News and Insights

ESG Ranking Method

by Peter Brooke, Head of Quantitative Investments, Platypus Asset Management

Article |
12 February 2018

Incorporating environmental, social and governance (ESG) considerations into investment processes is not straightforward. Methods and techniques vary across investment styles, and different portfolio managers can treat similar ESG issues in different ways. In addition, the literature examining ESG strategies over multiple market cycles is not as broad as it is for more traditional strategies (e.g., momentum, value, size), making it difficult to build consensus around systematic methods.

We start from first principles, and use available data to construct a model that outperforms. We test the alpha against standard strategies, and argue that it comes directly from the ESG data.

Data

We collate all company reported data from a global data provider. Unlike with accounting data, there is no universal calculation standard. We expect this to improve over time, and for more standardised reporting to become commonplace. For now, though, we make no allowance for this in our factor model.

The data has a ten year history, and most companies report at least one metric over this period. Data is reported annually, but we update our database more regularly, ensuring we capture all changes immediately. Coverage, however, can vary substantially between metrics.

Number of companies with at least one reported ESG metric

Chart 1: Number of companies in our database with at least one ESG metric. Prior to 2007, the data does not exist in its current form, so all models can only be constructed using ten years of data.

The number of metrics from the data provider total 45. We investigate all of these in the model building process.

We choose to penalise companies for missing data. Collecting ESG data requires time and money, and so reporting more metrics in our view reflects positively on the ESG culture. Not reporting is both going against the prevailing trend, and leads us to assume that information is either not known or is being hidden. Both instances reflect negatively.

Model

We test the efficacy of the ESG metrics. One test involves creating five portfolios, sorted by the metric of interest. We look at the relative performance of each portfolio. If the metric contains information, we would expect to see the top quintile outperform the bottom quintile. For the percentage of women on the board, this is exactly what we observe.

Percentage of women on the board

Chart 2: Performance of two portfolios constructed from the top 20% of stocks ranked by the amount of women members of the board. Companies with more female representation on the board outperformed by 16% since August 2008.

After going through this process, we select a model based on six factors. We have a belief that succinct factor models are more robust. We include factors that we are comfortable with from a fundamental investors perspective. The model can be separated into three parts: diversity, board efficiency, and sustainability.

The model performs well, outperforming the index by 4.8% per annum with similar volatility. Due to the rate at which companies report, the average turnover of the strategy (68% per annum) is low enough to be investible.

ESG 6-factor model compared to S&P/ASX 300

Chart 3: Cumulative performance of a long portfolio, based on our six factor ESG model. The annual return of the strategy is 8.6%, with volatility of 14.0%, compared to an index return of 3.8% and volatility of 14.1% over the same period.

With all models, however, there are a number of tests that confirm we are investing in ESG and not unintentionally something else. Stocks can be grouped in multiple ways. The three ways (or ‘factors’) with the most supporting evidence across different markets and time periods are momentum, value, and size. Stocks with good momentum (strong past performance) outperform those with poor momentum (weak past performance), stocks that are cheaper (as measured by price-to-book) outperform those that are expensive, and small stocks out-perform large ones. We can dis-aggregate the returns of our ESG model into these three different groups.

The model adds 23 basis points a month in addition to the returns that can be generated by momentum, value, and size. The model tilts towards stocks that are more expensive, larger and that have good momentum. The beta is less than one, which means that on this measure the portfolio is less risky than the index as whole.

12 month rolling returns

Chart 4: Rolling returns of ESG 6-factor model compared to a portfolio based on size, value, and momentum.

Characteristics

Looking further into the return profile of the portfolio details how the ESG 6-factor model trades through time. Sometimes the model is trading the same way as momentum, and sometimes it is not. Similar correlations are shown for value and size. This variation through time adds to our confidence that the model is genuinely trading ESG factors, and not something else.

We look at the sector weights of the model. We do not want a larger overweight to one sector – a diversified portfolio is more statistically robust than one that makes very specific bets.

Average sector weight through time of the ESG 6 factor model

Table 1: Average sector weight through time of the ESG 6-factor model.

The most significant sector bias is to Health Care. We expect this given the growth (or quality) tilt of the factor model. We do not think bias invalidates the model, especially because the weight in both the Materials and Financials sector is very close to market weight.

It is important to remember that the model is a based on reported factors, and ranks the stocks relative to the rest of the market. Stocks in the Energy sector may rank better on our Social and Governance metrics than stocks in other sectors, which could lead to the overweight seen in Table 1. Emissions are not part of out model.

In terms of valuation metrics, the portfolio trades at an average price-to earnings of 14.09x (average index is 13.91x), a price-to-book of 2.38x (index 2.04x), and a return-on-equity of 17.2% (index 14.9%). These metrics confirm our analysis – the portfolio has a tilt towards quality.

Summary

Our 6-factor ESG model has a number of characteristics:

  • It outperforms the index by 4.8% per annum with similar volatility
  • It has low turnover
  • It has a quality bias
The outperformance of our model is pleasing. It supports that idea that ESG factor investing can form the basis for a factor portfolio, and can be considered as more than just a compensation for risk.


Important Information

Information provided here is general information only and current at the time of publication and does not take into account your objectives, financial situation or needs. In deciding whether to acquire, hold or dispose of the product you should obtain a copy of the Product Disclosure Statement which is made available on this website and seek professional financial and taxation advice. This information is intended for recipients in Australia only. Past performance is not a reliable indicator of future performance.

For more information please contact

MarkHearne

Mark Hearne


Distribution Director – Platypus Asset Management

mhearne@platypus.com.au

Latest News

View all articles