The news at the start of May that Scientific Beta had launched two new multi-factor indices in order to satisfy the demands of a French pension fund, SACRA, with regard to integrating ESG funds threw the spotlight on how ESG mixes and matches up with factor investing.
ETF Stream caught up with Frederic Ducoulombier, director of risk and compliance for ERI at Scientific Beta, to run through some of the issues raised by marrying factor investing with low carbon aim and got around to the question of whether ESG is a factor in and of itself (spoiler alert – certainly not).
But we started with the basics of how ESG and factor investing can marry up and questions need to be addressed.
ETF Stream: When it comes to factor investing and ESG, what are the issues that have to be addressed? What are the correlations between ESG and acknowledged factors?
Frederic Ducoulombier: As with any other form of investing, one must decide upon the ESG incorporation approach. Are the ESG and financial risk and performance dimensions to be approached jointly for security selection and weighting or independently? The answer depends both on the nature of the investor and on how the strategy designer views the potential contribution of ESG data to expected financial risks and performances.
Does the investor have deontological requirements (ESG constraints) or a progressive ESG agenda (ESG objectives)? Or is this business as usual whereby the only mandate is financial and the ESG data are to be exploited to that end, possibly with no regard to the resulting performance of the strategy on an ESG plane? An approach that deals with ESG constraints and objectives through exclusions performed as a first step guarantees that financial considerations will not be a threat to the integrity of the ESG policy and that exclusions will send clear signals to stakeholders and will be justifiable purely on ESG grounds.
Approaches that allow compensation between potential financial performance and ESG performance at the level of each portfolio constituent (e.g. via use of composite signals or sequential tilting) or across constituents (e.g. optimisation using a portfolio average ESG score as constraint) are not appropriate for ethical or socially responsible investors. However, this does not mean that they are appropriate for egoist responsible investors; for the latter, the integration of ESG data makes sense only if it can improve the financial profile of the strategy.
On a risk-adjusted basis, there is little evidence that this can happen and many ESG integration strategies that are being advocated come with significant risks of diluting time-tested and academically validated sources of performance.
ETF Stream: What are the correlations between ESG and the acknowledged factors?
Progressive ESG strategies typically result in lowered exposure to total risk and higher exposure to profitability.
ETF Stream: Is there enough academic research in this area yet? What is the impact of ESG on portfolio performance and characteristics?
Regarding the performance of strategies incorporating ESG dimensions, we have forty years of studies of live ESG funds and indices that conclude that ESG incorporation neither hurts nor improves risk-adjusted performance. See Friede et al (2015), where 60 studies were reviewed: “Investors in ESG mutual funds can expect to lose nothing compared to conventional fund investments.” Also see Revelli and Viviani, 2015, where 85 studies were reviewed: “The consideration of CSR in stock market portfolios is neither a weakness nor a strength.”
There have been all kinds of paper studies of strategies claiming to make smarter use of average and pillar ESG scores and some have showed back-tested performance; but effects vary over time and databases and effects that may have had economic relevance in the past seem to have faded away (which is consistent with learning on the market).
Altogether, there is little relevance for current investment. See Halbritter and Dorfleitner (2015): “Investors should no longer expect abnormal returns by trading a difference portfolio of high and low-rated firms with regard to ESG aspects.”
Strategies mining or reprocessing criterion-level data are the new hope of those searching for performance in ESG data; the new conventional wisdom marketed by the investment industry is that there is a link between the financial materiality of corporate ESG investments and investment performance; this however is based on a single study (Khan et al, 2016). Regarding the potential use of ESG data to improve the estimation of risk parameters, there is insufficient research.
ETF Stream: Is there necessarily a difference between how you implement ESG with passive funds and smart beta funds?
If the implementation concern the exclusion of companies with unacceptable products or conducts or poor ESG performers without concern for the financial impact, then there is no difference whatsoever. If you integrate ESG considerations in security selection and weighting beyond that, even if you anchor on capitalisation weighting, I do not think you should call this passive anymore and at the same time it need not be smart.
ETF Stream: Is there the possibility that ESG itself will one day be seen as a factor?
Traditional factors carrying long-term rewards in excess of the broad equity market premium are justified on a risk basis (higher performance is compensation for higher risk in certain circumstances) or behavioural/structural basis (human psychology or market player organisation lead to systematic errors in the processing of data, which are not immediately arbitraged away due to institutional constraints).
Progressive ESG investment is marketed as reducing long-term risks – if this goes beyond idiosyncratic risks and markets are efficient, then this should be associated with a negative premium. It is hard to comprehend what institutional factors would prevent arbitrage of mispricing in relation to progressive-ESG investments; if anything, depressed demand may cause mispricing of regressive ESG investments.
In the latter respect, note that the traditional literature on the over-performance of ‘sin’ stocks following Hong et al. (2009) has been challenged by recent work, (e.g. Blitz and Fabozzi, 2017; Richey, 2017, that finds that performance is explained away by loadings on factors that are now conventional (after Fama and French, 2015).