Several years ago, an Internet friend and physics professor, Bob Warasila, helped me test rebalancing target limits using my eight asset class database that begins in 1989. At that time we concluded that 30% is the optimum rebalancing percentage. For example, if the target for an asset class is 15% then 0.15 x 0.30 = 0.045 or 4.5%. Using this percentage, rebalancing is called into action if the asset class with the 15% target rises above 19.5% or moves below 10.5%.
Until recently we used the 30% limits in the AA-Mosaic.xls spreadsheet as well as upper and lower constraints in the mean-variance optimization (MVO) calculations. Bob recently reworked the numbers using the database that now includes figures through 2007. The new optimum rebalancing percentage is 35%. The higher percentage provides the following advantages to the investor.
- Rebalancing will not occur as frequently and as a result it will save commissions. While commissions are not high, any saving works to the bottom line of the portfolio performance.
- Perhaps even more significant is the inertia asset classes exhibit. Once in an upward trend, an asset class tends to run longer than we frequently expect. Of course the reverse is also true. When an asset class begins to trend down it will continue in that direction longer than we expect. With the 35% limit, asset classes will run longer before rebalancing is called into action.
Portfolio rebalancing, as you will recall, is a process that “requires” one to sell high and buy low. We sell shares of an ETF when the asset class rises above its upper target limit and we buy shares when an asset class is below target.
Here is a summary of Professor Warasila’s recent rebalancing analysis.
- Equal % in each class rebalanced every year $602k
- No rebalance on non-equal % classes $590k
- Non-equal % in each class rebalanced every year $605k
- 10% rebalance trigger $605k
- 20% rebalance trigger $624k
- 25% rebalance trigger $631k
- 30% rebalance trigger $609k
- 35% rebalance trigger $645k
- 40% rebalance trigger $632k
- 30% with 1/2 out-of-balance amount $645k
- 35% with 1/2 out-of-balance amount $613k
Professor Warasila found that rebalancing works. Here is a quote from a recent e-mail. “I totally rebuilt the 30% from scratch and get the same result. So there are no mistakes. My preliminary impression is that the “timing” of the rebalance signal can greatly effect things when we are experiencing large % changes in some or all assets. However that being said, for this data set (1989 to 2007) 35% is the optimum AND any rebalancing is better than none. Whether one achieves the maximum in a particular data series using a particular % is kind of a gamble on when you are living.”
I asked Bob to amplify what is mean by “1/2 out-of-balance” means. Here is his explanation.
“In those examples only 1/2 of the difference between actual and target % in the asset class was reset. For example, if in Large Growth we had $10,000, and the asset target was 10% of the total portfolio which was $150,000, we would be $5,000 low. We would only add $2,500 to LG. This notion is along lines of what we discussed in the past of making a “tactical” rebalance. What it effectively did was improve the 30% number but reduce the 35% significantly. I think all these variations tend to show rebalancing always helps but there is no set strategy to guarantee the optimal result. I think a massive study of rolling time intervals might deduce statistically what is the optimum % and/or strategy but since it will be a statistical outcome, when you are living also matters.”
“Again rebalancing always helps, and one must pick a strategy and stick to it. One the lessons I concluded after my 1st 10 years of investing is DON’T change strategies , rather focus on one you understand as you are doing with AA and concentrate on refining it.”
If you have any rebalancing questions, ask them in the comments area below.
Lowell Herr
Photograph: Art aboard Celebrity cruise ship.
Sphere: Related Content
