What Can Stocks Return?

?Achieving investment goals depends on good estimates of the potential range of long-term results. The abstractions of modern finance provide elegant insights into portfolio construction and the operation of markets, but they provide no easy answers about expected returns.

Although the Capital Asset Pricing Model (CAPM) does provide one framework for calculating expected returns, it also depends on estimating the unobservable equity-risk premium (the additional return investors demand above Treasury notes as compensation for risk). No doubt the risk premium has risen given recent events.

What constitutes a reasonable forward-looking equity-risk premium? Agreement among experts can be difficult to find. Papers with widely differing estimates are published in academic financial journals. One notable paper published in 2002 calculated a negative forward-looking risk premium for the U.S. stock market, while others estimated 3% to 5%.

Individual investors, presumably ignoring the academics, routinely show overly optimistic expectations. Participants in the Federal Reserve Bank’s annual Michigan Surveys during 2000-04 expected long-term stock returns of 10% to 12%. Other surveys suggest the expectations of individuals are highly correlated to recent market performance, with highest expectations during the high-tech bull market of the late 1990s.

But now, with the worst bear market in a generation under way, investors are focusing on short-term losses and losing sight of the long-term opportunities presented by the best valuations for the U.S. market since the 1980s.

The accompanying chart uses data compiled by Yale professor and author Robert Shiller, with the blue line being the current price of the S&P 500 divided by 10-year trailing-average inflation-adjusted earnings. This represents a method for smoothing the peaks and troughs of the economic cycle and creating a conservative valuation of the stock market. The red line represents the ratio of the S&P 500 price to the following year’s earnings, with analysts’ consensus operating-earnings estimates for the remainder of 2008 and 2009.

Remember that the contraction in earnings during a recession will lead to higher price-to-earnings (P/E) multiples, even though prices have declined. Consensus estimates for 2009 are likely overly optimistic, but they are useful as an estimate of what stocks are capable of earning—once the economy recovers.

Low P/Es
What is shocking is that these declines came off normal valuation levels, not the bubble valuations seen in the late 1990s. The Dow and the S&P 500 traded at levels below the 2002 bottom. The market’s levels fully reflect a serious recession through 2009 and perhaps longer.

The price of the S&P 500 peaked around 1,550 in 2007. At press time, Wall Street analysts expected companies in the index to earn an aggregated $100 in 2009. At today’s price levels of around 1,000 this translates into a P/E ratio of 10. What does this mean? If $100 is a reasonable estimate of what the index can earn once the global economy is growing normally again, and it trades at 16 times earnings, which has been a long-term average, then the S&P should recover to 1,600 (where it was in 2007).

A recovery in the S&P is unlikely next year, but it will happen at some point in the future. What if it takes five years for the S&P 500 to get back to 1,600? Assuming an annual dividend return of 2% (the current yield is closer to 3%), the five-year annualized return comes to 14.5%. By comparison, the risk-free option of U.S. Treasury bonds for the next five years offers a 2.5% return. If the recovery to 1,600 takes 10 years, the annualized S&P 500 return to that point comes to approximately 7%.

A look at history
The most common method to generate expected returns is to simply project past history into the future, with the rationale that data spanning several decades and business cycles will provide a robust estimate of what to expect.

Ibbotson Associates provides the most widely used data set with U.S. equity-market returns dating back to 1926. Shiller has compiled the historical earnings growth for the U.S. stock market. Putting these data together reveals the relationship between total stock performance, earnings growth, dividends, changes in valuation and inflation. From January 1926 through December 2007, the S&P 500 returned 10.36%, annualized.

The stock returns have three factors: dividends, real (inflation-adjusted) earnings growth and changes in P/E ratios. It is important to understand the relationship between these factors and their potential impact on forward-looking return estimates. The real rate of return, including dividends, was 7.10%.

The dividend component of the return was 4.16%, which is striking relative to the current S&P 500 dividend yield of 2.14%.

Does this mean one should expect total equity returns to be a full 2% lower during the next 80 years? Not necessarily. Dividends are paid out of earnings and ultimately shareholders are theoretically indifferent to how those earnings get paid—through dividends, share buybacks (a relatively recent development) or reinvestment and future earnings growth.

Payout ratios are lower now than they were in the past where 45% or more of earnings were paid as dividends. The forward-looking assumption therefore is simply that shareholders will get the economic benefit of corporate earnings in some fashion.

Over the long run, stocks can be expected to pass through moderate levels of inflation, making forecasts of changes in price levels unnecessary. It matters little whether inflation during the next 30 years averages 2% or 4%, as businesses will simply pass the inflation through to consumers. A volatile monetary environment with periods of very high inflation or severe deflation would translate into market disruptions and poor earnings performance, but this macroeconomic risk can be assumed to be part of the overall risk of investing in equities, and for which investors are compensated.

In January 1926, the stocks in the Ibbotson dataset traded at 10.2 times earnings while the S&P 500 at December 31, 2007, traded at 22.0 times trailing, fully diluted earnings. This 116% increase in the market multiple translated into an additional 0.96% of annual return in the period. Stripping this effect from historical data reduces real annualized return to 6.14%.

What to expect
How reasonable is a 6.14% real return as an estimate of future stock performance? Assume that, after a recovery, stocks trade at a P/E ratio of 20, using the more conservative Shiller methodology of 10-year average inflation-adjusted earnings. This translates (inverting the P/E ratio to E/P) to a 5% earnings yield.

So, if an investor could purchase 100% ownership in every company in the S&P 500 today at current prices, he or she could theoretically get a 5% annual dividend. This dividend should be expected to pass through inflation and also reflect real earnings growth.

Very few companies pay out 100% of their earnings in dividends, but many effectively do through share buybacks. Earnings that are not returned in buybacks or dividends are invested in new projects creating additional growth.

Disconnections do occur. For example, companies may reinvest earnings in poor investments rather than return capital to shareholders. But private-equity funds and corporate buyers help ensure that equity values realistically reflect earnings and valuable assets do not languish for long periods under poor management.

Earnings growth can be broken down between organic growth that comes from assets currently in place and reflects improved efficiencies, and growth in consumption and “extra” earnings growth, which accrue from share repurchases and reinvesting profits into successful new business ventures.

Putting the historical earnings growth data together with the current normalized earnings yield, the initial projected return is 6.93% real (1.93% plus 5% normalized earnings yield).

Real earnings growth of 1.93% may be overly optimistic, as it reflects the period where the United States went from being a primarily agricultural economy to a post-industrial one. Real earnings growth cannot exceed real GDP growth over long periods of time, but it may be surprising to see that the 1.93% real earnings growth is a fraction of the 3.35% real GDP growth over the period.

Stock indices are weighted toward large, mature companies while marginal GDP growth comes from smaller, faster-growing public companies and privately held businesses.

There are periods of more than 2% real earnings growth in recent history, but there have been periods of negative real growth. Placing this into a sensitivity analysis provides insight into the range of possible long-term outcomes.

Internally, Paul Comstock Partners has used 1.25% real earnings growth as the “organic” number, which, when combined with a 5% earnings yield, ties closely to the 6.14% real 1926-2007 Ibbotson returns, when the effects of P/E expansion are removed.

By incorporating the current valuation of the equity market into the expected return, this method provides a reasonable basis for making tactical allocation decisions. At its peak in 2000, the S&P 500 traded at a P/E ratio of 33, which translated to a 3.3% earnings yield. This was well below the then-4% yield on TIPS. An investor using this process would have had to think about the prudence of a large equity allocation at those valuation levels.

Asset allocation
A thoughtful investor in 1999 could have examined the equity markets further and discovered many attractively priced stocks outside the technology sector. The numbers today support a long-term rate of return comparable to historical averages, making tactical asset-allocation decisions more difficult.

Tactical asset-allocation decisions should be made on the basis of an over- or underweight to a strategic target. For example, an investor might have a strategic target of 25% in large-cap U.S. stocks with a range of plus or minus 5%, based on economic or valuation factors.

Full-blown market timing should be avoided.

The equity markets anticipate the future. Valuations reflect currently available economic data: the stock market already knows about the credit bubble, the election, the trade deficit, etc. A prospective market timer would have to possess a superior method of forecasting the impact of hundreds or even thousands of economic variables.

Volatility increases during periods of instability, leading to large and unpredictable short-term price movements. Large declines and rallies occur over the space of days or weeks rather than months. For example, in March 2003, there was a period of five trading days that saw a 7.5% increase in value and turned out to be the beginning of the recovery that ensued over the rest of the year. Throughout 2002 there were similar short rallies that were followed by declines that erased the gains. A market timer would have to know the March 2003 rally was “real” while the 2002 rallies were not. Similarly, 8% daily moves were common in October 2008.

There have been periods of more than 2% real earnings growth in recent history, but also periods of negative real growth. the most likely outcomes are highlighted.

Market timing requires a very high probability of success as the bets are large and infrequent. An investor who can select stocks that outperform the market 55% of the time will be successful, as he or she can diversify and make dozens of bets per year, whereas a market timer who is right 80% of the time risks ruin, as one of five bets will be wrong and it may take years to recover from the impact.

History is full of successful investors, but scant few market timers. A few pundits got famous in 1987 for making calls to sell out prior to the crash. (They did not predict the crash; they just thought stocks were overpriced.) Their subsequent history of market predictions was lackluster and they quickly faded from the spotlight in the 1990s.

Stocks trade currently at the lowest levels in more than 20 years, with the potential impact of a deep recession fully discounted into the share prices. An unprecedented level of fear now grips the markets, which paradoxically has eliminated most of the risk of owning stocks in a five- to 10-year horizon.

A recovery to normal levels of earnings and valuations over this timeframe will translate into rates of return that well surpass that of bonds or cash. We believe the global economy remains a growing concern, and that is about all that has to occur for stocks to outperform bonds and cash over the next five years.

The current normalized earnings combined with 1% to 2% real earnings growth provides a reasonable estimate of long-term real equity returns. This translates into 6% to 7% real rates of return off recovered stock values.

Equity markets can be expected to pass through moderate levels of inflation, but returns will suffer under periods of high inflation or deflation. While this method offers a framework for tactical asset-allocation decisions, market timing is exceedingly difficult, if not impossible, to implement successfully. Attempts at tactical asset allocation therefore should be done in moderation relative to strategic targets.

Stephen C. Browne, CFA, is chief investment officer of Paul Comstock Partners, a wealth-advisory firm in Houston.