At a time when the headline-grabbing antics of, say, Martin Shkreli, have highlighted rising drug prices, citizen petitions are a stealthier and innocuous-sounding way for companies to keep out competitors. It’s part of the usual, to use an industry term, “product lifecycle management.” Patents grant exclusive rights to companies as an incentive to develop new drugs. But once a drug patent has expired, the incentive is then to be creative about keeping out generic competitors.

Gaming the FDA’s citizen petition this way puts the agency in a tricky spot. “The FDA’s number one job is safety. That’s what they’re good at. They’re not designed as a competition policing agency,” says Robin Feldman, a law professor at UC Hastings and a co-author of the paper. For example, the FDA can summarily deny petitions that it finds frivolous—but it never has. The agency tends to err on the side of caution because the consequences of approving an unsafe drug are so high.

According to a study last year, brand-name pharma companies filed 92 percent of 505(q) citizen petitions, a specific type related to pending generic applications. And in turn, 92 percent of those petitions were ultimately rejected by the FDA.

For its part, the FDA is is aware of a problem. In its last annual report to Congress about citizen petitions, the agency wrote it was concerned that current law “may not be discouraging the submissions that are intended primarily to delay the approval of competing drug products and do not raise valid scientific issues.” (The FDA declined to comment specifically for this story.)

The scope of the problem is bigger than the FDA. Michael Carrier, a law professor at Rutgers, who has also studied the citizen petition process, noted that the issue sits at the junction of several areas of law: antitrust law, patent law, state drug laws, and the Hatch-Waxman Act that set up the generic approvals system. “When it gets really complicated,” he said, “there’s a chance to play in all those joints and have anticompetitive conduct.”

Carrier cites EpiPens has an example of where Mylan used multiple delay tactics against a competitor. First, Mylan and Teva, the company seeking to make a generic EpiPen, settled a lawsuit in 2012, in which Teva agreed not to enter the market until June 2015. Months before the settlement period was supposed to end, Mylan filed a citizen petition about Teva’s product. Then, weeks before the FDA was supposed to respond to that petition, it filed a supplement containing the Mylan-commissioned study questioning Teva’s version. Carrier notes that Mylan waited years to file its citizen petition. The last-minute timing raised his suspicions.

In a statement, a Mylan spokesperson said, “Our concern is that the product doesn’t meet the standards for being approved as a generic version of the EpiPen Auto-Injector. For that reason, we submitted a petition, which is a public process that gives anyone with a view the opportunity to voice an opinion by submitting comments, and puts the decision in FDA’s hands, where it belongs. For the record, FDA denied our petition.” (Teva’s generic EpiPen has since been delayed for other reasons. It does not expect to have a product out until late 2017 or 2018.)