PanAgora’s CIO & head of sustainable investing explain firm’s ESG framework, best practices

Waters Wavelength Podcast Interview Series: PanAgora’s George Mussalli and Mike Chen hit on topics including building predictive models using point-in-time data, and balancing ESG portfolios.

Podcast Timestamps

2:00 To start, George and Mike give background on PanAgora and the firm’s ultimate investment strategy.

6:00 Mike explains what ESG means to PanAgora.

9:00 George lays out the investment objective for a chief investment officer when incorporating ESG into a portfolio.

15:00 Let’s get hypothetical: A company is extremely climate efficient and has extremely high employee satisfaction, but they have terrible board diversity and make monetary contributions to organizations that are deemed poor when it comes to social justice issues—do you ditch that company even if it’s producing solid returns? How do you build a flexible framework to adjust an ESG portfolio easily?

23:00 Mike and George talk about how the firm builds predictive and forward-looking models in nature, even if the input data tends to be point-in-time.

29:00 Mike drills into avoiding data bias in the ESG space.

35:00 George explains how ESG metrics fared under the strain of the pandemic.

36:00 Mike says greenwashing is the greatest challenge facing ESG investors as we head on into 2021 and beyond.

PanAgora Asset Management is a Boston-based, quantitative investment manager that has built a framework to incorporate ESG metrics into the firm’s overall investment strategy.

George Mussalli, chief investment officer of equity investments at the firm, and Mike Chen, the firm’s director of portfolio management and sustainable investing, joined the Waters Wavelength Podcast to talk about a range of topics relating to ESG.


One of the topics broached looked at how a manager builds models that are predictive and forward-looking in nature, even if the data going into the model tends to be point-in-time (23:00).

This is a topic that was recently raised by Mary-Catherine Lader, who, at the beginning of 2020, was appointed to the newly-created role of head of Aladdin Sustainability at BlackRock. She told WatersTechnology that she expects to see sustainability data—which is just one piece of the overall ESG pie—“transition from being a point-in-time snapshot, to more predictive and forward-looking.”

She continued: “Today, we have a few facts about a company; in the future, we expect that you’ll have lots more unstructured data at your fingertips that an investor can use a software tool to predict—to model—how a company’s performance in a certain area might change over time.”

PanAgora is also looking to address this point-in-time data challenge to drive more future-looking insights that yield alpha. Chen said that one reason the firm can build more predictive models is advancement in the fields of natural language processing and, more generally, machine learning.

He gave the example of a company that emits 10 million tons of carbon into the atmosphere annually, which is not great, to say the least. But if you simply look at that piece of point-in-time information, you might miss the larger picture. Let’s say that the company’s management has also put out a very concrete plan that shows what they’re doing to reduce their emissions, and they set a firm percentage-reduction outlook by a specific date, perhaps by introducing a new type of technology into the manufacturing process. Perhaps then that company becomes more palatable to include in an ESG portfolio.

“If you can somehow read into that report—which is more descriptive rather than a pure number—you can actually gauge management on whether their plans are effective,” Chen says. “And more than that, you can actually gauge them on whether their plans are credible by looking at the words and the context of the words that they use. So you can actually gain a lot of forward-looking, predictive information if you apply some of these advanced technologies, such as NLP.”

  • Innovation Exchange: Mike Chen will be a speaker at this year’s Innovation Exchange, a virtual conference that will be held from March 22-25. To listen to Chen’s panel discussion and others, you can register here

The tech, essentially, allows a company like PanAgora to ingest more—and potentially better—data. But to get to that point, Mussalli said that it’s important for the humans to first think about what data they need—essentially, which company characteristics are most likely to lead to outperformance?

“After long discussions, we then go out and look for this data—a lot of times, quants tend to do the opposite,” Mussalli said.

Every morning, he added, he receives a flood of emails from data providers pitching “unique” offerings. Most recently, those pitches have tended to include data around Reddit forums like r/WallStreetBets. Mussalli said that that method is reactive. “If you give a data scientist a piece of data, they’re going to look for a signal and then make up the story after. What we do is kind of the opposite.”

At PanAgora, the equity investment team comes up with a fundamental idea, and then they go out and look for data that provides a full picture of that idea. While he acknowledged that they might miss out on some opportunities, the group has “a pretty good hit-rate” using this method for security selection.

“The challenge is, when it’s hard to find the data, the alpha potential is very high; and once everybody has the data, it goes away,” Mussalli said.

For example, about 15 years ago, PanAgora would manually collect same-store sales figures from retailers like Gap and Home Depot, and interns would type that data into a spreadsheet, which would then be loaded into the asset manager’s model for this type of investment vehicle.  

“It worked great; it was the biggest alpha producer in the model for a long time,” he said. “Then one day, Bloomberg has a field—same-source sales, you type it in, you download it, it’s gone.”

Fifteen years ago, ESG data was one dimensional and backward looking, because if you’re the only company that has that data and knows how to use it, it could generate alpha. Today, to calculate something like consumer strength, “it requires terabytes of data and a machine learning algorithm that’s run on the cloud because we don’t have enough computing power [on PanAgora’s internal servers],” he said. “The idea is the same over the years, but the amount of data that we need to capture to be ahead of the curve is exponentially bigger.”

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact info@waterstechnology.com or view our subscription options here: http://subscriptions.waterstechnology.com/subscribe

You are currently unable to copy this content. Please contact info@waterstechnology.com to find out more.

Most read articles loading...

You need to sign in to use this feature. If you don’t have a WatersTechnology account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here