Environmental, Social, and Governance (ESG)

ESG refers to a set of criteria that investors, fund managers, and research analysts use to evaluate a company’s practices and strategies from three different perspectives: environmental impact, relationship with communities and employees, and financial and managerial ethics (shareholders’ rights, insider trading). Investors and professional asset managers use ESG research and systems of ESG ‘ratings’ to screen for ESG investing opportunities — both for themselves and for clients.

Public (ie. free) ESG data is typically very hard to come by, as it’s often used by professional and retail investors to help make profitable investment decisions. As a result, ESG research companies, like Sustainalytics generally charge a (several hundred dollar) subscription fee to gain access to their large sets of ESG data and research reports. However, Sustainalytics does have a free ESG Rating search tool that returns ESG Risk Ratings for any publicly-traded company.

Sustainalytics ESG Report for Exxon Mobil

To collect these ESG scores, we used a web-scraping bot that utilizes Selenium software. The bot enters the desired company name into the ESG tool’s search bar and then scrapes the corresponding ESG Risk Rating. This provided us with a method to collect ESG scores for a collection of companies of our choosing.

Top 75 U.S. Companies

We pulled revenue, revenue growth, industry name, and headquarters data for the largest revenue generating companies in the U.S., with the exception of 13 companies which are not publicly traded (and thus did not have ESG scores available). We chose to investigate top-revenue-earning companies because they are (arguably) the most successful companies in the U.S. and therefore likely have the most influence and presence among U.S. corporate activity. Furthermore, these companies make up many factions of American life from E-Commerce to Oil to Food and Beverages. These companies have far reaching economic, social, and environmental implications, thus it’s essential to consider their efforts to combat climate change, foster equality, and impart positive changes, not only in America, but globally.

Once we had our dataset of U.S. companies, we passed each of their names into the Selenium bot to their respective ESG Risk Ratings.

ESG Across U.S. Sectors

To address our first question — How does Sustainability vary among industries and sectors? — we assigned each of the top-75 companies to one of five sector ‘groups.’ The sector group assignments followed the criteria below:

# group similar sectors together for stat analyses
esg_grouped <- esg %>%
  mutate(sector_group = case_when(
    sector == 'Communication Services' |
      sector == 'Telecommunications' |
      sector == 'Technology' ~ "Tech, Telecomm, and Communications",
    sector == 'Consumer Discretionary' |
      sector == 'Consumer Staples' ~ 'Consumer Disc. and Staples',
    sector == 'Financials' ~ 'Financials',
    sector == 'Energy' |
      sector == 'Materials' |
      sector == 'Industrials' ~ 'Energy, Materials, and Industrials',
    sector == 'Health Care' ~ 'Health Care'))

Sectors were grouped together if their respective companies carried out similar functions or provided similar goods/services. This was done because, originally, we had some individual sectors (such as Materials) that had very few observations, and this would have prevented us from moving forward with any statistical analyses.

To start, we compared the medians and spreads of ESG Risk Ratings among our five sector groups using a box plot. The plot and table below reveal that the Energy, Materials, and Industrials group has the highest mean and median risk rating, while the Tech, Telecomm, and Communications group has the lowest mean and median rating.

For our next question — whether Sustainability efforts have an impact on corporate revenues — we conducted a linear regression with ESG Risk Rating as a predictor variable and revenue growth as the response variable. When checking linearity and normality assumptions, we found that the relationship between revenue growth and esg rating is not very linear, regardless if log or square root transformations are applied. Similarly, we found that applying log and square root transformations did not resolve the lack of normality in our predictor and response variables; the log transformation appeared to improve normality in for revenue growth in some sector groups, but we reasoned that this improvement was marginal, and ultimately decided to proceed with the original, untransformed data. We recognize that these assumptions — normality and linearity — are not fully met and recognize the limitations of our results going forward.

Impact on Revenue Growth

Our regressions on each sector group yielded positive estimates for the ESG coefficient for three groups and negative estimates for two groups; however, only the estimate for Tech, Telecomm, adn Communications was found to be significant at the 0.05 confidence level. This was fairly expected since we had a relatively small number of observations within each sector group, as well difficulty meeting the linearity and normality assumptions. An aggregate/ungrouped regression on all of the companies yielded a positive coefficient estimate and was significant at the 0.10 confidence level.

An interesting result is that our significant coefficient estimate for Tech, Telecomm, and Communications was positive. Indeed, even though most of our estimates were insignificant at the 0.05 level, the majority of them were positive, rather than negative. This seems to contradict the literature that sustainability efforts are a driver of market share and sales growth. However, the companies in our dataset are top-earners and are therefore not very representative of the average corporation in the U.S. With this and our assumption violations in mind, we recognize that these results may not be indicative of a negative association between sustainability efforts and prospective revenue growth.

To address our last question regarding the location of sustainable companies in the U.S., we constructed a choropleth map colored by average ESG Risk Rating. That is, the color of each state on the map below reflects the average ESG Risk Rating of all the top-75 companies that are headquartered in that state.

ESG in the U.S.

From the above map, it appears that Virginia and Arkansas house top-75 companies of high ESG risk, while Pennsylvania and Michigan house companies of low ESG risk. The tables below provide insight into the highest and lowest-risk states, as well as the companies contributing to their high/low ESG risk averages. It appears that the risk averages in Virginia and Illinois are brought up by companies that belong to the the Energy, Materials, and Industrial sector group. This corroborates our earlier finding that top-75 companies in this sector group demonstrate higher ESG risks, on average, than those in other sector groups.

Sources