Analyzing Relationships Between ESG Data and Stock Performance, Part 2

Back in Part 1 of this exploration, we talked about including our values in investment decisions. Corporations listen to our money more than they listen to us, so can we vote with our money by investing in companies that adhere to our values? The answer is maybe. (The answer to any question like this is always ‘maybe.’ It’s highly unsatisfying).

There’s a lot to figure out: what are our values, do we have accurate metrics for companies adhering to them, et cetera. In this series, we’re pretending that we know the KLD environmental, social, and governmental metrics to accurately measure companies’ adherence to various values. We’re using them to explore yet another question about investing with our values: would we pay a price for doing it? That is, would we forego investment earnings?

We’re going through this in three parts:

  1. Get our tools in place to clean the data sets and pull out the information we need
  2. Run some correlations between ESG metrics and stock performance (this post)
  3. Assess the significance of our results

Let’s take a look at what we found.

If you love the intuition behind how we transform numbers into meaning, you might like the text portions of this notebook. I explain why I choose the variables I choose and how I represent them in the investigation. We talk about aggregation techniques, separation techniques, and the nuts and bolts of imposing order on inherently messy real-world data. This particular data isn’t all that messy, so it’s a good introductory case study.

If you couldn’t care less about the data or the code and you want to skip straight to what we saw, here is the picture you’re looking for:

ESG Scores v Stock Returns

In the top left, we see how stock returns change as employment policy strength scores rise. In the bottom left, we see stock returns as employment policy concerns totals rise. Top right, returns versus environmental impact strengths. Bottom  right, returns versus environmental impact concerns. I see what might be some downward trends, but I’ll withhold on any definitive analysis until we get to part 3: assessing significance.

Leave a Reply