Do Past 10-Year Returns Forecast Future 10-Year Returns?
Bill Hester, CFA
All rights reserved and actively enforced.
Can the process of forecasting long-term stock market returns be simplified to include just one step calculating the prior decade's return? This is one of the arguments that analysts are currently using to convince investors that the coming decade will offer above-average returns. It's important to take a closer look at the argument because it's become widely discussed and reported. A strategist at a major investment bank argued recently that poor 10-year trailing returns is reason enough to expect lofty returns over the next decade. A similar argument was recently made in Barron's, and by various mutual fund companies.
Some of the research is structurally flawed. One piece examines the 50 worst 10-year returns since 1871 to construct a sample to calculate subsequent 10-year returns. The study used monthly data, so many of the observations clustered within a single year. 11 of those "worst" returns occurred during the past decade. Of the remaining 39 months for which the subsequent 10-year stretch of returns is known, 32 of them occurred around 1920, so the study is heavily influenced by a single period. The implicit argument is that the next decade will look much like the Roaring 1920's.
But even using a broader set of periods with poor trailing returns, the average return during the decade that followed has typically been slightly above average. In fact, some of the individual periods have provided strong returns, especially when they marked the beginning of secular bull markets, like in 1942 and the early 1980's. The graph below shows the 10-year rolling total return of stocks since 1929.
Looking at the graph, it is clear that 10-year returns for the market have varied a great deal, creating long "secular" periods of above-average and below-average returns. There are just few periods where 10-year trailing returns fell to very low levels after the stock market crash of the early 1930's, in the early 1940's, and again during the late 1970's and the early 1980's (on an inflation-adjusted basis, the 10-year returns during these last two periods were also negative). A cyclical bull market followed the low returns of 1933, and secular bull markets followed the low returns of the early 1940's and early 1980's. When looking at the data below, we'll include the 100 worst months of 10-year trailing total returns beginning in 1929. This group of months will capture the bulk of the periods just mentioned.
The Drivers of Returns
Once the observation is made that long periods of negative trailing returns have been followed by strong subsequent returns, it's logical to ask whether there are other characteristics that those strong decades shared. To that end, it makes sense to begin with the fundamental underpinnings of stock prices: growth and valuations.
The valuation argument is relatively straight forward. On a cyclically adjusted earnings basis (where profits are averaged over the prior decade), the average cyclically adjusted P/E ratio following periods of poor long-term returns was 12. That compares with the long-term average of 16 using the entire historical period, or 14 if you exclude the Bubble Years surrounding the year 2000. Either way, not surprisingly, valuations tended to be quite low and below average following decades of poor stock performance. The cyclically adjusted P/E ratio is graphed below, with the red line showing the long-term average following poor long-term returns.
At the low in 2009, the CAPE was close to the average level of prior periods that followed poor 10-year returns. But the current argument being made by analysts is not about investing at last year's low. It is about investing at current levels of valuation. And here, the graph shows that today's CAPE of over 21 sits far above the levels that typically lead to strong long-term returns.
To drive earnings over the long-term, economic growth matters too. Here the gap between today's characteristics and prior periods widens further. In the 10-year periods that followed low return decades, the US economy grew at an average nominal rate of 10.5 percent a year (using annual GDP data through 1946 and quarterly data thereafter). So, clearly the economy has eventually grown rapidly following periods where stocks suffered poor long-term returns. This makes sense. It was a long, hard slog from the depths of the depression until full recovery. It took until 1941 for output to climb above 1929's level of GDP. But once it did, the economy grew quickly, both during the build-up to the War and during its aftermath. The economy also grew quickly coming out the early 1980's recession, after a long stretch of mediocre economic performance and poor stock market returns.
This data suggests that we should modify the assumption that poor past returns, in and of themselves, reliably lead to strong subsequent long-term returns. It is more accurate to argue that following poor 10-year returns, provided that valuations are depressed based on normalized earnings and the economy is likely to grow at double digits rates of nominal growth - investors can probably anticipate higher subsequent long-term returns.
There are a couple of arguments that put that latter assumption into question today. One is the secular downshift of nominal economic growth that has occurred during the last two-and-half decades. The graph below plots the rolling 10-year change in nominal GDP. Although a long-term drop in inflation explains part of the decline, the inflation-adjusted also data shows a downshift. Ned Davis typically shows this chart to his subscribers along with one that depicts the increased levels of debt in the economy, making the case that higher levels of debt have been producing lower levels of GDP growth. Another reason for this downshift is that according to Commerce Department data, demographics and other factors have caused the growth rate of "potential GDP" to slow persistently in recent decades. Regardless of the cause, the graph does show that in order to attain high rates of nominal GDP growth from current trends, very high levels of inflation would likely be needed. And historically, abrupt shifts from low levels of inflation to high levels of inflation have delivered investors poor returns.
Estimating Long-Term GDP Growth
There are a couple of ways that we can obtain an estimate of GDP growth over the next decade. One is from combining estimates from economists and bond investors. The Livingston Survey, which is collected by the Philadelphia Federal Reserve Bank, periodically asks economists for their 10-year forecast for GDP growth. The most recent survey showed that economists expect the economy to grow at an average rate of 2.80 percent a year over the next decade, adjusted for inflation. The current spread between 10-Yr nominal Treasury bonds and 10-Yr Treasury inflation-protected bonds is currently about 1.8 percent. That gives us an estimate for nominal GDP to grow at about 4.6 percent a year over the next decade.
We can also obtain an estimate of the likely growth rate of the economy by relying on historical precedent. A paper that was presented at the Kansas City Fed's most recent economic policy symposium can help. The paper After the Fall' was written by Carmen and Vincent Reinhart and it discusses common economic outcomes following major credit crises. The work is an extension of This Time is Different, Carmen Reinhart's collaboration with Ken Rogoff on the periods leading up to and immediately following credit crises. In the paper the Reinhart's extend the window of the research and ask the question: what are the long-term effects on inflation, unemployment, and GDP growth following severe credit crises?
Their findings are sobering. For developed countries, around a standard severe credit crisis those that were generally country specific the unemployment rate averaged 2.7 percent in the decade prior to the beginning of the crisis. In the decade that followed, the jobless rate averaged nearly 8 percent. Stunningly, in all five advanced economies and in four out of five emerging economies they studied, the unemployment rate has failed to decline below the levels reached prior to each crisis (even though most of crises in developed economies occurred 20 years ago).
Their data on GDP growth following credit crises was equally uninspiring. For developed countries, GDP growth was typically 1 percentage point lower than the decade prior to the peak. Developed economies grew at an average real rate of 3.1 percent in the decade prior to each crisis, and at 2.1 percent in the subsequent decade. The outcome was even worse for periods that followed severe credit crises which were global in nature. The Reinhart's put the decade of the 1930's and the 10-year period following the 1973 oil shock into this group. Here the rate of economic growth in the decade that followed these global credit crises was cut by half. Developed economies grew at an average of just 1.8 percent a year following global credit crises.
The Reinhart's research also sheds some light on the tendencies of inflation following a collapse in credit. While they note the well documented periods following the 1929 stock market crash (severe deflation) and the 1973 oil shock (high rates of inflation), they also draw a distinction between these periods and the standard (but still severe) credit crises. For developed countries, inflation averaged 6.5 percent in the decade prior to each crisis and 2.2 percent in the decade that followed. Crises in emerging countries showed similar patterns. Inflation in these countries averaged 5.9 percent in the decade leading up to the crisis and 3.6 percent in the decade that followed. As the Reinhart's point out, this is all the more remarkable considering that the emerging market countries all sustained massive devaluations/depreciations in their currencies at the height of the economic turmoil.
Within this framework, we can construct an estimate for GDP growth over the next decade. In the 10-year period up to the peak of the 2008 credit crises, real GDP growth averaged 3.1 percent a year. Based on the Reinhart's research, it is reasonable to expect the economy to grow at somewhere between 1.5 percent and 2 percent over the next decade. Inflation averaged 2.5 percent leading up to the crises. That's low by historical standards, so a rough estimate based on their work might be about 2 percent. So, if the economy takes a decade to recover which is standard for post credit crisis periods then the US economy would be expected to grow at a 3.5 percent to 4 percent nominal rate over the next seven or eight years.
Following the worst 10-year returns for the S&P 500, the average cyclically-adjusted P/E was just 12, GDP growth over the following decade average 10.5%, earnings growth averaged 8.63%, and the S&P 500 return over the following decade averaged 11.33. Presently, the cyclically adjusted P/E is above 21, while the prospects for earnings growth are depressed, both based on potential GDP and on the typical aftermath of a credit crisis. It is worth remembering that investors in the Japanese stock market have had rolling negative 10-year returns since 1997.
The argument that above-average long-term returns typically follow periods of poor past long-term returns is not wrong, it's just incomplete. The more complete argument is above-average long-term returns can be expected to follow long periods of low or negative, provided that they end with low P/E multiples on smoothed earnings and precede a period where the economy can be expected to enjoy robust growth. Today, valuations are at levels that have normally been followed by 10-year returns that are well below average. At the same time, based on a template from more than a dozen prior credit crises, the argument that the economy will grow strongly over the coming decade finds little support.
|Home | The Funds | Open an Account | Account Access | Research & Insight | Site Map|
For more information about investing in the Hussman Funds, please call us at
513-326-3551 outside the United States
Site and site contents © copyright 2003 Hussman Funds. Brief quotations including attribution and a direct link to this site (www.hussmanfunds.com) are authorized. All other rights reserved and actively enforced. Extensive or unattributed reproduction of text or research findings are violations of copyright law.
Site design by 1WebsiteDesigners.