In a prior blog, we demonstrated that unconstrained carbon-efficient portfolios have significant unintended (and unfavorable) sector and risk factor tilts that can drag down performance. In this follow-up blog, we explore potential ways sector-relative, carbon-efficient portfolios can address the drawbacks of sector-unconstrained, carbon-efficient portfolios. To form sector-relative, carbon-efficient quintile portfolios, we ranked and grouped securities within each sector based on individual companies’ carbon intensity.
Sector-relative, carbon-efficient portfolios were evaluated from several metrics: weighted average carbon intensity, risk/return profile, sector allocation, and active risk exposures. Carbon intensity data were provided by Trucost, part of S&P Dow Jones Indices, and are defined as greenhouse gas emissions measured in tons of carbon dioxide equivalent per USD 1 million of revenue (CO2e/USD 1 million).
The most sector-relative, carbon-efficient portfolio, Quintile 1, substantially reduced the carbon intensity by nearly 80% to 66 CO2e/USD 1 million compared with the benchmark, the S&P United States LargeMidCap (see Exhibit 1). Moreover, on an absolute return basis, Quintile 1 outperformed most of its peers and the underlying benchmark, except for Quintile 2. On the other hand, Quintile 1 had marginally higher volatility among all the portfolios, thereby resulting in comparable risk-adjusted return to Quintiles 2, 3, and 5, but higher risk-adjusted returns than Quintile 4 and the benchmark. The sector-relative, carbon-efficient portfolio improved its efficiency in terms of carbon efficiency and risk-adjusted returns over its underlying benchmark.
We further analyzed sector composition and sector attribution of the sector-relative, carbon-efficient portfolio to explore potential sector biases relative to the underlying benchmark and their impacts on portfolio efficiency.
The sector-relative, carbon-efficient portfolio displayed minor sector deviations from the underlying benchmark. During the period studied, average active sector weights maximized at about 5% (see Exhibit 2). Moreover, with the sector-relative, carbon-efficient portfolio, the sector allocation effect was positive (with an annualized return of 0.32% on a monthly average basis; see Exhibit 3) and contributed positively to its active return over the benchmark. In contrast, the unconstrained carbon-efficient portfolio had a sector allocation effect of an annualized underperformance of -1.51% on a monthly average basis (see Exhibit 3) due to its large sector bias in financials, as shown in the previous blog.
Analysis of active risk exposures showed that the sector-relative, carbon-efficient portfolio had much less positive active exposures in beta, value, liquidity, and price volatility. On the other hand, it had large negative exposure to yield, size, earnings variability, and high leverage (see Exhibit 4). Therefore, during the period studied, the sector-relative, carbon-efficient portfolio tended to have higher exposure to quality and limited exposure to value than the benchmark. In addition, the portfolio had some positive exposure to EPS growth rate and relative strength (momentum).
The results from Exhibits 2, 3, and 4 showed that the sector-relative, carbon-efficient portfolio had limited sector biases and more favorable risk factor tilts. Such unrewarded systematic risk reduction made the portfolio more efficient, while simultaneously meeting the carbon reduction objective.
Based on the findings of the two blogs, we advocate forming sector-relative, carbon-efficient portfolios over unconstrained ones when portfolio decarbonization is the main goal. In the next blog, we will explore how to integrate carbon risk with factor portfolios.
 We used the Northfield U.S. Fundamental Risk Model to estimate the risk exposure.
If you enjoyed this content, join us for our Seminar Discover the ESG Advantage in
London on May 17, 2018.