The US stock market (SPY) is about twelve and a half years into what will likely be the worst seventeen year period in US market history. The two substantial declines of the last twelve years should be dwarfed by the decline likely to unfold in the next one to five years. Investors should start looking for exits from the stock market now and/or consider shorting.
Warren Buffett in the past has recommended buying a stock at a good value and holding it forever. That has always sounded like an exaggeration to me, unless forever means about 17 years. This is the length of period which return has the strongest correlation with the measure of value I call the PEses. The PEses is a PE ratio that uses the monthly average of the daily closing price of the S&P 500 adjusted for inflation for "P" and simple exponential smoothing of real S&P 500 earnings for "E".
Constructing the PEses builds on the work of Benjamin Graham, Robert Shiller and Bob Bronson. Based on Graham's idea of using multiple years of earnings to establish value, Shiller uses a 10 year moving average of real (inflation adjusted) earnings of the S&P 500. His measure of value, sometimes called PE10, is spelled out in his book Irrational Exuberance. The PE10 and PEses use the same "P".
The 10 year moving average of earnings controls for much of the problem where volatile earnings make the stock market look cheaper than it is when earnings have had a huge run up, or more expensive than it is when earnings have plummeted. However, the 10 year moving average of earnings is still modestly susceptible to this problem especially in the last 12 years during which earnings have been more volatile than ever before.
The following two charts show corporate earnings and real earnings with pre and post 1937 trend lines. I date the modern inflation trend to 1937 because this was the first major recession where the Fed prevented significant deflation. Prior to 1937 there were periods of rapid inflation intermixed with extended periods of deflation. During this time the peaks and valleys in the consumer price index ("CPI") did not lead to an upward trend. Since 1937 there have been periods of rapid inflation interspersed with periods of modest inflation and when deflation did occur it was modest and short lived. Apparently, after the 43 month business contraction that ended March 1933, the Fed developed an institutional conviction to not allow deflation. During the sharp 13 month contraction that began in mid 1937, the Fed only allowed modest deflation which quickly gave way to inflation. Since 1937 the CPI has trended up and corporate earnings reflect this inflation trend.
Since 1937 the best fit growth rate for real earnings is 1.9%. If you calculated the best fit growth rate from 1937 to 2000 it would be 2.1%. Ominously, the growth rate has declined even as earnings have become the most volatile in history. What many people believe are strong earnings in the last 3 years I believe are more accurately described as volatile earnings which may soon decline toward or below the long term growth trend.
The blue line in the chart is the simple exponential smoothing ("ses") of the earnings. With ses each data-point has a higher weight than the one before it. So the smoothed earnings shown actually responds quicker to a change in trend than a 10 year moving average, even though it is roughly as smooth as a 50 year moving average.
The Wall Street adage of, "buy low and sell high" depends on having a measure of value that corresponds with return in the future. Shiller in Irrational Exuberance showed that when his PE10 was at a high level the stock market tended to have a weak return over the next 10 years and when the PE10 was low the market had a good return over the next 10 years. The correlation of PE10 with return is much better than the correlation using a PE based on 12 month trailing earnings or estimated future earnings. A PE based on estimated future earnings typically gives its worst calls right before a recession, showing the market to be very cheap just as a major bear market starts. This was the case in 2000 and in 2007. Its indication that today's market is cheap should not be trusted.
Looking at a 10 year period of return is useful, but there is nothing magic about 10 years. Bronson started looking at the relationship between value and returns for periods from 1 to 20 years. The PE has almost no impact on what the market will do in the next year. Value begins to have a meaningful impact for periods longer than 5 years and becomes quite significant for periods of 12 to 20 years.
The PEses has a stronger correlation with future return than the PE10 and 17 years is the period length with the strongest correlation.
The chart above shows the correlation between the total real return for 17 year periods and the PEses. The green scales are for the PEses and are moved forward 17 years to show the return for the 17 year period aligned with the PEses at the beginning of the period. The scale is inverted to show that when the PEses is high the next 17 years have a low return and when the PEses is low the market has a high return.
The strongest ever 17 year period began August 1982 when the PEses hit a low of 10.9. So far the weakest 17 year period ended in August 1982; it began with a PEses of 37.5 in August 1965. I do not expect this record to last; the correlation implies the 17 year period from December 1999 to December 2016 will annualize declining 11.65%, for a total decline of 88% from December 1999.
However, the 88% decline is based on the relationship being linear and there is not yet data to confirm a linear relationship with such a high PEses. It may turn out the relationship is logarithmic or of some non linear form. So the decline could be less, but there is not historical data to make an estimate.
The PEses of 35.6 for July 2012, on the other hand, is in a range with historical precedent and implies the stock market after inflation and dividends will annualize, returning 0.8% for the next 17 years. This is a higher valuation than 11 of the 12 months in 1929.
The stock market is not dead, but there is a lot of air yet to come out of the bubble. As the air comes out in the next few years Bill Gross may appear closer to reality than Jeremy Siegel. But, when the PEses falls to 21.5, which is about its average prior to 1998, the estimated return after inflation for the next 17 years will be Sieglel's constant 6.6%.
With the current economic crisis I expect the PEses to fall at least as low as it did in 1982 to 10.9, which would represent about a 70% drop from here. Of course, I believe it would be an extraordinary buying opportunity at that level.
The current high level in the PEses suggests the downside risk of the next few years dwarfs any upside potential. While the PEses may be the best measure of value I have seen it is not a timing tool and does not indicate what will happen in the next year. Yet, if the goal is to sell high it definitely shows the market is high.
Disclosure: I am short SPY.
Additional disclosure: There is no guarantee analysis of historical data and trends enable accurate forecasts. The data presented is from sources believed to be reliable, but its accuracy cannot be guaranteed. Past performance does not indicate future results. This is not a recommendation to buy or sell specific securities. This is not an offer to manage money.