In case you haven’t heard, millenials are killing everything from diamonds to department stores to designer crap to grocery chains.
Why? Sure, the recession had an impact. But also, millenials pay more attention to ethics than many multinational corporations bargain for. They cite the blood diamond trade as a major reason to spring for non-traditional engagement rings. They opt for grocery providers that can tell them where their food is coming from and under what conditions it was produced. They’re ditching the fast fashion industry for higher-priced items purchased secondhand on sites like Poshmark and ThreadUp.
Millenials aren’t teenagers anymore: the youngest are 24 and the eldest are 38. As they reach the age where they might accrue some savings, it makes sense that they would care about where that is going, too. In addition to millenial attendance at the NoDAPL protests, we saw thousands of millenials divest from Western Union, Bank of America, and other banks that loaned money to the project. Maybe megacorps won’t change their tunes because a few thousand people stood in a field to get mowed down by water cannons, but they’re likely to sit up and listen when those same people take their hard-earned doll hairs to another playhouse.
So we see that millenials are surveying their options to spend and save according to their values. What about investing? Any personal finance 101 that isn’t taught by a financial advisor will recommend a low cost index as the place to stick extra money so it can grow with the market. Most index funds, including the most recommended one (Vanguard), decide their investments via index-matching: matching their holdings to the S&P500 by market cap, with no other variables. Thing is, plenty of investors are expressing interest in taking ethical considerations into account. Some portfolios do this by blanket blocking investments in certain industries like tobacco or porn. Other more advanced options, like Betterment’s AutoSRI portfolio, use environmental, social, and governmental factors data (ESG data) to determine where they invest the money. There isn’t (yet) a fully customizable option to allow folks to automatically invest their funds based on a checklist of their individual values. For a while, I’ve thought about building a toy version of what that might look like.
When I talk about the idea with friends and relatives, I get the following objection: ‘What about the returns?’
Touche. Nobody wants to lose out on their potential earnings. At first, I figured I’d build a tolerance into the system that allowed investors to say ‘These are my values, but please don’t invest in a way that will trail general market performance by more than x percent.’ The algorithm would then predict stock performance for each company, somehow blend that with ESG rating, and come up with a combined weight for divvying up investment money.
Before I build that, though, I need to test the assumption that high ESG ratings correlate negatively with returns. If they don’t, there’s no need for the tolerance measure in the first place.
I’m not the first person to run correlations along these lines. Dorfleitner, Utz, and Wimmer published a paper on this just last year. Their analysis suggests that higher corporate social responsibility ratings increase returns over a long period of time (“long” being a 12 year period from 2002-2014). They even identify three specific areas that correlate with higher than average returns: emission and resource reduction, workforce, and society. So in my exploration, I’ll dig into some specific CSR breakdowns with the data I have on S&P 500 companies.
We’ll do this in 3 steps:
- Get our tools in place to clean the data sets and pull out the information we need (this post)
- Run some correlations between ESG metrics and stock performance
- Assess the significance of our results
Let’s get started.
What we’ve done here is graphed the returns on stock prices in the 1990’s for companies that had a non-minority CEO in 1999 and those that had a minority CEO in 1999. This particular ESG variable is unlikely to significantly affect returns, not least because it occurs at the very end of the testing period. The point here is not the results. The point of this notebook was to figure out which columns we need from the data. The next step is to combine several of the annual KLD spreadsheets together and get data throughout a decade, then compare that to returns throughout that decade.