Returns Don't Mean Revert, Fundamentals Do

Newfound Research |

We read a lot of blogs. They are a great source of alternative thoughts and one way to shock ourselves out of confirmation bias.

One of our favorites is The Reformed Broker, by Josh Brown. Josh is a pre-eminent financial personality and his blog has a tremendous amount of variety.

One of his posts this week, re: Expected Stock Returns – from the RWM Client Conference Call, caught my eye.

To quote the post,

But the number one reason investors believe returns will be muted is the fact that we’ve rallied 200% over the last 5 years.

This reasoning is dead wrong and is symptomatic of the Gambler’s Fallacy, as in “the roulette wheel landed on black five times, so the next one just has to be red!”

But something interesting happens when you broaden out the time horizon and focus on rolling returns as opposed to calendar-year returns, which are pedestrian and meaningless in real life. We know that stock market returns are mean-reverting in the long run, even if we don’t know how far they’ll be stretched in one direction or another over the short run.

In the post he shows a graph of realized 10-year returns, which we’ve recreated here.

The idea is that while the last 5-years have been spectacular in terms of total return, the last 10-years have been sub-par. Since long-term returns are mean reverting, our expectation should be that the next 10 years would be better.

Except this logic is wrong for a number of reasons.

First, it invokes the very same Gambler’s Fallacy that Josh argues against. Re-consider the first sentence I highlighted: “[b]ut the number one reason investors believe returns will be muted is the fact that we’ve rallied 200% over the last 5 years.” Now just swap in the results over the last 10-year period and adjust some of the language: “[b]ut the number one reason investors believe returns will be better is the fact that we’ve only rallied 121% over the last 10 years.” The logic is faulty no matter the time horizon or the order of gains and losses.

Second, the argument fails to hold up in data. Let’s examine a scatter plot of past 10-year returns versus forward 10-year returns.

If long-term returns were mean reverting, we should see a sharply negative relationship. While a slightly negative relationship exists, the variance around it is so large that it is impossible to rely on, especially since investors only have a few decades in their investment horizon.

Finally, it has no economic merit. Let’s ask ourselves this: why would returns be mean-reverting over the long run? Theoretically, over the long run the market should be tied to a fundamental measure like earnings growth and the discount rate. So if the market returns 600% over the last decade, the question we should ask is “was the growth justified?” If it was, then why would we expect the long-term returns to mean-revert?

So how can we reconcile the slightly negative relationship we saw in the prior 10-year vs. forward 10-year scatter plot? Surely that means there is some mean reversion?

The answer lies in the final point. Returns are not mean-reverting, but valuations are. Below we plot Shiller PE’s versus forward change in Shiller PE. Notice the much, much tighter negative relationship.



And this relationship makes economic sense: if stocks are underpriced relative to fundamentals, their valuations should increase and if stocks are overpriced, their valuations should decrease.

Now this is all complicated by the fact that fundamental metrics usually have two levers: the numerator and the denominator. In this case, the numerator is price and the denominator is earnings. So a high PE could revert back to a low PE because (a) price falls, (b) earnings grow, or (c) some combination of the two. But when stocks are expensive, it doesn’t matter whether prices fall or prices sit still while earnings grow: forward returns will still be muted.

The reversion of fundamentals, therefore, has historically been a great guide for future returns – especially since prices tend to be the volatile piece of the equation.

Below we plot Shiller PE versus forward 10-year returns.

So let’s go back to our slightly mean-reverting long-term returns. Why does that happen? The answer is that in the extremes, high valuations have coincided with strong prior returns, and low valuations have coincided with weak prior returns.

The takeaway here is that just because the prior 10-years were nothing noteworthy from a return perspective does not imply that the market necessarily has room to run. The slight mean-reversion we saw in historical returns was a factor of coincidental data: the extremes in returns lined up with fundamental extremes.

Quite simply, prior returns are not predictive of future returns. The fundamental data is the driver.

What we should be asking ourselves is whether we believe the market is currently cheap, expensive, or somewhere in between. Based on the Shiller PE, fundamentals are in the 89th percentile: a fairly expensive reading.

Based on this data, we wouldn’t bet on U.S. equities being the engine of growth for the next 10 years.

DISCLOSURE: The views and opinions expressed in this article are those of the authors, and do not represent the views of equities.com. Readers should not consider statements made by the author as formal recommendations and should consult their financial advisor before making any investment decisions. To read our full disclosure, please go to: http://www.equities.com/disclaimer

Comments

Emerging Growth

Sack Lunch Productions Inc

Sack Lunch Productions Inc is an entertainment company. The Company, through its subsidiaries, is engaged in event management, production service, film production and distribution services.

Private Markets

MyForce, Inc.

As parents, we constantly worry about the safety of our loved ones. The media bombards us with incidents from across the nation school shootings, frequent assaults on campuses, and crimes…

Initial State

Initial State is an Internet of Things (IoT) data analytics & data management platform company. We turn sensor and event data into information that matters by making it easy to…