In Focus: Country Tilts

As you may remember, when we rebalanced our recommended portfolios at the end of last year, we used a fairly broad definition of an asset class. Specifically, because the benefit from diversification comes from risk reduction, we required that the "asset classes" we used could have no more than a .60 correlation of returns with each other. That definition eliminated from use a number of groupings of stocks and bonds that other commentators call "asset classes." Examples of these include small cap stocks or large cap growth stocks, and short-term bonds. In our view, all of these represent various "tilts" that one can make in order to enhance the risk/return trade-off within an asset class. At the time of our rebalancing, we promised that we would be taking a closer look at these "tilts" to see which, if any of them, made sense. Last month we looked at tilts based on economic sectors. This month we look at country fund investing. Next month we’ll combine our insights about sector and country tilts, and reach some conclusions about their implications for asset allocation. In September, we'll look at investing in different bond maturities, and in October we’ll look at momentum investing. As was the case last month, the fundamental question we’re trying to answer is whether or not you can improve on the risk/return trade-off for the asset class as a whole by making a country tilt in your portfolio.

The data set for our analysis covered the period from January, 1988 to December, 2000. During this period, the EAFE generated average annual returns (in U.S. dollars) of 8.77%, with a standard deviation of 18.32%, or .48% of return per unit of risk taken on. Within the EAFE, the Europe Index had average annual returns of 15.01%, with a standard deviation of 17.06%, or .88% of return per unit of risk, while the Pacific Index had returns of 3.26% with a standard deviation of 23.79, or a measly .14% of return per unit of risk. Finally, the combined EAFE+EMF Index (our proxy for the Vanguard Total International Market Fund) generated returns of 8.56% with a standard deviation of 18.19%, or .47% of return per unit of risk. Our challenge was to see if it was possible to do better than these results, while taking on no more risk, and investing only in iShares.

First we did an obvious analysis, using relative GDP to weight the available Europe and Pacific iShares (for example, we calculated the percentage of Spain's GDP relative to the total GDP of all the iShares included in the Europe Index, and used that as our allocation to Spain). In the case of Europe, this had a very beneficial impact: average annual returns increased to 17.88%, while standard deviation grew to only 18.67%. As a result, return per unit of risk grew to .96%. The main difference from market cap weighting was that we had relatively more exposure to Germany and Italy, and less to Switzerland and the U.K.

In the Pacific region, this approach didn't improve things much. Average annual returns were only 3.54%, with a standard deviation of 19.21%, or .18% of return per unit of risk. The fundamental problem here is that in both GDP and market cap terms, Japan dominates this index.

We then got a bit more adventuresome, and decided to take advantage of the fact that Canada is not included in the EAFE, and therefore typically gets left out of many portfolios. We decided to include it in our GDP weighted Europe portfolio to see what would happen. At the same time, we substituted the EMU iShare for the iShares of countries included in the EMU, not only to hold down transaction costs, but also to gain exposure to Ireland, Greece, Portugal and Finland, which lack iShares of their own. The results were encouraging. Average annual returns were 15.69% with a standard deviation of 16.69%, or .94% of return per unit of risk.

We then took the same approach to the EAFE, using GDP weights, and including Canada and the EMU iShare. In this case, average annual returns were 11.96%, with a standard deviation of 15.45%, or .77% of return per unit of risk. This was a considerable improvement over the market cap EAFE (without Canada) index.

We then applied the same approach to the full EAFE+EMF, using the emerging market iShares, the EMU iShare, the Canada iShare, and the other EAFE iShares all at their respective relative GDP weights. The results were impressive: average annual returns were 16.31%, with a standard deviation of 17.65%, or .92% of return per unit of risk (almost double the result for the market cap weighted EAFE+EMF index).

Given the consistent improvement in results we were able to achieve, we need to ask the same question we've asked in previous chapters in this series. Are these results just an historical accident, or is there something more fundamental, and possibly sustainable, going on here?

First, the good news. Diversification across countries can improve risk adjusted performance because the correlations of returns between them are often low. Moreover, the cause of these low correlations goes beyond simple exchange rate relationships, as evidenced by the low correlation in many countries between exchange rate changes and equity market performance. In short, many of the differences between countries that may give rise to low equity market correlations -- such as differences in culture, language, ways of doing business, accounting rules, and the like -- are not going to go away any time soon. Equally as important, research has demonstrated that investors tend to have a "home country bias" -- that is, regardless of the potential for improving risk adjusted returns through international diversification, they still prefer to invest the bulk of their portfolios in the country in which they live. Taken together, these factors form a basis for concluding that the potential benefits from country tilts will continue to exist in the future.

Now for the bad news. The period covered by the data was one during which a lot of fundamental changes took place, including the formation of the Euro bloc, a substantial increase in the integration of the Mexican and United States economies, the introduction of significant structural reforms in Brazil, and a sharp increase in the global reach of the Nordic economies. This raises the possibility that some of the processes that gave rise to the improvement in results from simple GDP weighing may no longer be operating (e.g., economic policies are more similar across countries, and exchange rates are relatively more stable), or may be operating more weakly than in the past. Finally, because the relative importance of different industry sectors differs across countries, some of the difference in country performance may really have been driven by underlying differences in the structure of economic activity. For example, at the end of 2000, the energy sector accounted for about 26% of the Netherland's market capitalization (based on MSCI data), and 22% of Argentina's. In contrast, it accounted for only .6% of Japan's market cap, and 11% of Canada's and the U.K.'s. More complicated statistical analysis would undoubtedly show that industry effects were interacting with true country to produce the differences in country index returns that we observe (we'll look at the implications of this analysis next month).

Still, on balance, it would seem that you could do worse than take a simple GDP weighing approach when it comes to dividing your assets if you want to make country tilts. However, caution is in order: carried to its logical conclusion, this approach would generate relatively high exposures to countries like China where respect for foreign investors' and minority shareholders' rights is still spotty at best. Anybody adopting GDP weighting is therefore well advised to set as a pre-condition that the countries receiving a GDP based allocation have well functioning equity markets.

GDP, of course, is but one approach to weighting countries in ways that differ from their market cap weights in an index. Another approach is to use an optimization model to help you determine the country weights. When we did this, we came up with some interesting results.

Our first analysis focused on the MSCI Europe Index, which had achieved average annual returns of 15.01% (in U.S. dollars) between 1/88 and 12/00, with a standard deviation of 17.06%, or .88% of return per unit of risk. As possible investments, we used the country iShares contained in the Europe Index, as well as the Canada iShare. We set two constraints: (1) no more than 15% of the portfolio could be invested in any single country, and (2) the expected standard deviation of our optimized portfolio had to equal 17.06%. The portfolio that resulted had an expected annual return of 19.47%, or 1.14% per unit of risk taken on -- quite an improvement over the index, assuming, of course, that history is a good guide to the future (which sometimes it isn't!). The weightings behind this result were as follows: Belgium, France, the Netherlands, Sweden, Switzerland, and the U.K., all 15% each; and 10% to Canada.

However, when we looked at this result, we were concerned that some of it may have been driven by historical exchange rate factors that have moderated or disappeared since Belgium, France and the Netherlands signed up to use the Euro. To test this, we ran another optimization, this time using the MSCI EMU Index in place of separate indexes for each EMU member country. However, we again included Canada, and the non-EMU European countries for which iShares are available -- Sweden, Switzerland, and the U.K. As before, we set some constraints on our optimization: No more than 20% weighting to any country, except the EMU group, which could be weighted up to 50%, and expected standard deviation equal to 17.06%. This time our optimized portfolio had an expected annual return of 18.53%, or 1.09% per unit of risk taken on -- still an impressive result.

We next attempted to apply our methodology to the Pacific Index, and came up against one of its major shortcomings. In short, our model said, in effect, 50% Hong Kong, 40% Australia, 10% Singapore, and nothing in Japan. Of course, if history repeats itself in the future, this is undoubtedly the right way to maximize return per unit of risk. But what are the chances that the Japanese equity market will perform as poorly over the next ten years as it has over the past ten years? This is a perfect example of the dangers of blindly following the dictates of an optimization model, without first applying the common sense test to its recommendations. In this case, history is probably at best an imperfect guide to the future, not only because of the prospects for change in Japan, but also because of the tendency for investors and markets to overreact in both directions. In this case, the best route to take would probably be to set an additional constraint on the optimization that specified the minimum percentage of the portfolio that had to be allocated to Japan. As long as that percentage is less than its roughly 80% weighting in the market cap based Pacific Index, the resulting expected returns will be higher than what has been achieved historically (again, assuming the past is a good guide to the future).

This is the approach we took in our next analysis, which was focused on the EAFE+EMF combined index, which had an average annual return of 8.56% between January, 1988 and December, 2000, with a standard deviation of 18.19%. In this case, we set a minimum investment in Japan at 12% of the portfolio, which is equal to the country's share of the total GDP of the countries included in the index. Again, we used the EMU instead of the individual country indexes, and set a constraint that it could account for no more than 50% of the portfolio. Again, we included Canada, Sweden, Switzerland, the U.K., Australia, Hong Kong, Japan, Singapore, Brazil, South Korea, Malaysia, Mexico, and Taiwan as potential investments, with a constraint that no more than 15% of the portfolio could be invested in any one of these countries. Finally, we set the required standard deviation to 18.19%. Once again, our results were impressive: we achieved an expected average return of 24.55%, or 1.35% of return per unit of risk, based on these portfolio weights: Australia, Canada, Switzerland and the U.K., 15% each; 13% to the EMU, 12% to Japan; 10% to Mexico, 3% to Brazil, and 2% to Sweden. Of even more interest was the fact that the underlying industry weightings of our optimized portfolio were very similar to those for the EAFE+EMF Index as a whole.

The similarity in industry sector weightings between the two portfolios suggests that the expected performance we achieved through optimization was largely due to country effects, rather than an underlying industry tilt.

The next question we asked was whether these apparent country effects were mostly due to the diversification of exchange rate risks. To test this, we looked at the standard deviation of returns (in U.S. dollars) over the January, 1988 to December, 2000 period for both exchange rate changes and for the equity market as a whole in Australia, Canada, the EMU Group, Japan, Mexico, Sweden, Switzerland, and the U.K. In every case, the standard deviation of returns for the equity market as a whole was significantly larger than for the exchange rate alone. The multiple ranged from a low of 1.67 for the EMU and 1.66 for the U.K. to a high of 4.15 for Canada. This seems to indicate that the benefits of diversifying across countries go beyond simple reduction of exchange rate risks.

Overall, when they are combined with the improved performance we were previously able to achieve through sector tilts alone (as detailed in last month's Index Investor), our country tilt results suggest that still further improvements in expected risk adjusted portfolio returns might be achieved through a combination of sector and country tilts, rather than the simple use of either of these in isolation. This is the topic we will explore next month.

In the meantime, it bears repeating that the factors which gave rise to our country tilt results may not operate in the future as they have in the past. To cite but one example, governments and economic policies can and do change over time, and not always in ways you like. Still, having said that, the results remain intriguing, and well worth the consideration of an investor searching for ways to boost risk adjusted returns on the international portion

| Uses and Abuses of Country Funds and ETFs | In Focus: Country Tilts | Performance Update |