When And How Should You Rebalance?

Recent swings in asset classes have inevitably caused many investors to think about rebalancing their portfolios. As we have noted in the past, we distinguish between four different approaches to rebalancing, based on the objective being pursued and whether it is carried out automatically or episodically. These four logics are summed up in the following table:

Automatic rebalancing to limit risk usually involves adjusting asset class allocations back to their target weights at regular time intervals. Episodic rebalancing to limit risk is based on an investor's belief that he or she can identify - and reduce exposure to - asset classes that have become visibly and substantially overvalued. In contrast, market timing involves episodic adjustment of asset class weights based on a more comprehensive belief that an investor can consistently identify asset classes that are both over and undervalued. The global macro hedge fund style reflects this logic. Finally, automatic rebalancing can also be used to pursue higher returns, by (a) rebalancing only when at least one asset class has become significantly over or underweight relative to its target, and (b) adjusting the most overweight asset class to a specified percentage below its target, and the most underweight asset class to an equal percentage above its target. This approach aims to systematically earn higher returns by taking advantage of financial markets' tendency to overshoot and then revert back to the mean. We recommend this approach, along with episodic rebalancing aimed at limited the risk posed by situations of severe overvaluation.

Let's look more closely at the logic that lies behind systematic over and underweighting. We assume that an investor's primary goals are maximizing the probability of fully funding a long-term liability (say, accumulating a target amount of funds by his or her expected retirement date), while staying within some type of risk constraint. In order to achieve these goals, our investor has to make two decisions: (1) how much of his portfolio to allocate to different asset classes, and (2) the rebalancing rule to use. This investor's views on market efficiency should logically drive his or her decision regarding rebalancing.

If you believe that markets are strongly efficient - that is, that they are always fairly valued, given the information available - then the only reason to rebalance is to stay within your risk constraint; the rebalancing strategy you choose cannot logically be a source of excess returns. However, staying within your risk constraint involves costs, in the form of transaction charges and, possibly, taxes as well. Since rebalancing involves selling investments in those asset classes whose weights are above their targets, while purchasing investments in asset classes below their targets, it leads to the realization of capital gains. If the assets in question are held in taxable accounts, this will trigger an additional cost.

However, while the costs of rebalancing are quite real, the potential risk reduction gains may not be, as they are based on the assumption that asset class standard deviations and correlations remain stable over time. An examination of historical time series data shows that this is not the case; both tend to vary, depending on the period studied. Taken together, this leads to an argument against any rebalancing at all in a market that is strongly efficient.

Now let us consider the situation facing an investor who does not assume that markets are strongly efficient. Let us say he or she shares our view that financial markets are a complex adaptive system. In this view, active managers are constantly searching for superior information and superior models that help them to generate superior forecasts of future asset prices. From time to time, the information and models used by active managers can become self-reinforcing, as, for example, when many investors are using a momentum-based approach. These periods will tend to generate overconfidence, and overvaluation of some asset prices, or, possibly, overvaluation of an entire asset class.

Eventually, these overvaluations will reverse, and possibly lead to excessive negative momentum. In this system, while markets are attracted to equilibrium and efficiency (as evidenced by their tendency to mean revert), they seldom attain it. Under these conditioons, not only can systematic rebalancing limit risk, but it may also generate additional returns, by selling overvalued assets and buying undervalued ones. However, the achievement of these potential benefits once again requires an investor to incur transaction, and potentially tax costs as well.

We performed a simulation analysis to quantify the potential impact of different approaches to systematic portfolio rebalancing. Because our focus is on rebalancing strategy, we assume that the investor does not believe that asset class risks and returns can be successfully forecasted. As a result, he allocates equal portions of his portfolio to ten asset classes: (1) real return bonds; (2) investment grade bonds; (3) foreign currency bonds; (4) domestic commercial property securities; (5) foreign commercial property securities; (6) a commodity index; (7) timber; (8) domestic equity; (9) foreign developed market equity; and (10) emerging markets equity.

We analyzed six different rebalancing strategies. The first rebalances back to the target asset class weights (i.e., ten percent each) at the end of each year. The next four only rebalance back to the target weights when one or more asset classes is at least 5%, 10%, 15%, or 20% above its target weight. For example, the 20% strategy would not rebalance unless one asset class accounted for more than 30% of the portfolio. The last strategy never rebalances.

In our simulation analysis, we used historical distributions of real returns for each asset class over the 1989 to 2004 period. We used a correlation matrix calculated from the same data. Finally, to (roughly) capture the tendency of markets to trend, the return on real return bonds has a .2 serial correlation with the return on this asset class in the previous year. We also included one-way rebalancing costs that varied from 35 basis points to 135 basis points.

For each rebalancing strategy, we ran 10,000 simulations, and calculated the expected value of an initial 1,000 dollar investment 20 years hence. We also calculated the standard deviation of that final portfolio value, as well as the probability that it would exceed a target of 3,000 dollars in year 20 (e.g., assume the investor is trying to fund a liability with an expected value of 3,000 in year 20). The following table shows the results of our analysis:

This table makes clear the potential confusion that can arise in discussions about rebalancing. The first column shows the different rebalancing strategies we tested. If we look only at the second column, we would conclude that rebalancing doesn't make sense, as never rebalancing results in the highest expected portfolio value at the end of 20 years.

However, a look at the third column shows that reduced rebalancing frequency also leads to higher portfolio risk, as measured by the variability of the portfolio's value in twenty years time. And as column four shows, the trade-off between return and risk actually worsens the longer rebalancing is put off.

As previously noted, another way of defining risk is in terms of the probability of achieving the specified accumulation target at the end of twenty years (the probability of falling short of this goal is 1 minus the probability of success). The fifth column in the table shows that, because it affects both expected return and risk, the impact of rebalancing strategy on the probability of achieving the accumulation goal is, while positive, marginal at best. To put it differently, the beneficial impact of delayed rebalancing on returns tends to be offset by its detrimental impact on risk. As you can see, most of the improvement from rebalancing is achieved by setting a trigger of 10% to 15% above a portfolio's target weights. Completely eliminating rebalancing adds considerably more risk, for a minimal additional increase in the probability of achieving the investor's accumulation target.

| This Month's Letters to the Editor: Investing in Commodities/Legg Mason and Index Investor's Opinion on Leveraged Index Products | Fundamental and Dividend Weighted Indexes | Product and Strategy Notes: Depressing Dalbar Study; More 401(k) Fund Choice; Residential Propert: Is a Crash on the Horizon; Poor Man's Alpha; New RYDEX Foreign Currency ETF | Economic Warning Indicators Update | Asset Class Valuation Update - Revised - Includes Property, Commodities and Volatility and Updates to Sector Rotation Watch | Global Asset Class Returns | When And How Should You Rebalance? | What Do Bond Spreads Tell Us? | Another Month, Another Crop of New Commodity Products | This Month's Issue: Key Points | Volatility is Up: So What? |