The term "portfolio rebalancing" refers to the process of returning a portfolio to the desired asset allocation mix.
While the strategic asset allocation benchmarks adopted by pension plans tend to remain unchanged over long periods of time (typically three to 10 years), the asset mix of their actual portfolios fluctuates daily, reflecting the differential returns of each asset class. There will be a loss of efficiency when the portfolio allocation strays from the desired mix, or exceeds the plan trustee's risk tolerance. Rebalancing, therefore, is necessary to return a portfolio to its optimal configuration.
The question of when and how rebalancing should be conducted, however, remains one of the most controversial issues in the area of asset allocation.
This article seeks to analyze the importance of portfolio rebalancing and the effectiveness of various rebalancing strategies.
REBALANCING'S ROLE
An asset allocation process can be divided into three distinct stages:
* Strategic asset allocation;
* Tactical asset allocation; and
* Portfolio rebalancing.
Effective asset allocation requires an integration of all three stages. Suboptimal portfolios result when one or more of the stages are ignored or are applied ineffectively.
Strategic asset allocation is the process of deriving an efficient long-term asset allocation mix, which is commonly called the strategic benchmark.
It is based on the analysis of the expected risk-return of asset classes, as well as a plan sponsor's asset-liability configurations over three or more years. The process seeks to determine a point on the efficient frontier that optimizes the plan's risk-adjusted return subject to the constraint of the plan sponsor's risk tolerance.
Because it is focused on long-term market returns, it cannot take into consideration asset classes' short-term deviations from their long-term trends. Therefore, the strategic benchmark by itself does not offer a comprehensive or optimal solution to asset allocation.
Tactical asset allocation is a process whereby short-term deviations from long-term strategic benchmarks are seen as offering opportunities to add value through a specific investment process. It focuses on exploiting asset classes' tendencies to overshoot and undershoot their long-term trends in the short run. The objective is to generate returns in addition to those of the strategic portfolio.
Deviations of portfolio asset mix from the strategic benchmark could occur as a result of tactical investment decisions or the passage of time.
While the former case represents a conscious effort to add value based on risk-reward considerations and is under the control of selected managers, the latter case represents an unconscious deviation from the optimal allocation that is outside the control of investment managers. Significant loss of efficiency could result when this deviation exceeds the risk tolerance of the plan sponsor.
The key objective of portfolio rebalancing is to minimize the loss of efficiency resulting from this undesired deviation of the asset allocation. Rebalancing, therefore, forms an indispensable part of the asset allocation process and the decisions on when and how to rebalance are critical in retaining the integrity of the strategic benchmark.
Despite its significance, portfolio rebalancing often only occurs under one of the following circumstances:
* When there is a significant cash injection or withdrawal;
* As a coincidence, when a revision to the strategic benchmark takes place;
* When there is a decision to cut losses or take profits after a substantial decline or rally in the value of an asset class; and
* At the beginning or the end of an accounting period selected by the sponsor.
Unfortunately, while some of the above events might make operational sense, none constitutes an efficient or optimal rebalancing strategy.
Over recent years, trustees and plan sponsors have become increasingly aware of the significance of portfolio rebalancing. As a result, there has been a more intensive search for an optimal rebalancing strategy.
OPTIMAL STRATEGY
Doug McCalla of the San Diego City Employees' Retirement System has produced an interesting and rather comprehensive analysis on the "conventional methods" of rebalancing. These include calendar-period rebalancing rules and percentage-of-portfolio rebalancing methods. Mr. McCalla's findings, which were consistent with earlier work done by Mark A. Hurrell of Pittsburgh-based Yanni-Bilkey Investment Consulting, suggest that:
* Various methods of rebalancing a portfolio included in the test produced superior returns compared to a buy-and-hold strategy.
* Of the calendar-period-based rebalancing rules, quarterly rebalancing produced the best enhancement to return.
* Compared to calendar-period methods, percentage-of-portfolio rebalancing methods usually produced superior enhancements to total portfolio return; and
* The optimal return enhancement was associated with a wide rebalancing range such as +/-10%
In addition, Mr. McCalla, in his paper "Breaking Through the Efficient Frontiers with Portfolio Rebalancing," proposed a new and theoretically more robust rebalancing approach -- the volatility-based approach. He pointed out that because volatility of asset classes varies, percentage-based rebalancing thresholds "tend not to be the most efficient" because they imply "unequal probability of occurrence." That is to say, given the same percentage band, asset classes with a higher level of volatility tend to trigger rebalancing more frequently than those with a lower level of volatility. Theoretically, this is inferior relative to a "normalized" approach such as the volatility-based approach in which each asset class has "an equal probability of benefiting from the opportunity to be sold high and purchased low."
Mr. McCalla performed a statistical test on the efficiency of the volatility-based approach against the other approaches. (A rebalancing is considered to be efficient if it adds to the risk-adjusted return of the portfolio.) His findings indicate that:
* All approaches to rebalancing included in the test resulted in enhanced portfolio returns, net of transaction costs.
* All approaches included in the test reduce the portfolio's volatility compared to that experienced by the buy-and-hold strategy.
* "The narrower equal-probability constraints on the allocation ranges (i.e. volatility-based methods) had a greater potential to enhance the gains associated with rebalancing."
* The optimal, volatility-based, threshold in his test ranges from +2.6 to +3.6 standard deviations.
Mr. McCalla pointed out, however, that, "No one rebalancing rule will be the optimal approach for all asset structures." The selection of the rebalancing trigger point should be optimized with regard to the correlation between asset class returns, their volatility and their relative weighting in the portfolio.
We have replicated the statistical test on those rebalancing methods. Our results confirm Mr. McCalla's point that "no one optimal rebalancing rule is appropriate for all asset mixes."
But our test has gone one step further, and shown that even if only one single portfolio or one asset mix is concerned, there still is no single "winning" rebalancing rule. Different rebalancing rules win in different market scenarios. There is no inherent merit to any single rebalancing rule that enables it to generate consistent return enhancement across different market conditions. There is, in fact, no theoretical basis to expect such a rule exists.
Our fundamental belief is that effective investment decisions could not be achieved without intensive analysis of market risk-reward conditions. This applies equally to all asset allocation decisions, including rebalancing decisions. In addition, we believe mechanical rules tend to be inferior to active management of rebalancing decisions for the reason that they only allow the manager to control one of the two key factors determining the efficiency of portfolio rebalancing -- frequency and timing.
Since, by definition, the strategic benchmark is the optimized asset mix, given the plan sponsor's specific risk-reward preference, any unintended deviation from it will necessarily result in inefficiency. (That is, deviations caused by factors other than TAA decisions).
KEY FACTORS
The purpose of rebalancing is to minimize such inefficiency. Theoretically, in the absence of transaction costs, rebalancing should be conducted as often as possible. However, when we take transaction costs into account, finding the optimal frequency becomes cost-benefit analysis. The statistical tests listed earlier essentially are efforts to identify the optimal frequency based on a calculation of the marginal gain and marginal cost. The marginal gain, the expected reduction in inefficiency, tends to diminish as the frequency of rebalancing increases. The marginal cost, the transaction cost of rebalancing trades, tends to remain flat regardless of the frequency. The optimal frequency is the point at which the marginal gain equals the marginal cost. Nevertheless, we believe frequency is neither the only nor the most critical factor determining the efficiency of rebalancing.
Generally, in a low-volatility trending market, portfolios with frequent rebalancing tend to underperform those with infrequent or no rebalancing. This can best be illustrated by an example portfolio of two asset classes -- treasury bills and equities. As stock prices go down, rebalancing requires the purchase of stocks, and vice versa. If stock prices enter a sustained downtrend, rebalancing would accelerate portfolio capital losses by calling for repeated shifts of capital out of bills into stocks. However, as stock prices go up in a steady trend, rebalancing trades would slow or reduce the portfolio capital gain because they require repeated sales of stocks as their prices appreciate.
By contrast, in a volatile, range-bound market environment, portfolios subject to frequent rebalancing tend to outperform those with infrequent or no rebalancing. Under such market conditions, rebalancing achieves the same effect as mean-reversion trades, i.e. "buy low and sell high." This could significantly enhance the portfolio return even if the market return over the period turns out to be flat.
Therefore, whether the chosen frequency for rebalancing would result in a gain or loss is critically dependent on market condition. Unfortunately, expected market behavior is not an input into the mechanical rebalancing rules.
For this reason, we believe an argument can be made that rebalancing decisions should be actively managed based upon a forward-looking analysis of market risk and reward conditions. Indeed, for practical purposes, it is difficult to separate this process from the TAA function.
Active management of rebalancing timing is considered desirable because unintended deviations from the strategic benchmark (caused by factors other than TAA decisions) usually occur as a result of factors that have not been, or could not be, taken into account by the optimization exercise that generates the strategic benchmark in the first place. The plan sponsor could choose to ignore such information and focus on the strategic benchmark by adopting a mechanical rebalancing rule. A more pro-active approach, however, is to appoint a specialist manager with the appropriate skills to leverage such information to achieve more effective timing of rebalancing trades. We believe that, given their focus on market timing, TAA managers should have the most suitable skills to carry out the rebalancing task.
Rebalancing could have significant impact on the volatility of portfolio returns. Any return enhancement achieved by rebalancing, therefore, has to be measured in terms of risk-adjusted return.
CONCLUSION
Effective asset allocation requires efficient implementation of:
* The strategic asset allocation process;
* The tactical asset allocation process; and
* A portfolio rebalancing process.
While the strategic and tactical processes focus on the derivation of the optimal asset allocation mix on long and short time frames, the rebalancing process aims to minimize the efficiency loss resulting from the deviation of asset allocation from its pre-determined optimal mix.
The results of our study confirm the application of rebalancing rules can result in significant return enhancement. However, our results also indicate the return enhancement achieved by these mechanical rebalancing rules depends on the specific market conditions. In a volatile range-bound environment, rules requiring more frequent rebalancing tend to add more value. In a steady trending environment, however, rules requiring less frequent rebalancing tend to do better. Expected market conditions dictate which mechanical method should be chosen as optimal.
This is not to say selection of a particular mechanical rebalancing method is inappropriate. Selection achieves a key objective of managing the risk profile of the portfolio to avoid a breach of the risk tolerance limit of the plan sponsor.
A middle ground might be to combine the risk profile limitation of the sponsor with the skill set of a manager reasonably able to determine expected market conditions. The rebalancing trigger point could be precisely set to reflect the maximum risk tolerance of the plan sponsor. A pension plan with relatively high risk tolerance, for instance, could set a wide percentage or volatility band around the strategic benchmark as the trigger points for rebalancing. A conservative plan with relatively low risk tolerance would, by contrast, opt for a narrower band, resulting in more frequent rebalancing. The rebalancing would be triggered automatically when the plan sponsor's specified risk limit is reached.
The TAA manager would be required to optimize the frequency and timing of rebalancing subject to the constraint of these bands. The performance of the TAA manager could be measured under two methods. First, all investment decisions made by the TAA manager could be classified into long-term rebalancing value added and short-term tactical asset value added. The performance of each category being measured separately and summed to a total added plan value.
Alternatively, deviations from the strategic benchmark, whether driven by the differential performance of underlying asset classes or tactical decisions, are considered conscious moves by the TAA manager to add value. Any addition or deduction of return resulting from such deviations are counted as the TAA manager's performance.
