Back in 2011, when AT&T and T-Mobile were attempting to gain approval from the Federal Communications Commission for an ultimately doomed $39 billion merger deal, an unusual coalition of interest groups submitted comment to the agency in support of the merger: the NAACP. The Gay & Lesbian Alliance Against Defamation (GLAAD). A homeless shelter in Louisiana. The Asian Pacific Islander American Scholarship Fund.

The groups were united by two common threads. The first was their lack of any apparent stake in telecommunications policy. The second was the fact that they had all recently received donations from AT&T, in some cases totaling six figures or more.

Although AT&T denied any quid pro quo over its contributions to nonprofit groups, the donations raised eyebrows on Capitol Hill and elsewhere. The president of GLAAD resigned over the contributions, as did a number of board members. Ultimately, the merger fell through.

While it’s tempting to write off the incident as a one-time blunder — heavy-handed corporate lobbying gone amok — new research from Marianne Bertrand at the University of Chicago and others finds that’s not the case. Merging charitable-giving data of Fortune 500 companies with a complete record of public comments submitted to the federal government on proposed regulations between 2003 and 2015, Bertrand and her colleagues are able to trace how individual corporations influence the rulemaking process via nonprofit donations.

The data set compiled by the researchers demonstrates three crucial findings: First, after a firm donates to a nonprofit organization, that group becomes more likely to comment on rules that the firm has also commented on. Second, the organization’s comments in those cases have more similarities with the firm’s comments than with comments from other nonprofit organizations not receiving money from the firm. And finally, when a firm and its grantees comment on a rule together, regulators' final remarks on the rule are more likely to be in line with the firm’s comments on the rule.

The research “unearth[s] a channel of influence going from corporations to policymaking that was not measured before,” co-author Matilde Bombardini said in an email, “both in terms of monetary magnitude and impact on the agency’s views on the various regulations.”

The paper exclusively considers federal rulemaking, which is the process by which agencies deliberate on how laws and policies are interpreted and carried out in the real world. When an agency proposes a rule (for instance, a telecom merger approval, the addition of a new drug to a list of controlled substances or the applicability of mandatory rest requirements for different types of workers) it typically allows public comment on the proposal for a period of 30 to 60 days. Individuals and other interested parties — including lawmakers, businesses and nonprofit organizations — are allowed to weigh in on the proposal.

At the end of the period, the agency issues a final rule. The rule may be unchanged from its initial iteration, or it may be altered or even withdrawn completely based on the public input received. A conceptual diagram of the process would look something like this:

But Bertrand and her colleagues found evidence that, in many cases, the process is not that simple. Comments received from businesses and ostensibly independent entities may be linked by money that flows well outside the process. When that happens, the final rule is more likely to reflect the donating corporation’s views.

“Naturally firms favor profits — i.e., private benefit — over the social good,” study co-author Raymond Fisman said in an email. “So Coke wants to sell more soft drinks even if [there are] increases in obesity. Supposedly independent nonprofits are meant to provide input that acts as a counterweight. If they’re co-opted, they won’t.”

To draw these conclusions, Bertrand and her colleagues first identified 629 charitable foundations operated by 474 firms appearing on the Fortune 500 or Standard & Poor’s 500 composite index lists at any point from 1995 to 2016. Using IRS data, they then identified all 225,180 nonprofit entities receiving gifts of greater than $5,000 from these charitable foundations from 1998 to 2015.

Next they pulled the complete set of public comments on proposed rules submitted to between 2003 and 2016. They were then able to identify not only when individual corporations commented on a proposed rule, but also when nonprofit organizations that firm had donated to commented on the same rule.

They found that when a firm donates to a nonprofit group, it’s associated with a two- to four-fold increase in the likelihood that the nonprofit group will comment on the same proposed rule as the firm. “The magnitude of this effect is large,” the researchers wrote, with some degree of understatement.

When relationships like these surface publicly, as in the case of the proposed AT&T/T-Mobile merger, the nonprofit organizations involved usually say that donations do not influence their policy work. “One of the unique things about the NAACP is that financial support does not determine our civil rights positions,” the head of the North Carolina chapter of the NAACP told Politico in 2011. Similarly, a GLAAD spokesman said, “We do not make policy decisions based on what’s best for our corporate sponsors.”

Smaller groups make similar claims. “Their money that they gave was in no way connected with what we did,” the director of the Louisiana homeless shelter told the Center for Public Integrity about AT&T’s $50,000 donation to the group five months before it commented on the merger. “No one leveraged me or anything,” said the director of the Asian Pacific Islander American Scholarship Fund.

But the researchers found that when a nonprofit organization comments on a rule that a donor company also comments on, the language of the comments tends to mirror each other. Using machine learning techniques, the researchers analyzed the language of these comment pairs and found that they were significantly more similar to each other than pairs of comments selected at random from a given docket.

A corporate donation, in other words, made a nonprofit group considerably more likely to support that corporation’s view in regulatory matters.

Using similar techniques, they at last turned to the language used by regulators in drawing up and discussing the final rules. They find, again, that when a firm and nonprofit groups it has donated to comment on a proposed rule, the language used in these regulatory discussions tends to more closely mirror the language of the firm’s comments than it would otherwise.

What this work does, in effect, is trace the path of influence from a corporate donation to an ostensibly independent nonprofit group’s comment and all the way through to a final regulatory outcome. “It’s presenting an important link between corporate donors and the public policy conversation,” John Wonderlich, executive director of the Sunlight Foundation, said in an interview.

Nonprofit organizations often present themselves as independent entities not aligned with any particular for-profit interests. They are “providers of nonpartisan, technical expertise and are commonly expected to offer more neutral input into the lawmaking and rulemaking process, with a focus on cost-benefit analysis and broader societal interests,” Bertrand and her colleagues write.

But their research shows how corporations can influence nonprofit organizations' input into the policymaking process, which can “distort the outcome of the political process away from the public good and towards private interests,” as they put it. Policymaking ends up reflecting the interests of the people who have enough money to make their voices stand out. Government becomes less representative. Authorities make decisions that benefit the wealthy few, rather than the broader society.

“The biggest losers are constituents who end up getting their perspectives drowned,” Wonderlich said.