On Sept. 19, a federal judge approved UnitedHealth Group’s acquisition of Change Healthcare over the concerns of the U.S. Department of Justice. The judge’s approval rested on a false assessment of the risk that United/Change will abuse its access to consumer and rival data to increase profits at the expense of market competition and consumer welfare, writes Professor Theodosia Stavroulaki.
The Affordable Care Act (ACA) of 2010 prevented health insurers from discriminating against citizens on the basis of their pre-existing conditions, gender, race, or ethnicity and from charging unhealthy individuals’ higher premiums or refusing them coverage altogether. Before the ACA, fifty million Americans were uninsured. By 2016, nearly 20 million more Americans had health coverage.
However, health insurers still discriminate against people suffering from pre-existing health conditions, often by using data collection and big data analytics to identify and avoid offering coverage to consumer groups that may suffer from chronic conditions, such as diabetes, or that are predisposed to such conditions in the future. The recent merger between UnitedHealth Group, a giant health insurer, and Change Healthcare, the largest U.S. electronic data interchange (EDI) clearinghouse, illustrates how health insurers are looking to big data to develop more precise tools for assessing individual health risks. The deal will give United access to millions of consumers’ health data and rival insurers’ sensitive business data, which it can then use to disadvantage its rivals and, ultimately, the most at-risk individuals.
The U.S. government challenged the United/Change merger on the basis that, among other anti-competitive behaviors, United could use Change’s data to learn how rivals control demand for healthcare services, negotiate with providers, shape their insurance networks, and pay claims. United could also apply machine learning to gain insights into rivals’ business strategies and innovations, reducing its own and its rivals’ incentives to innovate. United would also be better equipped to identify healthier consumer groups that entail lower risks and, thus, higher profits for itself. Cherry-picking the most profitable consumer groups would give United a strong competitive advantage, especially if United’s major rivals are deprived of access to a similar range and quality of data. Ultimately, competition for the least healthy consumer groups that urgently need access to healthcare would be eliminated, crippling the ability of rival health insurers to compete, harming consumers, and diminishing innovation in the health insurance industry.
Sadly, the presiding district court rejected the government’s data concerns and blessed the merger. The lower court explained that to prove a Section 7 violation of the Clayton Act, upon which the government made its case, the government must show “that the proposed merger is likely to substantially lessen competition, which encompasses a concept of reasonable probability.” However, the court said, the government’s concerns that United would misuse Change’s data to imitate its rivals’ innovations and “pick-off” the “healthiest” and “more profitable” consumer groups, thus undermining competition for people most in need of healthcare, was unrealistic. Second, the court claimed the government’s core argument that the United/Change deal may undermine insurers’ incentives to innovate was unconvincing.
Specifically, the court rejected the government’s stance as too speculative. The court claimed any harm to competition would occur only if United/Change was willing to uproot its entire corporate culture; “intentionally” breach or alter its “longstanding firewall policies”; violate “existing contractual commitments”; and ruin its reputation. Because the government did not convincingly argue that all these “extreme actions” may realistically occur, the court concluded that the merger did not necessarily raise anticompetitive concerns.
The court’s decision does not imply that the government did not show how the United/Change merger could harm competition in the health insurance industry. The government even submitted several “hot documents” showing that United was motivated to acquire Change so that it could learn its rivals’ “custom claim edits,” and piggyback “on its rival efforts to develop proprietary edits” that “differentiate them” and “give them a competitive advantage.” The court dismissed these documents as “mere references to data and data rights” that do not imply that any harm to competition may occur in the real world. The court also said any misuse of rivals’ data would violate United/Change’s firewall policies and harm the firm’s reputation. As such, the court alleged that the merged firm’s incentives to protect its rivals’ data outweighed its incentives to misuse this data.
In an ideal world, where the merged firm’s customers can identify any misuse of their data without incurring significant search costs, the court’s line of thinking may make sense. Indeed, any violation of United’s firewall policies may harm the merged firm’s reputation and cost it customers. But this analysis relies on two presumptions: that United/Change’s rivals can detect any misuse of their data and, even if they can, that they can easily shift business to a rival EDI clearinghouse.
But these presumptions do not necessarily hold. First, United/Change would have more information than its rivals about how their data is used. Given this information asymmetry, rival health insurers may never identify any violation of the merged firm’s firewall policies or any “data misuse.” Second, and as the government argued, prohibitively high costs of migration may prevent rivals from switching to a new EDI clearinghouse even in the event of known data misuse.
Consider Amazon. The recent EU antitrust investigation against the giant tech firm contends that Amazon favors its own goods and services by using rivals’ data to adjust its offers and undermine their position in the market. Amazon also promotes Amazon Basics products over third party-sellers’ products through targeted price discrimination or biased rankings (search and recommendations). Nevertheless, Amazon’s business strategies do not discourage independent sellers from using its platform because its market dominance leaves them no other choice. To leave would cost these sellers visibility and sales.
The court did not only overestimate the reputation costs United/Change would incur if it breached its firewall policies. More importantly, the court underestimated how much United/Change would benefit from a potential violation of these policies.
Health insurers like Cigna and Aetna have started merging with drug companies to increase their access to our health data. They want to record our individual biology, our medical history, our levels of well-being, how much we sleep, our rates of sugar, junk food, and nicotine consumption, if we exercise and how often. In other words, they want to shape for each of us a unique digital health ID. Health insurers may use consumers’ health-related information to reduce their costs in various ways. They may nudge consumers towards healthier habits. But they may also try to reduce their costs by using healthcare algorithms to identify the types of customers that they are likely to attract with preconditions such as diabetes or depression and move the associated treatments to a higher cost-sharing tier.
For example, an official complaint filed with the Department of Health and Human Services in May 2014 revealed that health insurers in Florida that were providing plans through the federal marketplace had classified all HIV drugs, including generics, in the highest cost-sharing tier to dissuade people with HIV from choosing their plans. Alternatively, health insurers may stop cooperating with providers that have a strong reputation in treating diabetes or HIV. By effectively preventing high-risk groups from accessing their services, health insurers reduce their risks and increase their profits. They also raise rivals’ costs if the rivals do not have the range or quality of health data to identify and likewise avoid servicing high-risk consumer groups. To avoid exiting the market, competing insurers may then have to merge with drug suppliers, EDI clearinghouses, or other health tech companies to identify these groups and even further limit their access to healthcare.
United benefits from acquiring Change by using Change’s massive variety and volume of health data to make predictions about consumers’ future health and denying rivals the same opportunity to acquire a major health tech company and its data on millions of consumers. These advantages allow United to expand greatly its share of the health insurance industry and increase premiums over the long term for all consumer groups, not just the least healthy. The significant profits to be made also outstrip any costs from potential misuse of data.
The government only partially raised these concerns. For instance, it explained why the merger may allow United/Change to identify the least healthy consumer groups but did not delve into (a) the discriminatory practices United/Change would in reality apply to avoid offering them coverage; (b) the discriminatory practices health insurers already apply against the least healthy consumers to further increase or maintain their dominance; or (c) the ways in which vertical data-driven mergers in the healthcare field may facilitate health insurers’ efforts to evade the anti-discrimination requirements imposed by the ACA and monopolize the industry. By omitting to adequately analyze these concerns, the government missed the opportunity to convince the district court that United/Change’s benefits of misusing rivals’ data may far outweigh the costs. The court also missed the opportunity to prohibit a healthcare merger that may deprive the least healthy Americans from accessing affordable health insurance. The consequence is that a merger between UnitedHealth Group and Chance Healthcare may unravel the earnings of the ACA and further increase health disparities in the U.S.