[Editor's note: The text is from Canadian Investment Review, Winter 1998 (10th Anniversary Edition).
The correlation matrices in Table 1 were added Aug00. Figures 1 and 2 are unavailable....BS]

Also George$, a participant on The Wealthy Boomer forum, adds this caveat:

Date: 10-Aug-2000 - 8:15 PM
Subject: Re: The U.S. is the only market worth investing in
From: George$

Oh Bylo. I simply cannot resist. You seldom give us any cause to question your posts but I must do so here.

Oh, I forgot to add that the average return and volatilities for the period from 1978-89 were TSE 300 16.64%/18.13, S&P 500 17.24%/15.60 and MSCI EAFE 21.94%/15.85 respectively

This is so misleading! Typing 16.64 would imply all four figures have some significance. I would bet a 1000 to 1 they do not. I think that the number is more likely to be only significant to two [or to only one?] digits.

In first year physics labs we sometimes catch hapless students giving 12 "significant" figures in their lab manual from some measurement that involved dividing [1 by 3 say] because that is what their calculator showed.

One must never ever put any meaning into extra "significant" figures that have no basis in reality.

End of lecture. Sorry about that.

For many years, foreign investing meant buying U.S. stocks. Harry Marmer noted in 1991 that this was a non-optimal strategy, and that foreign investing should be viewed as global investing. Has this changed?

by Harry S. Marmer

1991
International Investing: A New Canadian Perspective examined the benefits of diversification outside a domestic market. It reviewed two issues:

• From a Canadian pension fund asset allocation perspective, is there a role for international equities in the portfolio?

• How does the asset allocation puzzle change over different economic environments?

UPDATING THE ANALYSIS
We see that "the more things change, the more they stay the same." Table 1 compares the average annual Canadian dollar total returns, volatilities and correlations for six asset classes over the earlier period of analysis, 1978­1989 inclusive, with the data updated to the period ending 1997 inclusive.¹ Since the last analysis, U.S. stocks replaced international stocks as the best-performing asset class. No surprise here, given the extraordinary bull market the U.S. has experienced over the past six years and the financial market implosion in Japan over the same period of time.

Time-Varying Results

Eyeballing the volatility and correlation results, one might hastily conclude that there has been no significant shift in these statistics over time. However, as the wise statistician said, "statistics don't lie, but know your statistician." Examining these statistics over monthly rolling five-year windows (Figures 1,2) is a more realistic assessment of the strategic planning horizon of investors as opposed to the 20-year view. We see in fact that there have been significant shifts in both volatility and correlation data. For example, over the 1990s, asset class volatility dropped relative to the experience of the 1980s.

The Impact of the October 1987 Crash

The October 1987 crash left an indelible impression not only with investors but also on markets. Figure 1 reveals that the shock of the crash caused a kinked smile across most correlations except between Canadian and international stocks and U.S. equities and international equities. The relative stability of this relationship supports the view global market integration has yet not eliminated the benefits of international diversification.

Surprisingly, over the 1990's the correlation between Canadian and U.S. equity markets actually declined relative to the long-term average. One might have expected the opposite, given the increase in free trade between Canada & the U.S. Unfortunately, this view confuses economics with financial market performance. As Canadian investors, we are all too aware of the decoupling in performance that has occurred between Canadian and U.S equities, and the declining correlation represents this decoupling.

The crash left an undeniable "frown" on the monthly rolling five-year standard deviations (Figure 2), reflecting the spike involatility caused by the sudden market drop. In addition, during this decade, volatility has trended down relative to the historical average. If regression to themean is an observed phenomenon, we should expect future volatility to spike up -- which has happened in 1998.

In conclusion, these results support the overwhelming academic research that has concluded that both return, risk and correlation are time-varying-dependent.²

Efficient Frontier Analysis

No analysis is complete without trotting out the efficient frontier to help put into perspective the historical return, risk and correlation statistics.

Efficient frontiers were constructed for Canadian assets only, Canadian with 20% foreign, and unconstrained. In the unconstrained analysis, the optimal asset allocations selected from the following asset classes: cash, Canadian bonds, U.S. equities and international equities. Canadian equities failed to enter the efficient frontier.

From a Canadian investor's perspective, what is particularly disturbing about these results is that even after generously allowing for 20% non-Canadian investments over the past 20 years, the unconstrained efficient frontier provided the greatest opportunity for selecting optimal asset mixes. It is clear that the financial consequences of misguided government policy have had significant opportunity costs for Canadians saving for retirement.

The inputs for efficient frontier analysis are time-period-sensitive, reflecting particular economic regimes or environments. Since the previous analysis, we have been living through a disinflationary environment which has seen yearly inflation downtrend significantly below the 20-year average. Using the statistics from 1990-1997, a disinflationary efficient frontier was generated with the optimal mixes containing allocations from cash, Canadian and global bonds and U.S. stocks. Once again, the dynamics of the efficient frontier inputs were reflected with the exclusion of international stocks and the significant role of global bonds in the optimal mixes. ³

The Real Messages

In summary, are we smarter than we were ten years ago concerning the asset allocation decision? Absolutely, for with a high confidence level we can today conclude the following: