Page 1

ssrn.com/abstract=1798386

Portfolio Rebalancing Revisited: What Strategy Optimally Triggers an

Adjustment to the Asset Weights?

- A New Story in Hong Kong Market

Hui Hong 1

Accounting and Finance Department. University of Limerick, Ireland

ABSTRUCT

This paper examines the robustness of the evidence on rebalancing effects within Hong Kong

context and addresses the issue of whether there is an optimal rebalancing strategy that can be

historically exploited by wealth managers to earn constant profits in excess of a buy-and-hold

strategy in the market index. It finds that although superior to buy-and-hold, no single

rebalancing strategy dominates one another under all conditions. The general rule is that

patient rather than quick-trigger rebalancing benefits portfolio management. In particular, the

superiority of rebalancing mainly comes from market conditions.

KEYWORDS: Asset Allocation, Rebalancing, Strategic Target

Successful investment always starts with a sound asset allocation plan which complies with

the client’s long term objectives and constraints2. The importance of asset allocation is

indicated in the seminal study of Brinson, Hood and Beebower (1986) which shows that

among three fundamental determinants3 of portfolio performance, asset allocation accounts

for over 90% of return variations. The similar conclusion is drawn by Blake, Lehmann and

Timmermann (1999), using the UK market data.

However asset allocation is not a once-for-all practice, instead, a continuously dynamic

process. As the paper by Smith Barney Citigroup4 (2005) describes, ‘letting a portfolio

untended is like leaving a toddler alone in a room with a hot stove – the outcome depends far

too much on the forces of chance.’ Any variations of market conditions would make asset

weights stray from the initial target. Therefore on one hand, an investor might hold

overpriced securities with potentially inferior future returns and deviations from his

investment constraints. On the other, the diversification benefits might be eroded as the initial

1 Accounting and Finance Department, University of Limerick, Ireland. Tel: +353-87-6256535. Email:

hui.hong@ul.ie

2 In general, the client’s objectives concern return requirement and risk tolerance; the constraints correspond to

liquidity, credit, horizon, taxes and unique needs.

3 These are asset allocation, market timing and security selection. Specially, asset allocation concerns the

determination of asset components and their relative weights in the portfolio; market timing is about under- or

over-weighting asset classes relative to the benchmark portfolio; and security selection is the active selection

within an asset class.

4 It is a report ‘the art of rebalancing – how to tell when your portfolio needs a tune-up’ by the consulting group

Smith Barney 2005.

Page 2

ssrn.com/abstract=1798386

asset weights are changed due to market forces. An example of U.S. stock market in the

1990s evidences the danger of letting a portfolio to its own device. The meteoric price

increases led to a great proportion of portfolio value tilted toward large cap stocks. With the

continuation of upward market trend, the portfolio was equity-concentrated and fragile to any

price volatility. A substantial loss was incurred to investors’ wealth when the market roller

coasters occurred in 2000. Yet the negative impact might have been mitigated by applying

appropriate asset allocation strategy when the stock market soared.

This paper aims to provide a theoretical framework for developing an appropriate asset

allocation strategy, i.e. rebalancing strategy within Hong Kong context. This is done by

empirical comparisons among different rebalancing strategies proposed in the literature,

examining what strategy optimally triggers an adjustment to the asset weights. In particular,

rebalancing is ‘the process of buying and selling portion of the portfolio in order to reset each

asset backs its original asset allocation policy’ (O’Brien 2006).

For the purpose of the analysis, the paper focuses on the performance of rebalancing

strategies starting with 60/40 stock/bond portfolio. In addition, it also considers the sensitivity

of the results to different market conditions and investment horizons.

In contrast to previous research, this paper reinforces the literature by two approaches: First,

it compares portfolio performance before and after transaction costs to examine how

rebalancing effects would be impacted, which more explicitly explores the importance of

transaction costs in the formation of optimal trading strategy. Second, it attempts to make

comparisons among a range of rebalancing strategies and therefore is able to examine

rebalancing issue to a greater extent.

The structure of the paper proceeds as follows: Section 1 reviews related studies in the

portfolio rebalancing area. A detailed description of data and methodology is presented in

Page 3

prevailing currents would end up with increasing risk. Rebalancing contributes to the

balanced risk, compatible with the client’s objectives and constraints.

The pioneering study in asset allocation is Perold and Sharpe (1988). They propose four

fundamental strategies5: buy-and-hold, constant-mix, constant-proportion and option-based

portfolio insurance (CPPI). Different strategies prevail in different market conditions. The

relevant mechanism to this paper is constant-mix, which requires purchase (sale) of assets as

they depreciate (appreciate) and sale (purchase) of other assets to ensure the asset weights in

accordance with the target. It will end up ahead relative to buy-and-hold in a flat (but

oscillating) market condition.

The following sections focus on reviewing topics in the literature on rebalancing, the

development from constant-mix in Perold and Sharpe (1988).

1.1.

Asset Characteristics

In general, asset characteristics include four basic components (Pliska and Suzuki 2004):

correlation, volatility, expected return and time horizon. Their deterministic effects on

rebalancing are explained as follows:

1) Correlation: Higher positive correlations among assets indicate higher co-movements.

Therefore, the higher the correlations, the less frequently will asset weights drift away

4) Time Horizon: A longer time horizon leads to greater likelihood that portfolio weights

deviate from the target and increases the need for rebalancing.

Similar attitudes are from Tasi (2001), Buetow (2002), Liu (2002) and Zakamouline (2002).

1.2.

Asset Weights

In contrast to asset characteristics, asset weights also have non-trivial impact on rebalancing

effects. The more one chooses to shift to risky assets, the less return enhancement that

rebalancing would bring (Plaxco 2001). By using an 80% equity benchmark, the author finds

that the return of the initial 50% equity benchmark has been far exceeded under the same

5 Buy-and-hold is similar to sitting back and drifting with the market tide. Both constant-proportion and option-

based portfolio insurance (CPPI) strategies focus on selling (purchasing) assets as they are temporarily out of

favour (in favour) and purchasing (selling) other assets to comply with the client’s objectives. In a trending

market condition, the continual purchase of assets with appreciation in value facilitate return enhancement while

the constant sale of assets out of favour prevents return deterioration.

Page 4

rebalancing strategy. Therefore the compensation for the higher risk associated with the

portfolio dwarves any value-added that changing the rebalancing strategy might provide.

Tasi (2001) proposes that higher percentage equity portfolio does not benefit as much from

rebalancing as do other low proportion equity portfolios. The reason might be attributed to

the likelihood that assets tend to move synchronously with each other, which results in the

least benefits from rebalancing especially when the equity market is declining.

Rebalancing Triggers and Targets

Rebalancing triggers and targets are another two factors that contribute to rebalancing

effects. In general, there are three categorizations of rebalancing triggers in the portfolio

management area (Riepe and Swerbenski 2007):

1) Time-based Triggers: The triggers force rebalancing to take place at specific time

intervals regardless of the extent to which current asset allocation strays from the

strategic target.

2) Threshold-based Triggers: Rebalancing is initiated when any asset weight exceeds the

threshold pre-defined which allows the deviation from the target.

3) Tolerance-based Triggers: The portfolio is reviewed at given time intervals while

rebalancing is only needed when asset allocation deviates from the target by a degree

above the stated tolerance level.

There is no consensus on the superiority of these rebalancing triggers. It is recognized that

periodic rebalancing (time-based triggers) is straightforward however it may not well time the

large market movements as it is independent of market conditions. By contrast, percentage

range rebalancing (threshold-based triggers) is easy to implement however there is difficulty

to set an effective percentage range (Donohune and Yip 2003). Both rebalancing strategies

have a significant effect on investment results (Smith and Desormeau 2006). However,

percentage range rebalancing is not as effective as the periodic one (Arnott and Lovel 1993).

Page 5

1) Back to Strategic Target: All assets are brought back to its long-term strategic weights

when rebalancing is triggered.

2) Back to Rebalance Band: Rebalance band is denoted as a percentage of the target

asset weights. Rebalancing is activated when any asset goes beyond the rebalance

band and allocation is restored to the edge of the band.

3) Back to Tolerance Band: Assets are rebalanced to the tolerance band which is a

proportion of rebalance band only if any asset return rises beyond the rebalance band.

Further illustration is provided in Exhibit one.

Exhibit 1 Rebalancing Targets

BH is the upper edge of rebalance band;

PB is tolerance band;

BL is the lower edge of rebalance band.

Source: Daryanani, G. (2008)

Rebalancing Costs 1.4.

It is also important to be aware that rebalancing is associated with costs, such as trading and

tax costs (Brunei 1999, Lynch and Balduzzi 2000, Horvitz 2002, and Tokat and Wicas 2007).

costs are proportional to the trade size, rebalancing to the tolerance band is a better choice as

it minimizes the transaction size. When both of the costs are dictated, the moderate strategy is

to restore to some intermediate point (Zakamouline 2002).