In past notes we’ve referenced long-term, 10-year normalized P/E ratios and the relationship to future 10-year annualized returns. With the 10-year P/E in the United States currently near 23x, we thought it might be interesting to look at outcomes when P/E ratios fell in a narrower range. In this case, we’ve focused on monthly P/E outcomes between 20x and 25x in an attempt to see how markets performed over the following 10 years on an annualized basis in valuation environments similar to the current one.
As in the past, we’ve used the Shiller earnings database as the basis for the normalized earnings calculation. For consistency, we’ve used real, total annualized 10-year returns (i.e. adjusted for inflation and including dividends). Finally, we use monthly observations from January 1925 to July 2003, the last month for which we can derive a 10-year performance number. Over that time period, there were 154 monthly observations with a P/E ratio between 20x and 25x out of a total of 943 monthly observations.
First, let’s look at the basic descriptive statistics for 10-year annualized returns following a P/E observation in this range:
AVERAGE 2.01%
MEDIAN 1.05%
MAX 8.72%
MIN -3.23%
STDEV 3.78%
The results above compare to a median total return of 6.32% and an average total real return of 5.72% across the entire period from 1925 to July, 2003. Therefore, we can make a very surface observation that valuation environments similar to the current one in the US have generated decidedly subpar performance over the ensuing 10 years.
However, looking at a basic histogram of the performance data points for valuations between 20x and 25x presents a more complicated view point:
As you can see, the histogram shows a number of results clustered in the “tails” so to speak. Instead of seeing a distribution with the bulk of the outcomes in the middle, results between 20x and 25x have tended to come out somewhat extreme. Approximately 75% of the observations either come in outright negative, or greater than 5.5% annualized.
Does this mean that making forward performance conclusions based upon valuations in this range is a fruitless exercise, i.e. it’s easy to say performance could be really decent or really bad?
Not necessarily, in our view. Digging even deeper, we find it instructive to look at the time periods in which the observations occurred.
Every single P/E observation between 20x and 25x generating future 10-year returns greater than 5.5% per annum (the right side of the histogram) occurred in the early to mid-1990s. Of course, the 10-year return profile for each one of these observations captured what may be the robust bull market run in history during the late 90s. Even though markets declined from 2000-2002, the positive effect of the late 90s bull outweighs the negative effects of the subsequent bear market.
On the flip side, many of the most negative 10-year return observations followed P/Es between 20x and 25x observed in the mid to late 1960s. Thus, those return profiles capture the effects of the ugly “stagflation” period experienced in the 1970s.
The “middle” observations in the histogram occurred across a number of years in the 20s, 30s, 60s, and 2000s, capturing a wider range of economic environments, from deflation to normal inflation, robust economic growth to depression, and geopolitical uncertainty to relative normalcy.
What do we conclude in the end? Excluding the data points from the 1990s incorporating the massive late 1990s bull market spike, every other P/E observation between 20x and 25x produced sub-median/sub-average 10-year forward annualized returns. Unless the markets are getting ready to experience another massive market spike like the one experienced in the late 1990s, a result of an incredibly unique set of market circumstances such as the emergence of the internet, a robust nominal and real GDP environment, and other factors, we’d wager that the real return profile for the next 10-years will look more like the less than inspiring outcomes in the middle or left side of the histogram. Of course, as we’ve observed in the past, return patters within those 10-year periods can be quite lumpy and can produce many opportunities for bulls and bears alike. For instance, real total annualized returns from June 2003 to June 2013 came in at a sub-median 4.19% annualized. Over that period, though, we observed a solid cyclical bull market from 2003 to 2007/2008, a massively devastating bear market during 2008/2009, then a robust bull market from 2009 to the present. Others periods in that return range produced similar patterns. Thus, it’s not a stretch to believe markets will face some more lumpy down and up periods over the next 10-years producing “ok” but not fantastic real returns. And, in light of the fact that long-term valuations are within the higher range of the historical record, we’re skeptical of talk that markets have entered a new robust, multi-decade secular bull phase. Finally, as a reminder, ignore skeptics that tell you the 10-year P/E ratio is “invalid” because it incorporates the massive profit down-spike of ‘08/’09. In reality, the current real 10-year trailing earnings number is currently at an all-time high, not to mention it is currently at a level significantly above the long-term trend line, as shown below.
