Natural Consequences

John P. Hussman, Ph.D.
All rights reserved and actively enforced.

You don't tug on Superman's cape. You don't spit into the wind. Jim Croce had it right. Some things just have natural consequences. The same is true for buying stocks as long-term investments when they're trading at 20 times peak earnings. Sooner or later, bad things are bound to happen.

Increasingly, the market's level of valuation is treated as if it's somehow a matter of opinion - as if one can look at current inflation and interest rates and declare "Well, it looks 'bout right to me." But except for a very limited range that you can wiggle the assumptions, valuations aren't art, or music, or subjects of opinion. You can look straight at them and work out the consequences with basic math.

Part of the problem is that investors don't really seem to understand what a P/E ratio means. Sure, it's the ratio of price to earnings, but the relationship of that ratio to the value of stocks is not straightforward. Even "corrections" to P/E ratios using earnings growth rates (e.g. PEG ratios) or interest rates (e.g. the Fed Model) are awkward and theoretically unsound approaches.

Stocks are a claim to a stream of free cash flows

To get at the significance of the P/E, you have to start by understanding that stocks are not a claim to earnings anyway. Stocks are a claim to a future stream of free cash flows - the cash that can actually be delivered to shareholders over time after all other obligations have been satisfied, including the provision for future growth.

Knowing this already tells us a lot. For example, price/earnings ratios based on operating earnings are inherently misleading, since that "earnings" figure does not deduct interest owed to bondholders nor taxes owed to the government. The price/operating earnings ratio makes high debt companies look misleadingly cheap. Since the debt burden of U.S. corporations has increased substantially in recent years, this also means that current price/operating earnings ratios are incomparable to historical ones.

This isn't to say that P/E ratios are useless, but it's important for the "E" chosen by an investor to have a reasonably stable relationship to what matters, which is the long-term stream of free cash flows. Operating earnings fail that test because they are too easily distorted by debt levels. Raw, net earnings are also a problem, because they typically plunge during recessions. Since stocks are a claim to a long-term stream of future cash flows, not simply one year of earnings, these short-term earnings fluctuations have remarkably little impact on the discounted value of a company's cash flows. For that reason, our favored earnings measure for market valuation (though it can't be used for individual stocks) is "peak earnings" - the highest level of net earnings achieved to date.

Once we settle on an earnings measure that has a reasonably stable relationship with free cash flows, the problem then becomes interpreting what P/E ratios imply for future returns.

High valuations imply disappointing long-term returns

Tape a $1 bill in the corner of the room, above your head, and take about 10 steps back. Now you've got a security - the promise of getting $1 in the future. Hold another $1 above your head, at the same level as the one in the corner. Clearly, if you pay this $1 today for that $1 in the future, your long-term return (the slope of the line between your current money and the future money) is zero. Drop your hand a little. If you pay just 68 cents today for that $1 in the future (say a decade from now), your long-term return (the slope of the line between your current money and the future money) goes up to about 4% annually. Drop your hand a lot, say to 32 cents, and your long-term return goes up to about 12% annually. So the lower the price you pay, the steeper the slope between present money and future money, and the higher your long-term return.

That's how to interpret valuations. For any given set of future cash flows, the higher the valuation you pay, the lower the long-term return you can expect. The lower the valuation you pay, the higher the long-term return you can expect.

Over the years, I've frequently noted that the long-term, peak-to-peak rate of earnings growth for the S&P 500 has been 6% annually. This is true regardless of whether one looks back 10, 20, 50 or 100 years. Given that the average historical price/peak earnings ratio for the S&P 500 is 14 (with the bulk of that history represented by periods of lower inflation and interest rates than currently), and that the median ratio is 11, it is fairly simple to combine this information to produce long-term expectations for stock returns.

For example, if valuations remain at present levels for eternity, stock prices will grow at the same 6% rate as peak-earnings. Add in a 1.7% dividend yield, and stocks appear priced to deliver a 7.7% long-term return, provided again, that valuations remain at current levels indefinitely. I don't think that's a useful assumption, but there it is.

Alternatively, if the price/peak-earnings ratio declines from its current multiple of 20 and touches its long-term average of 14 even a decade from now, the annual capital gain for the S&P 500 would be [(1.06)x(14/20)^(1/10)-1] = 2.3% annually. Combined with a 1.7% dividend yield, the resulting total return would be about 4%, without even assuming that valuations ever move below their historical average.

Real long-term earnings growth isn't a large number

But what about low inflation and interest rates? Don't they support high valuations and high long-term returns? Absolutely not. Unless the effect is to increase the growth rate of fundamentals, whatever raises valuations also lowers long-term returns. And again, the bulk of historical data, particularly prior to 1965, reflects lower inflation and interest rates than currently.

Let's look at stock returns in real, inflation adjusted terms.

Measured from peak-to-peak since 1950, real after-inflation growth in S&P 500 earnings has averaged 1.96% annually. Meanwhile, CPI inflation averaged 3.99% annually. The resulting growth in nominal peak-to-peak earnings has been just under 6% annually for the S&P.

During the low inflation 90's, we briefly saw faster real growth in earnings of 3.14% annually, while CPI inflation averaged 2.93% annually. Again, the nominal growth in earnings was roughly 6%, but the real component was clearly stronger.

That's actually something of a rule. Periods of higher inflation tend to be associated with slightly lower real earnings growth, while periods of lower inflation tend to be associated with slightly higher real earnings growth. The negative correlation between the two for 10-year growth rates is -0.42.

Inflation tends to raise the share of GDP represented by wages, and lowers the share represented by profits. One can't push this relationship too far, however, or expect it to have much effect on long-term growth rates. Ultimately, long-term growth in real earnings is limited by long-term growth in real GDP, which has historically averaged less than 3% annually. In fact, real earnings for existing companies tend to grow at a measurably lower rate than real GDP due to the fact that new companies comprise a great deal of GDP growth. Even when we talk about extraordinary long-term productivity growth, we're talking about variations of a few tenths of 1% around this level. A range between 1-3% annually is as wide as one can reasonably assume for long-term real earnings growth, and the higher portion of that range is very improbable.

Estimating the total return that stocks are priced to deliver

Now that we have the growth piece, let's look at the income piece. Dividends have historically represented about 45% of peak earnings. This ratio plunged during the late 1990's as companies increased their debt levels and made heavy stock repurchases in order to offset the dilution from options grants to management and employees. Growth in per-share earnings reflects any remaining impact of these repurchases, so there is no need to count them twice. Despite the fact that dividends have represented far less than 45% of peak earnings in recent years, we still think that a 45% payout of peak earnings is a reasonable proxy for the stream of free cash flows that the S&P 500 is likely to deliver to investors over time.

Let's put it all together.

A market P/E ratio of say, 20, translates into an "earnings yield" of 5%. Of that, we can expect about 45% of this to be paid to shareholders as income. In addition, shareholders can expect real long-term earnings growth of between 1-3%, say 3% to be firmly optimistic. If valuations remain constant, that's also your long-term real price appreciation. Total returns are simply the sum of the two.

Assuming that current valuations are maintained indefinitely, this means that on the basis of implausibly optimistic assumptions, stocks are priced to deliver long-term real total returns of about [.45 x .05 + .03 = ] 5.25%. It's a difficult situation when even such optimistic assumptions result in unimpressive conclusions.

To translate this figure to nominal terms, add in the inflation expectation of about 2.5% currently priced into long-term Treasury Inflation Protected Securities, and stocks appear priced to deliver long-term nominal total returns of about 7.75% annually if valuations remain at current levels indefinitely and real earnings grow at an improbably high rate over the long-term. Of course, if valuations decline toward more normal levels, returns could be substantially lower than this during the transition.

Stock valuations are largely independent of inflation

Now notice the effect of inflation on P/E valuations:

Nominal total return on stocks = payout ratio * E/P + real earnings growth + inflation.

P/E = payout ratio / (real total return on stocks - real earnings growth)

Except to the very small extent that inflation variations affect real earnings growth, the effect of inflation on the P/E ratio simply drops out of the calculation altogether. Lower inflation reduces not only the rate that you use to discount future cash flows, but also the nominal growth of those cash flows themselves. These two effects cancel out.

Word to the wise. Never use the preceding equation as a valuation model, where you make assumptions about both future returns and future growth. If you do, it's possible to justify virtually any P/E ratio you want by arbitrarily choosing figures that are sufficiently close to each other.

10% long-term stock returns? Not from these levels

Given that long-term nominal earnings growth has averaged about 6% annually, we might turn the question around a different way. What P/E multiple would be required in order for stocks to be priced to deliver long-term total returns of 10% annually in nominal terms?

The answer is straightforward: We need .45 x earnings yield = .04, which is satisfied by a price-earnings ratio of 11.25. That's just about the historical median for the S&P 500.

So now we see why stocks have historically delivered 10% long-term returns from a median price/peak-earnings ratio of 11. Unfortunately, investors expecting anything similar from a price/peak-earnings ratio of 20 aren't doing the math.

Market Climate

As of last week, the Market Climate for stocks remained characterized by unusually unfavorable valuations and modestly favorable market action. So far, we have some early deteriorations in market action, but overall conditions have not shifted enough to justify a strongly defensive position. For now, the Strategic Growth Fund is fully invested in stocks that we believe reflect favorable valuations and market action, and we have just over half of that exposure hedged against the impact of market fluctuations.

The Market Climate in bonds has turned hostile, and the duration of the Strategic Total Return Fund has been reduced to less than 2 years. This means that a 1% (100 basis point) move in interest rates would be expected to affect net asset value by less than 2%. The prospect for an abrupt increase in short-term interest rates remains problematic, and with it, the prospect for an unwinding of the "carry trade" in Treasuries. The overall result is that there is a strong prospect for a flattening of the yield curve, with short-term interest rates experiencing the majority of the pressure, but enough potential for pressure on the long part of the curve to wipe out the relatively low interest income accrued by holding these bonds. With both valuations and market action unfavorable, we don't rely on forecasts of this sort, and would be in a defensive position regardless. From an analytical standpoint, however, the prospect for an abrupt flattening of the yield curve is something that we continue to watch closely, because it could have an unusually negative effect on the credit markets and financial intermediaries that depend on wide interest rate spreads.

Finally, I am pleased to note that the expense ratio of the Hussman Strategic Growth Fund has been lowered again, to an annual rate of 1.35% as of Friday. The expense ratio is affected by a number of factors, including total net assets of the Fund, and may decrease or increase over time.