In this article published in The IRS Report in August 2010, Chris Gilchrist talks about rebalancing your portfolio

Rebalance your portfolio - but not too often

Practitioners of asset allocation investment methods usually assume that after an appropriate portfolio has been created, it will be periodically rebalanced. The argument is that there is little point in determining what the "right" asset allocation is to match your risk-return objectives - or "risk tolerance" - and then allowing the portfolio to drift away from this. If you had decided that 70% in equities was right for you, then allowing that proportion to rise to 85% over a two-year period just because there was raging bull market in equities would represent an acceptance of a much higher degree of risk than you originally considered desirable.

Academics have for many years debated the optimal strategy to use when rebalancing, but frequency is only one possible factor to use. You could also set a threshold, requiring that the allocation to any one asset class had to deviate by 5%, 10% or some other figure from its original proportion to trigger the rebalancing exercise.

A recent paper by US mutual fund giant Vanguard has worked through the numbers, providing some usefull guidelines for investors using asset allocation methods. Though it is based on historic US returns for stocks and bonds, there is no reason to think the results would be significantly different if UK data was used.

Automatically rebalancing at regular intervals is fine in theory, but as the table shows, the number of transactions becomes very large. This is only a two-stock portfolio, and rebalancing for portfolios with many more holdings would become ruinously expensive. But what is interesting is that the net equity exposure of the portfolio is not that different using annual (60.5%) as opposed to monthly (60.1%) rebalancing. As theory predicts, all the rebalanced portfolios produce a lower return than the unchanged portfolio, but with a reduction in volatility much larger than the reduction in return - they are delivering more bang for the buck.

The alternative to automatic regular rebalancing is using a threshold, but since you need to monitor a portfolio daily, this is not a practicable option for most, and Vanguard instead tests a time-plus threshold approach. Here you review the portfolio - monthly, quarterly or annually - and only rebalance if the allocation has deviated - by 1%, 5% or 10%.

The results show that applying a 10% threshold quarterly or annually allows equity allocation to average at a higher level (63%) than with more frequent rebalancing, which produces a modest increase in annualised return without greater volatility.

So the research seems to show that irregular rebalancing - half-yearly or yearly - will avoid your portfolio getting out of whack, but that to cut balancing transactions to the minimum, the best way is to add a threshold filter as well based on 5% or 10% deviation form the original allocation.

So far so good, at the macro level. Now, how about actually implementing this? After all, you won't hold a two-stock portfolio. Say you hold a portfolio of ten funds in three major asset classes - shares, bonds, property - and several sub-classes - UK shares, international developed market shares, international emerging market shares, corporate bonds, government bonds. If you aim to rebalance the major asset classes, you have to do so by adjusting the sub-classes. So, if you had to increase your equity allocation - as in 2008 - you would have to choose to top up all your equity funds or to allocate the top-up cash to one or two of them - a process some would call tactical asset allocation.

Obviously one motive will be to minimise transaction costs. Unless, that is, you hold a fund portfolio in a "wrap" account where rebalancing costs can be minimal - as little as £2 per holding. Here, the process adopted by some financial advisors is to automatically rebalance the lot. If the costs really are that low, then this is fine. But for portfolios containing actual equities, or where costs will be a more realistic 1% - the bid-ask spread - plus £30 per pair of transactions, then rebalancing through just six transactions - the minimum for a portfolio with three asset classes - is enough of a cost hit that you won't want to incur it that often.

There is another way, which is to rebalance from a cash account, into which all the portfolio income is paid. You might well want to hold a regular float there too. By the time you need to rebalance, you hope the portfolio has generated enough income to do most of the heavy lifting.

Vanguard's exercise is limited in scope, but enables us to set a useful strategic guideline - annual or half-yearly rebalancing with a threshold filter. But I suspect the big gains from doing this will come from the tactical choices of what to buy, at a time when clearly at least one asset class is likely to be offering plentiful bargains.

You can see all of Chris Gilchrist's articles at www.TheIRSReport.com

Chris is the editor of The IRS Report every month

Call 0800 756 5437 or click here for more information