February 21, 2013 By Rick Ferri

Diversification in investing is a mainstay of prudent portfolio management. Portfolio risk can be reduced with proper asset class selection. This isn’t to say that you won’t lose money. The purpose is to reduce the risk of a large portfolio loss by balancing portfolio risk among different investment types.

Markets are always moving, sometimes together and sometimes not. This unpredictable interaction leads to natural changes in a portfolio’s allocation and forces action if we wish to control risk. Asset class rebalancing is used to return a portfolio back to its fixed target allocation. If a portfolio has a target of 50% in stocks and 50% in bonds and market movements result in different allocation, a rebalancing would correct this mismatch back to its 50-50 target.

There are many different techniques used to rebalance a portfolio. The two most common methods for rebalancing are by a calendar date or by using percentage bands. The calendar means simply picking a particular day (or days) to rebalance during a year and doing it on that day, regardless of current events or the market status. The percentage band method requires monitoring the portfolio allocation vis-à-vis a target allocation and rebalancing when the two become meaningfully different.

People often ask me if there is a best method.  My answer is always the same. The best method is the one you’re able to implement religiously without emotion. After that, the method that results in the best risk-adjusted return going forward is more luck than science.

To show this, I created and rebalanced several portfolios using the Vanguard Total Stock Market Index Fund Admiral Shares (VTSAX) and the Vanguard Total Bond Market Index Fund Admiral Shares (VBTLX). Each portfolio’s starting allocation was 50% VTSAX and 50% VBTLX as of January 1, 2000. I ran the numbers forward for 13 years using various rebalancing techniques to get the portfolio back to 50-50. I also included one portfolio that was never rebalanced over the entire period.

Table 1 shows the portfolio results based on different annual rebalancing techniques and includes the never rebalanced portfolio. The table includes an annualized return, annual standard deviation of the technique, best and worst calendar year return, and the high and low amount in equity over the period.

Table 1: Annual rebalancing or longer (January 1, 2000 December 2012)

Source: Vanguard .com for mutual fund returns data

Never rebalancing produced the lowest returns and had the lowest best year. It also held the lowest allocation to equity at some point during the 13 years. Never rebalancing also had the smallest losing year and had the lowest standard deviation in annual returns.

I included two annual rebalancing dates (Jan 1 and Apr 1) to illustrate that portfolio returns are dependent on the month chosen to rebalance. During this particular study, an investor would have earned more rebalancing starting in the second calendar quarter rather than the first.

The best rebalancing method from a return standpoint was once every three years. But don’t read too much into Table 1 or any of this data. These results are period specific. The best method going forward might be quite different than the past 13 years.

Table 2 highlights the same study using quarterly rebalancing. I created three portfolios with different starting months coinciding with the beginning of each quarter; January 1, February 1, and March 1.

Table 2: Quarterly rebalancing methods (January 1, 2000 December 2012)

Source: Vanguard .com for mutual fund returns data

The month chosen for rebalancing had an effect on performance. January 1 rebalancing outperformed both February 1 and March 1. Again, there is no usable information in these numbers.  The higher return for a Jan. 1 start is not statistically significant. Don’t be fooled by randomness as Nassim Nicholas Taleb would say.

Table 3 highlights the same study using percentage bands.  Portfolios were rebalanced back to the 50-50 target when the allocation to stocks landed above or below the band.

I chose three common bands, 5%, 10%, and 20%. A 5% band means a portfolio would be rebalanced if the percentage in equity was over 52.5% or below 47.5% at the end of a month. The 10% band triggered rebalancing when equity over 55% or below 45% at the end of a month. The 20% band triggered rebalancing when equity over 60% or below 40% at the end of a month.

Table 3: Rebalancing using percentage bands (January 1, 2000 December 2012)

Source: Vanguard .com for mutual fund returns data

Portfolio risk and return using bands was slightly better than using quarterly rebalancing and on par with annual rebalancing.  Table 3 also shows the number of rebalancing events in each of the three portfolios.

The cost of trading was not included in this study, although it is important. The number of rebalancing events does have an impact on portfolio performance due to transaction costs. This could make quarterly rebalancing and rebalancing using a narrow band less desirable.

One way to look at the benefit derived from each rebalancing technique is to use Sharpe Ratios. This calculation subtracts the Treasury bill return from the annualized portfolio return and then divides the difference by risk (standard deviation). A higher Sharpe ratio is a better risk-adjusted portfolio. Figure 1 illustrates the Sharpe ratios of each strategy in order from worst (left) to best (right).

Figure 1: Rebalancing method Sharpe ratios (January 1, 2000 December 2012)

Source: Vanguard.com for mutual fund returns data

The highest Sharpe Ratio was derived from 3-year rebalancing and never rebalancing. The worst was derived from quarterly rebalancing and a 5% band (both also required frequent trading, which was not counted). The 10% and 20% band method fell in the middle.

On balance, it appears that the less rebalancing occurs the better off your portfolio will be. Don’t be misled by the data.  The return numbers hide the true purpose for rebalancing to start with, and that is to control portfolio risk.

Equities historically have had about three times as much price risk as bonds.  Accordingly, it is helpful to look at the maximum and minimum allocations to equity that each method produced to gauge the risk in a portfolio at any given time. Bad things may not happen when there is more risk, but if they do happen, you won’t be in control. Figure 2 highlights the maximum and minimum equity risks in each portfolio.

Figure 2: Minimum and maximum equity exposures (January 1, 2000 December 2012)

Source: Vanguard.com for mutual fund returns data

The target allocation for this exercise was 50% in equity. Figure 2 shows that the greatest deviations from 50% in equity occurred by never rebalancing, rebalancing only once every three years, and rebalancing using a 20% band. This is the consequence of not rebalancing.

Less frequent rebalancing may result in greater risk or less risk in a portfolio than you desire. The consequences for this might be having too much equity at the beginning of a bear market or not having enough equity at the beginning of a bull market. Neither of these allocations were part of your plan.

In contrast, quarterly rebalancing and the 5% band had the tightest fit to 50% in equity, albeit with greater turnover. As noted earlier, more frequent rebalancing lowers portfolios return.

Annual rebalancing on January 1 and rebalancing using 10% bands appear to have been the best compromise between holding true to an allocation and achieving a fair Sharpe ratio. The annual portfolio traded once per year and the 10% band portfolio traded about the same. This is a reasonable amount of turnover.

I’ll still hold to my original point on the best rebalancing method – it’s the one that you’ll stay disciplined doing. The 10% percentage band may be your best long-term option if you’re a hound-dog who watches your portfolio closely. On the other hand, for those less interested, annual rebalancing works just fine.