united statesen
Reducing the carbon intensity of multi-factor credits strategies

Reducing the carbon intensity of multi-factor credits strategies

27-05-2020 | Insight
Our research into sustainability integration in quantitative investment strategies shows that such strategies lend themselves well to integrating secondary objectives, such as reducing carbon intensity. Our research has shown that reducing the carbon intensity of the multi-factor strategies results in only a limited reduction in the backtested outperformance.
  • Robbert-Jan 't Hoen
    Robbert-Jan
    't Hoen
    Researcher
  • Patrick  Houweling
    Patrick
    Houweling
    Head of Quant Credits

Speed read

  • Large differences in carbon intensities across the investment universe
  • Carbon constraint has little impact on expected performance of strategies
  • Portfolios tilt away from (companies within) carbon-intensive sectors

Integrating a carbon constraint in multi-factor credit and high yield strategies

Our research into sustainability integration in quantitative investment strategies shows that such strategies lend themselves well to integrating secondary objectives, such as reducing carbon intensity, next to the primary objective of realizing attractive risk-adjusted returns. This is due to the large investment universe that the strategy can screen for what we view as the most attractive opportunities.

If a company with high carbon emissions issues a bond with attractive factor scores, the portfolio construction algorithm is often able to find a bond with similar factor scores, but with lower carbon intensity. This efficiently tilts the portfolio towards bonds of more sustainable companies, without having to forgo a lot of factor exposure.

We assessed the performance impact of imposing a constraint on the carbon intensity of our global multi-factor credit and global multi-factor high yield strategies, over a 1994-2019 backtest period.2 The constraint specified that the weighted average carbon intensity of the portfolio should be lower than that of the benchmark. The impact on the backtested performance was limited to a few basis points per annum. We have also researched more ambitious carbon reduction targets, e.g. 10% lower carbon intensity than the benchmark, or 20%, 30%, etc., and found that these are associated with modest reductions in the expected outperformance of the strategies.

Contact

230 Park Avenue, Ste 3330
New York, NY 10169
United States
Tel: +1 646 690 9397 or + 1 646 761 5484

Our backtested model showed only a limited reduction in outperformance

One may wonder why, in this backtested model, reducing the carbon intensity of the multi-factor strategies came with only a limited reduction in outperformance, even though imposing constraints in a factor strategy generally led to lower factor exposures. We think there are three reasons for this:

  1. The distribution of carbon intensity is right-skewed, so that avoiding only a few of the most carbon-intensive companies results in a relatively large reduction in the weighted average carbon intensity of the portfolio.

  2. For (close to) half of the backtest sample, the portfolios already have a lower carbon intensity than their benchmark, so that no adjustments need to be made in those periods.

  3. In the periods that the carbon intensity of the portfolios was higher than their benchmark, the large investment universe enabled the portfolio construction algorithm to replace bonds from companies with high carbon emissions with bonds that have similar factor scores, but lower carbon intensity. This efficiently tilted the portfolio towards bonds of more sustainable companies, without having to forgo a lot of factor exposure.

We believe that the described approach helps to reduce transition risks in the portfolio, while still providing attractive risk-adjusted return potential.

Download the white paper with the detailed findings

Disclaimer

Your details are not shared with third parties. This information is exclusively intended for professional investors. All requests are checked.