Over the past six months, I've come to the conclusion that the US markets are overheated. This realization came to me not because of a broad overview of the market, but rather, because of a dearth of attractive investment opportunities amongst major US companies. Nevertheless, this has led me to look at historical indicators of overheated markets.
In my prior article, "Now Is the Time To Be Fearful", I examined the P/E ratio of the S&P 500 (SPY) all the way back to 1900. I came to an interesting conclusion: if we define a "stock market bubble" to be an event where the P/E ratio exceeds 25 (in a normal environment) then full-on stock market bubbles in the US are extraordinarily rare. In fact from 1870 - 2013, I can only find one instance: the late 1990's tech bubble.
The 1920's boom, of course, is the prototypical "bubble" in the public perception. However, the evidence is less than convincing. In mid 1929, on the eve of the Great Depression, the P/E of the S&P 500 was only at 17. This is slightly above the historic norm of 15, but it's not exactly in the stratosphere either.
The two other greatest stock market corrections in the past century fail to support a "stock market bubble" hypothesis either. The Depression of 1920 - 21 was one of the worst deflationary events in American history; by some accounts even worse than the Great Depression. The P/E ratio of the S&P 500 at the 1919 market peak: 10.1. That's well below the historical average of 15x.
Even with the most recent recession, earnings peaked in June 2007, while the S&P P/E ratio was only 17.8. Once again, this is slightly above the norm, but far from "bubble" valuation territory.
Examining US markets all the way back to 1870, it becomes clear that valuations have generally made sense before stock market corrections if one assumes a reasonable earnings growth rate. Market corrections tend to happen not because of high valuation multiples, but rather because of rapid earnings crashes that undermine valuation assumptions.
In this series of articles, I want to examine the earnings crashes and market corrections of the past 142 years. I'll also do a bit of a study on the causes of the crashes, which as you will see, can vary quite widely. In this article, I will focus on the 1871 - 1900 period and the three major earnings crashes of that era.
What is a "Bubble"?
Before starting, however, "bubble" is a term of art that can be defined very differently from one person to the next. A bubble, by my definition, is an event where the valuation of a particular asset becomes so egregiously high, that there is virtually no way that the investors could earn positive returns moving forward.
I mentioned that true "stock market bubbles" are rare, but sector-, company-, and asset-specific bubbles are much more common. That's because a "bubble", at its core, is a misallocation of resources. It's a condition where investors overbid the prices for one group of assets at the expense of other groups. Hence, a bubble is a form of malinvestment.
The reason "stock market bubbles" are rare in the US is because the American stock markets represent a wide and diverse range of assets. That said, sector-specific bubbles can have wide repercussions on the broader market. We don't need to look any further back than the deflation of the US housing bubble to see this. That deflation began around late 2005 and slowly morphed into a broader US economic crisis, that severely crimped the US banking system. As the banks struggled and credit condition tightened, the stock market became much more volatile and eventually crashed, alongside of earnings for many major US companies.
For 2007 / 8, the real error committed by investors was assuming that "reasonable growth" would continue into the future. Instead, the housing crash worked to rapidly undermine credit, and resulted in negative earnings growth.
Now that we're through the boring theoretical aspects of this, let's take a look at the hard data.
Earnings Crashes from 1871 - 1900
Let's start off during the great railroad boom era of the late 19th Century. Before I begin, I'll note that most of this data comes two sources: multpl.com and economist Robert Shiller's website. Shiller, thankfully, has market data going back all the way to the 1870's.
There are three major earnings crashes from 1871 - 1900, which you may be able to identify in the chart below.
The next chart shows more detailed stats on these events, including the peak and trough months, earnings per share ["EPS"] during those months, the overall decline in earnings, and the duration of the contraction.
The first crash at the end of 1874 was a 24-month decline where earnings fell 39%. The next crash beginning at the end of 1880 produced a 45% total decline, but lasted much longer at 60 months. Then, the 1892 crash was the most severe with a 57% peak-to-trough plunge over 24 months.
Finally, we can take a look at the market reaction. Interestingly, the market declined the most with the first crash (which was the least severe) and the least with the last crash (which was the most severe).
One final thing to note is that market valuations in the 1870's look rather depressed when compared to the 1880's and 1890's, with average P/E ratios hovering around 11, as opposed to the long-run average of 15.
It's not clear why 1870's valuations were so low. Perhaps investors were more reluctant to bid up prices in the post-Civil War period.
Growth and the Flaws of Valuation
Now, let's jump back to my theme of growth. One of the major reasons for market crashes is that investors tend to almost always assume "reasonable earnings growth" continues into the future. If you've ever taken a valuation course, you might recall doing a DCF analysis and applying a 5% growth for the first three years (as an example) and then a "terminal growth rate" afterwards (normally 3% - 4%). While these concepts make for simple and pretty valuation models, they are almost universally flawed.
The problem is that earnings growth in the US economy has almost always been "lumpy." That is, it comes it dramatic spurts, and then it's often followed by significant contractions. The chart below looks at the 5-yr compounded annual earnings growth rate for the US market in our examined time period.
As you can see, forward earnings rates vary wildly. In October 1873, the forwards earnings growth rate was -7.5%. Three years later in December 1876, it surged upwards to +9.5%. Let's also take a look at the 10-year forward earnings growth rates.
(click to enlarge)Even with the more expansive time period, we still see a ton of volatility. For instance, in December 1884, the forward earnings growth rate was -6.4%. If you invested in December 1890 instead, however, you would have seen forward earnings growth of +5.2% over the next 10 years.
If we chart the 5-yr forwards earnings growth rates alongside the 5-yr forward market returns, we can see a fairly strong correlation. In fact, I came up with a correlation coefficient of +0.795 between the two data sets.
With that, let's take a look at some of themes in our three depressions of the late 19th Century.
The Three Depressions of the Late 19th Century
Let's start with "The Long Depression." Officially, it lasted from 1873 - 1879 and is generally considered one of the worst economic events in US history. There are a few causes historically attributed to it. The first is the after-effects of a series of wars, including the US Civil War (which ended in 1865), as well as the Franco-Prussian War (which ended in 1871). Another explanation, favored by Monetarists is that the return to the gold standard led to tight credit conditions and resulted in price depression. These explanations can be seen as related because the gold standard was frequently abandoned during war time and then re-implemented at some point afterwards.
Major depressions and recessions are extremely common following wars. Economically speaking, wars typically lead to a surge in government spending. This leads to full-employment and temporary economic growth. However, since most of the money is being "invested" in the destruction of assets, little new wealth is being created. Instead, it's simply inevitable that once the war ends, governments draw down spending. This rapid decline in government spending results in a major recession or depression. This was particularly true in the late 19th Century, as nations would come off the gold standard during times of war and re-enter it afterwards, creating an episode of rapid inflation, followed by a harsh period of monetary contraction.
The Franco-Prussian War added a few other issues during the Long Depression. After France was defeated, it was forced to pay a $5 billion indemnity, as well as give up Alsace and Lorraine to Germany. These territories were particularly important because of their iron ore and steel production. Undoubtedly, this would have caused a hit to the French economy and could've created more economic instability in Europe.
While the Long Depression officially lasted till 1879 (six long years), US corporate earnings appear to have started a rebound in 1877. In spite of its duration, it also appears to have been more of a "price depression", than a depression in production, at least in the US. The Long Depression was followed by another railroad boom that lasted from around 1879 to 1882.
The next depression (The Depression of 1882 - 85) was a result of railroad overbuilding. It was actually not nearly as severe as the name would indicate, and like the Long Depression, was more of a "price depression" than a depression in production. Not much else to note about this depression except that its end seems to coincide with the South African gold rush.
The Panic of 1893 had its roots in railroad overbuilding and agricultural troubles in Argentina. The Sherman Silver Purchase Act was also repealed during this time frame, perhaps resulted in tighter monetary conditions.
The late 19th Century depressions have a few themes in common:
1. Wars and war inflation
2. Government changes in currency pegs (e.g. gold, silver)
3. Rapid monetary tightening (often to re-establish gold peg),
4. Market changes in gold and silver supply, such as the South African gold rush,
5. American railroad booms and busts,
6. Agricultural cycles
The past informs the present. In my view, we're actually more likely to find insight 120 years ago than we are 5 years ago, precisely because the average person remembers 5 years ago so clearly, but has virtually no knowledge of the happenings of 1892, for instance.
From examining the crashes from 1871 - 1900, we see a few themes emerge. The first theme is that returns are highly related to earnings growth, which can be extremely volatile due to macroeconomic factors. Common themes driving earnings "booms" and earnings "busts" include wars and war inflation, as well as monetary tightening that often followed wars during that time period.
While the "war" theme might initially not seem to have as much applicability today, many of the ingredients of 19th Century wars are in place in 21st Century America and Europe. Governments typically tried to balance budgets before wars (due to the gold standard's harsh discipline), but once war began, dramatic fiscal excesses were common.
The current budget deficits in US and through most of Europe are actually fairly reminiscent of war-time deficits of the 19th Century, at around 8% - 12% of GDP. The circumstances, of course, vary a bit since few nations have currencies pegged to commodities in 2013; but this still implies elevated risk factors. However, it could manifest itself in the form of inflation and stagflation, as opposed to deflation (the norm in the 19th Century).
The railroad booms might also be instructive. Keep in mind that these booms were government-aided and driven heavily by debt, not unlike the modern American housing sector. Changes in 'railroad policy' from the US government have dramatically altered the market, in the same way one might argue that housing-related policies might.
Overall, there are some interesting parallels between the late 19th Century and the early 21st Century. For my next article in this series, I'll examine the period running from 1900 to 1925.