Laying A Foundation for Asset Allocation

(Originally written December 2013, some of the conclusions here are timely.)

Over the past several months, I have been performing due diligence on private wealth managers, weighing the benefits and costs of professionally managed strategies. Along the way, I met with some very talented people, advocates of both active and passive management philosophies. Researching these approaches in greater detail piqued my intellectual curiosity, and I ultimately decided that I wanted to maintain full control over our investment process.

This was not an easy decision. Away from the asset selection question (which I plan to cover in a future essay), came the larger question of asset allocation. Professional money managers love to warn that individuals lack the dispassionate discipline needed to stick to an asset allocation plan in the face of dynamic markets that often challenge one’s senses of risk and greed. My own past experience certainly indicated that this was no hollow warning.

To that end, I set out to better understand the primary competing theories governing asset price behavior, with the intention of finding principals that resonated with me and could therefore be more naturally (and stringently!) followed. Coincidentally, during this time, the champions of two of these competing theories, Eugene Fama and Robert Shiller (along with Lars Hansen), shared the 2013 Nobel Prize for Economics. This essay details some of what I’ve learned, and how I plan to incorporate these ideas into the asset allocation decision.

Stock ownership represents a claim on a company’s future stream of income. So the price of that stock should simply be the sum of those expected payments divided by a discount rate. That discount rate represents our required rate of return, which in turn consists of a risk free rate plus the demanded equity risk premium.

There are two basic schools of thought regarding why equity prices move. The Efficient Market Hypothesis assumes that prices are informationally efficient; that is, prices instantaneously reflect all known public data and are therefore “correct”. A behavioralist approach argues that prices are far from rational, and are significantly influenced by the herd mentality.

Shiller, who authored Irrational Exuberance, conducted a fascinating study that illustrates the wide gap between these theories. With a robust data series at his disposal, Shiller observed the actual stream of dividend payments from a basket of stocks, and their long-term return (the discount rate). It follows that for a given point in time, one could derive the theoretical price of that basket.

What Shiller observed was that equity prices in theory should be represented by a fairly smooth line, with a ~ +1.5% (in real terms) upward slope, and that “fair value” (as derived above) fell within ±1% of this trend-line two-thirds of the time. However, in the real world, the S&P 500 varied a whopping ±19% from its trend line two-thirds of the time!

How to explain this massive amount of price volatility (19x expected!) relative to the underlying stability in the dividend income stream? To a subscriber of the Efficient Market Hypothesis, these price swings come from changes in the risk-premium component of the required return. Under that scenario, prices can increase relative to earnings if the discount rate (risk premium) demanded by investors decreases. This argument comes up often in today’s Quantitative-Easing-driven low interest rate investment climate, where “there is no alternative” to equities. Its premise is also found in the Fed Model of investing, where equity yields are compared to interest rates. More on this later.

To behavioralists, the volatility of actual price movements relative to what would have been predicted by their underlying earnings proves the existence of irrational (or herd-driven) price movements. To me, this feels far more intuitive. Take October 1987 as an example: can one truly believe that risk premium demands changed instantaneously enough to cause a 30%+ erosion in prices over a few day span?

Examples of this herd-like behavior are readily observed and documented. One of the most famous (and comical) was Holland’s Tulip Mania of 1637. And to the degree that these booms and busts repeat themselves, the market becomes preconditioned by past experience for expectations of bubbles. We tend to jump to conclusions that have very poor grounding in probability. This can be referred to as “extrapolation”, or what behavioralists call a “representativeness heuristic” in which predictions are made using the closest match to past patterns, regardless of the probability of that pattern asserting itself.

In the financial markets, these periods tend to be accompanied by new theories which try to provide underpinnings to the price action that make them different – a.k.a. sustainable – this time around. To be dramatic, keep in mind the dynamics of a Ponzi scheme: initial investors must be convinced of some superficially plausible, if somewhat unverifiable, thesis. It is the subsequent investors that allow the scheme to proliferate when they abdicate the responsibility to perform any due diligence, rather, gauging the integrity of the thesis by the sizable returns accruing to other investors.

All of this is not necessarily to imply that prices are completely random and devoid of some ability to anticipate expected future returns. Shiller pioneered the use of the Cyclically Adjusted Price to Earnings Ratio (CAPE) to provide a useful valuation metric to the market. In comparing prices to rolling 10yr earnings series, Shiller aimed to better reflect prices as a function of the more consistent earnings power of a stock or index. (The expected future real yield of that instrument can then be expressed as 1/CAPE.)

Indeed, an October 2012 Vanguard research piece looked at many popular proposed expected future return indicators and found that CAPE carried an r-squared of 0.43 to 10yr-ahead real returns. Sadly, that correlation was by far the highest of any metric tested. (https://personal.vanguard.com/pdf/s338.pdf)

So where does all of this theory leave the investor who attempts to approach asset allocation in a prudent manner? I’ve discussed the distinct likeliness that investment performance will be strongly affected by herd-driven volatility. I then pointed out one metric that attempts to smooth some of these bumps to provide a useful barometer of expected future returns. Can we employ these concepts to arrive at a more opportunistic asset allocation process than a static equity/fixed income ratio? WARNING: money managers caution you will probably end up worse off for trying – i.e. they view this as “market timing”. But when one observes that CAPE’s long-term mean is 16.5, that CAPE stood at a record-high 42.87 when the S&P 500 (in real terms) peaked in Aug 2000, and then registered a cyclically low 13.32 in March 2009, one can’t help but wonder whether there is some value in trying to design a more dynamic approach to balancing your portfolio.

But here we encounter a problem: while the extremes noted above may have provided historically strong signals to alter your asset allocation, in general the CAPE data series has been cyclically quite sticky – that is, we observe 20+ year periods where it remained both below and above its historical mean. For instance, CAPE has currently remained above its long-term mean since June 1989, with the exception of just fifteen months. And before that, it remained below its mean from May 1973 – May 1987. (Find the data set here: http://www.multpl.com/shiller-pe/)

A popular argument, then, is to say that CAPE would have kept you out of the market for much of the past three decades. But that argument is simply off point. The purpose of CAPE is to provide some insight into future expected returns, not to signal an on/off switch as to whether one should have exposure to equities. One then needs to judge for themselves the adequacy of those returns. But to the extent that CAPE deviates significantly from some average, perhaps it can be used to fine-tune our asset allocation decision.

To that end, I aimed to look at the mean CAPE, distinctly, during these separate long “sticky” periods (which I call “streaks”). For instance, since the current streak began in June 1989, mean CAPE stands at 25.09. I believe this gives us some nearer-term reference point to gauge current valuations.

Well, timely enough, just in November 2013 CAPE crossed over its “streak mean” for the first time since December 2007. Again, this is not a signal to eliminate one’s equity holdings, but perhaps should be used as a warning that closer attention is now warranted. Specifically, there is now less room for error should earnings begin to roll back over from the torrid pace of growth they exhibited off the 2009 lows. Certainly much has been reported regarding the sustainability potential of current profit margins. Remember from above, rationales for “why things are different this time around” need to be met skeptically! Monitor the data.

The current real earnings yield of the S&P 500 using CAPE stands at 3.95% (1/25.36). When weighed against interest rates, this looks pretty attractive, right? Perhaps. This is the type of comparison that underscores the so-called Fed Model for equity valuation. But this is a poor comparison, since the duration of equities is considerably longer than that of bonds (price equals the PV of a long series of future cash flows). Moreover, according to Shiller’s research, CAPE would be expected to mean revert towards its long-term average of 16.5 over long cycles, implying losses for equities and eroding the future yield. Indeed, according to a 2012 research piece by Clifford Asness of AQR Capital Management, the average real 10yr return for CAPE values greater than 25.1 (since 1926) is just 0.5%.

Still, this sort of relative yield comparison – TINA, there is no alternative! – combined with the impressive gains to date (and the corresponding jealousy on the part of those who have missed the move) continues to funnel cash into the equity market (and recently, at a historic pace). These flows are not at all surprising, though I had actually assumed we were going to see them far earlier this year.

Meanwhile, for the true naysayers out there, this type of price action and capital inflow is the perpetuation phase of a scam.

So, for the prudent investor – one that is neither blindly chasing the herd nor intractably shunning the market – what are some current actionable themes? Some will argue that so long as expected real yields are positive, one should be fully invested. I just think that as projected yields decrease, especially when mean-reversion in the multiples is considered, additional caution is warranted. This may include some cash exposure, i.e. a 70%/30% asset allocation could shift to 60%/30% with 10% cash, etc. At a minimum, even if you stick with a static allocation, diligent portfolio rebalancing, on a regular basis, is absolutely mandated.

No doubt, we find ourselves in an incredibly challenging investment environment. Rarely has the expected real yield of a combined equity and fixed income portfolio been lower. This can make the asset allocation process seem even more complex than ever, perhaps even bordering on futile. But maintaining a disciplined approach to portfolio management should still serve you well over the long term. Stick with it.