Beyond The Abstract: Alternative Weighting

March 24, 2008

Does alternative weighting work for the MSCI Global Index? A look at recent research in the index and ETF space.

The validity of alternatively weighted indexing has been one of the chief debates of the indexing industry during the last few years, and panels on the topic have been de rigueur on the agendas of industry conferences. 

In "Alternative Indexing with the MSCI World Index," Dr. Thomas Neukirch, a senior quantitative analyst with Germany's Harald Quandt Holding GmbH, looks at the MSCI World Developed Index using a variety of weighting schemes to see if alternative weighting might improve returns.

You can access "Alternative Indexing with the MSCI World Index," by Dr. Neukirch here. One word of warning: Apparently the article was originally written in German, so some of the syntax is a bit off, to the point that it sounds as if Jedi Master Yoda is explaining how the Force applies to alternatively weighted indexes. ("High average weights have the US, Japan and UK.")

The study looks at the performance of the components of the MSCI World Developed Index in a variety of permutations that weight sectors, countries and individual components using the original index weighting, equal weighting, and equal risk weighting methods (only components are equal-risk weighted, not sectors or countries). The original index weighting is, of course, float-adjusted market capitalization, which represents the total available shares. Equal weighting, the article notes, acknowledges the forces of mean reversion at work in the stock market, while equal risk weighting takes the method a step further by giving lighter weightings to more volatile components and creating more stable returns. The time period it covers is somewhat limited, looking at just February 2001 through January 2008.

So what were the results? Well, consider that the index itself fell 2.86% during that period and was the worst-performer of the bunch. By contrast, the best-performing permutation in terms of raw numbers was the index that weighted components by equal risk with equally weighted individual countries; it had a positive return of 8.27% and also the second-lowest standard deviation of the different index versions. The next-best version in terms of performance was the one that equal-weighted both components and countries; it was up 8.19%. All of the alternatively weighted versions outperform the original index over time; however, that outperformance is more pronounced in a rising market than in a falling one.

Equal-risk weighting the broad index's components (up 6.27%) outperforms equally weighting (up 5.56%) the components, but equal weighting the countries appears to be a performance booster. It greatly reduces the weights of the U.S., which represents nearly 53% of the index on average, and the U.K. and Japan, which represent 10.9% and 9.9%, respectively.  These countries also have some of the worst average monthly returns for the time period. The U.S., in particular, has underperformed international markets in general in recent years; it had the lowest monthly return at -0.3%. The U.K. had an average monthly return of -0.1%, while Japan's monthly return was -0.2%.  By contrast, Austria had a weight of just 0.2% in the index, on average, but the highest average monthly return of any country at 1.1%. The smallest country, New Zealand, had a 0.1% weighting in the index and an average monthly return of 0.6%.  

The article also looks at the costs of implementing an equal weighting strategy. Before expenses, the equal weighting strategy returns 1.45% per annum versus a -3.17% return for the original index. After expenses, the equal-weighted index returns 0.70%, still outperforming the original index. However, the interesting part is that they haven't bothered to say what the performance of the original index is after costs are taken into account, which would seem to be fairly relevant, even if it would only widen the performance differential. It would be nice to have that number to compare the associated costs of alternative weighting.

Overall, the article adds fodder to the alternative weighting side of the argument despite its relatively narrow time frame. The data is fairly simple to understand, and the international approach is somewhat unique. It's definitely worth a look.

Bonus Article

And once again, this week we have another bonus for all the volatility geeks out there. Nelson Areal of Portugal's University of Minho has written an article that looks at different methodologies for the construction of a volatility index for the U.K. based on the FTSE 100. The article ultimately concludes that its "Alternative Interpolation Scheme" is the optimal means of constructing a volatility index for the U.K. market; the methodology is designed to accommodate the fact that the U.K. options market is less liquid than that of the U.S. You can access the article here.

Happy reading!


Find your next ETF

Reset All