What if the entire edifice of portfolio construction is built on a myth? That the axiom which underpins risk-matching a risk score to an asset allocation is a fairy tale?
While this sounds like the opening for Dan Brown’s next pot-boiler, it is nonetheless true that a fundamental theory of investment management cannot be recognised in the real world. Increasing return does not require increasing risk.
Fund managers adhere to the capital asset pricing model – a fundamental tenet of investment theory. The model boldly declares that by knowing the risk-free compensation for lending money, the expected market return and the relative-to-market risk of the proposed investment, we can assess whether any investment is worth undertaking.
The relative risk measure is beta. For maths buffs, it is the co-variance of the investment and the benchmark, divided by the variance of the benchmark. For the rest of us, on a chart with return on the vertical axis and beta on the horizontal, the risk-free return has a beta of zero and the market return a beta of one. Plot a line between the two coordinates and the result shows a linear relationship between risk and return: as relative risk rises, so does expected relative return.
This seems so obvious as to be indisputable. Those who take higher risks do so because they expect to be compensated for it. While returns might be volatile short term, long-term risk pays off.
But as statistician George Box said, while all models are wrong, they can be measured on a scale of usefulness. Models are not about truth – they are about application.
The fact is, the capital asset pricing model is nothing like the real world, and the empirical evidence is quite stark: higher beta does not lead to higher returns in the long term. On the contrary, low volatility strategies deliver superior risk-adjusted returns compared to the market (as a market cap-weighted index) over a long-term investment horizon.
A variety of academic studies demonstrate this but if you simply run a 10- or 20-year alpha versus beta chart of any equity or bond sector you like, with benchmarks either the sector average or the relevant index, the result is the same: lower beta produces higher returns. The slope is down from left to right, the exact opposite of what the model predicts.
Advisers using Markowitz-style mean variance optimisation tools understand a portfolio’s variance is minimised by considering the correlations among the constituent assets. Equity portfolios can be similarly constructed. Indeed, when this technique is applied to stock selection, the result is the same: lower beta portfolios tend produce higher alpha over time. Diversification, even within UK smaller companies, tends to produce higher long-term returns than focused portfolios.
Most fund managers are aware of this low volatility anomaly, so why does the risk approach persist? I suspect it is a behavioural issue.
Imagine buying an investment with a year in year out return of 5 per cent, while the market goes up by 25 per cent this year and down 11.8 per cent the next. Both investments would have grown by 10.25 per cent.
Problem is, this year you were behind the market by 20 per cent. You got no bonus and no new investors.
Assuming you have not been shown the door, you will outperform by nearly 17 per cent next year but have simply recovered your previous underperformance. Nobody benefited. From a career point of view, your “safe” investment was very risky; avoiding lower-risk opportunities makes sense.
There is also the lottery effect – a high probability of poor returns ignored versus the small chance of earning very large returns. Equity analysts starting on the career ladder can have a preference for shares with a perceived higher potential to deliver outperformance. In other words, they could place career risk ahead of the evidence.
The corollary of the low volatility anomaly is that its attractions are boosted by the fact that losses hurt more than gains help. If your portfolio falls by 20 per cent, you are going to need 25 per cent to recover. If you lose 50 per cent, you need 100 per cent. The bigger the draw-down, the more on the way up you need. Low volatility equity portfolios obviously do even better in an environment where the threat of a significant downturn seems to be increasing.
Equity portfolios can be constructed using a Markowitz approach – i.e. optimising return, volatility and correlation to build efficiency. They can also be built as a passive option by a ranking process – i.e. apportioning the highest weight to the lowest volatility constituent. Indeed, ETFs are available that offer just this approach.
This approach is less than smart and it is certainly not beta. However, with markets being in their current vertiginous state, considering a low volatility approach for equity portfolios – by the ETF route for passive fans, or by a mean variance process for active adherents – might not be so dumb either.
Graham Bentley is managing director of gbi2