Great Depression Economics 101: What Historical Numbers And Charts From The Great Depression Foretell About The Economy And Stock Market

Has another Great Depression begun?

A Washington Post headline from last Thursday answers in the affirmative, reading: America is in a depression. The challenge now is to make it short-lived.

In this post I discuss the relevance of historical trends in the U.S. economy for assessing the uncertainty we face about the depth and duration of the current contraction. While this is a standalone post, it also comprises Part III of a blog series of tutorials on Great Depression economics. Notably, this post serves to infuse data into the conceptual discussion in Part I, about negative shocks to aggregate supply and aggregate demand, and the discussion in Part II, which focuses on the stock market’s reaction to theses shocks.

A quick summary of the takeaways from Part III appears in a Conclusion at the end of the post. Readers who prefer short posts can scroll down at this point. For everyone else, here we go, beginning with a look back at the last twenty years, with which we are familiar, and then taking a similar look at the years marking the Great Depression.

Historical Trends in the Last Twenty Years

Consider the twenty year experience of the economy and the stock market prior to the onset of the pandemic. Figure 1 below displays a graph of real GDP from 1999-2019.¹ The graph provides a quick sense of how the U.S. economy performed over the last two decades. Notice that Figure 1 depicts robust growth at the tail end of the dot.com bubble, followed by a recession in 2001 marking the bursting of the bubble, a robust recovery leading to the Great Recession that began in 2007, and subsequent recovery.

You can see in Figure 1 that both recessions correspond to a flattening of the curve, occurring at the points with small dips. In the case of the Great Recession, the trajectory of real GDP lies below a line extrapolating the GDP-trajectory during the expansion between 2002 and 2007, with a slower pace of growth.

Next, consider Figure 2 below, which depicts the time paths of the S&P 500 and associated earnings between 1999 and 2019.² What this figure tells us is that broadly speaking, the trend in stock values reflected the trend in earnings. If anything stands out in this figure, it is the accentuated path of earnings relative to stocks, especially during the Great Recession. Stocks declined during the Great Recession, but not to the same degree as earnings.

In Part II of this blog series, I described the role of market sentiment in driving stock prices. One way to get a handle on market sentiment during the last twenty years is to consider the time path of price-to-earnings (P/E) relative to that of earnings. Figure 3 below displays how nominal earnings and the cyclically adjusted real P/E ratio CAPE (developed by economists John Campbell and Robert Shiller) co-varied over the period.

The average P/E in the period 1999-2019 was almost 27, much higher than its historical average (from 1926) of 18.25. A high P/E reflects optimism. During the dot.com bubble, P/E peaked over 40, reflecting a degree of optimism that former Fed chair Alan Greenspan and economist Robert Shiller characterized as “irrational exuberance.”

Figure 3 clearly shows that during this period, P/E tracked earnings. When earnings fell, during a recession, P/E fell. When earnings rose during an expansion, P/E rose.

My academic work studying the period between these two recessions, with colleagues Giovanni Barone-Adesi and Loriano Mancini, quantifies the degree of excessive optimism and overconfidence over time. In line with the discussion in in Part II about the psychology of booms and busts, our analysis led us to conclude that the market’s judgement about expected returns was pessimistic by 2% during the recession of 2001, and about 5% during the Great Recession. During the expansion between these recessions, excessive optimism rose to 2%, which it reached in mid-2007, before declining.

Historical Trends During the Great Depression

With the past twenty years as prologue, consider next the time paths of real GDP, the S&P 500, earnings, and CAPE during the Depression. Figure 4 below shows the time path of real GDP. GDP fell dramatically from 1929 through 1932, and leveled off in 1933. From peak to trough, real GDP fell by 18.4%. For the rest of the decade, GDP rose, except for a recession in 1937. The largest decline was from 1931 to 1932, when GDP fell by almost 13%. During the recession of 1937-38, GDP fell by 3.3%.

The recession of 1937-38 was especially interesting, as the U.S. economy had by then begun to recover robustly. This recovery appears to have produced a sense of overconfidence in the economy, coupled with a concern about the magnitude of the government’s budget deficit. U.S. policymakers at the time moved to try and balance the federal budget, and in the course of doing so, stalled the economy’s progress.

During the Depression, the U.S. stock market was extremely volatile. Figure 5 below displays the time paths of the S&P 500 and associated earnings during the period 1929-1939. As can be seen, stocks tracked earnings very closely during the period, much more closely than has been the case in the recent twenty years (or as I discuss below during the twenty years that preceded the Depression).

Figure 6 below displays the time paths for earnings and P/E. The message from this figure is that P/E also tracked earnings very closely; and P/E peaked in October 1929, the month the stock market crashed. In textbook finance, P/E is relatively stable, reflecting the cost of equity capital and the net present value of firms’ projects. During the Depression, investor psychology was absolutely whipsawed by earnings growth.

The COVID-19 Induced Economic Contraction

A new paper by economists Martin Eichenbaum, Sergio Rebelo, and Mathias Trabandt describes how the performance of the U.S. economy over the next few years will depend on the strength of the country’s containment response to the COVID-19 pandemic. Their paper considers a variety of situations, including the situation where the country’s medical system does not have the capacity to deal effectively with the surge in infections, but with the possibility that medical researchers will develop vaccines and/or treatments. In this situation, if the country’s response to the virus is appropriate, efficient, and based on the historical measures it has used to value life, then it will take almost two years for the U.S. economy to return to normal growth, with a peak-to-trough decline roughly around 25 or 30%. I would note that such a decline is larger than that which occurred during the Depression.

Eichenbaum, Rebelo, and Trabandt analyze the impact on the economy of having weak containment instead of appropriate containment. When the medical system lacks adequate capacity, as above, they suggest that with weak containment, the economy would experience a much smaller decline peak-to-trough, between 15 and 20%, with a recovery period that is about fifteen months. In the most recent version of their paper, they suggest that stronger containment will save roughly 500,000 lives, with total deaths being about 930,000 instead of 1.43 million.

Stock Market Risk

Economic forecasting is notoriously difficult, and forecasting financial markets is even more so. The range of economic and financial outcomes is wide. The length of the contraction that has just begun might be similar to that of the Great Recession, roughly a year and a half; or it might be similar to that of the Depression, which Depression scholar and former Fed chair Ben Bernanke tells us is twelve years. The range of potential outcomes is wide, and much depends on containment policy and economic policy.

Future outcomes also depend on vulnerability to psychological pitfalls such as overconfidence. Overconfidence might well have lengthened the Depression by removing stimulus too early, thereby creating the recession of 1937-38. By the same token, overconfidence might lengthen the current contraction by easing pandemic containment measures too early, and as a result experiencing a second wave.

Rather than offering forecasts, let me instead focus on patterns from the Depression that will plausibly be manifest in the stock market as the current contraction unfolds. These patterns pertain to the time paths of earnings, stock prices, and market sentiment as reflected in P/E.

Figure 2 above illustrates the time paths of the S&P 500 and its associated earnings in the twenty year period 1999-2019. Figure 3 illustrates the time paths of earnings and CAPE over the same period. Consider the counterpart twenty year period that preceded the Depression. Figure 7 below is the counterpart to Figure 2.

Figure 8 is the counterpart to Figure 3.

By comparing Figures 2 and 7, we can see that stock prices tracked earnings during both twenty year periods, with stock price changes being more muted than corresponding earnings changes. By comparing Figures 3 and 8, we can see that for the most part, sentiment (measured by CAPE) also tracked earnings, although not during the part of World War I that preceded the U.S. entry into the war. I would also point out that the major stock market bubble associated with 1999-2019 occurred at the beginning of the period, whereas the major stock market bubble associated with 1909-1929 occurred at the end of the period.

With the above comparisons in mind, consider the extent to which the relationships among earnings, stock prices, and sentiment that play out over the next few years will be similar to those experienced during the Depression. In other words, to what extent will be patterns similar to those displayed in Figure 5 and Figure 6?

My sense is that it is plausible to expect the general character of these relationships to be similar to those that were manifest during the Depression. Plausibly, stock prices and sentiment will both track earnings, and do so more closely than they did during the two decades preceding the beginning of the contractions.

With these observations in mind, consider how stocks might move as the current contraction unfolds, as economic activity reaches bottom out and then rebounds.

Earnings growth reflects economic growth. At the moment, the outlook for economic growth is negative. As I write this post, the S&P 500 is down from its peak by 17.6%, and had been down 35% just a few weeks ago. During the Depression, the S&P 500 fell 85% from its monthly peak in 1929.  During the Great Recession, the S&P 500 fell almost 51% from its peak in October 2007. Although possible, my sense is that the peak-to-trough decline for the current contraction will not be worse than it was in the Depression. Therefore, the question in my mind is where in the range 35% to 85% it will fall.

Inflation

Inflation is a factor to consider. Between 1974 and 1982 stocks were depressed, and there is reason to believe that runaway inflation was a prime contributor. During the last twenty years, U.S. inflation has been modest. During the period 1909-1929, inflation was also modest, except for World War I when it was high. What might we expect to happen to the price level during a contraction? Figure 9 below displays the time path of the Consumer Price Index during the Depression.³

In my previous post, I discussed theoretical reasons why the price level might decline during the initial phase of a contraction. As can be seen in Figure 9, at the outset of the Depression the price level did fall; and it continued to fall until 1932 even though GDP and earnings were still declining. However, the stock market reached its trough in mid-1932. After mid-1932, for the rest of the decade, the price level stabilized and inflation was low. Stocks then rose, with earnings, and earnings generally rose until the recession of 1937-38.

What if the Contraction Does Turn Out to be a Depression?

During the past three weeks, more than 16 million people have filed for unemployment insurance. Job losses are already twice the size that they were during the Great Recession, and are echoing the job loss trajectory from the Depression.

Nevertheless, there are important differences between the Depression and the current contraction, and it is important to keep these in mind when contemplating the future. Economic stimulus from both fiscal policy and monetary policy has begun much earlier during this contraction than occurred in the contraction of the 1930s. In addition, there are social safety nets in place today that were not in existence during the 1930s. Institutions such as the FDIC, SEC, and Social Security Administration, and programs such as unemployment insurance, all came into existence several years into the Depression.

Because of these differences, there is reason to anticipate that the current contraction will not be as prolonged as was the Depression. Nevertheless, for readers wondering what if — what if the the current contraction of the U.S. economy ends up being similar to the Depression — Figures 10 and 11 below offer a glimpse of what the first three decades of the twenty first century might look like in respect to GDP, earnings, and stock prices.

Figure 10 below displays the time path of real GDP, with past actual historical values in solid blue and subsequent future values displayed in dotted blue.

Figure 11 below displays the time paths of the S&P 500 and associated earnings, with past actual historical values in solid colors and subsequent future values displayed in dotted colors.

Figures 10 and 11 make clear that a repeat of the Depression would be much more severe than the Great Recession. My sense is that the risk is low that the current contraction will turn out to be comparable in both magnitude and duration to the Depression. That said, it is still conceivable that the current contraction will feature a comparable, if not larger decline in real GDP, from peak-to-trough. Indeed the work of economists Eichenbaum, Rebelo, and Trabandt, discussed above, suggests that this is likely.

Less likely is that the duration of the contraction will be as long,⁴ although with human psychology being what it is, mistakes can happen. Rising levels of debt, upon which consumers, firms, and governments will need to rely during the contraction, hold the potential to prolong the downturn by suppressing future aggregate demand. If the trajectory suggested by Eichenbaurm, Rebelo, and Trabandt were to occur, then the dotted portion of the GDP curve in Figure 10 will turn out to be more compressed at the right end, with the duration of the U-shaped portion being perhaps between 15 and 18 months, with the peak-to-trough decline being comparable if not larger than that depicted in Figure 10. Just keep in mind that these figures pertain to aggregate supply shocks only. The follow on shocks to aggregate demand are likely to amplify these effects, extending the duration of the downturn. In addition, the left end of the U-pattern will feature an extremely steep decline, as a result of the containment shock to aggregate supply.

As for the dip at the right most portion of Figure 10, if a psychological phenomenon such as “decision fatigue” induces the country to declare victory over the pandemic prematurely, then we might be in for a double dip as happened during the Depression and for that matter again during the beginning of Ronald Reagan’s Presidency.

In respect to the stock market, the remarks about a compressed time frame for Figure 10 (GDP), as discussed above, also apply to Figure 11 (stock prices and earnings). With luck, the compressed time frame in Figure 11 will also feature reduced amplitude; but, with the recent rise in algorithmic trading, the probabilities associated with high volatility and flash crashes are higher than they were in the past.

I propose Figures 10 and 11 as plausible scenarios to help guide our intuition about stock market risk, in respect to how the future might unfold; however, I do not propose them as forecasts. In this regard, some of the key assumptions about the severity of the pandemic that are built into the model developed by Eichenbaum, Rebelo, and Trabandt differ from comparable assumptions used in the epidemic models upon which U.S. policy makers rely. It strikes me that these disparate assumptions provide reasonable forecast confidence intervals.

Conclusion

I think it is conceivable but unlikely that the the actual time paths for GDP, earnings, and stock prices will all feature low amplitude, low duration versions of the dotted tail ends displayed in Figures 10 and 11.

In regard to stock market risk, I urge readers to remember the lessons described in Part II about psychology, optimism, confidence, and market sentiment. As of last Thursday, the stock market was not in a bear market. However, historical trends suggest that stock prices strongly track earnings during a deep contraction; and earnings announcements of U.S. firms for the period since the pandemic began are yet to come!

If historical patterns repeat, then stock investors will continue to experience a rough ride accentuated by swings in sentiment, as earnings decline sharply and rebound, and market sentiment moves from excessive optimism to excessive pessimism and back again.