Chicago-born Harry Markowitz revolutionised finance with the publication in 1952 of his theory about modern portfolio investing. But the global financial crisis has put paid to his seminal work. Now, 60 years later, Goldman Sachs is telling clients that it’s time to try something new.
Markowitz’s insight was that investors could figure out a unique optimal portfolio that would either maximise returns for a given level of risk, or minimise risk for a given return target. More than anyone else, he put the concept of risk at the centre of the hunt for returns in the stock market.
However, his theory rested on several assumptions: that the markets behave as a ‘random walk’, that returns on one asset class were independent of behaviour in others; and that correlations among, and volatility within, asset classes was constant.
Even if those assumptions were untrue, the theory could have survived. What killed it was the reality that very small tweaks to inputs or assumptions have led to massive differences in outcomes.
Subsequent efforts by academics and quants focused on shoring up Markowitz, usually by fixing the data inputs. Put the supercomputers to work, crunch the data, and the rest would fall into place.
The effort to bring modern IT to portfolio theory looked impressive, but also rested on two assumptions: that global markets were in equilibrium, and that there really was such a thing as an optimal portfolio. It’s a sort of ‘wisdom of crowds’ approach, but one that accepted market prices as some kind of received knowledge: if the market said X, then X must be the optimum.
Behavioural science, however, teaches us that markets are probably never in equilibrium and no portfolio is optimal. If it were, then bonds wouldn’t be outperforming equities so soundly for so long.
Given that Goldman Sachs alumni such as Fischer Black and Bob Litterman were major proponents of the ‘new Markowitz’ models in the 1990s, it’s telling that today the firm is telling institutional investors that they should give up trying to discover the optimal portfolio.
Mark Evans, managing director and head of global portfolio solutions at Goldman Sachs Asset Management, says the goal is a more modest one: if the optimal portfolio is impossible to derive, then use diversification to figure out what asset classes to include in a portfolio.
We’ve gone from the optimism of the can-do 1950s to the ‘well it’s better than the alternative’ triage of the post-Lehman world.
The new approach, outlined by Evans at AsianInvestor’s institutional investor conference in Singapore this week, is officially called factor-based risk budgeting. It’s not that new: anyone familiar with consultant-speak will recognise the idea that the only free lunch in portfolio management is diversification (an insight into which we owe much to, yes, Harry Markowitz).
Risk budgeting has been around for decades, but Evans puts it at the front of an investor’s thought process, rather than making it an afterthought. The Markowitz capital asset pricing model tries to ID an asset class’s rate of return if it is added to a diversified portfolio.
Evans says this approach assumes you have a risk-free rate of return, whereas risk budgets seek to evaluate the amount of risk added to a total portfolio with the inclusion of a new asset class. The idea is to spread the sources of risk more widely, because it’s too difficult to predict accurately what kind of return you’ll get.
Without going into the full details, Evans’ explanatory models show that a traditional investor portfolio split 60% to equities and 40% to bonds actually has a risk exposure that is 97.5% accrued from equities and only 2.5% from interest rates.
The correlation between risk has nothing to do with correlation of returns. The 2008 crash showed that the only true diversifier from equities was Treasuries, cash and gold, and Evans’s model is designed to reflect this truth.
His practical recommendation is to cut up the equity portion of the portfolio into an array of equity-like assets, from high yield to Reits to emerging-market bonds.
That’s the first part. The second part is to take advantage of market dislocations – those inefficiencies ignored by Markowitz – to generate alpha. Specifically he recommends that investors should write options (sell insurance) on asset classes where there is plenty of demand and no sellers. Done carefully, the premiums from selling options can provide a meaningful boost to returns.
(This is already happening in one part of Asia: Japan’s retail universe, where the likes of Nomura sell options for mutual funds to generate enough income to continue to pay monthly dividends.)
The result is a portfolio that, instead of allocating to three or four asset classes, is exposed to a dozen, and uses futures and options both to provide access as well as generate income.
So what are the underlying assumptions of factor-based risk budgeting? Institutional behaviour is probably one: getting such a scheme past technocrats and their trustees or ministries will not be easy.
Evans says the actual asset class exposures are simple. He has an even stronger argument on his side: the old model failed virtually everyone. Time to remind investors that it may not be perfect, but it’s better than the alternatives.