The MSCI EAFE Index is arguably the most popular international index on the block. Many active fund managers’ returns are benchmarked to this index (and the majority of them fail to outperform it). Couch potato investors also tend to allocate a portion of their portfolios to ETFs that track the MSCI EAFE or a similar index.
If we pop the hood on the MSCI EAFE Index, we can see that it currently allocates 63.8% to European companies, 35.4% to Pacific companies and 0.8% to Middle Eastern companies. These weights are free to fluctuate over time, based on the performance of the underlying currencies and stocks.
Underlying country weights of the MSCI EAFE Index
Source: MSCI Index Fact Sheets as of June 30, 2016
A reader recently asked me whether they should attempt to outperform the MSCI EAFE Index by allocating 50 percent to a Europe ETF and 50 percent to a Pacific ETF. Although this may sound like active management, many notable index investors, like Rick Ferri and Ben Carlson, have advocated this approach in the past. Most of their analysis indicates that a 50-50 split between European and Pacific stocks has historically outperformed the MSCI EAFE Index.
Whenever I read arguments in favour of tinkering with a plain-vanilla index approach, I like to run the numbers myself, both from a Canadian investor perspective and an overall portfolio perspective. To do this, I’ve created two balanced portfolios for comparison; one has a 20% allocation to the MSCI EAFE Index, while the other allocates 10% to the MSCI Europe Index and 10% to the MSCI Pacific Index. The portfolios are then rebalanced annually at the end of each year.
MSCI EAFE and 50-50 Portfolios
| Index | Portfolio 1 (100% MSCI EAFE Index) | Portfolio 2 (50% MSCI Europe Index + 50% MSCI Pacific Index) |
| FTSE TMX Canada Universe Bond Index | 40% | 40% |
| S&P/TSX Composite Index | 20% | 20% |
| S&P 500 Index | 20% | 20% |
| MSCI EAFE Index | 20% | |
| MSCI Europe Index | 10% | |
| MSCI Pacific Index | 10% | |
| Total | 100% | 100% |
*Portfolios are rebalanced annually at year-end.
Splitting hairs
The results will likely surprise many readers. The returns were not only close, but identical for the 10 and 20 year periods. The 50-50 split portfolio did manage to outperform the 100 percent EAFE portfolio since inception with a lower standard deviation, but the differences were immaterial. Also, the analysis did not factor in the additional capital gains taxes that would have likely been realized by taxable investors while rebalancing the Europe and Pacific component back to its 50-50 target. Based on the results below, I see no compelling evidence that suggests splitting the EAFE allocation has resulted in anything more than additional taxes and portfolio complexity.
Performance Results as of June 30, 2016
| Annualized Return | Portfolio 1 (100% MSCI EAFE Index) | Portfolio 2 (50% MSCI Europe Index + 50% MSCI Pacific Index) |
| 1 Year | 2.77% | 2.86% |
| 3 Years | 10.27% | 10.25% |
| 5 Years | 8.75% | 8.80% |
| 10 Years | 6.33% | 6.33% |
| 20 Years | 7.00% | 7.00% |
| Since Inception (01/1980 to 06/2016) | 10.12% | 10.16% |
| Std Dev Since Inception | 8.48% | 8.44% |
Is there a rebalancing bonus, when you rebalance between stock subasset classes? If two or more stock subasset classes have the same return and volatility, but a correlation of less than 1, then rebalancing should reduced volatility and increase compound return. Why didn’t you find that result?
AQR published a paper on rebalancing:
https://www.aqr.com/Insights/Research/White-Papers/Portfolio-Rebalancing-Part-1-Strategic-Asset-Allocation
The paper gives autocorrelations for asset classes between 1972-2014.
Positive autocorrelation favors momentum. Negative autocorrelation favors reversion to mean.
US stocks: 12 months 0.05, 3 years -.25, 5 years -0.13
NonUS stocks: 0.11, -0.37, -.45
US Bonds: 0.21, -0.18, 0.01
NonUS Bonds: 0.12, –0.51, -0.65
Reversion to mean is an important part of any increased return from rebalancing. In this analysis, there was yearly rebalancing. At one year intervals, there is still some positive autocorrelation (momentum). What if the analysis was repeated with rebalancing at 3 or 4 or 5 year intervals? In these time periods, one has allowed reversion to mean to occur.
Justin, that’s very interesting. Is that just due to Canadian currency effects? Also, I see that Rick Ferri was looking at just a 10 year period ending in 2011, so the different sample may have some effect.
@Grant: It could be due to a number of factors:
– The measurement period differed (although when I ran the analysis with an ending measurement period of December 2011, the EAFE/50-50 10-year return changed to 4.25%/4.30% and the 20-year return changed to 7.60%/7.61% respectively).
– The currency interactions were different from a Canadian investor perspective
– I used a balanced portfolio (which included Canadian bonds, Canadian equities, and US equities), where he just viewed the 2 equity regions in isolation – I always recommend viewing these types of decisions in the context of your overall portfolio.
Justin,
An excellent article in regards to investing in international markets. Interesting how the results are almost identical regardless of which strategy one chose. The extra effort in holding two separate geographic regions instead of one doesn’t seem to be worth the effort.
@Constantine Kostarakis: Thanks! :) It was interesting that the results differed from Rick Ferri’s analysis (likely because they were from a Canadian investor perspective).