No. of Recommendations: 1
Actually over a quarter of my rule-of-thumb 1% "cap weight index" drag for the S&P 500 is due to the reconstitution problem: changes are (a) ill advised, and (b) pre-announced and therefore front-run.
(a) is a problem because recent winners tend to get added while recent losers tend to get demoted, and there is on average mean reversion.
(b) is a problem because the index underperforms the actual stocks. It lets funds track the index accurately, at the expense of the holders of the funds in a hidden way.
Both problems are substantially mitigated simply by using an index list that's at least 6 months old, preferably 12 or more. (minus anything actually delisted since then, of course, most often from M&A).
Another theoretical advantage of the equal weight system, (although I don't know what the empirical evidence is), is that the index is rebalanced every quarter. True, that means new editions get put in at high prices, but it also means that stocks whose prices have gone down have to be purchased, to keep the equal weight, and stocks that have gone up have to be sold.
If that is true, here's a question, taking the rebalancing to an extreme: what if you could rebalance more frequently? For instance, if every stock in the RSP had a weight of 0.2% at rebalancing, what if you sold any stock whenever its weight was more than 0.21% (i.e. a relative 5% move up in weighting). Ignoring transaction costs, would you expect such a strategy to outperform the RSP? Has anyone (I am of course thinking of Jim) done such a simulation to see if it would work?
dtb
