Using 5-Year Shiller PEs To Influence Risk Profiles

Summary

• We look at data from 1871 through 2009 to examine how valuation quintile affects subsequent long-term returns.
• For a long-term investor (10-years or longer) dollar-cost averaging tends to dominate all strategies, even in the richest valuation quintile, primarily due to the late 1990s period.
• Over 1 and 3 year time horizons, the impact of valuation is more pronounced on the distribution of subsequent returns.

Editor's note: Seeking Alpha is proud to welcome Long-Short Manager as a new contributor. It's easy to become a Seeking Alpha contributor and earn money for your best investment ideas. Active contributors also get free access to the SA PRO archive. Click here to find out more »

Below (Figure 1) we look at something called a 5-year P/E, using the S&P500. This basically takes the average of the last 5-years of earnings on the index as the denominator, and the price of the index at the end of that 5-year period as the numerator. If you think about the index as representing the entire economy, this basically tells you what multiple the market is offering economic profits at. The reason to use a 5-year P/E is that this tends to smoothen out the economic and profit cycles so you get a more stable sense of where profits are selling. For example, a P/E multiple of 10 would correspond to economic profits of $1 selling for $10; said another way, you could spend $10 to earn $1 of profits per year, meaning a 10% earnings yield (so E/P, the reciprocal, would be the 5-year earnings yield).

This is the historical pattern, with percentile lines drawn in for the 20^th,40^th,60^th and 80^th percentiles of the valuation level.

A P/E(5) of 10.6 or cheaper has only been offered 1/5 of the time; interestingly, these are all times of economic turbulence and worry: the great deflation and panics of the 1870s, multiple recessions, WWI, the great depression, WWII, the inflationary early 1980s, and most recently, March 2009. The other percentile lines correspond to P/Es of 13.8, 16, and 19.3. There is considerable evidence that sales are followed by mark-ups and mark-ups by sales, with the timescales of transitions on the order of a decade or so (between the cheapest and most expensive quintiles, anyway).

So an obvious question to ask is this: given that you were inside a particular valuation band, did your distribution of returns look any different than the overall average distribution?

Conditional Returns

The “conditional” above comes from the language of probability theory (used at least as far back as Bayes in the 1700s), and basically involves splitting up our earlier return data based on valuation band. Unfortunately, looking at many percentiles becomes more difficult given that there are only 27 10-year periods beginning within each valuation band (or quintile). We can, however, see that there is a significant difference by just looking at how the average 10-year return would have been.

The following table gives the boundaries of the quintiles in both earnings yield and P/E terms, and looks at the subsequent 10-year returns for each quintile. The mean wealth and std(standard deviation) columns give the ten-year return in terms of the value of $1 ten years later while the annualized numbers give the same return in terms most of us are used to seeing – as a percent return which compounded for 10-years would produce the dollar value shown in the mean wealth row.

The results show that if we are in the middle 3 quintiles, we can’t really improve upon the dollar cost averaging method, at least with index investing. But when valuations are in their richest quintile, we can expect to earn only 6.4% a year over the next ten years. This clearly suggests that we need to think about whether we should follow dollar-cost averaging in the richest quintile. Furthermore, as we age, the percentage that we are saving and investing in relation to our existing investment portfolio necessarily goes down (mostly because our incomes tend not to grow as fast as our stock of compounded wealth). This means that the cost of any investing mistakes is much higher than when our annual savings is a relatively large percentage of our net worth (which is why advisors have been recommending age-based asset allocation for decades).

The question is – what can we do about those time periods when the market is in the richest quintile? What alternatives do we have to staying invested in the market through those time periods? Clearly if bonds or cash are likely to return greater than the 6.4% mean of that quintile, there is an easy alternative. But based on the historical data, bond yields beat 6.4% only about 20% of the time. So on the surface it looks like the conventional wisdom of always staying invested (at least with a 7-10 year horizon) might be right. However, this ignores the point that for individuals with substantial net worth in relation to income or savings, if we don’t have a substantial amount of net worth in bonds or cash already, when the rich market eventually transitions to a cheap market, we will have very little cash available to put to use at those cheap levels. If bonds are yielding, say, 4%, then the potential 2.4% we are giving up for the 10-years that it takes for equities to return to cheap levels can be more than made up for by the 13.8% prospective returns (which we can think of as about 4% out-performance relative to the long-term average). And that’s if the valuation reversion takes the full 10-years – there are many cases where it took quite a bit less time. It is also worth noting that a sizable part of the good returns for the richest valuation quintiles come from the late 1990s internet bubble - not necessarily something we'd want to bet on happening again in our lifetime.

Digging Deeper

Looking at all equity investment decisions using 10-year horizons is good in terms of having sufficient time for all economic cycle and sentiment related effects to wash out, and for business fundamentals to dominate (it is the time frame in which the market is a weighing machine rather than a voting machine[1]). Nevertheless, for the purposes of rebalancing strategy analysis, it is more useful to look at how valuation affected shorter-term returns, for the process of valuation reversion is inextricably linked to underperfomance (it is by under-performing the mean for long-periods that expensive markets becomes cheap, after all). Looking at 1-year and 3-year returns conditional on being in our extreme quintiles we get the following graphs (we include the performance of bonds in the 3-year graph):

Interestingly, the outperformance of cheap equity markets is apparent even at 1- and 3- year horizons. In a cheap (Q5) market, a relatively higher than normal (for ones age) allocation to equities appears much safer even with a 3-year horizon, whereas in a rich (Q1) market even the median return over 3 years appears not much better than bonds, with substantial downside risk.

Our analysis of bonds has been simplistic, because we only looked at yields and hold-to-maturity. Given that transitions to cheap markets from rich markets appear to involve various forms of financial and economic turmoil, and often these involve significant bond price increases (ie interest rate reductions going into economic slowdowns), there is potential upside to the idea of shifting asset allocation based on both which quintile we are in and our investment horizon for the various dollars in our portfolios.

“Timing” the Market & Rebalancing Frequency

We've all heard advice telling people not to try to “time the market”. If by time the market one means trying in the short-run to find peaks and valleys, we agree that it is generally futile, and DCA will dominate this strategy. But there is a rather wide spectrum of strategies one can follow between day trading & technical graph trading, quarterly rebalancing and annual rebalancing, and also a spectrum of reasonable weights we can put on the valuation component when deciding our asset allocation. It seems as illogical to not be able to buy more of something that is cheap (or own less of it when it is expensive) due to an ideological compulsion to stick to an established purist position as to believe that one can divine useful patterns from the almost random daily and weekly fluctuations of the market.

---

[1] “In the short-run the market is a voting machine; in the long run it is a weighing machine” is a quote attributed to Benjamin Graham, who was Warren Buffett’s professor and mentor at Columbia.

Analyst’s Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.