Monday, June 15, 2009

Portfolio Rebalancing

Many commentators advocate an investing strategy based on fixed percentage asset allocations among different types of stocks (large cap, small cap, domestic, foreign), bonds, cash, and possibly other things like commodities, real estate, or precious metals. This brings us to the question of what to do when assets grow beyond or shrink below their target percentages.

The main debate is between rebalancing periodically based on time, such as quarterly, or rebalancing based on percentages, such as when an asset is more than 5% off its target percentage. I don’t particularly like either approach. I prefer a focus on costs.

Suppose that an investor doesn’t have enough savings to qualify for low commissions on ETF purchases, and pays $25 per trade. If this investor doesn’t want to spend more than 1% on the sell and buy commissions, then all trades for rebalancing should be $5000 or more (and trades with new money should be $2500 or more).

This leads to a simple rule: rebalance when the allocation of some asset class is off by at least $5000. This may mean that for small portfolios, allocations can be very far off on a percentage basis. The fact that the portfolio is small means that this is not a major problem. As new contributions come in, the balance will be improved.

With this strategy, larger portfolios will be kept closer to the chosen asset allocation. As portfolios become very large (and qualify for lower commissions), a 1% rule might lead to frequent trading. In this case, the investor can switch to a rule based on a combination of trade size and deviation from the desired allocation.

By this I mean that rebalancing would take place when the allocation to an asset class deviates from the desired percentage by some amount (say 5%) AND the rebalancing would involve trades of at least a minimum size (say $5000).

I’m not a big fan of rebalancing at fixed times because it allow investors to ignore the need to rebalance when the stock market gets “scary”. If stocks drop by large amounts, this is the best time to rebalance. However, most investors ignore their asset allocations until stocks rebound. In this way, they miss the opportunity to profit from the plunge in stocks.

Much of the value of fixed asset allocations comes from buying asset classes when they are down. Failing to do so eliminates most of the upside of the asset allocation strategy.


  1. Hi Michael,

    I did a simulation where I rebalanced using a percentage deviation threshold. I found the results interesting, and I was wondering if you think they agree with theory.

    Suppose you have a portfolio that is 60% stocks and 40% cash, and you rebalance when your allocations are more than 5% off. As a rule of thumb, it seemed that the returns you get would be about equal to what you'd expect if you had a fixed 65% allocation to stocks. So you seemed to end up with the returns of a 65%-stock portfolio with the volatility of a 60%-stock portfolio. Basically this allows you to cheat a bit on the risk premium.

    I also found no way for someone to beat the expected returns from a portfolio of 100% stocks if the only other investment option is cash. (This is what I was actually looking for. I had hoped that, say, a 95%-stock portfolio with proper rebalancing might actually beat 100% stocks.)

    I haven't confirmed this in any rigorous way; I meant to get back to it but I haven't yet. I may have missed something, or I may also be just noticing patterns in historical S&P returns that aren't true in general.

  2. Patrick: Your results will depend on what period of time you use for returns. Over long periods of time, we expect that fixed allocations would beat rebalancing. This is because a fixed allocation will have its stock percentage drift upwards while bonds and cash fall behind. However, the rebalanced portfolio will have lower volatility. My guess is that your results are based on returns from the past 20 years where stocks didn't perform all that well. Is this right?

  3. Hi Michael,

    You're right, my returns were based on a period that included the last 20 years and ended some time earlier this year when stocks had tanked. I don't recall the exact period; it was probably the entire history of the S&P500 at the time.

    I see what you mean about the fixed allocation. I guess what I was comparing was the portfolio with rebalancing against a hypothetical portfolio that achieved X% of the returns of a pure-stock portfolio. For example, the rebalanced portfolio that's 60% stock with a 5% threshold achieved 65% of the compound annual growth of the S&P index. Maybe this is what you would get with daily (or continuous) rebalancing?

    Anyway, thanks for your thoughts.

  4. Yes, I like the idea of "nudging" the portfolio towards balanced anytime new money comes in. This would take some of the emotion out of the problem. New money would include bond coupons, dividends, and new capital from salary, etc.