Judgement day for smart beta

Judgement day for smart beta

Considering how much money continues to be invested in smart beta by pension funds and via the explosive global growth of exchange traded funds, there is regrettably little independent research into how well strategies perform. 

The asset managers and others promoting the strategies do of course have back-tested models, which show prospective clients what they might expect in terms of out-performance, but it is hard for the client to tell whether this is smart investing or smart marketing.

Objective independent research is scarce but it is scarce for a reason. To be useful it has to look at performance over the very long run. Few academics have easy access to the required data sets.

Luckily there are exceptions. When all three were at London Business School Professors Elroy Dimson, Mike Staunton and Paul Marsh began to compile what is now the world’s most comprehensive study of the long-run behaviour of equities and bonds. 

They burst on the scene with Triumph of the Optimists published in 2000, which charted how equities had trounced bonds over the last 100 years in London and the other major markets. The database has been extended each year since and their work now takes in most of the world’s major markets.

Testing claims

They set out to answer two questions: is smart beta really smart and is smart beta persistent? And, can the factors that tilt a portfolio deliver lasting outperformance? The results appear in the recently published Credit Suisse Investment Returns Yearbook.

Over 300 different factors have been isolated over the years and obviously they could not test every single one. Instead they focussed on the five most commonly used – value, momentum, size, low volatility and income.

First of all they plotted these on a grid to grade performance since 2008. These last 10 years show considerable volatility. What they found however was that a factor which drives performance one year may well fail to do much the next. 

Thus, over the 10 analysed years the low volatility factor delivered three years of top quintile performance, three years in the bottom quintile and four around the middle. The size factor delivered three years at the top but only one year at the bottom. 

Momentum also had three years at the top and one year at the bottom. Value had one year at the top and five years at the bottom. Income never made it to the top, but was never at the bottom either. It spent all 10 years in the middle three of the quintile bands with a bias towards the lower.

Over the 10 years most factors had their time in the sun. But there was no persistency and no pattern and no way of knowing in advance which was likely to do well. 

And, of course, the analysis ignores the impact of costs. Some smart beta strategies require a large amount of dealing, others in the long-short space can be difficult to construct. Theoretical outperformance may not be achievable.

Over periods of 50 years or more they discovered similar uncertainty. All factors make a difference – most are generally benign but there is an ever-present danger of factor reversal. 

For example, over 50 years value significantly outperforms growth. But there are periods when it goes painfully the other way and that is when all but the most resolute clients get shaken out.

Similarly small caps significantly outperform large caps in a 50-year run but there will be periods with durations of 10 years or more where the opposite is the case. It is worth considering too whether the outperformance is simply the rewards for extra risk and illiquidity.

Overall the data shows that portfolios tilted by individual factors or a combination of them will behave differently from the market as a whole and this is something investors cannot ignore. 

But that effect can be down as well as up; and that is also something investors cannot ignore. Smart beta is not smart all of the time.