Wedding ESG with Modern Portfolio Theory

Wedding ESG with Modern Portfolio Theory

Three executives at AQR Capital have proposed a new variant on modern portfolio theory, an elaboration of the classical idea of an efficient frontier.

The idea, going back to Markowitz’ 1952 paper, is that there exists a set of possible portfolios such that each achieves the best possible risk-return tradeoff. For any given portfolio on this frontier, there is no other that would get greater return at that level of risk.

One inference from Markowitz’ model was known as “two-fund separation.” This is the idea that given certain idealizing assumptions (such as the availability of a risk-free asset) every investor will end up holding one or both of two assets—the risk-free asset and/or the market portfolio.

The AQR authors suggest the extension of the underlying idea to the world of environmental, social, and governance-conscious investing. They propose a frontier that consists of the set of all portfolios that achieve the highest possible Sharpe ratio given any specific ESG score.

In their theory, each stock has an ESG score, and that score plays two roles. It provides information about the fundamentals and it has an impact on the investor preferences.

The authors are Lasse Heje Pedersen, both a principal at AQR and a finance professor at Copenhagen Business School; Shaun Fitzgibbons, managing director at AQR and an alum of Goldman Sachs; and Lukasz Pomorski, a managing director at AQR and its head of ESG research.

Four Funds—Three Types of Investors

The authors advocate a “four-fund separation” view of the portfolio problem. This means that they believe as ESG funds become more common they become more desirable as part of a portfolio. The portfolios on the efficiency frontier are all composed of four elements: the risk-free asset, the minimum-variance portfolio, the tangency portfolio, and the ESG-tangency portfolio.

Their model involves three types of investors. Type U investors are unaware of ESG scores and simply look to maximize their unconditional mean-variance utility. Type A investors are aware and use ESG to maximize their anticipations of risk and expected return. Type M investors, finally, are motivated by ESG scores, looking to invest responsibly in the ways those scores identify.

Much of ESG scholarship discusses the relationship between various ESG metrics of corporate or fund performance. Some metrics positively predict performance, others negative. Still others in this space don’t offer any prediction at all. The AQR authors seek to make a contribution to discussion of that point.

“When there are many type-U investors and when high ESG predicts high future profits … high-ESG stocks deliver high expected returns. This is because high-ESG stocks are profitable, yet their prices are not bid up by type-U investors, leading to high future returns,” they write. On the other hand,  “when the economy has many type-A investors, then these investors bid up the prices of high ESG stocks to exactly reflect their expected profits, thus eliminating the connection between ESG and expected returns.” On the third hand, if an economy “has many type-M investors, then high-ESG stocks actually deliver low expected returns, because ESG-motivated investors are willing to accept a lower return for a higher ESG portfolio.”

Negative, Positive, and Uncorrelated

There are also more micro effects. For example, some scholars have produced evidence that S scores, scores related to the social utility of a company’s behavior, are negatively correlated with performance. Though the negative correlation is weak, this research has been used to promote “sin stocks” in alcohol, tobacco and gambling. But this relationship is not a mystery: people are willing to accept a discount in performance in order to avoid investing in practices they regard as anti-social or perhaps forbidden by their religion. A higher demand from ESG investors for non-sin stocks lowers the expected return, even where potentially stronger flows from investors pump the price short term.

Under the heading of “E,” the article observes that a measure of low carbon emissions doesn’t work as a predictor of performance at all, either positive or negative. The authors tentatively attribute this to the brevity of the sample period; that is, not enough data.

Finally, the “G” in ESG does well in predicting performance positively. Proper governance standards do boost performance, and do so in ways which much of the investment universe is unaware of, so this boost is not discounted.

As the authors acknowledge, many of their predictions from the model are qualitative rather than quantitative in nature,  but the model may serve as a platform for further work and refinement.

Be Sociable, Share!

Leave A Reply

← Using Alternative Data and Machine Learning in Alternative Asset Classes Long Alpha and Activist Strategies  →