Friday, May 22, 2009

A few thoughts on rebalancing

We attended our regional DFA Learning Group session yesterday. It's always an informative and enjoyable trip. Many of you have been encouraged to seek out a group in your area and we reiterate that suggestion here. We're also aware that several of you aren't able to access a group due to DFA capacity constraints. We've communicated our desire to have membership available to all our advisor alliances to DFA and have been assured that forming new groups in over-subscribed areas is being actively considered.

At yesterday's session we heard from Marlena Lee, PhD., one of DFA's newest Research Assistants (only at DFA would a PhD be an "assistant"). Marlena previewed an upcoming white paper she's authored on rebalancing ("Rebalancing and Returns"). What prompted her to address the issue was a recent paper asserting that more frequent rebalancing (as frequently as daily or weekly) can be demonstrated to have produced statistically significant increases in returns.

Marlena's analysis found flaws in the paper's reasoning and conclusions. First, the time period was relatively short at about 8 years. As the whole exercise is essentially data mining, it makes sense to examine as large a sample as possible. Second, she was able to demonstrate that there is in fact NO statistically significant benefit to returns as a result of frequent rebalancing. After discussing the original paper and establishing that the conclusions were not valid, she presented a similar analysis of out of sample data from 1926 through early 2009. It is clear from that exercise that there is no incremental return associated with rebalancing whether daily, weekly, monthly, quarterly, annually or some multiple of years is used as a trigger. It's important to remember when considering client exposures compared to their IPS allocations that rebalancing's main purpose is to control risk levels within the portfolio. Expenses and tax impacts can easily negatively impact portfolio returns if rebalancing is too frequent. Tax loss harvesting can be another trigger but there still wouldn't be any incremental returns expected. While there wasn't any specific time period identified that seems better or worse, annually seems to make sense within the context of controlling equity exposure risks and overall portfolio volatility when balanced against the negative impact of transaction costs and taxes. Guess we could call it the Goldilocks approach...

No comments:

Post a Comment