Challenging the Calculator: The Capricious Bug in the Affordable Care Act

By: Chris Kozak

The Affordable Care Act is probably the most controversial piece of legislation since the Civil Rights Act. To date, the Act's mandates have been challenged mainly on substantive grounds.[1] By rebuffing most of these challenges, the Court has signaled it is done ruling on the Act's substance. However, opponents of the Act have overlooked a potentially fatal weakness in the Act's design:

It is a nightmare from the depths of Administrative Law hell.

Professor Seth Chandler, in his article Regulation by Calculator,[2] unearthed a striking set of facts that could be used to bring the Act to a grinding halt. The Act relies on a so-called “actuarial value calculator,” a pre-programmed Excel spreadsheet that allows insurers to input over 200 plan variables. The calculator then uses embedded coding to decide if the plan is legal—and if it is, how valuable it is to purchasers.

If done properly, this is an ingenious way for the government to conduct large-scale regulation. Instead of holding individual adjudications to determine a plan's viability, an insurer can download the spreadsheet, pop in its proposed data, and get an answer. The calculator's substantive decisions may then receive Chevron-style deference if an insurer dislikes the result. This sort of deference is desirable if the efficiency of “regulation by calculator” is to be preserved, and the courts would probably be willing to grant it.

The problem? No one knows how the calculator works. The code underlying the calculator and its origins are opaque and poorly documented. Even worse, there seems to be no meaningful way to test the calculator against itself. Professor Chandler identified multiple internal inconsistencies within the calculator's code indicating that either (a) the data underlying the code is incorrect, or (b) the documentation describing the code is flawed. As he concludes, “Neither alternative is particularly comforting.”[3] 

An unflinching axiom of Administrative Law is that agencies must explain the rules they issue. As the Supreme Court has famously said, an agency must "articulate a satisfactory explanation for its action, including ‘a rational connection between the facts found and the choice made.”’[4] Anything less is arbitrary, capricious, and unlawful.  Moreover, an agency that decides similar cases differently is equally guilty of caprice, since inconsistency is the hallmark of bad logic (and bad coding).[5] These principles are rooted in the reasons we are even comfortable with having agencies in a democracy: If we force them to be transparent, we can root out deceit, incompetence, and corruption.   

All this lays the groundwork for a claim by an insurer that the spreadsheet is arbitrary and capricious under APA § 706(2)(A). If the Center for Medicare and Medicaid Services (CMS) is using a spreadsheet with opaque and inconsistent coding to decide which plans are lawful, then the spreadsheet loses both our trust and its legal force. Although a favorable ruling from the courts would simply remand the spreadsheet to CMS, the case could serve as the tip of the political spear for those who want to kill the Act. It is beyond doubt that humans can create reliable code to govern insurance plans. This case, however, would create a question about whether it can be done on a national scale with a public budget. Future failures by CMS to produce consistent coding and reliable documentation[6] may convince the median voter that the “invisible hand”—with all its externalities and vices—is the only beast capable of producing the capital necessary to sustain regulation by calculator.

[1] Zubik v. Burwell, 136 S. Ct. 1557, 1559 (2016), https://www.supremecourt.gov/opinions/15pdf/14-1418_8758.pdf; Julie Rovner, As Supreme Court Sends Back Birth Control Case, Both Sides Claim Victory, NPR.org, (May 16, 2016, 3:55pm), http://www.npr.org/sections/health-shots/2016/05/16/ 478262940/supreme-court-sends-obamacare-birth-control-case-back-to-lower-courts; King v. Burwell, 135 S. Ct. 2480 (2015), https://www.supremecourt.gov/opinions/14pdf/14-114_qol1.pdf; Nat'l Fed'n of Indep. Bus. v. Sebelius, 132 S. Ct. 2566, 2585, 2593 (2012), available at https://www.supremecourt.gov/ opinions/11pdf/11-393c3a2.pdf.

[2] Seth J. Chandler, Regulation by Calculator: Experience Under the Affordable Care Act, 2016 Mich. St. L. Rev. 465.

[3] Chandler, supra note 2, at 479-87.

[4] Motor Vehicles Mfg. Ass'n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983), https://supreme. justia.com/cases/federal/us/463/29/case.html.

[5] See, e.g., FCC v. Fox Television Stations, Inc., 556 U.S. 502, 535-36 (2009) (Kennedy, J., concurring in part and in the judgment), https://www.law.cornell.edu/supct/pdf/07-582P.ZC1.

[6] This is not beyond the realm of possibility. In 2013, a graduate student published a paper demonstrating that an influential economic study on austerity—written by two Harvard Professors—was seriously flawed because of a simple coding error. Chandler, supra note 2, at 484 n.59.