Medicare — the large government program that provides healthcare for disabled Americans and those 65 and over — suffers from a significant fraud problem.

While healthcare providers are subject to complicated Medicare reimbursement criteria, the unscrupulous find ways around these rules, making fraudulent claims to dip into public coffers and boost profits. According to the Centers for Medicare and Medicaid Services, $31.2 billion was spent on improper payments in 2019 alone; this accounts for more than seven percent of all Medicare spending.

How best to combat this fraud? Historically, Medicare has used a pay-and-chase approach, where it first “pays” claims submitted by healthcare providers and then “chases” improper payments by referring suspected fraud to the FBI and Department of Justice for investigation. Providers receiving these improper payments may be subject to both civil and criminal enforcement, and guilty verdicts can result in large civil penalties and even jail time.

Another approach would be to require providers to show detailed documentation of a medical need before paying out for a claim. So-called “prior authorization” has been criticized for adding paperwork, red tape, and bureaucracy to an already heavily regulated healthcare system.

In new research forthcoming in the Journal of Political Economy, David Eccles School of Business economist Paul Eliason and co-authors examine this question using Medicare’s End-Stage Renal Disease program as a laboratory. “Patients with failing kidneys typically visit a dialysis center three times a week to have their blood cleaned,” Eliason said. “And while many patients are healthy enough to arrange their own transportation to and from dialysis, the very sick must sometimes be transported by ambulance.”

Medicare rules allow reimbursement to ambulance providers for dialysis transport, but only when the patient’s medical conditions make the ambulance the only safe way to travel. The problem? Because Medicare reimbursement rates for ambulance travel are generous and dialysis patients must visit centers regularly, ambulance companies can make big money by operating as a taxi service, shuttling sick — and maybe-not-so-sick — dialysis patients to and from their appointments.

“Our data show that the monthly number of ambulance rides to centers more than tripled between 2003 and 2014,” Eliason noted. “However, the number of dialysis patients grew only 54% over this period, suggesting that Medicare’s rules were widely ignored. And we know they were being flouted because over this period the DOJ prosecuted over 100 firms committing this fraud, winning virtually every case.”

To compare the effects of regulation and litigation on healthcare fraud, Eliason and co-authors rely on two main facts to generate quasi-experimental variation in Medicare data. First, federal law enforcement officials have considerable discretion over whether to pursue criminal and civil litigation for Medicare fraud, and prosecutors were more active in some federal court districts than others. The announcement of litigation against an ambulance company in a given district should put other companies in that district on notice that they are being watched and could therefore provide a powerful deterrent effect.

“To analyze the causal effect of this litigation,” Eliason explained, “we calculated how both rides and billing changed after the litigation was announced, using the federal court districts where there was no litigation as a control group. The existence of this control group allows us to statistically net out other factors that were changing rides and billing nationally, and really nail down the impact of litigation.”

The second key fact is that in 2014, Medicare implemented prior authorization for non-emergency ambulance claims using a staggered rollout starting with three states — New Jersey, South Carolina, and Pennsylvania — that had unusually high rates of non-emergency ambulance transport. This requirement was then extended to five other eastern states and the District of Columbia in 2016 and was finally implemented nationwide in 2022.

“The staggered rollout was vital for our research,” Eliason emphasized. “In 2015, only three states were subject to prior authorization, so we calculate a treatment effect of regulation comparing trends in these three states to the rest of the nation. Then by 2017, we have an additional six states added subject to this requirement, so we again use the rest of the nation as a control group.”

The results are lopsided in favor of regulation over litigation as a fraud-fighting measure.

In states where ambulance providers were required to obtain prior authorization before driving a patient to dialysis, Eliason and co-authors find a massive 68% reduction in non-emergency rides. “This effect was immediate and persistent and suggests very large savings associated with this form of regulation,” Eliason said. “If Medicare had simply implemented prior authorization in 2003 rather than waiting until 2014 to pilot the program, the federal government would have saved $4.8 billion.”

In contrast, Eliason and co-authors found civil litigation to have no statistically identifiable effect whatsoever. That is, in federal districts where prosecutors were actively pursuing civil litigation for ambulance-taxi-related fraud, ambulance companies behaved the same as in districts where there was no such enforcement. Criminal litigation was shown to have a modest deterrence effect, with reductions in monthly rides and monthly billing of 24 and 19 percent, respectively.

But what of the effects on dialysis patients? Are they harmed when bureaucrats must pre-approve expenditures for services as simple as ambulance rides? “The 68% drop in ambulance rides made us really worry about the impact on patients,” Eliason said. “It’s possible that increased fraud enforcement could cause some patients to skip dialysis sessions, harming their health and quality of life, and leading to more hospitalizations and even deaths. This effect could lead to higher costs for Medicare and eat into the $4.8 billion in potential savings.”

Examining more than 15 million observations of monthly patient-level data from Medicare’s United States Renal Data System, Eliason and co-authors find no evidence of adverse effects on patients. “Prior authorization had very close to zero effect on patients,” Eliason noted. “We see no statistically significant impact on monthly dialysis sessions, suggesting patients are still able to get to their appointments, and no apparent impact on hospitalizations or deaths. This holds true even when we restrict attention to patients who relied on ambulance rides the most.”

Prior authorization also had large effects on the market for ambulance services. Eliason and co-authors show that the number of ambulance companies fell by almost 25% in the month immediately following prior authorization and that the firms exiting the market were typically those that had offered a higher share of non-emergency rides. “We also found an interesting pattern of specialization in the wake of prior authorization,” Eliason said. “Firms that offered very few non-emergency rides seem to have stopped offering this service entirely, while other firms appear to specialize by offering only non-emergency rides. This is consistent with economies of scale in administrative capabilities, where firms are willing to invest in learning how to gain prior approval for ambulance services only if they can spread that cost over a sufficiently large number of patients.”

The study also contributes to a long-running debate among economists over whether litigation or regulation is the most cost-effective tool to enforce anti-fraud rules. Litigation concentrates enforcement resources, which in this context would be investigative and court costs, on only those cases where fraud is suspected. “What makes litigation really appealing is that it can deter would-be fraudsters. Once you get deterrence, it drives down enforcement costs because firms govern themselves to avoid prosecution,” Eliason explained. Regulation, on the other hand, usually requires up-front enforcement costs in the form of claim review, documentation, and monitoring. These can be smaller than the litigation costs, but they must be incurred by everyone, not only those suspected of fraud. “Our study is the first large-scale empirical analysis to look at the relative effectiveness of these approaches,” Eliason said.

Eliason highlighted two potential reasons why regulation may outperform litigation in the ambulance-taxi setting. For ambulance companies, he explained, the expected cost of committing fraud depends on the probability of being caught and the size of the penalty associated with being caught. “Both are probably low in this context,” he said. “Investigative resources are limited, and most investigators don’t see enough healthcare fraud cases to become experts. This reduces the probability of detection. Further, many ambulance companies are small, and limited liability makes it easy for them to spend down ill-gotten gains, leaving only a small asset base for the government to recover in civil litigation.”

Paul Eliason, Riley League, Jetson Leder-Luis, Ryan C. McDevitt, and James W. Roberts, 2024, “Ambulance Taxis: The Impact of Regulation and Litigation of Health Care Fraud,” Journal of Political Economy, forthcoming.