From the November 2006 issue of Wealth Manager Web • Subscribe!

November 1, 2006

Prediction Friction

What's past is prologue, Shakespeare wrote, although the Bard never had to test his theory with an investment forecast. If he had, he would have discovered the hard way that history can mislead when it comes to searching the past for clues about the future of risk and return. Looking at, say, a simple average equity performance for the past 20 years is hopelessly naive for projecting what awaits in the next 20. Yet for all its flaws, the past is still a building block for modeling future performance of asset classes for the simple reason that history carries some information about the future. Nonetheless, an enlightened strategist recognizes yesteryear's limits along with its insights, and adjusts his outlook accordingly.

Among the veterans who are paid to do just that is EnnisKnupp & Associates (EnnisKnupp.com), a Chicago institutional-investment consultancy that oversees $700-billion-plus in assets for the likes of pension funds and foundations. The 25-year-old firm is a respected force in the business of advising big money, and so it has learned much about making forecasts that are prudent yet practical for clients.

With the U.S. economy weakening and a variety of global trends buffeting confidence in the prediction game, we wondered how the big picture looked to a veteran consultant with lots of capital at stake when it comes to designing investment strategies. Our queries were answered by Armand Yambao, who heads up financial modeling at EnnisKnupp. In a recent interview with Wealth Manager, Yambao advised that the outlook for returns is relatively modest compared with the past, which means that the margin for error is shrinking when it comes to asset allocation. For details, read on.

What is EnnisKnupp's strategic outlook for equities?

As a long-term investment, U.S. equities will provide enough compensation for the risk underlying it. In one respect, the outlook for U.S. equities is based on faith in the U.S. economy and corporations in general. Basically, there are no red flags that the U.S. economy's in big trouble. You can't guarantee short-term performance, or that the economy won't soften in the near term, of course. But as we see it, there are no fundamental changes for long-term investors.

What's your time horizon?

Usually, we're looking at a 15-year time frame. That's arbitrary, but looking at less than 10 years is too short in terms of having patience with different asset classes, and looking beyond 20 years is too uncertain. We're surveying current economic measures when forecasting our assumptions, and somewhere between a 10- and 20-year time horizon is reasonable.

EnnisKnupp's latest semi-annual outlook (see tables on page 70), published midway in 2006, has a forecast for compounded expected return for U.S. equities at 7.5 percent a year. How does that compare with history?

If you just look at historical equity returns, you'll come up with something like a 10 percent to 11 percent annual return, and so our 7.5 percent forecast for U.S. equities is less than what backward-looking expectations suggest.

How do you model expected equity returns?

We use dividend income, nominal growth in corporate earnings, and changes in equity valuation levels. The dividend yield for stocks was recently 1.8 percent. The historic dividend yield is much higher, which partly explains why our 7.5 percent equity performance outlook is lower than history's. We also consider inflation, and our expectation for inflation is only 2.5 percent. Depending on how you slice the past, inflation might be 3 percent over time. But we expect a somewhat modest 2.5 percent inflation, and that's another reason for projecting a 7.5 percent equity return instead of 10 percent or 11 percent.

What convinces you that inflation will be that low?

Recent economic indicators. The market's implied outlook for inflation for the next 10 to 30 years was recently around 2.5 percent based on inflation-indexed Treasuries. Another data point comes from the Blue Chip Economic Indicators newsletter, which also had a forecast of 2.5 percent inflation.

What's your outlook for bonds?

We're expecting a 5.6 percent compounded annual return for bonds over time. We came up with that by looking at the current yield for the broad bond market index. Midway in 2006, the current yield was about 5.8 percent [based on Lehman Brothers Aggregate Bond Index]. That's consistent with our expectations. We develop our fixed-income return expectations based on the current yield, and then consider the outlook for yields going forward.

Your expected return of 5.6 percent for bonds is a bit lower than the current yield of 5.8 percent when you ran the numbers. Why the difference?

One reason is volatility. If bond prices remained unchanged, the current yield of 5.8 percent would become the compounded return over time. But volatility, or uncertainty, will affect bond prices, and our expectations account for that by reducing the performance outlook down a bit to 5.6 percent. Overall, our outlook for bond volatility is average compared with history. We think that volatility for bonds will be a standard deviation of 6.6 percent, meaning that in two out of three years, bond returns could be plus or minus 6.6 percent of the average return number.

What about foreign stocks?

For non-U.S. stocks, our long-term expectation is a compounded annual return of about 7.2 percent, or a bit lower than the 7.5 percent for U.S. equities. The reason for the slightly lower performance forecast is higher volatility. If you backtrack, the risk premium for non-U.S. equities is slightly higher than for U.S. equities. But non-U.S. equities are more volatile, and so they carry greater uncertainty. We compensate for that by projecting a 7.2 percent return. Nonetheless, we encourage clients to diversify internationally. One reason: There's a diversification benefit with foreign stocks, even though the returns are expected to be less than for U.S. equities.

How does the future for real estate look?

For a generic real estate investment, the outlook is in the range of 6.5 percent on a long-term basis. Our long-term return expectation for real estate is somewhere between U.S. bonds and U.S. equities, and that's how we come up with a 6.5 percent compounded return that's between those two asset classes. Basically, we look at the historical experience--how much market risk premium was rewarded for real estate relative to U.S. bonds and U.S. stocks. We find that real estate's outlook is somewhere between the two.

If you look at real estate returns, it's a combination of income--to the extent there's rent--and appreciation. The rent is a bond-like characteristic, while the appreciation is more equity-like in terms of uncertainty and fluctuation in relationship with the economy. That's how we rationalize an outlook for real estate returns that's between those of stocks and bonds.

Based on your projections, U.S. bonds in the years ahead will post a much lower correlation with U.S. stocks than will real estate or foreign equities. If so, that means that bonds will continue to deliver substantial diversification benefits. Why do you think that low correlation between U.S. stocks and bonds will persist?

The main component of fixed-income return is yield. If yields go higher than expected, that will be an attractive investment for many investors. To the extent that investors buy more bonds, they tend to sell equities. So, in a tradeoff, equity returns might drop because of the selling. The inverse will be true, too. If the current yield or current return expectation for bonds is low, equities will be more attractive, and people will buy more stocks. In general, market equilibrium tends to keep correlations between stocks and bonds lower than with other asset classes.

On the flip side, what do you think of the rising correlation between foreign and domestic stocks in recent years?

Some predict that the higher correlation will persist because of, say, an increasingly globalized economy. But history shows that rising correlations don't necessarily persist. It's a subject we've discussed internally. Our conclusion is that for a long-term investor, there's still not enough compelling evidence for expecting a lower diversification benefit between foreign and U.S. equities. The markets are still different, and there's a currency effect that might dampen the correlation. We're skeptical that the higher correlation of late will persist for someone with a long-term outlook because higher correlations in the past eventually dipped.

Some strategists say that history is a more reliable guide for predicting standard deviation and correlation than it is for forecasting returns.

Correct. That's the reason why we rely more on history for developing our assumptions there. Nonetheless, U.S. equity standard deviations could be very different from our forecasts, which we estimate will be close to 17 percent [about two-and-a-half times as high as bonds]. Of course, if you focus on shorter time frames for, say, standard deviation, you have more volatility. But for a long horizon, it's more stable.

How far back do you look into history for your estimates generally?

The late 1970s. That's how far back confidence extends for bond-market data, for instance. And in real estate, you have limited history.

Do you use the straight histories for coming up with expectations for standard deviation and correlation?

We take the straight history from 1978. The only exception is the standard deviation for U.S. bonds. If you just take the standard deviation for U.S. bonds from 1978, it'll be higher--roughly 7.5 percent versus the 6.6 percent we expect. The problem with blindly using the history from 1978 is that past volatility, especially during the late 1970s and early 1980s, might be overstated. The U.S. bond market is tamer now, and so the volatility of the 1970s and early 1980s is less likely to continue. That said, the 6.6 percent standard deviation for bonds that we come up with isn't totally arbitrary. We ran a Monte Carlo simulation of interest rates over the next 15 years.

Why is the past so risky for forecasting returns?

Historical returns aren't a very good predictor of future returns. If you did a back test and relied only on history, you'd do a poor job. Let's say that equity performance skyrockets in 2007. What happens in 2008 is that the straight history from the previous year makes your return assumption higher than it was before. But rationally, market expectations might say that if you just had a good year, you might want to temper expectations because equities could be overvalued at that point. Stocks, as a result, might be more prone to a correction. It's sort of counterintuitive, but if you increase return assumptions, you're probably doing so at the wrong time in the market cycle, given the cyclical nature of returns.

Wouldn't a recent jump in volatility lead you to temper expectations for standard deviation as well?

Not really. If you look at different slices in history, standard deviation is more stable. The cyclicality of the market returns isn't really present in standard deviation.

What about correlations?

It's a similar kind of thing. Correlations basically mean that returns tend to move in the same direction. It's not based on whether recent history was a down or up market.

You expect a risk premium in U.S. equities of 190 basis points. How does that compare to history?

History gave you a higher risk premium. Our forecast is lower.

What's the bottom line?

Investors shouldn't blindly rely on equities to satisfy investment goals. We're not saying avoid equities; it's more about understanding the risks.

James Picerno (jpicerno@highlinemedia.com) is senior writer at Wealth Manager.

Reprints Discuss this story
This is where the comments go.