|
Fundamental Indexation: Rebalancing Assumptions and Performance Authors: David Blitz, Bart van der Grient, Pim van Vliet Source: Journal of Index Investing, volume 1, number 2 Date: Fall 2010 |
In this article, the authors investigate the impact of the choice of a rebalancing date on index performance. They especially focus on fundamental indices proposed by Arnott, Hsu and Moore (2005) which are rebalanced once a year. In the present paper, Blitz, van der Grient and van Vliet show that the performance of fundamental indices is highly sensitive to the choice of the annual rebalancing date.
For example, in 2009, using different rebalancing dates, they observe differences in returns exceeding 10%. In addition, these differences remain permanent over time, as no mean reversion is observed. As a result, the choice of the rebalancing date appears not to be neutral for a fundamental index, which is a drawback for an index that serves as a benchmark. To solve this problem, Blitz, van der Grient and van Vliet propose computing an index made of four fundamental indices with various rebalancing dates. Following FTSE/RAFI Indices, the four underlying indices were based on book value, sales, cash flow and dividends respectively, and made up of the 1,000 largest U.S. stocks according to fundamental indicators. The performance of the resulting fundamental index, as well as the those of the four fundamental indices considered on their own, were compared with that of a cap-weighted index made up of the 1,000 largest U.S. stocks according to market capitalisation. The indices were considered over the period from 1991 to 2009, with a monthly frequency. The four fundamental indices were rebalanced once a year, in March, June, September and December, respectively.
Looking at index performance results, it appears that the fundamental index rebalanced in December outperforms the market-capitalisation index by about 200 basis points per year, a similar result to that obtained by Arnott, Hsu, and Moore (2005). Quite similar results are obtained for the fundamental indices rebalanced in March and September. However, the index rebalanced in June achieves lower performance than the other three, with an outperformance of only 140 basis points compared to the cap-weighted index. All indices exhibit similar absolute risk. Looking in detail at the results obtained for each year in turn, it appears that the differences in performance between the indices rebalanced at various dates may be quite large. The extreme figures were obtained in 2009 with a 10% outperformance for the index rebalanced in March, compared to the cap-weighted index, while the index rebalanced in September slightly underperformed the cap-weighted index. 2003 is another example of a significant return spread between indices, with a 5% spread between the December and June indices.
In view of these results, the authors made the assumption that this difference in performance between indices using different rebalancing date will be permanent and that no mean-reversion will occur. In order to prove this assumption, they performed Augmented Dickey-Fuller (ADF) tests on the difference between the cumulative returns of each pair of fundamental indices. The result they obtained shows no support for mean-reverting behaviour, leading to the conclusion that the difference in performance between the various indices will persist over time. The authors argue that increasing index rebalancing frequency does not appear to be a good solution, due to turnover fees. Alternatively, they conclude that a blended index, made up of a fundamental index with various rebalancing dates, may be a possible solution to reduce index sensitivity to the choice of the rebalancing date.
Reference
Arnott, R. D., J. Hsu, and P. Moore. 2005. Fundamental Indexation. Financial Analysts Journal 61 (2): 83-99.
Related links
