation will tend to revert to its long-term average of around 2.1% (data since 1971) then nominal bond yields will certainly rise from current levels. Assuming that the real rate of return demanded by holders of 10Y Treasury bonds reverts to its historical level of 2.6%, nominal 10Y Treasury bond yields will eventually revert back to their historical mean of around 4.7% - a increase of 250 basis points from current levels.

Thus, 250 basis points of the 625 basis points observed "great divergence" in earnings yield spreads could be explained on this basis.

Furthermore, if we assume that inflation actually rises above its long-term historical average of 2.1%, then the earnings yield spread will only tend to contract further.

Hypothesis: Stocks Are Extremely Undervalued

It could be plausibly argued that stocks are extremely undervalued. Indeed, if the 12-month forward PE of the S&P 500 were to revert to its mean since 1985, the current PE of around 11.7x could expand to around 18.0x. This implies that the earnings yield for the S&P 500 could come down to around 5.6% compared to the earnings yield of 8.5% at the time of publication of Salmon's original article.

Thus 2.9%, or 290 basis points of the anomaly observed in Salmon's article could be explained by the undervaluation of stocks relative to their mean valuations since 1985.

However, this argument must confront three formidable and interrelated objections:

Historical PEs. Since 1871, the average 12-month forward PE on stocks has averaged roughly 14.5x on an as-reported basis, and roughly 13.0x on an operating basis. The bear argument is that the average forward PE since 1980 have been a historical anomaly and that PEs will tend to revert towards their longer-term mean. On this basis of this argument, it must be acknowledged that stocks currently seem reasonably attractive - although they are not extremely undervalued as bulls have claimed. Furthermore, there have been several historical periods in which stocks have traded at 10x earnings or less for considerable lengths of time - particularly during times of economic volatility or troubles associated with wars, financial panics or depressions. Thus, under current distressed economic and financial conditions it is hardly reasonable to assume that there will be significant price gains from PE expansion towards the historically anomalous levels of the 1980-2000 period.

Normalized earnings growth. Since 1980, several non-recurring historical factors have contributed to faster-than-trend earnings growth for the S&P 500. First, the end of the cold war liberated vast capital and human resources and opened global markets. This was a one-off affair that juiced earnings growth in the decade or two following the end of the cold war and will not be repeated. Second, a technological revolution enhanced productivity growth and earnings growth during this period. The marginal productivity gains from the information revolution must eventually level off, as they always do.

Debt deleveraging. Unsustainable debt accumulation accounted for much of the above-trend GDP growth and EPS growth in the period from 1980 through 2011. Just as debt growth juiced up the GDP growth and EPS growth numbers during the boom years, the process of debt deleveraging will impair GDP growth and earnings growth going forward.

Normalized profit margins. Since 1980, various factors have conspired to drive profit margins to unprecedented heights that will ultimately prove to have been a historical anomaly. One of those factors was the secular decline in inflation and interest rates. In the context of "sticky prices" companies were able to cut production costs faster than the rate at which companies were forced to bring prices down (or restrain the inertial rate of price increases) to reflect marginal costs. Given that inflation and interest rates have collided against the zero-bound limit, companies will not be able to benefit from this "sticky price" phenomenon as much going forward. To the contrary, the potential increase in raw materials and labor costs driven by secular trends in developing nations could drive US production costs upward faster than companies are able to raise prices in the highly competitive and depressed economic environment in the U.S.

Normalized earnings. Robert Shiller has popularized a method of calculating "normalized earnings" referred to as PE10 which suggests that stocks are currently overvalued. I will address this issue in a subsequent article. However, for now, I will merely say that I believe that this methodology is very flawed and that its application at the present time leads to a gross underestimation of the normalized earnings of the S&P 500.

Notwithstanding these objections, and taking all of them into proper account, there are several plausible arguments that support the notion that current valuations should be higher than their averages prior to 1960 and that such higher valuations are sustainable. Consider a few:

International profits. Prior to 1960, the S&P 500 was essentially a domestic stock index. Today, well over 35% of sales and over 50% of profits of the S&P 500 come from outside the U,S. Furthermore, the highest growing and most profitable business segments of S&P 500 companies are in emerging markets, which prior to 1960 represented a negligible portion of S&P 500 earnings or reasonably projectable earnings growth. The opening up of these new markets justify higher PE ratios and lower earnings yields for S&P 500 companies as a whole because of the increased rate of long-term EPS growth that can be reasonably projected due to non-U.S. earnings.

Technology. The productivity of capital has been revolutionized and this has increased the profit margins on the production of virtually all goods and services. Increased productivity enhances the return on capital invested, thereby justifying higher PEs. There are no signs that the productivity gains derived from this technological revolution are leveling off. To the contrary, technological advancements and communications are producing a virtuous cycle of innovation around the world that is gathering steam on a historically unprecedented scale.

International cost arbitrage. Revolutions in international trade, transportation, communications and investment allow companies to geographically move production to wherever it is cheapest and most efficient to produce goods and/or services. This is a revolutionary development supporting increased profit margins that was simply not available to the same extent prior to 1960. Globalization and the arbitrage opportunities that it affords producers is a factor that is not likely to be transitory: This is likely to be a sustained advantage to the owners of capital and producers in the very long term.

Changes in capital markets increase demand for stocks. The ease with which individuals and institutions are able to invest in stocks today has permanently shifted the indifference point between stock ownership and other outlays on investment and/or consumer goods and services. Prior to 1960 investing in stocks was a time-consuming and costly affair. Only a few wealthy people and enthusiasts were willing and/or able to overcome the obstacles. Today, virtually anybody, no matter how poor or ignorant, can invest in stocks with virtually no effort. 401K and other such plans developed and popularized after 1960 further incentivize people to invest in stocks. Finally, due to the ease with which investments can be made, stocks currently attract savings that historically might have been spent on other investment goods such as rental properties, farming or CDs. Furthermore, the ease of equity investment persuades many individuals, particularly at the lower income levels, to invest (or speculate) in stocks rather than spend income on consumer goods. All things being equal, the increased demand pool for equities increases the value of stocks relative to their earnings, and with respect to to historical parameters of PEs or earnings yields.

Hypothesis: Stocks Are Undervalued And Bonds Are Undervalued

Please note that if one takes the 250 basis points that result from the apparent overvaluation of bonds relative to historical averages and adds the 290 basis points that result from the apparent undervaluation of stocks, the result is a 640 basis point divergence.

Thus, the combination of stocks being undervalued and bonds being overvalued can hypothetically explain the entirety of the 625 basis point "great divergence" described by Salmon.

This is a very enticing possibility that, if true, would provide both bond and stock investors with a historical opportunity to earn abnormal returns on their investments.

This possibility is not easily dismissed. The theoretical relationship between stock and bond yields is one that will allow only so much divergence in liquid and global market for financial assets. Unlike the case prior to 1960, market inefficiencies are much more likely to be arbitraged away today than they were at that time.

Note that I am not arguing that the correct theoretical spread between the S&P 500 earnings yield and the 10Y bond yield is equal to 0%. The "correct" theoretical spread is not my concern here. My concern in this article is merely to show that there is a strong theoretical relationship between the yield on stocks and the yield on bonds and that this relationship provides an intellectual framework from which we can observe the empirical behavior of earnings yields relative to stock yields and compare this behavior to marginal changes in fundamental developments that affect the economy and the prospects for equity earnings.

Indeed, the theoretical link between bond yields and earnings yields robust enough that virtually any argument that can be thought of to posit the widening of the earnings yield spread beyond a certain point will result in a countervailing fundamental factors that will tend to cause the earnings yield spread to contract. Consider a few such arguments:

Lower profitability. One argument against earnings yield spread contraction might be that that deteriorating economic and financial conditions will cause the long-term rate of return on invested capital to fall, thereby depressing PEs and expanding earnings yields. However, if this is the case, all else being equal, the equity risk premium may also fall to offset this development. After all, the equity risk premium is simply the excess return that stocks earn above the risk free rate. If returns on capital will fall, then the return on all financial assets will fall relative to the stock of investable funds. This suggests that equity risk premiums will fall as investors will no longer be able to demand historical excess returns of 4.5%-5.0% above the risk free rate.

Inflation. Some argue that high inflation could erode the PE on stocks. Indeed, this is conventional wisdom. However, this is far from clear in the current environment. Inflation is defined by a rise in the price of goods and services. And whom, may we ask, produces and sells the goods and services that are rising in price? The answer is that it is largely S&P 500 companies. Thus, it is far from clear that inflation should cause PEs to decline. EPS growth could conceivably accelerate with the return of pricing power (e.g. prices rise faster than wages which comprise 70% of all costs). Furthermore, the negative effects of higher interest rates that are associated with inflation can be counterbalanced by the lower cost of capital enjoyed by companies in real (inflation-adjusted) terms. Indeed inflation can be expected to cause nominal bond yields to rise while PEs need not be overly affected. This would cause a compression of the earnings yield ratio. Finally, a moderate dose of inflation could conceivably be perceived by market participants as a viable long-term means of achieving debt deleveraging. Such a view would suggest PE expansion for stocks.

Deleveraging. It is true that deleveraging may affect the rate of GDP growth in the U.S. and the EPS growth that derives from this source. However, not all countries around the world are deleveraging. Quite to the contrary, the citizens and businesses of developing countries are for the first time in their history gaining access to credit. This implies super-charged rates of EPS growth for S&P 500 companies that are doing business in these countries. Furthermore, deleveraging in the U.S. implies a lower demand for capital in the U.S. This implies a lower equity risk premium applied to U.S. productive assets as investors demand lower excess returns on capital on those assets relative to a risk-free benchmark.


The hypothesis that stocks and bonds are currently mispriced due to a financial panic and that their values will revert to levels seen in the 1980s and 1990s must be analyzed within a broad historical context of bond and stock valuation. The analysis must begin by acknowledging that, historically speaking, the 1980s through the present have represented an economic and financial departure from longer-term trends.

The question is therefore not whether stocks and bonds might revert to observed patterns established between 1960 to 2011, but why the historically unusual trends observed during this period (particularly during the 1980s and 1990s), might persist?

As it turns out, there are very strong arguments that support substantially higher valuations for equities in terms of PEs and earnings yields than the historical mean that prevailed prior to 1960. The world has unquestionably changed since 1960.

In many ways, considering the revolutionary changes that have transpired in technology, international commerce, cross-border finance and other developments it is perhaps more difficult to make the argument that things are going to go back to the way they were prior to 1960 than to argue that they will go back to how they were in the 1980s or 1990s.

Improved productivity of capital, sustainably higher profit margins, sustainably higher returns on capital, opportunities for growth in developing global markets and profound changes in capital markets and equity ownership discussed in this article suggest that PEs and earnings yields for the S&P 500 should be substantially higher than the mean prior to the 1960s.

Furthermore, it is very difficult to make the case that the value of 10Y Treasury Bonds is not too high and their yields are not too low. In a fiat monetary system, it is extremely difficult to countenance the expectation that inflation rates can be sustained near or below zero. The opposite seems far more likely. Central bankers with a mandate to promote full employment may well be forced to run the risk of provoking inflation above historical norms. Indeed, this structural constraint confronted by central bankers suggests that, if anything, nominal bond yields in coming years should rise above the historical mean.

Therefore, whether you look at it from the point of view of stocks, and even more so from the point of view of bonds, it seems that earnings yield spreads are highly likely to contract significantly in coming years.

In my next article in this series, I will lay out the contrary case.

Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.