Collared Weighting: A Hybrid Approach to Indexing
Authors: Sanjay Arya, Paul Kaplan
Source: Morningstar working paper
Date: July 2006
Arya and Kaplan note that capitalisation weighting, which appears to be a good weighting system from a theoretical point of view, is not without drawbacks. One of the drawbacks of cap weighting indices is that they often end up exposed to the more overpriced assets, which can possibly reverse and cause losses. As an alternative, some have turned to fundamental indexation, in which assets are weighted by measures of size such as revenue, book value, and dividends. Fundamental weighting may remedy some of the drawbacks of cap weighting, but it can also lead to value and small-cap biases, higher turnover, higher costs, problems of scalability, and macro inconsistency.
In this article, the authors posit a new methodology, named collared weighting, and argue that it is possible to enjoy the benefits of both cap weighting and fundamental weighting, while minimising their drawbacks. Their approach consists of weighting assets by market capitalisation, if it does not deviate significantly from fundamental weights. If markets are overvalued or undervalued, they use fundamental weights. The limits, or collars, are established by constructing a band around the fundamental weights.
To illustrate this methodology, the authors construct a fundamental and a collar-weighted version of the Morningstar US Market Index over the period July 1997-December 2005. At the outset, cap-weighted assets accounted for 67% of the collared index. By the end of 1999, this proportion falls to 39%, with the growth of the technology bubble. After the market collapse, market weighting once again predominated. Fundamental weighting yielded the highest total returns. Collared weighting came in second, while cap weighting produced the lowest returns. The authors attribute this result to the value bias of the fundamental index, as value stocks outperformed growth stocks during the period considered. During the two years 1999 and 2000, a declining period for value stocks, the cap-weighted index dominated the two other indices and the fundamental index was the worst performing one.
The collared methodology tends to smooth index returns, which, for the entire period, results in a standard deviation for this index lower than that of the two others. The authors also investigate the impact of the various weighting schemes with regards to turnover and costs. As expected, the cap-weighted index exhibited the lowest turnover, while the fundamental index exhibited the highest. The turnover of the collared index is in between. After taking into account the costs related to turnover, it turns out that, for the entire period, the performance of the collared index performs only slightly less well than the fundamental index.
The authors conclude that collared weighting seems to be a good alternative, as it makes it possible to limit disproportionate weighting of overvalued stocks, as well as value biases and turnover—the respective drawbacks of cap weighting and fundamental indices—thus leading to an index with moderate volatility. This should result in better index products for investors.