Cambridge, MA, July 22, 2026 (GLOBE NEWSWIRE) — Many institutional investors depend on the Environmental, Social, and Governance (ESG) ratings they pay for to assess risk and build resilient portfolios. They also don’t want the scores that judge how environmentally, socially, and governance-conscious a company is to shift too often or sharply since it’s costly to reshuffle portfolios.
In “The Traffic Light Effect in ESG Ratings,” a new working paper, MIT Sloan School of Management principal research scientist Florian Berg has assessed how raters approach this delicate balance between accurate information and stability.
Berg and his co-researchers discovered evidence that the world’s largest ESG ratings provider held ratings steady even when underlying data suggested a change was warranted. Along with Jess Cornaggia from Pennsylvania State University, Cristian Foroni from European University Institute, and Francesco Tripoli from Harvard Business School, Berg found that companies’ ESG scores, as provided by MSCI, pile up just before the line that would change their letter grade, as if they were held at a red traffic light.
Researchers found that this “traffic light effect” happens when routine adjustments made by internal MSCI review committees create ratings that differ from what the provider’s methodology otherwise would have dictated. These suppressed downgrades have a real cost too: when ratings don’t move, a company’s stock price stays flat when it likely should have fallen, leaving investors holding an asset that’s probably worth less than they think.
“Because the investor pays for the data set, you would think that what they’re interested in is the most accurate information,” Berg said.
Why scores don’t cross the line
The researchers set out to examine whether a conflict of interest is baked into the ESG ratings business itself, as ESG ratings now determine which companies make it into major indexes, which funds are allowed to hold which stocks, and ultimately where trillions of dollars flow. Unlike credit ratings, where a company defaults or doesn’t, there’s no clean way to verify whether an ESG rating is “correct,” giving raters significant wiggle room.
Berg and colleagues looked at MSCI ESG Ratings monthly, from 2014 to 2022, covering around 16,000 companies a year and focusing on ratings decisions that triggered review by committee. Four conditions could trigger review: when a company is proposed for an upgrade to AAA (highest rating) or downgrade to CCC (lowest rating); when a change of two or more notches is suggested; when the analyst believes the model-determined rating doesn’t reflect the company’s actual situation; or when there’s uncertainty about data quality or coverage.
The researchers reverse-engineered MSCI’s internal scoring process, working backward from published ratings to figure out what the companies’ scores would have been if MSCI had followed its own publicly documented rules, and then they compared those numbers to published ratings.
They found that MSCI’s scores piled up just below the thresholds for higher letter grades and just above the thresholds for lower ones, Adjustments tended to be larger when a company’s rating would have broken from what its competitor showed, and smaller when raters disagreed with each other anyway. The bunching was absent in the raw underlying scores before committee review, meaning it was being introduced by human intervention, not math.
“When scores would cross the threshold, committees keep them below that threshold,” said Berg, with one interpretation being they acted as such to avoid ratings volatility, according to the researchers.
Results suggest MSCI’s internal committees adjust scores to keep companies from getting upgraded or downgraded too often — which could trigger costly portfolio reshuffles for its clients — rather than letting only methodology dictate decisions.
Indeed, the researchers found that only 24.3% of the 129,702 firm-years — the data representing a specific company’s financial, operational, or economic metrics during a single fiscal year — that underwent a committee reassessment led to a letter grade change, a striking sign of how rarely MSCI allows ratings to move. Among those that did change, upgrades outnumbered downgrades nearly two to one (20,588 vs. 10,944), suggesting an upward bias in the changes that do get implemented.
Testing whether MSCI reacted to peers
To rule out the possibility that MSCI and its peers were reacting to the same underlying information about companies, the researchers exploited a quirk in the data: a competing company and another major ESG rater overhauled its methodology in 2020 and retroactively rewrote its entire historical database.
The researchers had access to both the original pre-2020 scores and rewritten ones, creating a rare natural experiment. This allowed them to separate whether MSCI’s adjustments were responding to what peer raters were actually saying at the time, or simply reflecting commonly available information about the underlying companies.
Results confirmed the pattern: MSCI’s adjustments tracked peer ratings that existed at the time of each decision, not the rewritten scores that came later, ruling out the possibility that both were simply reacting to the same fundamentals.
What is the price of a buried downgrade?
Berg and colleagues also found a tangible real-world financial impact. When a ratings downgrade does go through, the affected company’s stock drops an average of 2.8% over the following year. But when a warranted downgrade gets suppressed, no such signal emerges. The information never reaches the market, and investors are left in the dark.
In the months immediately following a suppressed downgrade, the gap between what the stock does and what it would have done had the downgrade gone through represented roughly $300 million to $400 million in avoided market-value loss for a company with a $10 billion market cap, a concrete measure of what’s at stake.
Investors should understand that the ESG ratings they pay for, and build portfolios around, are not purely the product of MSCI’s stated methodology, Berg said. They may be getting stability instead of accuracy, and missing out on signals that would affect returns.
“You have to be aware as an investor,” said Berg. “ESG ratings don’t necessarily reflect the most timely information because there’s this incentive to slow down the timeliness to reduce portfolio turnover, and that’s impacting investors’ returns. If this is what investors want, they should be able to buy it, but the question is, does every investor want it?”
- A new MIT Sloan School of Management study finds that the ratings investors rely on are quietly optimized for stability by MSCI.