Library

The myth of market timing

By Bob Tattersall

TORONTO (GlobeinvestorGOLD) – Many investors believe that it is a portfolio manager's job to be in the stock market when it is going up and out of the market when it is going down. As an afterthought, they usually concede that 100 per cent predictive accuracy is too much to expect and they would be content if they could be out of the market within 10 per cent of its peak and back in again within 10 per cent of the low.

In investment terminology, this is known as "market timing," and the rewards for successful practioners are thought to be immense. We know that stockbrokers and writers of investment newsletters favour market timing because it creates headlines and generates trading activity, but does it really pay off for the average investor? An article which I wrote in the mid 1970s suggested that the rewards were surprisingly modest and could be quickly eroded by transaction costs, taxes and inaccurate forecasts. My conclusion was that a market timer had to be right 75 per cent of the time in order to beat a passive buy and hold investor.

There was little interest in market timing during the 1990s as the declines in 1992, 1994 and 1998 were each less than 2 per cent for the full year. This was an era when it made sense not just to buy and hold, but to hold and buy more whenever the market dipped. Stock market volatility of the past two years has renewed interest in market timing, however, so our assistant portfolio manager, Scott Carscallen, updated the old study through 2001. Our conclusions are not very different: if your level of predictive accuracy is less than 70 per cent, only your broker and the tax collector will thank you for trading in and out of the market.

Let us assume that a market-timing strategist starts each year from 1950 to 2001 with a prediction as to whether the Canadian equity market or three-month Treasury bills will provide the greatest total return over the next 12 months. He or she then places the entire portfolio in the asset with the highest predicted return and waits a full year before making another forecast. If we assume that our perfect forecaster has started out with cash, he or she would have had to buy or sell the entire stock portfolio 30 times during this 52-year period. For the purpose of this analysis, the transaction cost to buy or sell the equity portfolio is set at 2 per cent, although this may be on the low side for a large institutional portfolio.

The following table summarizes the rate of return achieved by the perfect market timer (after transaction costs) over the 52-year period and compares the record with a “buy and hold” equity strategy.

For the full period under review, the final column indicates that a perfect track record of predictive accuracy would have yielded a 4 per cent incremental return over and above a buy-and-hold equity strategy. On a large fund, this would represent a sizeable benefit, but it may be useful to recall the original assumptions at this point:

1. Perfect forecasts for 52 consecutive years.
2. A portfolio which is 100 per cent stocks or 100 per cent treasury bills depending on the forecast.
3. Position held for a full calendar year before re-evaluating.
4. No taxes on investment income or realized capital gains.

It is difficult to imagine a portfolio manager, or even a client, accepting the first three assumptions as a realistic description of their investment management style. Most investors who pursue a market timing strategy recognize that they have less than perfect forecasting ability and so normally consider a 30- to 40-per cent cash position to be extremely defensive. As a result, they tend to suffer from the worst of both worlds: if they are right and the market declines, there is still 60- to 70-per cent of the fund exposed to a decline and, when the market recovers, less than 100 per cent of the fund participates.

Market timing strategies rarely perform better than a buy-and-hold portfolio simply because the stock market typically outperforms treasury bills over the long haul. In 64 per cent of the years covered by this study, treasury bills either lagged the stock market or failed to exceed it by the 2 per cent that would be necessary for a market timer to recoup his transaction costs. As a result, a buy-and-hold investor automatically has a 64 per cent predictive accuracy record.

In essence, this type of investor assumes that every year will be a good one for the stock market and 64 per cent of the time he is right. The market timer, therefore, must demonstrate even greater accuracy if he is to beat the buy-and-hold strategy and recoup his trading costs.

What level of accuracy is required under these circumstances? Based on the period under review, the market timer has to be right almost 70 per cent of the time if he is to equal a buy-and-hold strategy. Even higher levels of forecasting accuracy are required if any incremental return is to be earned.

The assumption regarding a zero tax rate on investment income and realized capital gains is entirely appropriate for a corporate pension fund or other tax-exempt portfolio. However, most enthusiasts of market timing tend to be individuals in a taxable position, so they have additional factors to consider. The buy-and-hold investor receives only dividend income and realizes capital gains infrequently as a result of takeovers etc. The market timer, in contrast, receives interest income for 20 of the 52 years and realizes a capital gain every time the stock portfolio is switched back into treasury bills. As a result, it is impossible to estimate the incremental return to a taxable market timer, but it is obviously well below 4 per cent per year.

What conclusions can we draw from this study?

1. It is clear that a successful market timing strategy can materially enhance the portfolio return over and above a simple buy-and-hold approach.
2. Once transaction costs, taxes and less-than-perfect forecasting ability are introduced, the differential rapidly shrinks to 4 per cent or less.
3. Based on the period under review, the stock market outpaces the return on treasury bills in 64 per cent of all calendar years. As a result, the market timer must have a predictive accuracy record approaching 70 per cent if he or she is to recoup the transaction costs and stay ahead of the market.
4. Finally, these conclusions are based not only on 100 per cent forecasting accuracy but also an extreme portfolio position of 100 per cent stocks or 100 per cent treasury bills.

Investors who are unwilling to adopt this "all-or-nothing" attitude will probably find that the combination of inaccuracy and indecision holds their rate of return below that of a less active investor.