Within the investment consulting community, it has long been preached that periodically rebalancing a multi-asset class portfolio is a requirement for an effective investment policy. However, little research is available to help in determining what is the optimal rebalancing frequency. From an efficiency and simplicity standpoint, rebalancing at a set time period, such as annually, may be easiest. However, if you consider the taxes and transaction costs that result from rebalancing, we question whether an investor is better off with such a simplified structure.
In this analysis we seek to interpret what we believe to be the optimal rebalancing frequency after considering taxes and transaction costs. In this analysis, we will discuss our rationale for believing that investors are better off establishing a disciplined rebalancing frequency.For those investors with taxable assets and an intermediate to long-term investment horizon, we find a rebalancing strategy that uses a 5 percent rebalance trigger (essentially rebalancing a portfolio whenever an allocation deviates 5 percent from its target weight) is the most optimal when considering return, risk and the costs associated with rebalancing.
The visible, hard costs of rebalancing are transaction costs and realized taxable gains. Whether an investor pays transaction commissions or pays for services on a fee basis,transaction costs occur at some level. Additionally, all clients are affected by the bid/ask spread that must be paid when selling and purchasing securities in order to rebalance. While the bid/ ask spread is typically insignificant (often as low as $0.01 a share), this cost can add up when rebalancing an entire portfolio. For clients with taxable investments, after-tax returns are just as important, if not more important, than pre-tax returns. Therefore, controlling the unnecessary recognition of capital gains is important in maximizing after-tax returns. Due to the nature of rebalancing, those assets that have appreciated the most, relative to the others within the portfolio, are sold in favor of increasing the allocation to those assets that have not performed as well on an absolute basis. This naturally creates a tendency to sell gains, therefore increasing your tax liability.
Collectively, taxes and transaction costs must be considered when setting a rebalancing frequency. Taxes may not apply to all investors or all aspects of a portfolio. For example, investors with tax-deferred investments in annuities, IRAs or 401(k) plans will not have to consider tax costs when determining a rebalancing discipline. Transaction costs will always be a constant consideration when determining a rebalancing frequency.
There are five basic forms of rebalancing used by individual and institutional investors:
With the first method, no rebalancing, an investor essentially is allowing their portfolio to drift and will allow the market to rebalance the portfolio for them. As large-company stocks appreciate and bonds decline, the allocation to bonds will fall and the allocation to large-company stocks will rise.
Rebalancing on demand does not establish a set rebalancing guideline; rather, it gives the investor the freedom to make tactical decisions on when they believe it is appropriate to rebalance their portfolio. For example, in 2003, the stock market regained its bullish position with large-company stocks rising 28.7 percent1 and small-company stocks rising 47.3 percent2. After a year such as that, a natural reaction may be have been to take some of these gains off of the table and reallocate them to bonds. However, with expectations of rising interest rates in 2004 and a negative outlook for bonds, an investor rebalancing on demand may have made the tactical decision to delay rebalancing to bonds until rates had risen.
Rebalancing frequencies is the most common and most disciplined rebalancing method.An investor chooses a rate of recurrence to rebalance,such as quarterly, semiannually or annually. Regardless of market direction or expectations for the market, a portfolio is rebalanced based on a predetermined frequency.
Allocation triggers set boundaries on an asset allocation, thereby forcing a portfolio to be rebalanced when a boundary is violated. For example, a 5 percent trigger set on a 50/50 stock and bond allocation would rebalance the portfolio whenever one asset class reached 55 percent or 45 percent. While less disciplined than a set rebalancing frequency, this method allows portfolios to shift with the market and does not rebalance unless there has been a significant move.
Finally, the hybrid method combines both a rebalance frequency with an allocation trigger.An example would be setting an annual rebalance frequency with a 5 percent trigger and stating that the portfolio will be rebalanced no less than annually. However if the portfolio deviates by 5 percent from its target during the year, the portfolio will be rebalanced at that time as well.
Collectively, these methods provide investors with latitude in determining to what extent they wish to let market movements rebalance their portfolios. In addition, investors also retain the freedom of determining the discipline they wish to instill within their investment policy based on their comfort with the markets. The lower the number of occurrences to rebalance the portfolio, the lower the transaction costs and taxable effects of rebalancing will be.
At the individual level, investors should set a rebalancing frequency based on what is optimal to them in order to control risk and enhance return, while at the same time minimizing their costs (transaction costs and taxes).
Research on the subject of selecting an optimal rebalancing frequency is relatively scarce. In the research that has been conducted, an overwhelmingly consistent message throughout is the idea that it doesn't matter what your method of rebalancing is, all that is important is that you rebalance.
To test these findings, we conducted our own research on two different scenarios to determine the optimal rebalancing method and frequency. To construct our conclusions, we compared (back-tested) a number of rebalancing methods against historical investment returns. We began this comparison by first looking at a simple portfolio that was invested 60 percent in equities and 40 percent in bonds. The methods considered for this simple comparison included: not rebalancing, annual rebalancing, quarterly rebalancing and rebalancing based on a 5 percent trigger. In addition, we ran a second comparison of a more diversified allocation that included large cap, small cap and foreign stocks along with bonds. With this scenario we added two additional rebalancing methods: a 10 percent trigger and a hybrid 5 percent trigger with automatic annual rebalancing.
In comparing the effects of rebalancing a 60 percent stock/40 percent bond allocation3 from 1979 through 2003 (Figure 1), it was quite apparent that rebalancing alone reduced volatility by over 18 percent, dropping the long-term standard deviation down from 12.2 percent down to 10.3 percent. The difference in volatility (as measured by standard deviation) among the three rebalancing frequencies (annual, quarterly and a 5 percent trigger) was minimal, signaling that it is not the method of rebalancing that matters, but just the decision to rebalance that makes a difference. When comparing the returns of these methods, not rebalancing the portfolio provided a marginally better return than the various rebalancing methods.
| Return | Standard Deviation | Sharpe Ratio* | |
| Annual Rebalance | 12.39% | 10.30% | 1.20 |
| Quarterly Rebalance | 12.33% | 10.22% | 1.21 |
| 5 Percent Trigger | 12.46% | 10.28% | 1.21 |
| No Rebalance | 12.55% | 12.22% | 1.03 |
*Sharpe ratio calculated based on a simple Sharpe calculation (Return/Standard Deviation)
When considering return and risk together, we use a simplified version of the Sharpe Ratio. This ratio takes the return of the portfolio divided by its risk (standard deviation). For example, taking the annual rebalancing method, the return (12.39 percent) is divided by the standard deviation (10.30 percent) to produce a Sharpe Ratio of 1.20. This can be translated as saying for each unit of risk (standard deviation) an investor earns 1.20 percent in additional return.
Using these results on a risk-adjusted basis (i.e., Sharpe Ratio), an investor is better off rebalancing. All three rebalancing frequencies produced a Sharpe Ratio that was 16.5 percent higher than was produced without rebalancing. If you take transaction costs and taxes into consideration, an investor would be better off selecting the 5 percent rebalance trigger because it yields the smallest number of rebalancing events. Over the 25-year period analyzed, a 5 percent rebalance trigger would have created 17 fewer rebalancing events versus 100 events using quarterly rebalancing.
| 2003 | 2002 | 2001 | 2000 | 1999 | 1998 | 1997 | 1996 | 1995 | |
| Annual Rebalance | 18.9% | -9.2% | -3.8% | -0.8% | 12.3% | 20.6% | 23.9% | 15.2% | 29.9% |
| Quarterly Rebalance | 18.5% | -10.1% | -3.7% | -1.2% | 12.0% | 20.6% | 23.8% | 15.0% | 29.7% |
| 5 Percent Trigger | 18.8% | -9.5% | -3.7% | -0.9% | 12.3% | 20.8% | 24.0% | 15.3% | 30.1% |
| No Rebalance | 22.9% | -16.3% | -8.9% | -6.5% | 17.8% | 25.2% | 28.6% | 18.5% | 32.7% |
What is intriguing about this analysis is that while the pure returns for not rebalancing the portfolio are mildly better than the other options, there is a noticeable difference in the volatility of the rebalanced portfolios versus the non-rebalanced portfolio. A great deal of the decline in standard deviation comes with less volatility in down markets. The lower volatility comes from rebalancing back to bonds after stocks have run, therefore creating a better cushion in a declining stock market. Calendar year performance from 1995-2003 is listed in above (Figure 2).
In our second scenario, we took this analysis a step further and compared the performance of a diversified mix of bonds along with large companies, smaller companies, foreign stocks and cash. This allocation was a moderate, Growth and Income allocation of 40 percent fixed income, 40 percent large companies, 7 percent small companies and 13 percent foreign stocks.4 This analysis was performed for the period beginning January 1988 and ending in December 2003. In this analysis we compared several rebalancing methods: quarterly, a 5 percent trigger, a 10 percent trigger and a 5 percent trigger with an automatic annual rebalancing.
The results of this analysis are displayed in Figure 3 on the next page, with the conclusions matching those in our first comparison. The choice of rebalancing frequency makes little impact on overall returns, but the decision to employ a rebalancing discipline meaningfully diminishes the overall volatility in an investor's portfolio. More importantly, the volatility decline results from more steady performance in down markets of 29 percent. Although the risk-adjusted results, as measured by the Sharpe Ratio, show all rebalancing methods to be similar, the use of a 5 percent trigger appears to be the most optimal in our opinion. After considering transaction costs and taxes, a strong argument could be made in favor of a 5 percent or 10 percent rebalancing trigger, due to the minor number of rebalancing events each incurred. As a means of better containing risk within the portfolio, we recommend a 5 percent trigger, as it will keep the portfolio from gaining a posture that is too aggressive or too conservative.
| # of Rebalancing Events | Return | Standard Deviation | Sharpe Ratio | Average Return in Up Markets | Average Return in Down Markets | |
| Annual | 16 | 11.19% | 4.54% | 2.47 | 4.76% | -3.29% |
| Quarterly | 63 | 11.34% | 4.60% | 2.46 | 4.81% | -3.30% |
| 5 Percent Trigger | 7 | 11.44% | 4.61% | 2.48 | 4.84% | -3.31% |
| 5 Percent Trigger w/ Annual | 18 | 11.41% | 4.60% | 2.48 | 4.83% | -3.29% |
| 10 Percent Trigger | 3 | 11.40% | 4.65% | 2.45 | 4.85% | -3.36% |
| No Rebalancing | 0 | 11.56% | 5.44% | 2.12 | 5.18% | -4.65% |
We also looked at the sensitivity of these results to changes in the initial allocation. As the asset allocation was tweaked to be more conservative or more aggressive, the results remained the same.
Other industry research conducted on the topic of optimal rebalancing suggests similar conclusions to our analysis. In an October 2001 issue of the Journal of Financial Planning, Cindy Sin-Yi Tsai makes the conclusion that "Once an advisor and client have made the decision to rebalance, it does not matter much which strategy they adopt."6 An online article written by William Bernstein provides a case study on rebalancing. In this analysis, Bernstein compares rebalancing methods monthly, quarterly, annually, two years and four years.7 Bernstein concludes that monthly rebalancing is too frequent and that there are incremental benefits in return for stretching out a rebalancing frequency; however, the cost of doing so is increasing the risk profile of the portfolio. 6Sin-Yi Tsai, Cindy. "Rebalancing Diversified Portfolios of Various Risk Profiles." Journal of Financial Planning. October 2001. 7Bernstein,William J. "Case Studies in Rebalancing." www.efficientfrontier.com November 2002. Returns calculated in this analysis do not reflect transaction costs and taxes. These costs are specific to a client's individual financial situation and cannot be accurately estimated. Transaction costs could occur as an up-front load, back-end load, quarterly fee or a per transaction cost. Taxes vary based on the individual's federal, state and local tax rates. Investors should recognize that transaction costs and taxes would lower the returns shown in this analysis.
The analysis conducted for this research coincides with other industry research suggesting that the method and frequency of rebalancing is insignificant, as long as an investor employs a rebalancing discipline. Contrary to popular belief, rebalancing does not appear to enhanceinvestment returns. However, it has proven to significantly reduce volatility. This reduction in volatility has tended to occur in down markets.
We believe the most optimal rebalancing discipline, after considering taxes and transaction costs, is a 5 percent trigger. Based on our study this method has proven to perform in line with other methods and its smaller number of rebalancing instances (as compared with a quarterly or annual rebalancing method) will significantly reduce transaction costs and taxes. When considering taxes, a quarterly or annual rebalancing method is more apt to recognize short-term gains, a more costly taxable gain, than a 5 percent rebalancing trigger,which is less likely to recognize short-term gains. The negative side of a 5 percent rebalancing frequency is that it is not as static or automated and places greater responsibility on the investor and their financial advisor to periodically review their asset allocation to ensure a trigger has not been violated.
While we maintain an opinion on an optimal rebalancing frequency, it is important to recognize that each individual investor is different. Regardless of our opinion, what matters most is that investors rebalance, with their rebalancing method being only a secondary consideration.
1As measured by the S&P 500. Source: Zephyr Associates.
2As measured by the Russell 2000. Source: Zephyr Associates.
3Used the S&P 500 as a proxy for equities and the Lehman Aggregate Bond index as a proxy for bonds.
4Data from the Russell 1000 index was used to simulate the performance of large-company stocks. The Russell 2000 was used to simulate results of small-company stocks, with the MSCI EAFE index being used to simulate the results of foreign stocks. The Lehman Aggregate Bond index was used as a proxy for performance of bonds.
5Assuming that volatility decreased from 4.65 percent with not rebalancing to an average of 3.30 percent with rebalancing. This decline of 1.35 percent accounts for a 29 percent decrease in volatility as measured by standard deviation.
6Sin-Yi Tsai, Cindy."Rebalancing Diversified Portfolios of Various Risk Profiles." Journal of Financial Planning. October 2001.
7Bernstein,William J. "Case Studies in Rebalancing." www.efficientfrontier.com November 2002.
Returns calculated in this analysis do not reflect transaction costs and taxes. These costs are specific to a client'sindividual financial situation and cannot be accurately estimated. Transaction costs could occur as an up-front load, back-end load, quarterly fee or a per transaction cost. Taxes vary based on the individual's federal, state and local tax rates. Investors should recognize that transaction costs and taxes would lower the returns shown in this analysis.
Get registered and learn more
