To see this, imagine an investor had followed a very simple policy. At the end of each month, he compared total returns on equities with those on gilts, and he put all his money into whichever had done better in the previous 12 months.
Using this very simple momentum strategy, £100 at the end of 1989 would have grown to almost £886 by now. That's an annual return of 9.7 per cent. By contrast, £100 split evenly between gilts and equities (rebalanced each month) would have grown to just £701. And an anti-momentum strategy of buying the worse performer of gilts or equities in the previous 12 months would have grown to just £486.
Momentum, then, works for whole asset classes as well as for individual equities.
However, all this outperformance has come only since around 2001. During the 1990s, the three strategies' returns were very similar. And in fact, momentum did poorly in the late 90s because it was in equities for the crash of August 1998 and out of equities in the early part of 1999's tech bubble.
Since then, though, momentum has triumphed. It was in gilts from December 2000 to August 2003, and so missed the bear market of 2001 and 2002. And it got out of equities in January 2008 and so avoided the crash of 2008-09. Yes, momentum paid a price for this success in that it missed the early phases of equities' bounce-backs in 2003 and 2009. But this price was small compared with the benefit of avoiding the preceding big falls.
What should we make of this? One possibility is that investors' behaviour changed around the turn of the century to become more momentum-prone. For example, pension funds that fell into deficit because of falling share prices and bond yields became more cautious in their investment policies, with the result that selling of equities led to more selling. This process might have been exacerbated by the growth of hedge funds, whose use of apparently sophisticated algorithms can disguise what is basically momentum investing.
There is, though, another view. It could be that what we've seen since 2001 is just luck. It has always been the case that cheap assets can get cheaper and expensive ones more expensive. And maybe recent years have over-sampled such episodes and under-sampled episodes in which cheap assets quickly bounced back.
One fact is consistent with this - the difference between the momentum and the balanced strategy has been sufficiently small that it could be due to luck. Since December 2001, momentum has beaten the 50:50 strategy by an annualised 3.4 percentage points - a return of 9.8 per cent per year against 6.4 per cent. This is the sort of difference that would arise one-in-eight times if the two strategies in fact had the same returns and were subject to the sort of volatility we've seen in gilts and equities since 2001.
What's more, momentum's outperformance has come in part at the expense of higher volatility. Since 2001, the Sharpe ratio on momentum (return divided by annualised standard deviation) has been only slightly better than that on the 50:50 strategy. And since 1989, it has actually been worse.
Personally, I suspect the evidence for true, lasting momentum in assets isn't overwhelming - and is certainly less robust that the evidence for momentum in individual stocks. You might think the test of this is simply to run the momentum strategy in future. This, though, runs into a problem. Right now, the strategy is telling us that equities will outperform gilts in the next 12 months. But of course, such a prediction also arises from a more innocuous hypothesis - that risky assets should outperform safe ones. It will, therefore, be some time before we can get more evidence on whether momentum in assets really exists.
