Examining ESG data in more detail
Tuesday, March 12, 2019
Tuesday, March 12, 2019
A deep dive of sustainability topics with Ethic's lead data scientist
Examining ESG data in more detail
Tuesdays with Travis is a collection of monthly interviews with our data science lead, Travis Korte, that explores the complexities of expressing values through data.

Sustainable investing is only as good as the markers used to measure “sustainability.” That’s why at Ethic, the curation, analysis, and implementation of our data is of the utmost importance. It’s also one of the most dynamic parts of our job.

So how do you interpret data in the sustainable investing context? How do you make that information actionable? Is it costly to incorporate?

We’re launching a series of interviews with our lead data scientist, Travis Korte, to highlight our approach and underscore some of the decisions we make along the way. This first Q&A introduces some of the complexities involved in measuring impact. Be sure to check out the key takeaways at the end.

Comments have been edited and condensed.

Travis, tell us about your role. What does it mean to be a data scientist for an impact investing firm?

A lot of my job is about taking all the signals we can get from the global equities market and identifying which ones we think matter. Aside from the technical aspects of vetting, processing, and incorporating these data points into our models, I spend a lot of my time thinking about measuring impact and diving deep into academic research on different environmental and social issues. The goal is to put together a set of really thoughtful, useful, and intuitive tools for investing more sustainably.

How do you take big, if not opaque topics such as “social impact” or “governance” and translate them into actionable investment data?

It depends on the issue that we’re looking at. We use a broad range of ESG data providers, some of whom incorporate a sense of materiality into their data. So in those cases, you’re not just getting a data point that some company has, for example, poor toxic waste management policies; instead, you’re getting an estimate of how likely it is that that particular company’s toxic waste management will yield real financial impacts. Materiality is useful for pretty intuitive reasons: we don’t expect the toxic waste management policies at a software company to be as comprehensive as those at a metal mining company, because software companies don’t have a lot of toxic waste crises to prevent.

In other cases, we don’t get materiality estimates in the data, and it’s up to us to make some of those assessments. For example, it turns out that it’s fairly easy to collect a lot of useful indicators for corporate governance, because companies are publicly releasing a lot of it already—things like the number and affiliations of board members, how director elections work, or (in some countries) what the executive compensation looks like. Setting aside any inaccuracies or biases in self-reporting—which are topics for another time—there are dozens of these relatively well-defined and publicly reported governance indicators where you can generally say, ‘yes the company does this’, or ‘no it doesn’t.’ But that doesn’t capture whether or not an indicator really matters in a particular market or a particular industry. So we take all these signals and look deeply at which ones are the most predictive of real-world impacts, meaning both sustainability impacts and financial impacts. Some variables don’t make the cut.

If “sustainability” means different things to different clients, and it’s constantly evolving, how do you keep up with a potentially endless universe of metrics?

We like the endless growth of metrics. It makes our models better. We can better report on impact if there are new approaches to measurement that we hadn’t thought of, we can toss things out when they stop working as well, and we’re constantly developing new lenses through which to view the impacts of investments.

A lot of this comes from clients. For example, one issue we cover is military contracting; aside from manufacturing these broadly condemned weapons like cluster bombs and anti-personnel landmines, big defense contractors have business models predicated on the continued growth of global conflict. Traditional responsible investing approaches to military contracting aren’t very informative: either you have war profiteers in your portfolio or you don’t. But we recently got a data set that gives the percentage of revenues of the world’s top military contractors that come from contracts issued by departments of defense, and we can use that more granular data to paint a much more meaningful picture of impact. “You divested from six weapons manufacturers” is a lot less tangible than “You no longer have a stake in X billion dollars of defense contracts.” That data was brought to our attention by a client, and we love it when that happens because it gives us a head start on how to think about communicating that information.

One of the challenges of sustainable investing is implementing sustainability criteria while limiting tracking error*. How does Ethic approach the tradeoff between tracking error and causes?

It depends. Certain kinds of causes are much easier to account for and incorporate into a portfolio. It’s relatively easy, for example, to decrease the carbon emissions associated with a portfolio. There are a few companies that are the worst offenders and a longer tail of companies that are hardly emitting at all, so it’s fairly straightforward from a technical perspective to clean up. Other kinds of sustainable issues are much tougher to take a hard line on and still yield something that resembles the benchmark. For example, even though we think it’s a really important issue, we only have limited leverage against anticompetitive practices. The reason is that the companies most likely to be engaged in those practices tend to be among the bigger companies in any index, and so trying to reduce your exposure to anticompetitive practices is, in a way, equivalent to reducing your exposure to the index itself. This is kind of a disconcerting thought, and it points to larger issues in the financial system, but my hope is that we can surface some of these issues and inform broader debates in the community.

We try to be really transparent about those kinds of conundrums because the tradeoff between a sustainability emphasis and tracking error is a personal choice. One thing we get fairly often is clients asking for both an aggressive screen on the oil and gas industry and a strong favoring of companies that invest in clean energy technologies. The tricky part is that those are the same companies. ExxonMobil, for example, is one of the world’s biggest investors in clean energy technology. If you’re looking to satisfy both criteria, you have hardly any companies left on the table, and your tracking error gets unmanageable very quickly. Again, disconcerting thought. But these are the current realities of the market and so these are the decisions investors have to make. Does it feel more intuitive to you to divest, invest or some balance of the two? Do you want to get rid of all your exposure to the company, or do you want to keep some shares so you can have a voice in proxy votes to improve the company? We have these kinds of conversations all the time, and they differ from client to client.

*Tracking error is defined as the standard deviation of the difference between the returns of the portfolio and its underlying benchmark. The smaller the number, the more tightly bound the portfolio return should be to the Underlying Benchmark return. Tracking error is also known as “active risk”. Tracking error does not indicate whether returns are expected to be “above” or “below” a benchmarks return, rather the expected divergence (positive or negative) from the benchmark. Tracking error is a relative measure of risk versus the benchmark where volatility is the total fluctuation of the portfolio. For example: if a portfolio has a tracking error of 1%, its return is expected to be within 1% of its benchmark return approximately every two out of three years.

Are companies responding to the sustainable investing movement? Do they feel pressure to have goals beyond profit?

Yes and no. Companies tend to recognize the public relations and reputational risks associated with the worst and most egregious sustainability issues. This is Nike in the 2000s finally starting to clean up the sweatshops. This is Uber in 2017 pushing to reform its toxic governance structures. It’s clearer to companies now than ever before that the reputational and brand impacts of those kinds of behaviors will be felt strongly and immediately. What’s unclear is the extent to which companies are making these changes proactively, as opposed to waiting until some big scandal finally spurs change.

For some sustainability issues, there are cosmetic measures that companies can take immediately, programs they can create in a quarter or two that improve their ESG ratings. And, naturally, companies tend to gravitate to these kinds of reforms. But some issues are deeper, more systemic and tougher to fix, and the alternative to fixing them now is waiting until some big investigative report comes out and public outcry forces changes. So my hope is that we’re able to express the consequences of these sustainability issues that are less obvious. We’re still at the very beginning of that.

Key Takeaways

  • Not all criteria that may indicate sustainable behaviors have actual financial or sustainability impacts. The discussion of materiality, or relevance, is therefore key to converting sustainability criteria into actionable investment ideas.
  • The reality that sustainability means different things to different people only enhances the quality of sustainability data sets. More granular data allows for more meaningful translation: a shift from "you divested from six weapons manufacturers” to “you no longer have a stake in X billion dollars of defense contracts,” for example.
  • Implementing certain sustainability criteria can be costly from a tracking error perspective. While our job is to quantify that cost, the tradeoff between a cause and tracking error is a client’s personal choice.
Sources and footnotes
Contributors

Travis Korte is the Data Science Lead at Ethic. Previously, Travis organized civic-minded technologists at Hack for LA and advised a wide range of clients on data science, data policy, and quantitative methods. You can follow him on Twitter at @traviskorte.

Melissa Mittelman creates content at Ethic and is an alumna of Bloomberg News, where she covered private equity & deals. Melissa previously worked at Deutsche Bank, providing institutional, cross-asset sales coverage for ultra-high-net-worth investors.

Jay Lipman, a co-founder of Ethic, is driven by the need to address climate and environmental risks with the resources to which we each have unique access. He has been ranked among the Forbes 30 Under 30: Social Entrepreneurs.