This article was originally published in ChicagoBoothReview
People living in communities near the Washington Works DuPont factory in Parkersburg, West Virginia, in spring 1984 barely noticed employees on assignment surreptitiously gathering samples of their drinking water. Under analysis, the samples—taken from locations including public water fountains as well as tap water from one employee’s home—revealed bad news: perfluorooctanoic acid, a chemical used in Teflon manufacturing, could be detected in the nearby drinking water. Evidence suggested the chemical, also known as PFOA or C8, caused birth defects in lab animals.
DuPont executives discussed the situation. The pollution violated environmental rules and carried legal risk, to say nothing of health risks. But DuPont went on producing Teflon in the same way, as related in a new documentary, The Devil We Know. According to court documents the company increased production, stopping only in 2013, as lawsuits mounted.
DuPont, one of the most respectable companies in the United States, ended up causing environmental damage that cost it around $1 billion. Why did a “good” company act so badly? IDC Herzliya’s Roy Shapira and Chicago Booth’s Luigi Zingaless coured internal company documents disclosed in court to better understand what went wrong. The researchers rule out ignorance as well as bad governance. Rather, they argue, “the harmful pollution was a rational decision”—and they suggest ways to change that.
DuPont, founded in 1802, is one of the oldest US companies, created when President Thomas Jefferson called attention to the country’s need for gunpowder. It has been considered one of the most respected US companies, has scored highly on environmental and corporate governance metrics, and has employed many thousands of scientists.
One of those scientists created Teflon, which many consumers know as a nonstick coating used on pans and other cookware. For decades it was manufactured using PFOA, which allowed for Teflon to be less lumpy. Making PFOA requires combining carbon and fluorine, which creates a heat-stable and water-resistant chemical that we now know is toxic. PFOA is both biopersistent and bioaccumulative—it stays in the environment and accumulates in the bloodstream.
The dangers weren’t publicly discussed until the mid-1990s, when a family of farmers, the Tennants, who lived near DuPont’s West Virginia plant, suspected that emissions had caused their cattle to die after drinking water from a creek close to a DuPont landfill. The family hired an attorney, and information the plaintiff lawyers received in the family’s suit led to a class action and other trials. An independent science panel established that 3,500 of the 70,000 people living near the DuPont plant suffered from diseases related to PFOA. Last year DuPont settled all the cases against it for $670 million. By then it had spun off the division that made Teflon and would soon merge the remainder of its business with the Dow Chemical Company.
Today, Chemours, the company that emerged from DuPont’s performance chemicals business, still makes Teflon for a variety of uses beyond cookware, including furniture, paints, and fabrics. It no longer uses PFOA in the process. DuPont’s longtime PFOA supplier, 3M, phased out the chemical in 2000—and this past February settled its own pollution claims for $850 million. The US Environmental Protection Agency has established a nonenforceable health advisory about PFOA and a related fluorinated organic chemical but doesn’t regulate the concentration of PFOA in drinking water, promising that “as science on health effects of these chemicals evolves, EPA will continue to evaluate new evidence.” A summit “provided an unprecedented opportunity for stakeholders to share vital information and best practices,” according to a statement released by former EPA Administrator Scott Pruitt, who said a national management plan would be developed.
The massive disclosure of court documents interested Shapira and Zingales, who wanted to understand how the PFOA debacle could have happened at such a reputable company. Did DuPont’s managers not understand the potential health consequences? Did the company pollute by accident? Was poor corporate governance the issue?
The legal cases against DuPont made public thousands of papers and emails ranging from memos to water-sampling results. Zingales suspects that the company’s lawyers “didn’t know what they were dumping” in the first wave of documents. And because DuPont’s legal department didn’t initially realize personal emails sent from company laptops and BlackBerries were discoverable in court, the researchers ended up with internal DuPont correspondence that was perhaps painfully honest.
In particular, they point to a memo about a meeting that took place in 1984 at the company’s Delaware headquarters. By that point, DuPont executives had concerning information about PFOA. They knew, among other things, that it was toxic and needed to be handled with extreme care. Executives from 3M had alerted DuPont to a study where rat fetuses whose mothers had been exposed to PFOA developed eye defects. After DuPont scientists reviewed the 3M study and deemed it valid, DuPont started monitoring the babies of its own female workers who had worked with PFOA while pregnant. Two out of seven babies monitored were born with birth defects related to the eyes and nostrils.
The 1984 memo indicates executives considered several options. DuPont could stop using PFOA, continue to use it but invest in abatement measures, reduce production, or continue with no abatement. DuPont chose to continue using PFOA—and in fact to use twice as much of it. “The fundamental question is why,” write the researchers. Abatement was relatively cheap and could have prevented health problems to people and legal and reputational damages to the company. “How come the company chose the option that seems worse for society and for the company itself?”
Shapira and Zingales, 2017
The researchers quickly ruled out the possibility that DuPont executives were simply ignorant of the potential consequences and costs. While the 1984 memo doesn’t suggest that executives decided to consciously harm people, it does suggest there were many uncertainties about the effects of PFOA and safe level of exposure.
“Documents from after May 1984 complete the picture and help us further rule out ignorance,” write the researchers. By 1986, 3M had warned DuPont executives that PFOA should be incinerated or dumped in a commercial landfill. DuPont did neither. It then learned that PFOA is carcinogenic in animals and potentially humans. And after setting a threshold for PFOA of 1 part per billion in drinking water, DuPont didn’t alert the public when it detected PFOA levels vastly exceeding that level in several water samples. The documents also provide evidence that executives were warned that PFOA decisions could have legal and reputational considerations.
The researchers sought to replicate the executives’ cost-benefit calculation. On the benefit side of that equation, DuPont’s executives estimated that replacing or dropping PFOA would have put $100 million to $200 million in annual profits at risk, which the researchers say conservatively calculates to $1.1 billion in present-day dollars.
On the cost side, however, were $350 million in societal health costs. Shapira and Zingales arrive at that figure using the uptick in the incidence and prevalence rates of six diseases among the 70,000 people living near the West Virginia plant.
DuPont could have paid to incinerate PFOA, which would have lessened the societal costs but would have involved an up-front cost of $1 million, and an ongoing operating cost of $1 million annually. Following general principles of analysis used by the late economist Ronald Coase, Shapira and Zingales compared the present-value cost of abatement to that of the potential societal damage and find that from a social efficiency point of view, abatement was clearly the right solution. But economic incentives changed the picture significantly.
Executives weighed the potential benefit to shareholders against the potential costs of polluting, which were related to the probability of being caught. The executives couldn’t have known exactly what the costs could be, but they had some indications, and the researchers estimate the legal liabilities were around $100 million in 1984 dollars. That’s a far cry from the potential $1.1 billion in profits that were at stake. Executives, the researchers calculate, would have found it optimal to continue producing PFOA without abatement measures as long as the probability of being caught was less than 19 percent.
In fact, the perceived probability of being caught was far lower than that, they argue. The Tennant family hired a successful environmental lawyer with extensive corporate experience, an unlikely ally except that the lawyer’s grandmother had lived nearby. Many DuPont documents likely came to light only because its legal department hadn’t centralized the collection and storage of documents. The class action was filed in one of the few states where plaintiffs were able to continue a suit without proving actual physical harm. And DuPont agreed to accept the conclusions of an independent science panel, presumably knowing there was no hard evidence available tying PFOA emissions to health problems—but a bit of legal brilliance enabled the plaintiffs to gather such evidence. The Tennants’ lawyer used part of the class action settlement to pay thousands of people living near the plant to give blood, thereby gaining the data needed to establish a link between PFOA and six diseases.
“The combined probability of all four above-mentioned events alone is likely below 19 percent,” write the researchers. “In fact, the likelihood of the first event—landing an experienced defendant lawyer to try your case—is in itself an event with a probability of less than 19 percent.” The odds, then, were in the executives’ favor. It was entirely rational for them to decide to pollute.
A DuPont spokesman, when contacted about the research, issued this statement:
It is important to point out that documents cited in the paper reflect that actions taken by DuPont, at all relevant times, were guided by the company’s good faith belief that low levels of C8 exposure do not pose a health hazard. In fact, neither the C8 trials nor the settlement had anything to do with the actual scientific evidence relating to the health effects of C8 exposure. The juries were not permitted to hear scientific evidence concerning whether C8 is a proven cause of the conditions at issue in those trials or the magnitude of the risk, if any, at the levels of claimed exposure. As a result, juries lacked important context for evaluating DuPont’s decision-making.
The parties agreed to accept the conclusions of an independent scientific committee, notes Zingales, who calls the statement “highly misleading.”
Several other factors also worked in DuPont’s favor, the research indicates. The company benefited from a time lag. The fact that lawsuits can drag on for years, or decades, affected the cost side of the equation. It also takes years for chemical producers to establish evidence that a chemical has negative outcomes, Shapira and Zingales write. And chemical producers, as well as companies in all types of environmental and health-related cases, benefit from internal access to data before the public sees the information. Meanwhile regulators, faced with complexity, uncertainty, and potential conflicts of interest, can be weak enforcers.
As for reputational considerations, the DuPont case didn’t receive serious media attention until the government intervened in 2003, by which time 19 of 29 directors who had been on the company’s board in 1984 had died or retired. And media outlets didn’t blame the remaining directors for the PFOA debacle. “To the contrary, we found some articles celebrating these executives as environmental visionaries,” write Shapira and Zingales. Local journalists generally avoided the story, note the researchers, and national reporters can overlook stories considered too local. When lawsuit findings did finally damage DuPont’s brand, the company’s spin-off and merger helped the company distance itself from the problems.
Shapira and Zingales see generalizable lessons. With penalties low and regulators weak, regulation can fail to prevent bad behavior. Legal liability doesn’t necessarily work, either. Penalties take too long to arrive, and watchdogs don’t materialize. When called out publicly, companies can take steps to reduce reputational damages.
What could have changed the calculation in this case—and the calculations for executives in similar circumstances? Shapira and Zingales offer a few rough ideas for better aligning executives’ goals with societal ones. For starters, they suggest the federal government should offer monetary rewards to whistle-blowers. These types of reward systems already exist in corporate-fraud and tax-evasion cases and could be extended to environmental liability ones.
Gag-order settlements, which prohibit parties from discussing or commenting on details of a settled case, can keep information with societal consequences from circulating. Shapira and Zingales suggest that the government should levy a higher tax on companies that seal settlement documents than on those that do not. They also suggest penalizing companies for delay tactics—often used by defense attorneys—by increasing the rate of interest at which the awards for damages are paid out.
It could also help to assign personal accountability within companies, they argue. DuPont’s top managers benefited from a “faceless crime” dynamic in which individual executives were not held personally responsible, they write. In future cases, making specific people take a more personal stake in the company’s decisions, by having them vouch for policies and knowledge, could make it tougher to skirt consequences.
It would be a mistake to see this only as a story about PFOA or DuPont, the researchers stress. Rather, they consider it a case study of a well-established, highly regarded company whose strong reputation made its executives’ decision-making process particularly valuable to study. When pharmaceutical companies hide health data, or food companies mislead consumers about ingredients or health benefits (see “Consumer challenge: Spot the false claim,” Fall 2015), their decisions could be the rational products of similar decision processes. To change the equation, make it harder to suppress and distort potentially damaging information, the researchers argue. “If this can happen in the best possible company, because the rules in place don’t force good behavior,” says Zingales, “it can happen anywhere.”