In defence of optimisation

Sebastien Page, senior managing director of the portfolio and risk management group at State Street Associates is excited about his upcoming paper “In Defense of Optimization: The Fallacy of 1/N”, which responds to the increasingly popular notion that equal weighted portfolios outperform. He spoke with Amanda White about the “1/N paper”, and how he advises institutional investors to incorporate best practice macro issues in their decision making.

There is nothing like a bit of academic rivalry to get an industry flourishing, and Page believes his latest co-authored paper, which will be published in the Financial Analysts Journal early next year, will cause an “uproar”.

There is an increasingly popular notion, instigated in what has been known as the “1/N paper” by Victor DeMiguel, Lorenzo Garlappi and Raman Uppal (DGU), published in the Review of Financial Studies earlier this year, that equal-weighted portfolios outperform.

Now Page, with co-authors Mark Kritzman and David Turkington, counter that argument with a detailed study looking at 13 datasets comprising 1,028 data series from which more than 50,000 optimised portfolios were constructed.

The premise of their argument is that by relying on longer-term samples for estimating expected returns, optimised portfolios usually outperform equally weighted portfolios. And more specifically, the authors argue that poor optimisation results in previous studies arise from the reliance on 60 and 120-month historical returns to model expected returns.

“DGU for example report results for various models that rely on trailing 60-month returns. No thoughtful investor would blindly extrapolate historical means estimated over such short samples as expectations for the future, especially if they are outright implausible.”

The report concludes that: “In our view, 1/N is not a viable alternative to thoughtful optimisation but rather a capitulation to cynicism”.

This paper, highlighting the importance of academic study in the practical progression of investment strategy, will be Page’s 11th published paper, with another just published in the Journal of Portfolio Management, titled “Myth Diversification”.

This paper uses extensive empirical evidence to demonstrate that correlations are higher on the downside, and thus the inappropriateness of using the full sample coefficient as a measure of diversification.

“One example of this is the correlation between US equities and the world ex-US equities. When both are one standard deviation above the mean the correlation is -17 per cent, when they are both one standard deviation below the mean the correlation is 76 per cent,” Page says.

“This demonstrates it is extremely misleading to use the full sample co-efficient as a measure of diversification. It is the equivalent of saying: look at the average temperature throughout the year and wear clothes all year that are appropriate for that temperature. In Boston this would just simply not work.”

This research by Page is an adjunct to his work as head of the portfolio and risk management group at State Street Associates, which provides consulting research to 250 institutions globally, roughly split 70:30 between asset owners and investment managers.

His team, of only 15 people, cover the macro issues of risk management including asset allocation, management optimisation, currency hedging, optimal rebalancing, as well as a customised risk projects.

At the moment the group is conducting 38 simultaneous client projects, and last did 75 asset allocation studies for clients.

One of the key trends Page identifies in this area is the interest in “event sensitive portfolios”, and a whole body of research is due to come out on that soon.

“Clients are asking, for example, can you set up for me in advance the best optimal portfolio for high inflation environment,” he says.

“During the crisis CIOs and boards of pension funds wanted to review and change their asset mix but they couldn’t move quickly enough. This research will help them put in place policies, that will then enable them to act quickly when they want to.”

Page believes one of the key functions of his team is to contribute to the intellectual property of the industry as a whole and points to four innovations in the past 10 years of research: Risk regimes, such as correlation asymmetries and portfolio stress testing; within-horizon risk measurement, such that risk models shouldn’t look only at the end horizon but events along the way; full scale optimisation, or how to optimise a portfolio with non-normal return distributions; and optimal rebalancing.

The biggest issue on the minds of clients, he says, is how to deal with risk instability, liquidity risk and rebalancing policies.