Saturday, March 16, 2019

No Free Lunch: Valuation Determines Return

https://ggc-mauldin-images.s3.amazonaws.com/uploads/pdf/TFTF_Mar_15_2019.pdf

Valuation Matters

Markowitz advised us to use relevant assumptions. To assess relevance and reasonableness, let’s consider a period long enough to smooth short-term fluctuations, yet not so long that an investor loses the opportunity to adjust expectations, lifestyle, savings rate, etc. For most investors, that period is a decade or two. If you are age 55 or over, 20 years starts to sound like the long run.
As previously mentioned, the stock market’s nominal long-term annualized total return has been around 10%. Total return includes capital gains as well as dividends. The century-long 10% average is also close to the average annualized return across all 110 decade-long periods since 1900 (i.e., 1900–1909, 1901–1910, etc.). Yet none of those decades delivered exactly 10%.
To assess the reasonableness of using 10% as an assumption for future annualized returns, a range is more relevant than a single value. If a high percentage of the decades fall near the average, then it would be reasonable to assume that average is relevant and has a reasonable likelihood of occurring. However, if near-average is rare, then such an assumption would be foolish.
To be generous to the analysis, let’s say 8% to 12% represent a near-average range. Although 8% and 12% deliver quite different long-term return results, our purpose here is just to assess credibility.
As shown below, only 21% of the decade-long periods since 1900 delivered annualized total return from the S&P 500 Index between 8% and 12%, and strikingly few were close to the 10% average. Only about one-third of the periods showed a compounded rate over 12%. Almost half of the periods showed less than 8% annual returns!
With almost 80% of the decade-long periods not near-average, using 10% as a relevant assumption for the next decade or two is a long-shot bet. For investors patient enough to evaluate twenty-year periods, the incidence of near-average values between 8% and 12% increases to 35%. Thus, the odds-on bet—at least two-thirds of the time—is to assume nowhere close to 10%.
Taking the analysis of decade-long periods a step further, let’s explore the effect of relative valuation on returns. Stock market valuation is most often measured with the price/earnings ratio (P/E).
Across the 110 decade-long periods, total return for the S&P 500 Index ranged from an annualized loss of almost -2% to an annualized gain of just under 20%. Even a ten-year period wasn’t enough to ensure a gain. Four of the decade-long periods delivered losses, and even more when inflation is taken into account.
The next chart divides the series into five quintiles, each with twenty-two of the decades. The first quintile includes the twenty-two periods with the lowest starting value for P/E. The second quintile has the next lowest set and the fifth quintile includes the decades starting with the highest P/E values.
The graph and table present the average value for P/E in each of the quintiles as well as the corresponding average annualized total return. As the market’s valuation level rises, the level of return realized from the stock market declines.
For example, the average P/E for the lowest twenty-two decades is 8.5 and the average compounded annualized return is 13.5%. The average P/E across the highest twenty-two periods is 26.9, with return averaging 4.8%. As beginning P/E rises, the subsequent return slides.
There is some variation and occasional outliers within these quintiles. For example, some periods start with high valuation and end with even higher valuation (e.g., 1995). In other instances (e.g., 1974), relatively low valuation was even lower ten years later. Bull market and bear market cycles run for various lengths. But when assessed in the aggregate, the relationship of valuation and subsequent return is strong. A higher valuation is strongly associated with diminished returns over the next 10­–20 years.
This is intriguing because the results are counterintuitive. It raises a question about whether either or both of the extremes might be predictable. Is there a way to know at the start of the ten-year period whether it’s likely to deliver above-average or below-average returns?
The quintiles provide a hint to the underlying cause, but don’t provide all of the answers. It also doesn’t address whether the relationship of valuation and return is simple correlation or is causal. These answers and insights could significantly impact an investor’s decision about the most appropriate investment approach. It would even provide buy-and-hold investors with a better expectation for their likely outcome.

No comments:

Post a Comment