In this piece I examine various way in which an investor can think about their active market timing decisions, often labeled with the innocuous term “rebalancing.”
Rebalancing a portfolio is the finance version of “eat your vegetables” — the advice is taken as gospel, but very few people question the advice and/or how to implement the advice. But do the so-called sophisticated market participants (i.e., traditional advisers, robo-advisers, and target date funds) who tout rebalancing as a valuable service really think through their rebalancing process? Have these “passive” investors, who are engaged in selling an active market timing service, ever really considered how and why they rebalance the way they do? Many of these providers have certainly carefully considered the costs and benefits of their rebalancing process, but having reviewed many of the rebalancing services embedded in investment services/products available to the public, my guess is that some of these providers have not thought too deeply about rebalancing.
Why are rebalance decisions so important?
To reiterate, rebalancing is an active market timing consideration and should not be a decision that is taken lightly. The decision is just as active as deciding between buying a passive market-cap weighted index or a portfolio of concentrated value and momentum stocks. And like all active decisions, one should carefully consider the evidence and the actions of other market participants when rebalancing a portfolio. Rebalancing decisions and processes don’t need to be complex, but they do need to be considered.
If You Rebalance, You Are an Active Investor
There has been a lot of ink spilled in discussion about passive index investing, but in this sense passive index investing generally means a market cap weighted stock portfolio…but not very many people own only a market cap weighted stock portfolio (and far fewer probably own a global market cap weighted stock portfolio which is the only true market cap weighted stock portfolio but that is a discussion for another day). Most people probably own a portfolio comprised of stocks and bonds. The mix of stock and bonds in their portfolios is probably set by either their risk tolerance, time to retirement, or some similar planning concept. The resulting asset allocation is often referred to as their target portfolio (strategic allocation or policy portfolio).
One thing I find very interesting is that their portfolio spends very little time actually at the target allocation because the price of stocks and bonds is ever changing. Therefore, any asset allocation deviation from the target allocation is an active (i.e., market timing) decision.
The deviation from the target allocation can be made on accident (prices changed and I haven’t rebalanced yet) or on purpose, but the end result is the same — any deviation from the target allocation is an explicit market timing bet in the portfolio. For example, if you started the year at your target allocation of 50% stocks and 50% bonds and stocks outperformed bonds by 10% in the first week, then your portfolio is approximately 55% stocks and 45% bonds. If you decide not to rebalance your portfolio (tolerance bands haven’t been breached or it isn’t the end of the quarter or year) you are now making (a small) market timing bet that stocks are likely to outperform bonds until your next rebalance. If you do rebalance the portfolio, you are making a small timing bet that stocks are likely to underperform bonds.
Common Rebalancing Algorithms
With this (market timing) perspective of rebalancing in mind, let’s quickly review the most common rebalancing decision algorithms 1) calendar based or 2) tolerance range (tolerance band) based.
- Calendar based rebalancing means that the adviser rebalances a portfolio back to the strategic (or target) allocation at a set calendar frequency (monthly, quarterly or annually) and some even tout doing this on a real-time daily basis.
- Tolerance range rebalancing means that the investor has a predefined range of variation in asset allocation that is acceptable, but once an asset breaches the upper or lower bound of the range the portfolio is then rebalanced back to the strategic (or target) allocation. This may occur as frequently as daily or weekly and as infrequently as biannually (or even longer).
But does the generic rebalancing of a portfolio actually add value? This research paper by Vanguard points to the conclusion that there is no optimal rebalancing algorithm when calendar or tolerance ranges are considered.
A recent research paper by Baker, Dieschbourg, McIntyre and Muralidhar also found a similar conclusion that traditional rebalancing algorithms didn’t add much value (and actually hurt from the perspective of portfolio drawdowns). In addition, they went a step further and provided some specific market timing algorithms that in isolation and when combined provided additional returns and reduced drawdowns but not standard deviations.
Let’s take a more in-depth look at their paper.
The Mythology of Rebalancing Overview
The analysis in this paper uses daily data from 1/1/2000 through 12/31/14 for the following asset classes:
They used the following to judge the effectiveness of the rebalancing algorithms:
Finally, the looked at the following standard rebalancing algorithms:
- 3% Tolerance Band
- 5% Tolerance Band
- Volatility adjusted Tolerance Range (Bonds & Cash at 3% range all others at 5%)
- Naive Target Date Fund (reduce stock allocation by 1% per year and increase bond allocation by 1% per year)
- Benchmark – this is an unrealistic, costless and continuously rebalanced portfolio back to the target weights.
However, with the perspective that all rebalancing decisions are active market timing decisions, they evaluate the the following market timing algorithms in lieu of traditional rebalancing algorithms:
The following excerpt from their paper explains how they apply the rules and which asset class to underweight and which to overweight:
So do the explicit market timing algorithms (called “Intelligent Rebalancing” by the authors) outperform traditional rebalancing algorithms (called “Naive Rebalancing” by the authors)?
Here are the results from the paper: