ETF Investing Guide: Rebalancing Rules

Once you have resolved to rebalance your portfolio, the next question is: How often should you rebalance?

William Bernstein, in The Intelligent Asset Allocator (a great book, but written in the pre-ETF era), briefly discusses whether you should rebalance your portfolio every quarter, bi-annually, or annually.

There are two considerations here.

First, in a taxable account you don’t want to rebalance too often, as rebalancing involves the sale of assets that have appreciated, and thus the realization of capital gains. You can mitigate the tax effects of rebalancing by carefully choosing the highest cost-basis tax lots to sell (Bernstein doesn't discuss that), but nonetheless frequent rebalancing could result in higher tax bills and lower compounded returns.

How serious could the tax consequences of rebalancing be? If the market is broadly flat for the next few years but fluctuates wildly over shorter periods, you won’t mind the taxes. You’ll be happy to make some profits due to asset class price fluctuation, when a buy-and-hold strategy would leave you flat, even if you have to pay taxes on the profits. However, if the markets among which you’ve allocated your assets trend upwards at roughly the same rate over extended periods of time, frequent rebalancing could trigger unnecessary capital gains and reduce your compounded after-tax return. So frequent rebalancing is generally more suitable for tax-sheltered accounts than taxable accounts.

The second consideration when thinking about the frequency of rebalancing is that asset classes tend to move with enough momentum that deviations from their long-term growth trajectory can last for a number of years. (Dick Davis calls this The Durability of Major Trends.) This means that if you rebalance less frequently you can actually end up doing better.

Think about the multi-year out-performance of US technology stocks until 2000, for example. Imagine that you had divided your portfolio between only two asset classes - tech growth stocks and non-tech value stocks - and you frequently rebalanced the portfolio to maintain an equal weighting between the two. In that case, you would have limited your upside from the run-up in technology stocks, because you'd be constantly selling tech stocks too early to maintain your asset allocation. Instead, you should have rebalanced your portfolio only when it became heavily misaligned. That way, you would have benefited from the momentum run-up in technology stocks, but would have periodically corrected your portfolio to avoid a situation where 80% of your portfolio was in tech stocks that subsequently collapsed.

For this reason, the correct question may not be “How often should I rebalance?”, but rather “How far should I allow my asset classes to stray from their target allocations before I rebalance?”. Rebalancing only when an asset class reaches 150% of the target allocation, for example, will perhaps result in a more tax efficient and more profitable portfolio.

Here’s one intriguing rebalancing variation to consider. If an asset-class allocation reaches 150% of your original allocation, don’t just cut it back to the target allocation. Instead, cut it back to below the target allocation - say 75% of your target allocation. If that asset class then falls to 50% of your original allocation, restore it to 150% of the original allocation.

The rationale is as follows. If one asset class is appreciating much faster than the others in your portfolio, you want to ride the momentum to 150% of your target allocation. But when you are ready to trim back the asset class, it’s probably become overvalued relative to your other assets. So sell more of it than would be required to return to your “normal” asset allocation. Similarly, if the asset class then depreciates significantly, it has probably become cheap relative to other asset classes, in which case you can overweight it.

My gut feeling is that this approach to rebalancing would have saved us - and made us - a lot of money during the bubble and its subsequent bursting. You would have ridden the growth stock boom to the point where US large cap stocks were one and a half times your intended allocation. Then as the market kept rising you would have sold half of them. And two-and-a-half years into the market downturn you would have aggressively re-invested in US stocks when they became much cheaper. The key question about this strategy is how much momentum there is in asset class value appreciation. If there’s lots of momentum, this method works. If there’s little momentum, you may want to rebalance at smaller deviances from your benchmark asset allocations.

ETF Investing Guide Main Page
Previous: How to Make Money By Rebalancing
Next: Should You Use Sector ETFs?

Comments

[Augustus]

Most of the various rebalancing schemes are Calendar based. They usually suggest checking the allocations on an annual basis, maybe quarterly, and shifting the allocation.

This method of using a Percentage Out of Balance trigger may work much better but require a bit more monitoring. However, doing a weekly check of the allocation to the particular segment of the market would be pretty painless, and really not required if everything was in near balance the previous checkup. It would take a bit of backtesting to come up with a proper rebalance trigger percentage. My guess is that something in the 20% range (under or over) would do it.

One other feature suggested by the article is to set an Under allocation when shifting money out of an allocation. Generally, I find that a pretty appealing idea, although I would again want to look at some backtest results before committing to it. The concept that a particular segment has bbecome overvalued - causing the original Over allocation problem - also leads to the idea that it should be temporarlily under weighted. It would be interesting to see what levels have worked well in the past.