It's late in the evening. The children have gone to bed, and you are all alone with your IRA accounts and spreadsheet, wrestling with your asset allocation strategy. You nod off and reawake in an alternative universe. Strange tropical plants are growing out of your keyboard, and snow wafts down from the ceiling.

A genie appears, looking like some vague demented amalgam of Mario Vargas Llosa and Harry Markowitz. "The spirits inform me that you are a seeker of efficient portfolios," he intones. "Oy vey," you respond, "Excedrin headache number 11." The genie offers a sly smile. He tells you that in his pocket are the annual returns for for the next 27 years for the MSCI regional indexes (Europe, Japan, Pacific X Japan), US large and small stocks, and an index of gold mutual funds.

Can you see them? Of course not. The genie will, however, tell you what mix of these assets will produce the highest return over the next 27 years. Still interested? You bet.

The genie suddenly disappears, and along with him the snow and tropical plants. Rats! Just another asset allocation hallucination. But perhaps you have learned something. Your computer contains the returns for the past 27 years for these indexes. You quickly calculate the optimal mix of these six assets for the 1970-96 period. It consists of just 3 of the assets: 35% Japanese, 28% small US, and 37% gold. Very unconventional, but it yielded 16.85% over the 27 year period, versus 12.27% for the S&P500. (Annual rebalancing is assumed.) The annual returns for this optimal mix and the S&P are plotted below:






What is striking is just how different the annual returns of these two portfolios are. In order to better appreciate this, I've plotted the "tracking error" (optimal portfolio minus S&P return) of the portfolio:

Above the line, the "optimal" portfolio's annual return was higher, below it the S&P was better. Despite the fact that the optimal portfolio yielded 4.58% compounded more than the S&P, it lagged it in 12 out of 27 years, or 44% of the time. The underperformance was not trivial either, on one occasion 27%. In other words, international diversification is not a free lunch. You must be willing to feel like a fool for extended periods of time in order to reap its benefits. If you're a small investor, that is not such a high price to pay. For the professional money manager the cost is much higher; lagging the benchmark by such margins turns you into instant roadkill. The small investor definitely has an unfair advantage over the professional in the asset allocation arena.


The Coward's Portfolio

The "optimal" portfolio has a most peculiar allocation. Further, it is dubious that the same allocation will be anywhere near optimal during the next 27 years. Financial history doesn't repeat (or even rhyme, as Mark Twain suggested). Is it possible to predict the "optimal" portfolio allocation for the next 5, 10, or 30 years with any accuracy? Your chances of doing so are about the same as starting at point guard for the Bulls next season. Can an "investment professional" using state of the art software and a Cray 23 do any better? Hardly likely.

If you've perused these pages before, you know the answer. Pick an allocation, almost any allocation, and stick to it. How about equal amounts of all 6 of the above assets (Europe, Japan, Pacific X Japan, US large and small stocks, and gold equity)? The 27 year return of this "coward's strategy" was 15.33% -- just 1.5% less than the best possible strategy. Again, even though this strategy beat the S&P by 3% annualized, it lagged the S&P in 13 out of 27 years -- almost half of the time. Below are the annual returns for the equally weighted coward's portfolio and S&P, followed by the graph of its S&P tracking error.

Which brings us to the past decade. For those of you who've been living in a trappist monastery since 1987, the S&P 500 has been the only place to be. For the past 10 years the S&P has returned 15.26% versus 6.59% for the "optimal" portfolio and 10.44% for the "cowards" (equally weighted) portfolio. For internationally diversified investors it's been a very long decade indeed.

So which period do we rely on for guidance -- the past decade of US large cap outperformance or the longer period of global portfolio dominance? There is no sure answer to this dilemma. However, it is likely that longer periods provide more reliable data. Asset return divergences generally revert; the underperformance of foreign assets during the past decade is probably simply "payback" for their overperformance before 1987. If this is true then the next 10 years should again provide handsome rewards for foreign diversification.

Hence, the loneliness of the long distance asset allocator. Holding foreign assets during the past decade has been a lonely, frustrating experience. Few things offend human nature more than watching helplessly as your neighbors become effortlessly rich investing in the big blue chips. Keep the faith -- your day will come.

copyright (c) 1997, William J. Bernstein