'You can’t time the market.' This is a common claim. But it might well be wrong. In fact, it is probably easier to time the market than it is to be a stockpicker.
Here are six possible market-timing strategies.
- Use the dividend yield. My chart shows the point. It plots the dividend yield for each month between 1985 and 2016 against annualised total returns on the All-Share index since that month. It’s clear that there’s a massive correlation between the two. High yields predict high returns and low yields low returns. If you’d bought in 2000 or 2007 when the yield was low you would have earned only low returns since then, especially adjusted for inflation. If you’d bought in 2003 or 2009 you’d have done much better.
- Be a seasonal investor. Since 1966 the All-Share index has given an average real total return of 8.3 per cent from Halloween to May Day, but has lost 0.6 per cent on average from May Day to Halloween. Ben Jacobsen and Cherry Zhang show that a similar pattern exists in almost all national stock markets since they began trading.
- Use the 10-month average rule proposed by Meb Faber of Cambria Investment Management. This says you should buy when prices are above their 10-month (or 200-day) average and sell when they are below them. Had you applied this rule to the All-Share index since 1991 you’d have made much better risk-adjusted returns than you’d have got from a buy-and-hold strategy. The latter would have made 8.5 per cent a year with a standard deviation of 13.7 percentage points, giving a Sharpe ratio of 0.62. The 10-month rule would have made 8.4 per cent with a standard deviation of 10.1 percentage points, giving a Sharpe ratio of 0.83. This is because the rule would have got us out of equities early in the bear markets of 2000-03 and 2007-09 and the losses this would have avoided offset the fact that the rule fails when a 'buy on dips' strategy works.
- Watch foreign buying of US equities. When such buying is high – as it was in 2000 and 2007 – it is a sign that global investors are too optimistic about equities and hence a warning that prices will fall in the following 12 months. And when foreigners are net sellers, as they have been recently, it’s a sign of pessimism and hence of rising prices in coming months.
- Look at price-money ratios. When the ratio of share prices to the money stock in developed economies is high, it predicts falling prices. This is because it’s a sign that investors have a high weighting of equities in their portfolios and a low weighting of cash, so any rebalancing of these portfolios is likely to be away from equities.
- Watch consumer spending. Sydney Ludvigson and Martin Lettau have shown that high spending, relative to consumers’ wealth, can predict equity returns: research at the Bank of England has corroborated this. High spending, relative to wealth in 2003 and 2009, for example, led to bull markets while low spending, relative to share prices, in 2000 and 2007, led to bear markets. This is because there is sometimes wisdom in crowds: low consumer spending is a sign that people are anticipating bad times.
We have, therefore, lots of ways of helping us to time the market. But we’ve fewer reliable methods of picking stocks. Duke University’s Campbell Harvey has said that most of the ways of doing so discovered by academics are “likely false”. And John Cotter and Niall McGeevor at University College Dublin have shown that many successful strategies stop working after they have been publicised. The recent poor performance of momentum investing might be yet another example of this.
In truth, what I’m saying here shouldn’t be so surprising. It’s just a natural implication of a point made by the Nobel laureate Paul Samuelson – that markets are "micro efficient but macro inefficient". Back in 2002 Jeeman Jung and Robert Shiller provided simple evidence for this. They showed that for individual shares' dividend yields tended on average to predict subsequent dividend growth, with high yields predicting low growth and vice versa. That’s consistent with one reading of the efficient market hypothesis – that growth is already in the price. But they found that for the aggregate market this was not the case. A high yield doesn’t predict higher dividend growth but rather faster price growth. This suggests that the aggregate market overreacts. Prices can get too high or too low, which means that returns can be predicted by measures of investor sentiment such as the dividend yield, price-money ratios or foreign buying of US equities.
All this poses the question: why, given this evidence, does market timing have a bad reputation while stockpicking has a good one?
There are respectable reasons. It’s impractical to sell an equity portfolio and reassemble it months later. Doing so incurs dealing costs and might expose you to tax liabilities. This is why I implement market timing by switching between equity and money funds only within my pension, leaving my other equity investments unchanged.
Also, these methods only predict returns at longer horizons – 12 months or more. For shorter periods, the aggregate market is not so predictable, which might be why many macro hedge funds do badly.
And then there is the persistent risk that these past relationships that allow us to predict future returns might break down. In fact, they would do so if enough people tried to time the market. Market timing can only work if it's a minority occupation.
I suspect, though, that something else is going on. It’s called professional deformation. This is the idea that our professional training doesn’t just teach us skills but also inculcates ways of seeing the world that are biased and incomplete. (Yes, this is true of economists too). If you’ve spent years training to be a stockpicker you’ll naturally become unreceptive to other methods and reluctant to see that your training is of little use. And if you’ve trained as an accountant, you naturally overestimate the importance of accounts, just as lawyers overrate the importance of law and economists of economics. In these ways, sympathy for stockpicking and antipathy to market timing might not be based entirely upon a dispassionate assessment of the empirical evidence.
