US pension fund teams up with academics to cut through ESG fog

Ethically minded investors considering putting money into Porsche face a dilemma: one data provider deems the car maker’s environmental, social, and governance (ESG) credentials substantially below average, while another thinks the opposite.

Research originally published in 2019 found similar divergence in the ESG ratings of many other companies, including Johnson & Johnson and AT&T. Pension Reserves Investment Management believes such discrepancies are hampering ESG investing and so the fund has joined forces with MIT Sloan School of Management to find a better way to spot ESG stars and laggards.

“We think that [the divergence] poses a major challenge for investors trying to integrate ESG into the investment process,” says Michael Trotsky, chief investment officer of Prim, which manages the 60%-funded, $80 billion state pension fund for Massachusetts. “If the data isn’t good, then it’s hard to have an impact and construct a portfolio that reflects your views.”

The aptly named Aggregate Confusion Project was launched in September by MIT Sloan School of Management and it hopes to enlist more asset managers and asset owners.

The business school has analyzed the correlation between the ESG ratings from six well-known providers, including MSCI and Sustainalytics, and found that it is only 0.61 on average. By comparison, credit ratings from Moody’s and Standard & Poor’s are correlated at 0.92.

The wide divergence in ESG scores forms part of what Roberto Rigobon, a co-author of the research, calls “noise”—discrepancies in available information on a firm’s ESG performance. 

If the [ESG] data isn’t good, then it’s hard to have an impact and construct a portfolio that reflects your views

Michael Trotsky, Pension Reserves Investment Management

Another example of such noise is the difference in data coming from different sources, says the professor of applied economics at the MIT business school.

“If I have a sensor that measures CO2 emissions versus if the company tells me what their CO2 emissions were…there are different degrees of how accurate those measures are,” Rigobon says.

He adds that “it is not clear” the sensor is a better source of data than the company statement, because the sensor will measure emissions from one particular factory, which may not be representative of the whole company.

“So the noise comes from many different sources: the data that the different rating agencies are collecting, the procedures that those rating agencies have that are different, and the fact that some of them pay attention to different sets of [ESG] attributes,” Rigobon concludes.

One of the project’s aims then is to improve the measurement of specific aspects of companies’ ESG performance—for instance, carbon emissions and the treatment of employees.

Gaming ESG scores

Apart from divergence, another snag with vendor ESG ratings is that they often combine three separate groups of characteristics—environmental, social, and governance—to produce a single company score.

As a result, companies can manipulate their ESG scores by, for example, improving their environmental metrics to offset a slide on the labor aspect, Rigobon says.

And that is a problem for asset managers such as Prim.

“We believe companies that operate with consideration to ESG issues will make higher-quality investments and will perform better over the long term,” Trotsky says. “The question is identifying companies that are truly making an impact rather than just gaming a score.”

So another goal of the Aggregate Confusion Project is to develop better ways to aggregate separate ESG factors into composite indexes.

The discrepancies in vendor ESG scores are compounded by the different impacts they have on the relevant stocks. Rigobon and his colleagues have found that, for example, a 5% improvement in a rating from MSCI has almost twice the impact on a stock’s price as a 5% improvement in a rating for the same firm from Sustainalytics, albeit based on a methodology that Sustainalytics has since changed.

The findings will be published in an upcoming paper on how divergence in vendor ESG ratings affects equity returns.

Difficult choices

Investor interest in ESG principles is growing. Assets in sustainable funds worldwide reached a record high of $1.3 trillion by the end of September 2020, up around 60% since early 2019, according to the latest data from Morningstar.

But within that overall interest, there is a diversity of priorities. The Aggregate Confusion Project aims to find a reliable way to assess the ESG preferences of underlying investors.

Prim, for example, serves a wide spectrum of investors, from teachers to policemen who may hold different views.

“What we’re trying to do is establish the priorities of Prim’s constituents. We’re seeking common denominators to inform our investment strategy,” Trotsky says.

He notes that MIT Sloan School of Management has developed a system for extracting investors’ preferences and that Prim plans to use it with its constituents “to do a better job in reflecting their views”.

What we learned…is, first, how different people are. Second, we learned a huge amount about how your [ESG] preferences change with your income level, your gender, your religion

Roberto Rigobon, MIT Sloan School of Management

Rigobon explains that the business school has run an experiment asking participants to allocate money to three out of seven social causes, “with different trade-offs”. The causes were education, poverty alleviation, environment, women’s empowerment, civil rights, public health for the poor, and biodiversity. To make their choices, the participants had to play a specially designed game.

“What we learned from that is, first, how different people are,” he says. “Second, we learned a huge amount about how your preferences change with your income level, your gender, your religion.”

The question is how these different preferences should be aggregated, Rigobon says. It is a question the Aggregate Confusion Project also seeks to answer.   

“Rating agencies are not going to be able to construct [ESG] rankings that do not take into account the preferences of the investors and, more importantly, asset managers and asset holders will not be able to conduct investments without understanding the constraints imposed by their clients,” states a document describing the project, noting that asset managers already take risk preferences into account.

Rigobon says that, rather than rolling different investors’ views into a single portfolio, it may be necessary to construct a number of portfolios to suit different clusters of clients.