The term “benefit corporation” is becoming increasingly familiar as a label for companies with a hybrid commitment to the social good as well as profit-making. Interestingly, the term itself has taken on a hybrid meaning.
“Benefit corporation” is a legal structure recognized by most U.S. states, but the term is often used synonymously with “B Corp,” a certification granted by B Lab, a global NGO. The confusion is understandable, given that B Lab was the principal lobbying force behind the adoption of “benefit corporation” as an official legal entity.
“The legal framework is viewed by proponents as a way to help benefit corporations ‘lock in’ their social commitments. It layers on additional requirements that traditional for-profit legal forms don’t have related to non-financial reporting and the resolution of conflicts over these goals,” says David Lucas, assistant professor of management at the Costello College of Business at George Mason University.
Lucas’s recently published paper in Journal of Business Research uses AI to shed light on what happens when a benefit corporation becomes a B Corp. Does the dual affiliation equate to redoubled commitment, or “greenwashing” in stereo?
The paper was co-authored by Clara Scheve of Hamburg University of Technology and Joel Gehman of George Washington University.
The researchers took advantage of Minnesota’s unusually robust monitoring regime for benefit corporations. “A few studies now have documented relatively low compliance across states, and seemingly few penalties,” Lucas says. “Minnesota is the only state we could find where the Secretary of State published every benefit report that they received, and if they don’t receive one in the annual timeframe, they revoke benefit status.”
However, Lucas emphasizes that there is a difference between reporting regularly and reporting substantively. The disclosures were by no means equally meaningful or forthcoming. “There’s huge variation,” Lucas explains. “Some of the reports are just the cover sheet that’s required and a couple of sentences. Others have long, well-structured reports with a lot of detail and quantitative evidence.”
The researchers aimed to use report quality as a signal of sincerity and effectiveness in pursuing social or environmental impact goals. But first they would have to construct a balanced and comprehensive barometer of quality, over and above very coarse criteria such as length. That is where AI came in.
Lucas and his co-authors trained large language models (LLMs) to rate the reports according to what Lucas calls “the things we really care about: how they talk about stakeholders, how they talk about impact, how transparent they seem to be, etc.” All in all, the researchers accessed 766 benefit reports uploaded to Minnesota’s Secretary of State website from 2015-2023.
The reports submitted by benefit corporations that were also B Corps were significantly longer (by approximately 1.38 pages on average), and of markedly high quality. In particular, they provided far more evidence of real-world impact, greater transparency on challenges and more clarity on next steps.
“There are at least two mechanisms behind these findings,” Lucas says. “First, benefit corporations that also pursue B Corp certification tend to be more serious about the mission. They pay fees to get certified, and they go through the actual audit and process, which is extensive and costly to retain. Second, B Lab’s certification requirements yield a lot of tools to communicate impact. The process of being audited and having to account for these things gives firms measures, both qualitative and quantitative, that speak to these issues in detailed and specific ways.” Rather than undermining or duplicating one another, the legal form and the certification appeared complementary.
For Lucas, the study serves as an example of public and private entities filling in each other’s blind spots. NGOs, after all, have little legal pull, while governments often lack resources for sharp oversight and close collaboration with companies. “Redundancy can appear wasteful, but it has a real value, not just across public and private oversight, but even across different kinds of private certifications,” Lucas says.
Viewed in this light, Minnesota’s approach may be a happy combination of compulsion and flexibility. “The healthiest thing possible is to allow businesses to say as much or as little as they want, but they have to say something,” Lucas says. “That way, all stakeholders are able to see and assess for themselves, and the process is shielded from political swings about what counts as ‘social value’.”
Lucas’s analysis implies a unique role for AI in measuring quality abstracted from politics. “There are other legal forms that are emerging across different countries that are getting at the same or similar goals,” Lucas says. “My ongoing research on these new legal forms and their role in society uses AI in a similar fashion, pulling new dimensions and insights out of this reporting.”