Rebalancing a portfolio requires restoring asset allocations to their chosen percentages. Some investors do this by selling off part of the rising asset and buying more of the falling asset. Others may use new cash injections to bring the losers up to their desired relative weights.

It is said rebalancing controls for risk. If your chosen allocation to equities is 60 per cent and a bull market pulls it up to 80 per cent, youll be going into the next bear market with too much risk. Vice versa, if a bear market pulls it down to 40 per cent youll be going into the recovery with too little risk.

Rebalancing is also said to compel buying low and selling high. Your equity allocation usually falls below the chosen allocation during bear markets, compelling you to buy stocks in the midst of pessimism. It usually rises above the chosen level during bull markets, compelling you to sell stock in the midst of optimism.

Many people think rebalancing is done just once a year. There are actually many other possibilities and studies have shown they can generate better returns. Lets review main alternatives (what follows is a slightly modified version of an article I wrote a while back).

1. In their study published in the November, 2006 issue of the Journal of Financial Planning, David Smith and William Desormeau tried out different rebalancing rules on various U.S. bond-stock model portfolios over the 1926 to 2003 period. They found that rebalancing every three to four years outperformed monthly, quarterly and annual rebalancing.

2. John Bogle does not rebalance his portfolio. A few years ago, the Bogle Financial Markets Research Center found that annual rebalancing of a stock-bond portfolio in all the 25-year periods from 1826 onwards only outperformed an unadjusted portfolio 52 per cent of the time. This would seem to suggest it doesnt matter either way if one rebalances or not. However, Bogle discovered that when a portfolio was never rebalanced, the annual underperformance was never more than 0.5 percentage points but when it outperformed, the extra gains were often 2 to 3 percentage points annually.

3. Why might rebalancing less frequently than every year yield better performance (every 3 to 4 years versus annually or less, as found in the David Smith and William Desormeau study)? The likely explanation, at least in part, could be the tendency for financial markets to trend in the short term before mean reversion sets in. In other words, if market prices are characterized by momentum, it does not pay to shift often out of winning into losing investments. That increases the weight of the declining positions and lowers the weight of the appreciating positions, resulting in lower returns overall.

4. The threshold approach doesnt rebalance according to the calendar. It calls for adjusting weights only when assets diverge from the target allocation by some specified amount. A common variant is the 5/25 method, as described by Larry Swedroe in his book, The Only Guide to a Winning Investment Strategy Youll Ever Need. It calls for rebalancing if the change in an asset classs allocation is greater than either an absolute five percent or 25 percent of the original percentage allocation, whichever is less. For example, the threshold bands for a position with a 10-per-cent allocation would be 5 to 15 per cent or 7.5 to 12.5 per cent with a preference for the latter since it is smaller.

5. Then there is tactical rebalancing. Beside uncovering evidence that infrequent calendar rebalancing performed better, the Smith and Desormeau study mentioned above found that Federal Reserve monetary policy had a discernible impact on rebalancing returns. In fact, they found that the best rebalancing policy is dependent on, and can be planned around, the Feds prevailing monetary policy. In short, rebalancing every three to four years was better than rebalancing annually, and furthermore, synchronizing the three- to four-year frequency to the monetary cycle was better than doing it without reference to the monetary cycle.

6. Another rebalancing strategy with the potential to deliver extra returns is offered by Warren MacKenzie, president of Weigh House Investor Services. Its based on a price-to-earnings (P/E) valuation rule and goes as follows: say the preferred allocation to equities is 50 per cent and the P/E ratio for the stock market is higher than 25. Then shift to a lower allocation such as 40 per cent. If he market P/E is below 12, then shift to 60-per-cent equity.

7. Gobind Daryanani advocates opportunistic rebalancing in the January, 2008 edition of the Journal of Financial Planning. It involves closely monitoring asset allocations to see if they move outside threshold bands. He found the width of the bands and frequency of checking had an impact on returns. Specifically, monitoring a threshold band of 20 per cent around the chosen allocation on a bi-weekly basis constituted the optimal strategy in terms of capturing opportunities to buy low and sell high (i.e. increasing returns to rebalancing). The margin was reduced somewhat after factoring in all costs, taxes, and risk, but still positive.

8. Rebalancing has its fat tail too a situation where portfolio ruin occurs. A current illustration is the 25-year downward trend in Japanese equities. Rebalancing on a long-term declining trend can wreck havoc with a portfolio. As lengthy downtrends tend to happen after periods of mania and euphoria, investors should perhaps be wary of starting a rebalancing program in such environments (indeed to having any significant exposure).

9. Some rebalance by selling off part of the winners and buying more units in the losers, while others rebalance by funneling cash injections (and dividends, interest income) into the losers. Practitioners of the latter approach are keeping costs down by avoiding the commissions arising from selling the winning position. But they are letting, some say, the winning position go further into possibly overvalued territory, which increases the risk of experiencing asharp correction (buying low but not selling high).

10. Shorter investment periods and lower volatility reduce the need to rebalance, say Stanley Pliska and Kiyoshi Suzuki in their study published April, 2004 in Quantitative Finance. They also point out that if there is a high degree of correlation between holdings in a portfolio, no or little rebalancing may be necessary (if asset prices largely move in unison, the relative weights of the assets dont drift much)

Wrap-up: Many studies have examined different ways to rebalance and found that returns can be maximized in certain ways. While instructive, these results may be dependent upon the historical series used. Moreover, the less visible aspect of risk should not be overlooked. Some approaches may raise returns but the trade-off can be higher risk levels. Indeed, for many investors, rebalancing is not about boosting returns but controlling risk i.e. maintaining their portfolio allocations reasonably close to their risk tolerances. They are willing to accept the possibility of lower average returns for the sake of minimizing exposure to setbacks exceeding their risk tolerances. But then again, others may prefer going for higher returns.

Postscript: Just to further illustrate the diversity of rebalancing methods, the rule in the One-Minute Portfolio is to raise equity exposure when a moving average of the market index is below the average annual historical return on equities and lower exposure to equities when it is above.