Weighting by market capitalization is a common but by no means the only way to allocate among developed-country markets. Depending on investor wants and needs, it may not be the best way either. State Street Global Advisors explains the major options considered by its clients.
For the past decade, institutional investors have struggled with the question of how to structure international equity exposure. In the mid and late 1980s, investment committees were inundated with arguments proclaiming the virtues of international investing-higher returns and reduced equity risk due to increased diversification. For most of these elegant studies, the MSCI EAFE Index was the basis for analysis. The advice was simple: forget valuations, ignore the Japan weight, and put 5-20% of your portfolio into international. Along came 1990 and 1992 and we were reminded about the true meaning of diversification as the collapse of the Japanese equity market coincided with a strong period for US stocks. The Asian crisis of 1998 reminded us once again.
Partially in response to the Japan collapse and 1993-94 turnaround, we are often asked about an appropriate benchmark for the international equity markets. The answer, if our clients are an indication, is as varied as the markets themselves. Aclear understanding of the alternatives is particularly important since the choice can influence asset allocation policy, manager measurement, funding requirements, and most importantly, investment returns. We do not present any specific recommendations on the optimal benchmark here, but instead point out several important considerations in selecting a benchmark appropriate to each investor's needs.
There are many reasons why our client base has rapidly moved toward non-cap-weighted benchmarks in their developed market equity portfolios. There has long been a lack of consensus as to the extent that modern portfolio theory applies to a global equity portfolio, and it is clear that home investor biases and restrictions on cross-border capital flows exist. Most clients are motivated by purely practical reasons rather than disbelief in the theoretical side of global cap-weighting. In short, they are simply uncomfortable with a particularly large concentration in any single country. Historically, most benchmark alternatives focused on limiting Japan's long outsized weight in the overall portfolio. Today, with Japan's cap weight roughly equal to that of the UK plus Germany, alternative country weighting strategies are predominantly focused on creating broadly diversified portfolios that serve as logical, neutral benchmarks. Less common reasons for avoiding cap-weighting include the desire to neutralize the effect of cross-ownership on country capitalization, to capture anticipated reversion to the mean through fixed weighting or periodic rebalancing, or to lower variance within the manager universe. But in spite of the attractive historical mean/variance characteristics of several other benchmarks and the intuitive appeal of mean reverting strategies, it is the concentration issue that overwhelms all others in this matter.
In the exodus from the cap-weighted EAFE benchmark, most plan sponsors at one point or another consider eight different country weighting strategies. Two of these alternatives are rarely considered seriously: trade-weighted and minimum variance portfolios. The remaining six are more common alternatives:
• "Japan lite"
• static regional allocations
• equal weighting
• liquidity tiering
• composite weighting
It has been our experience that most plan sponsors in the 1990s chose either GDP-weighted, Japan-Lite, or static regional benchmarks as the solution to concerns over cap-weighting. As mentioned above, the Japan weight in the cap-weighted portfolio is no longer the main focus. Rather, it is avoiding a large single concentration in any market going forward.
|CORRELATION WITH S&P 500 1970-1988|
|50% Europe/50% Pac Basin||0.49|
This simplest approach holds all countries in a weighting equivalent to the respective size of their equity markets. Several features appeal to proponents of cap-weighted passive investing. First, it is the lowest cost benchmark approach in that a cap-weighted portfolio tends to be the most liquid portfolio and is essentially self-rebalancing with price changes. In the developed equity markets, capitalization is a fairly good proxy for liquidity, so investment is concentrated in the largest, most liquid markets. Japan and the UK, for example, tend to be significantly more liquid as well as much larger markets than Ireland, New Zealand, and Finland.
Second, country weights will change with the relative performance of each market without a need to trade among markets to maintain a balance. In other words, it is the lowest cost (lowest turnover) portfolio to maintain on an ongoing basis. Third,for those ardent Modern Portfolio Theory enthusiasts, the cap-weighted global portfolio has theoretical appeal. But while cap-weighted portfolios have historically had the lowest correlation with US stocks, they have also had lower returns and higher risk than several other alternatives we will examine. As commitment to international equity becomes greater within an institutional portfolio, the inferior historical risk/return characteristics can outweigh the advantages of low correlation (diversification) with US markets.
In years past, many investors have expressed concern over the weight of Japan in a cap-weighted index. To control the weight and enhance country diversification, they limited Japan to a set percentage. If the Japan weight (or the large weighting of any country in the future, for that matter) is the main reason for moving away from cap-weighting, then clearly this is the most direct way to eliminate the concentration issue. This reduction in country weight is typically done by either fixing an initial weight for Japan and then letting it vary with market movement or periodically rebalancing to the target weight as the Japanese market experiences relative performance variations. The latter alternative increases turnover within the portfolio and also requires that rebalancing rules be determined with the investment manager. On the other hand, the fixed-weight methodology with periodic rebalancing has experienced higher returns than cap-weighting, even after transaction costs. Furthermore, historical volatility is lower than with a cap-weighted approach with only a modest increase in correlation with the US market. In spite of these attractive attributes of the fixed-weight approach, it has been far more common among our client base to set Japan's weight to a specific percentage of its current capitalization, and then simply let it float with market movement from that point on. The main advantage of this approach is lower turnover, since no rebalancing of country weights is necessary.
Many investors have adjusted exposure based on an estimate of cross-ownership of stocks within Japan. This eliminates much of the subjectivity in picking the "optimal" reduction factor. Based on an average of several industry estimates of this effect, limiting Japan's weight to 75% of its capitalization has been a fairly common target. Going forward, as clients express concern over concentrations in other markets, the fixed weight or cap adjustment could be applied more broadly.
STATIC REGIONAL ALLOCATION
In this approach, an effort is made to control regional exposures within a specified range. Generally, clients have chosen to assign fixed percentage weights to Europe and the Pacific Basin areas, but Latin America and other emerging markets could be included in a broader mandate. Many of the same advantages of the Japan-Lite approach-higher historical returns and lower historical volatility-have accrued to the portfolio with this approach. This methodology has a greater rebalancing cost and increased correlation with US equity market returns. Aregional allocation approach also requires the creation of rebalancing rules, which adds another layer of subjectivity to the benchmark construction process. This strategy generally appeals to those investors who want to control specific country concentrations and who see the increased turnover as an opportunity to add value from a mean-reversion point of view.
GDP weighting has been used as an alternative weighting approach for almost a decade and has proven to be one of the most common non-cap-weighted benchmarks in the developed markets. While no specific theory of asset pricing supports the use of a GDP-weighted benchmark, the existence of a published third-party benchmark (the MSCI GDP world benchmark) has enhanced the method's appeal in the eyes of many investors. Using this technique, countries are simply weighted by the size of their gross domestic product. In a GDP-weighted portfolio, assets are allocated away from over-equitized markets such as Japan and the UK and into markets such as Germany, France, and Italy, where the equity markets are relatively small in relation to the size of the respective economies.
GDP weights among countries remain fairly stable over time. In fact, since the GDP numbers themselves are normally converted to U.S. dollars, it is the changing foreign exchange rates that cause most of the changes in the GDPweights among countries. This means that as the GDP weights are reset each year in July (benchmark weights float with changing capitalization throughout the year), the portfolio is rebalanced to the new target weights by selling the best performing markets and purchasing the worst markets. Historically, GDP weighting has exhibited favorable risk/return characteristics relative to a cap-weighted alternative. On the other hand, it also exhibits a higher correlation with the US market, and due to the peculiarities of the published benchmark and lags in the release of economic data it is always somewhat outdated. In the past, GDP-weighted benchmarks have appealed to several types of investors: those seeking to limit Japan's weight in the portfolio, those who were comfortable with mean-reverting strategies, and those who preferred to have a benchmark that was published by a third party vendor. GDP-weighted strategies declined through much of the 1990s, probably due to the steadily declining weight of Japan in the cap-weighted index.
The equal-weighted benchmark is the simplest way to avoid any country concentration issues in an international equity portfolio. With an EAFE universe of 20 developed market countries, an equal-weighted strategy invests 5% in each country. Due to the large allocation to the smallest, less liquid countries, this can be one of the most costly passive portfolios to acquire. Also, depending on the frequency of rebalancing, it can be quite costly to maintain due to high turnover. Similar to the fixed regional allocation strategy, the need to choose a rebalancing process adds a layer of subjectivity. In spite of the fact that equal-weighted strategies exhibit higher correlation with the US market, the strategies have historically exhibited attractive risk/return attributes. Due to the contrarian nature of most potential rebalancing rules, it is possible that mean reverting tendencies will overcome any costs of maintaining the index. Furthermore, it has been shown that similar to the well documented small-cap effect (typically higher returns), a small country effect has existed in the developed markets. If this is indeed the case, any strategy that is more highly concentrated in the smaller equity markets of the world can expect higher risk-adjusted returns than a cap-weighted portfolio over the long term. Investors who find the equal-weighted index appealing have concentration concerns on the top of their list, and find the mean reverting nature of the index attractive.
Aliquidity-tiering approach is a variation of a pure equal-weighting approach in that countries are divided into groups of similar liquidity and then equally weighted within each group. The objective is to create a diversified country portfolio that directly addresses both the concentration issues and the varying degrees of liquidity that exist among the world's equity markets. Typically, rebalancing rules are employed that allow country weights to fluctuate within preset bands. There is considerable discretion over how many tiers should be constructed, which markets belong in which tier, how to rebalance each tier, and how to handle the addition of new countries to the index. The subjective nature of a liquidity-tiered approach may not appeal to investors who seek a clearly defined benchmark that may be tracked month after month. Notwithstanding, the liquidity-tiered approach appears to display an attractive risk/return trade-off, which makes it a viable benchmark. It also may be a better active manager benchmark since it is consistent with the active manager's tendency to equally weight attractive instruments within a portfolio.
COMPOSITE MANAGER WEIGHTING
An interesting benchmark for institutional equity portfolios is the composite manager benchmark. It allocates country weights based on the average weights used by all managers in a specific universe. As a result, a portfolio is created that reflects the average investor's country preferences, or "bets." By definition, a portfolio constructed to match this benchmark will tend to exhibit stable performance within the manager universe. This may be an attractive feature for those concerned with the high volatility of cap-weighted (or close to cap-weighted) strategies within their universe of active managers. Indirectly, using a composite manager-weighted benchmark also reduces country concentration because the typical active manager will have less than a cap-weight in Japan. However, because the average manager weights remain fairly stable over time, it is somewhat analogous to fixed-weighted portfolios: Rebalancing a passive portfolio to track this benchmark results in high turnover and related transaction costs. Nevertheless, similar to other mean revert ing strategies, the composite manager benchmark has attractive historical risk/return characteristics. The composite manager benchmark may be ideal for investors who want to avoid excessive volatility within the manager universe and who find con-trarian rebalancing strategies appealing.
MINIMUM VARIANCE WEIGHTING
The goal of the minimum variance portfolio is to provide the maximum diversification benefit to an investor's overall plan. Utilizing historical risk and return characteristics for the various asset classes, and the resulting covariance matrix, an optimization process is employed to select the appropriate portfolio. In the case of creating a minimum variance international equity portfolio, all other asset classes are constrained to current weights. Then, assuming expected returns of all eligible countries are set to be equal, the optimal diversifying portfolio is selected. In the absence of concentration or liquidity constraints, this process will tend to overweight countries with the most independent returns, which have tended to be the smallest and least liquid markets. The historical evidence of a small country effect (as mentioned previously) may explain why historical minimum variance returns appear so attractive. In addition, there is the hindsight bias inherent in any historically derived minimum variance solution. This approach will obviously tend to produce a portfolio with risk/return attributes that are difficult-if not impossible-to replicate on a forward-looking basis unless the future closely duplicates the past. While the minimum variance portfolio is theoretically appealing, it too often results in high turnover and elevated transaction costs. It is, however, one of the few benchmarks considered that takes into account the risk to an investor's overall plan. In spite of the attractive features of this strategy, it is unusual for investors to rely blindly on an historically based quantitative process to allocate assets across countries.
The trade-weighted benchmark is another country weighting technique that is often discussed but rarely adopted. In a trade-weighted portfolio, country weights are chosen according to the ratio of imports into the country of domicile, from each of the countries eligible for inclusion in the benchmark. The goal of creating a trade-weighted portfolio is to hedge the future consumption needs of the beneficiaries of the assets. In theory, as an increasing percentage of beneficiaries' consumption takes place in foreign goods and services, exposure to foreign equities and currencies in roughly the same proportion will serve as a reasonable hedge against future purchasing needs. Based on current import patterns, trade weighting in a developed-market portfolio will tend to increase exposure to the Pacific Basin. While the trade-weighted approach is one of the only benchmarks frequently discussed that directly addresses the liabilities of beneficiaries, it is probably the least common international benchmark utilized.
It should be clear that there are several thoughtful ways to structure an international equity benchmark. However, because the objectives and constraints of institutional investors vary widely, there is no optimal choice for all investors. While historical risk/return data provide an interesting perspective on how various benchmarks compare, that should not be the deciding factor in choosing which strategy is "right" for a particular investor. More importantly, one should consider the different characteristics of the alternatives in light of the original reasons for departing from cap weighting. We have found that in most cases, there will be a few constraints (or strong opinions on the part of the investor) that will limit the choice to only a few benchmarks. Nevertheless, it is the process of considering all the alternatives that is most valuable to the investor.