Stock image of a stethescope

When it was signed into legislation 11 years ago, the Affordable Care Act (ACA) was designed to reduce health care costs, increase the availability of health insurance and provide consumers with a minimum level of insurance benefits.

Recent research, however, is shedding new light on one aspect of the ACA: how much policy premium revenue is actually spent on policyholder benefits. Specifically, the research looks at how insurers report the amount they spend on required benefits and whether that self-reported amount is accurate.

The paper, “Accounting-Based Regulation: Evidence from Health Insurers and the Affordable Care Act,” was authored by Andrew Van Buskirk, associate professor of accounting and management information systems at Fisher, and his colleagues, Evan M. Eastman, of Florida State University, and David L. Eckles of the University of Georgia. The paper is forthcoming in The Accounting Review.

They found that insurers’ reported claims estimates — used by the federal government to assess whether those insurers were meeting the ACA’s spending requirements — were consistently overstated in situations where more accurate reported estimates would have triggered rebate payments to customers. In short, insurance companies overestimated the reported cost of policyholder benefits, which reduced how much the companies would have had to pay back to individuals.

“It’s easy for these insurance companies to understand what they’re receiving from policyholders in terms of the number of individuals enrolled and their premium payments,” Van Buskirk said. “The challenge for them is figuring out the benefits they’ll be providing in return. These costs can extend out for many years and therefore require the insurers to make estimates about what those costs will ultimately be. Whenever you’re dealing with that kind of estimation, there are going to be estimation errors, but you also have the possibility for firms to be misleading and to bend or bias the estimates they report to regulators.”

“The ACA, and the way it’s written and its regulatory policies, provide a clear incentive to bias these reported estimates: If companies don’t report a sufficiently high level of benefits, they have to pay refunds.”

The researchers reviewed data that pre-dated the adoption of the ACA, as well as data from the first five years of the ACA (at the time, the most current information available from the Centers for Medicare & Medicaid Services). They focused on a figure called the medical loss ratio (MLR), a relative measure of claims to premiums. In all, the researchers looked at data from several hundred unique firms spanning 2003-2014.

To get a complete picture of the accuracy of companies’ estimates, the researchers used the reported data from the ACA’s rolling three-year MLR calculation. For example, in order to evaluate data from 2015, they looked at initial estimates filed in 2016 and assessed whether those 2015 estimates were revised (up or down) in the subsequent 2017 and 2018 filings.

“We looked at those revisions. When the companies revised their estimates, did it look like they were consistently revised upward or downward,” Van Buskirk said. “We found firms were consistently revising downward, which revealed that their initial estimates were too high.”

They also looked at companies that had incentives to manipulate their reports — such as avoiding paying out rebates ­— versus those with no incentive. They inferred that approximately 14% of firms with the incentive to manipulate did so. And publicly traded firms, particularly those with a history of financial restatements, tended to overestimate claims.

While the results of the study were significant, Van Buskirk cautioned against assumptions that overestimation is a widespread issue.

“If there is reporting manipulation occurring, it’s not universal,” he said. “I think many people would be surprised that only 14% of firms with the incentive to bias their reports actually do so.” 

He also noted that their data does not allow them to identify which individual insurers are likely to be biasing their reports. 

“If we look at an insurer that over-estimated its initial claims, we can’t tell whether that individual insurer intentionally biased their claims or, alternatively, whether they reported a good-faith estimate that turned out to be too high,” Van Buskirk said.

The researchers also used detailed MLR information to quantify the effect of these reporting errors on the policyholder refunds the insurers were required to pay. They estimated that rebates for their sample were understated by $219 million, or roughly 10%, during their five-year sample period, a statistic that was more prevalent in the individual policy market. Taking into account the insurers not in their sample, this 10% underpayment would translate to roughly $400 million in total underpaid rebates for the same period.

So what can be done?

Solving the issue of overreporting health care benefits wasn’t necessarily the focus of the paper, but its findings may be helpful in ongoing regulation of health insurance companies and the ACA.

The research shows the potential cost to policyholders created by inaccurate or inflated benefits estimation. The researchers also say that future regulation can be designed to eliminate the type of misreporting incentive provided by the ACA’s self-reporting structure. In this particular case, they theorize incorporating a clawback provision based on data already obtained by regulators. This lookback provision would largely eliminate firms’ incentives to bias their reports.

“Using financial information to regulate insurer behavior can be effective,” Van Buskirk said. “But it’s important for legislators and regulators to understand exactly what that financial information represents and how it might be manipulated. Accounting-based regulation will be less effective if it doesn’t anticipate that potential manipulation.”

The ACA, and the way it’s written and its regulatory policies, provide a clear incentive to bias these reported estimates: If companies don’t report a sufficiently high level of benefits, they have to pay refunds.

Andrew Van BuskirkAssociate Professor, Accounting and Management Information Systems


Andrew Van Buskirk FCOB Distinguished Professor
Faculty Profile