Abstract
Investors have yet to aggressively integrate Environment, Social and Governance (ESG) factors into portfolio construction due in part to concerns about data quality and availability or uncertainty around how ESG integration might affect returns. A chief obstacle to overcoming these concerns is the abundance of ESG indicators with a paucity of data-fill. In order to test the hypothesis that using ESG criteria for security selection need not be detrimental to returns, we review which ESG performance metrics are disclosed by a critical threshold of mid and large cap companies in relevant sectors, and then use this data to rate firms from the S&P 500. This information is used to construct ESG portfolios (comprised of firms which score in the top half) and test their performance against a market capitalization weighted benchmark. We then review how investors can reduce their investable universe by a random factor and still outperform the market capitalization weighted benchmark, using non-market capitalization weighting schemes. Finally we field test this hypothesis with a live risk-weighted ESG portfolio (superior resource productivity, responsible tax and compensation practices, diverse boards, clean revenue exposure) and test its performance against a market capitalization weighted benchmark. The results demonstrate it is not necessary for an ESG factor to have predictive power over future security prices in order to be applied in a fiduciary compliant context. The implication is that an ESG portfolio can be built in a manner to potentially outperform a market capitalization weighted index without requiring the strong view that ESG can provide a forecast of future returns, and only requiring the view that ESG has a non-significant relationship with future returns. The value of this implication to the broader field of study is that, with proper portfolio construction, investors can align their capital with their values, whether these be ESG, gender-lens, faith-based, or something else.
Introduction
While an increasing number of investment managers are incorporating sustainability criteria into portfolio construction, most are not doing so. The integration of environmental, social, and governance (ESG) data has presented challenges and obstacles to many portfolio managers for at least three, sometimes overlapping reasons. The first reason is that the data itself is incomplete and inaccurate, with diminishing corporate transparency in mid-cap and small public companies as well as emerging and frontier markets. The combination of these challenges, along with the fact that ESG data sources don’t always agree on the underlying facts has been an obstacle to adoption, as explored by Montiel et al. The second reason is that while simple exclusionary values (screens) provide comfort to clients who wish to avoid sectors of the economy they abhor, e.g. tobacco, in the view of some portfolio managers, any reduction in an investable universe increases risk of benchmark under- performance. Recent research by Bank of America (2017) suggests that rather than increasing risk, ESG screening may be the most effective means for reducing risk. The fear of underperformance is not borne out in the literature. In the most exhaustive review to date of over 2300 studies by Friede, Busch and Basse, they found that roughly 90 per cent of studies find a nonnegative ESG to corporate financial performance relationship.
And thirdly, portfolio managers have varying views on materiality. In a previous paper (Khan, Serafeim, Yoon, 2015) research showed that companies with good ratings on material sustainability issues outperformed companies with bad ratings on material issues over long time frames. At the same time, companies with good ratings on immaterial sustainability issues did not outperform companies with bad ratings on these issues.
For portfolio managers focused on short-term returns, taken together, data quality, availability, materiality, and diversification have given these managers ample reasons to avoid inclusion of ESG inputs. Despite these barriers as Amel-Zadeh and Serafiem (2017) find, there are a growing number of investors who are seeking to use ESG inputs which are linked to financial performance.
In this paper we review ESG data options that can be used by investors to split groups of companies based on whether they align or do not align with a specific values-based preference. We then demonstrate how ESG factor integration is no better or worse than random security selection allowing us to leverage academic research showing how random security selection combined with risk efficient weighting can lead to benchmark outperformance. We then review a field test of this approach in the US Equities (large cap) market with live product. The clear implication of this study extends beyond ESG, to a broader field of values-aligned investing.