Are there patterns in the performance of stock market investments? And can they be exploited? Elroy Dimson, Paul Marsh and Mike Staunton's latest publication seeks out the answers
At the turn of the millennium, the term had been scarcely heard of. Now, "smart beta" is in vogue. Major investors are either using the strategy or actively thinking of doing so. Smart beta has joined the mainstream.
But what does the idea mean? It is not simply a matter of following a well-known index – buying shares, for example, that mirror the makeup of the FTSE-100 or the S&P 500. Nor is it about entrusting a fund manager with the responsibility of analysing individual companies and trying to spot those whose shares he or she thinks will do well.
Smart beta is about building a portfolio of shares that have particular characteristics in the belief that, over time, investments with those attributes will deliver returns that beat the stock market as a whole. Proponents maintain that history shows that shares that tick a particular box have done better than others. And crucially, they are likely to continue to do so.
This investment approach – smart beta or factor investing – is one of the topics examined in the latest Credit Suisse Investment Returns Yearbook by Elroy Dimson, Paul Marsh and Mike Staunton. They have looked in particular at five different criteria against which stocks can be judged:
Size: do investments in small companies do better or worse than those in big companies?
Yield: Is it better to invest in companies whose shares offer a high dividend yield?
Value: are returns likely to be higher from companies which seem to have a low stock market valuation when measured by the gap between their book value and their share price?
Momentum: If, relative to the market, a share price has already risen strongly - say over three, six or twelve months – is it likely to continue to outperform?
Volatility: if a share price has been less volatile – again, compared to the market as a whole – is it likely to do better or worse in the future?
Data over the recent past show no strong pattern. Take some examples from the years since the 2008 financial crisis.
Shares whose prices had previously shown low volatility did well in 2008 – far better than the market average. But in 2010 they performed badly. They had a good year in 2011 but a bad 2013.
Shares in small companies did well in 2009 and 2010, but badly in 2014.
Investors in "value" shares – putting their money into companies whose book value relative to their market price was low – could congratulate themselves in 2012 and 2016, but returns trailed the market in 2011 and 2015.
High-yielding shares did moderately well in most years, but were behind the pack in 2012 and 2013.
And taking “momentum” as a guide – investing in shares that had performed well in the recent past – worked well in 2015, but delivered poor results the following year.
The above results are for the US market. But the pattern is similarly erratic – or, indeed non-existent ¬- for the UK. In both these cases and subsequent examples, overall returns assume that dividends are reinvested. Also, figures take no account of the dealing costs of buying and selling shares.
So is there really no pattern at all? In the words of Dimson, Marsh and Staunton, “are smart-beta strategies smart?” Or does the smart beta approach to investing actually give no real advantage? To find answers, look at the long term.
Take each of the five approaches analysed by Dimson, Marsh and Staunton.
Start with the case of small companies versus large ones. It is now well-established that returns from investing in small companies have, over several decades, brought higher returns than investing in large ones. In the UK, a £1 investment in large-capitalisation companies at the start of 1955 would be worth £1087 today. An investment in medium-sized companies would have given a better result and in smaller companies better still. But the real star performers over this 61-year period were the so-called micro-caps – the bottom 1 percent by value of the UK equity market. A £1 investment in 1955 would have grown to no less than £27,256 by the end of last year.
What about value investing, where money is put into companies whose share prices are relatively low compared to their book value? There have been periods when the strategy has yielded disappointing results: for example, in the 1990s, it was growth stocks that did well; value stocks trailed behind. (This pattern was reversed with the pricking of the dot-com bubble.) But again, over the very long term, there is a discernible pattern. Dimson, Marsh and Staunton compare the performance of two groups of equities – those whose book-to-market value is high versus those whose is low.
In the UK, the former group – those whose share prices are relatively low when measured against book value – saw an average annual return of 16 percent over the years 1955 to 2016. For shares with a high market price relative to their book value, the figure was 10.3 percent. Value investing worked.
A separate measure of “value” is the percentage dividend yield on a share. Here, Dimson, Marsh and Staunton have the advantage of a database that goes back to 1900.
They have calculated the returns on a “high yield” portfolio versus those on a “low-yield” portfolio. The 100 largest UK stocks at the start of each calendar year are split into two groups – the 50 offering the highest yield and the 50 with the lowest. The returns of each group are calculated over the following 12 months, and then the process is repeated annually. An investment of £1 in the low-yield group in 1900 would have grown to £6,810 by the end of 2016. But over the same period, £1 in the high-yield group would have swollen to £158,727 – more than 23 times greater.
Using the past momentum of share prices to predict future performance is counter-intuitive. If markets functioned with perfect efficiency, it should not be possible to secure above-average returns simply by buying past winners and selling losers.
Yet the strategy does seem to work.
There are many ways of putting this approach into practice. For example, when measuring past momentum, it is possible to choose share price performance over three, six or twelve months when identifying winners and losers. Similarly, once bought, past winners could be held for one month or two, or longer still.
Looking at the data since 1900, buying stocks that have outperformed over a prior period (winners) and selling those that have under-performed (losers) would have yielded substantial returns. Dimson, Marsh and Staunton do the calculation by segregating stocks according to their previous 12 month performance. They then allow a one-month waiting period before buying and holding for one month. Using that formula, and looking only at the top 100 UK stocks, the winners would have returned an average 14.1 percent a year; meanwhile, the losers would have returned only 3.6 percent.
Intuition suggests that investors should be rewarded for taking on higher risk: they will receive higher long-term returns from putting money into stocks whose prices are volatile, moving up or down more sharply than the market as a whole.
It’s not that simple. Several studies have suggested that the opposite is true: investing in stocks that have been volatile in the recent past has delivered returns inferior to those from investing in shares whose prices are steadier. In other words, low volatility and high investment returns go hand-in-hand.
But these findings are generally based on short-term price movements – daily returns over three months.
Dimson, Marsh and Staunton look at a longer-term measure of volatility, tracking shares’ performance relative to the market over a five-year period.
On that basis, there was little to choose between the returns from high-risk and low-risk shares in the years between 1960 and the turn of the millennium. Then came the bursting of the tech bubble in 2000. Prices of high-risk stock (for which read high volatility) shares collapsed. Since 2003, high-risk stocks have actually outperformed their low-risk rivals.
Certainly, over the full period from 1960 to 2016, high-risk stocks have underperformed. But that underperformance is entirely due to their collapse when the tech bubble burst.
So is smart beta important?
It certainly seems so. Dimson, Marsh and Staunton conclude: “Size, value, income, momentum and volatility have an important impact on portfolio returns. They should be monitored by all investors.”
The big picture
Investing in equities is not a one-way bet. As the advertisements constantly remind us, share prices can fall as well as rise.
The LBS share price database yields vivid reminders of that truth. Between 1929 and 1931, the real rate of return on equities was minus 54 percent for the world as a whole and minus 61 percent for the US. Between 1973 and 1974, following the oil price shock, real returns on UK equities were minus 71 percent. More recently, investors saw a real return of minus 41 percent globally in the wake of the 2008 banking crisis.
Of course, there were booms, too. Between 1980 and 1989, UK equities gave a total real return of no less than 337 percent. The Nineties tech boom helped bring a real return of 276 percent for investors in the US. And overall? A $1 investment in the US in 1900 would grow over the following 116 years to $39,524 by the end of last year. Inflation nibbled away quite a bit of that. But in real terms, the value of that $1 investment would still swell to $1,402.
Want to accelerate your career in finance? Discover how our Masters in Finance and portfolio of finance short courses could make a difference to your career and your organisation. Find out why.
“Predicting winners” was originally published on the London Business School site.
Elroy Dimson is chairman of the Centre for Endowment Asset Management at Cambridge Judge Business School and Emeritus Professor of Finance at London Business School.
Paul Marsh is Emeritus Professor of Finance at London Business School.
Mike Staunton is director of the London Share Price database at London Business School.