How to Build a Portfolio on Shaky Ground - TheStreet

The entire edifice of asset allocation, the art and science of building a portfolio to maximize return while minimizing risk, stands on three legs.

Unfortunately for anyone trying to build a portfolio right now, each of those three legs rests on very uncertain ground.

It happens. From time to time, the financial markets go through periods of extreme uncertainty that turn deciding how much money to put in categories like government and corporate bonds, domestic and foreign stocks, and cash, real estate or precious metals into a nightmare of confusion and doubt. And it doesn't help that investors' confidence in themselves or in the ability of the long-term historical record for the behavior of stocks and bonds to predict future results is at a low point.

This isn't a minor problem: More than a decade ago, academic research by Gary Brinson, Brian Singer and Gilbert Beebower showed that as much as 90% of the variability in a portfolio's performance over time is due to asset allocation. Nothing is more important -- not stock selection, not market timing -- than asset allocation in determining the movement in the value of a portfolio.

So given the difficulties in deciding how to allocate assets right now, how should investors go about building the best possible portfolio?

My suggestion: Start by understanding the three elements that determine an asset allocation decision. Go on to get a sense of the uncertainties surrounding each of these right now. Then ballpark how important the current degree of uncertainty is for making asset allocation decisions. And finally, identify and begin to track the one or two "wild cards" powerful enough to throw an entire portfolio out of whack.

Dust Off the Historical Record

The classic asset allocation model requires an investor to make four key estimates:

How long he or she will hold the investment.

What returns he expects for each asset class.

The expected potential variation from that return -- usually expressed as standard deviation.

How the performance of each class will correlate with every other asset class.

To do that, the standard operating procedure is to go to the historical record and look up past performance on these measures. So, for example, Ibbotson Associates Stocks, Bonds, Bills and Inflation 2002 Yearbook tells me that the average annual return from large-company stocks for 1926-2001 is 12.7%, the annual total return on long-term government bonds is 5.7% and the annual return on U.S. Treasury bills is 3.9%.

The same source gives me the standard deviation, a measure of the variability of these returns, for this long period. The standard deviation for large-company stocks is plus or minus 20.2%, for long-term government bonds plus or minus 9.4% and for Treasury bills just plus or minus 3.2%.

And finally Ibbotson's yearbook gives me the correlations -- the degree to which asset classes move up or down together or independently -- for the same time period. Large-company stocks historically don't correlate very well with long-term government bonds: The correlation is just 0.16. The correlation with Treasury bills is even less, actually a negative 0.03. Long-term government bonds, on the other hand, show a 0.91 correlation with intermediate-term government bonds.

Looking at this historical record, an investor can understand why adding government bonds to large-company stocks in a portfolio produces a portfolio with less volatility than a pure stock portfolio. It also shows why adding bonds gives an investor a chance to do well, even when stocks don't.

But these are all historical numbers that describe the average behavior of financial assets over a long period of time. They can't predict how these assets will do over the next five years or even over the next 10. That same historical record shows just how much the returns, variation and correlation of asset classes can differ over even relatively long time periods.

From recent experience, investors know how much returns can change from period to period. For 1960 to 1969, for example, large-company stocks returned an annual 7.8%. From 1990 to 1999, the annual return climbed to 18%.

Bonds show the same variation in return over periods as long as a decade. For that 1960-1969 period, long-term government bonds showed a total annual return of just 1.5%. For 1990-1999, the annual total return came to 8.8%.

But investors aren't as familiar with how other measures, such as the variability of returns, another word for risk, can shift over long periods of time. For example, Ibbotson Associates has calculated that the annualized standard deviation for large-company stocks was a whopping 42% in the decade of the 1930s, as the financial markets struggled with a bear market in equities and bonds and worldwide depression. That variability in returns shrank to just 13% in the 1950s before expanding again to hit 19% in the 1980s. For the first two years of the 2000s, standard deviation has been 17%.

Bonds show the same historical variation. Standard deviation for long-term government bonds hit a high note in the 1930s (5.3% for the decade), but in the 1980s, the end of double-digit inflation and double-digit interest rates caused the standard deviation to spike to 16%.

Three Issues That Set the Stage

So what kind of numbers can investors reasonably project going forward -- say for the next 10 years?

I think that the size of returns, the variability of those returns and the correlation of those returns -- the three elements needed to successfully determine an optimum asset allocation for a portfolio, you'll remember -- for the next 10 years depend on three key issues:

Will interest rates rise and how fast?

Will inflation kick up again and how hard?

And what will be the relative strength of U.S. financial assets vs. those of the rest of the world and against specific regions of the globe?

Why do I target these three questions? For one, because interest rate, inflation and the strength/weakness of dollar-denominated assets have the power to radically change the returns that investors will receive from their ownership of all asset classes. A rapid rise in interest rates, for example, will hurt stocks and lower the returns from owning stocks and it will absolutely crush returns from bonds.

But there's a second reason to pay special attention to these three factors. Not only will they affect the absolute level of future returns from all asset classes, but they have the power to shift the variability of those returns; that is the risk in each asset class. A rising rate of inflation would introduce new uncertainty into the bond market, as bond investors scramble to figure out how fast inflation is eroding the value of their assets and struggle to find strategies to limit the damage.

Finally, interest rates, inflation and the strength/weakness of dollar-denominated assets will change the correlations between asset classes. For example, at a moderate increase in interest rates and a moderate increase in inflation, bond prices and thus bond returns will fall and stock prices rise because modest increases in inflation and interest rates are likely to be signs of improved economic growth.

So at moderate rates of increase in these two factors, bond and stock returns would be likely to continue their historic low correlation and indeed might even move more in opposite directions. But at a higher rate of increase in inflation and interest rates, stocks are likely to suffer along with bonds as companies struggle with rapid rises in the cost of raw materials and as high interest rates lower the multiples that investors are willing to pay for stocks. At high rates of increase, the correlation between the two asset classes is likely to increase.

Confronting High Rates of Change, Lower Returns

So how likely is it that investors will see rates of change in these three factors high enough to change returns, the variability of returns and the correlation of returns? Rates of change so high, in other words, that they should reshape a portfolio's asset allocations for the next 10 years?

Unfortunately, rates of change that high are very likely.

In my column,

Economic Forecast: More Pain Than You Thought, I made my case for lower long-term returns on equities over the next decade because of the need for U.S. business to restructure to meet the dual challenges of global price competition and an aging U.S. workforce. There's good reason to plan for equity returns in the range of 8% annually for this restructuring period. Plan for 8% -- and hope they're higher. I for one won't give back a single extra dollar if I turn out to be too pessimistic.

Bond returns could easily be lower as well. All the signs say we're coming to an end of a 15-year bull market in bonds powered by the fall in interest rates from double digits in the early 1980s. With the yield on 10-year Treasury notes hanging just below 4% and inflation near 2%, it's clear that there is some more room on the downside, but not much. Real yields on long bonds -- that is, after inflation -- historically have come in around 3%. So going forward, it's reasonable to think that total returns on bonds might not be much higher than the coupon yield and that rising rates might even produce capital losses as bonds sink in price.

And finally, there is the worry, backed up only by anecdotal evidence at this point, that the world's appetite for dollars may not be endless after all. The anecdotes include a shift in the Russian economy, long used to doing business in dollars instead of rubles, toward the euro, and the perhaps temporary reluctance of foreign investors to buy U.S. securities. Certainly fears that the rising U.S. budget deficits will lead to a weaker dollar and hostility in many countries to U.S. policy in Iraq aren't making overseas investors any more eager to hold dollars.

Frankly, a period of lower returns for almost all asset classes isn't anything I look forward to, but it is a challenge that asset allocation can meet without too much struggle. If rates of return fall in absolute terms, but remain identical to historical norms in relative terms, then I may have trouble reaching my financial goals with any combination of assets. Nevertheless, the optimal combination of assets for producing the highest possible return with the lowest possible risk remains relatively unchanged.

And I can even find ways to deal with the shortfall by saving more or planning to do with less in retirement or working longer or taking a part-time job after retirement, etc. And I can even come up with strategies for maximizing the available return in any asset category. If I'm worried about rising interest rates pushing down the price of bonds, for example, I can build in protection for my capital by buying individual bonds (instead of bond mutual funds) in a laddered portfolio of maturities and holding each one to maturity -- when I get my original capital back no matter how the price has gyrated in the meantime.

How to Address the Uncertainty

But the possibility that high rates of change in interest rates, inflation and the value of the dollar could change the variability of returns for all asset classes -- for some asset classes more than others -- and also that high rates of change could shift the correlations between asset classes is a much tougher problem for asset allocation. What happens, for example, if interest and inflation rates spike up to such a degree that both stocks and bonds fall in unison so that a portfolio mixing those two asset classes isn't buffered from the problems in one class, but actually multiplies the negative effects?

I wish I knew how variability of returns and correlations were going to change, even if the news was bad, because then I could plan for it. But none of my crystal balls are providing much visibility on these issues right now.

In the absence of 20/20 foresight, I can think of two ways to address the uncertainty:

First, track the forces that will determine the direction and rate of change in interest rates, inflation and the global value of our currency. In the next week or so, for example, I'll be posting a column on the federal budget and the question "Do deficits matter?"

And second, open your thinking to include asset classes that you normally don't consider, but that could offer useful building blocks for allocating assets in a portfolio. I've recently read an interesting argument for treating TIPS, the inflation-indexed bonds issued by the U.S. Treasury, as a separate asset class. I'll be writing a column on that work and its implications in the coming weeks.