View Full Version : Omitted Variables by John P. Hussman, Ph.D.

08-22-2005, 11:33 PM
Omitted Variables August 22, 2005 John P. Hussman, Ph.D.[/i]

One of the nearly indelible assumptions about the current stock market is that valuations “deserve” to be high because interest rates and inflation are reasonably low. It accords with common sense that since stocks compete with bonds, lower interest rates should generally be accompanied by higher price/earnings multiples (and accordingly, lower future returns on stocks).

There's certainly some validity to this argument. We can observe, for instance, that as interest rates have declined since 1980, earnings yields (the inverse of P/E ratios) have also declined. This is conveniently summarized in the “Fed Model”, which assumes that the prospective earnings yield on the S&P 500 (on the basis of expected operating earnings) should be equal to the 10-year Treasury yield.

Unfortunately, the Fed Model dramatically overstates the relationship that exists between interest rates and justified stock valuations. Stock valuations began in the early 1980's at what were, objectively, very undervalued levels. Indeed, the price/peak earnings ratio for the S&P 500 Index fell below 7 in August 1982. Over the following two decades, stock valuations moved up to extremely high valuations, peaking at over 30 times earnings during the recent market bubble.

In other words, as interest rates declined, so did earnings yields. But a major portion of that decline in earnings yields represented a long movement from extreme undervaluation to extreme overvaluation, not a “fair value” relationship with interest rates.

What's really going on here is something called “omitted variables bias.” The Fed Model explains the decline in stock yields since 1980 with one variable – the 10-year Treasury bond yield. But in fact, there are two variables at work. One is the declining level of interest rates, sure. But the other is the move from deep undervaluation to extreme overvaluation – in other words, a profound decline in the “risk premium” that stocks have been priced to deliver, over and above what bonds could be expected to return. In 1982, the risk premium on stocks was huge. At present, it's close to zero, and possibly negative.

So what the Fed Model picks up as “the effect of interest rates on stock valuations” is actually the combined effect of interest rates and risk premiums. As it turns out, interest rates and risk premiums on stocks have moved in a huge cycle together since about 1965. Back then, interest rates were reasonably low, and stock valuations were high. As interest rates rose into the early 1980's stock valuations plunged. Then as interest rates declined back toward normal levels, stock valuations advanced. So to a casual observer, it would seem very obvious that stock valuations and interest rates go strongly hand-in-hand.

But here is the key: the relationship we've observed since 1965 between stock valuations and interest rates has not been a “fair value” relationship. To the contrary, stock valuations during this period have substantially “overshot” fair value in both directions, becoming substantially overvalued when interest rates were low, and becoming substantially undervalued when interest rates were high. Interest rates may have been correlated with stock valuations, but they haven't “justified” them.

..................various ranges for the 10-year Treasury bond yield since 1965, along with the actual total return that the S&P 500 achieved over the following 10-year period.

10-year Treasury Bond YieldSubsequent 10-yr total returnS&P 500

0-5% 4.34%
6-7% 10.54%
7-8% 13.18%
Notice that when interest rates have been low during this period, subsequent stock returns have also been low, and when interest rates have been high, subsequent stock returns have also been high. Though this accords with our basic intuition, the range of stock returns is far, far too wide. It doesn't make sense that stocks should have returned 5.46% on average, when interest rates were between 5-6%, but 13.18% on average, when interest rates were just two percent higher. There's nothing that implies “fair valuation” in these figures. Equally important, there's nothing “standard” in the relationship between interest rates and stock valuations that we've seen since 1965. In historical data before that point, interest rates and subsequent stock returns were negatively correlated.

If we don't entertain the possibility that stocks have been both undervalued and overvalued at various points over the past few decades, we end up attributing far, far too much importance to interest rate movements as drivers of "fair value." Worse, we risk overlooking the current, historically rich overvaluation of stocks because we imagine that these valuations are “justified” by low interest rates. Very simply, they are not. Stocks remain unusually overvalued, and are likely to deliver disappointing long-term returns to buy-and-hold investors purchasing at these prices.

Market Climate

As of last week, the Market Climate for stocks remained characterized by unusually unfavorable valuations and relatively neutral market action. We continue to observe evidence of distribution in large capitalization stocks. Market breadth has deteriorated modestly as well (as measured by advances versus declines), but as I've noted in recent comments, standard measures such as the NYSE advance decline line mask the extent of the internal turbulence in market action.

Still, there's very modest evidence that investors are still willing to accept risk, which is enough to warrant a small exposure to market fluctuations. In the Strategic Growth Fund, about 10-15% of our stock holdings are unhedged, with the remainder of the portfolio hedged against the impact of market fluctuations. That's not much exposure, of course, but it aligns proportionally with the average return/risk that the market has historically delivered in this market climate. Very slightly positive, but not significant. Suffice it to say that there's not enough evidence to warrant a fully hedged position, but to the extent that there are favorable aspects to market action, they are fairly weak.

In bonds, the Market Climate was characterized last week by modestly unfavorable valuations and modestly unfavorable market action. Bonds remain in something of a limbo. On one hand, we can observe short-term economic and inflation pressures, which are sufficient to prompt a continued normalization of short-term interest rates by the Fed. However, the economy still has substantial structural risks such as excessive leverage and debt, a large current account deficit, continued sluggishness in manufacturing, and employment growth that is positive but not impressive. The main factor that would tip the bond market toward greater attractiveness would be evidence of credit risk. A substantial widening in credit spreads between risky corporate bond yields and default-free Treasury yields would be a signal of increased recession risk. For now, we don't observe that, so the flatness of the yield curve, combined with prospects for further increases in short-term rates, are sufficient to hold the Strategic Total Return Fund to a relatively low duration of about 2 years.