CBO’s Record of Projecting Subsidies for Health Insurance Under the Affordable Care Act: 2014 to 2016

When it was enacted in March 2010, the Affordable Care Act (ACA) made major changes to the U.S. health care and health insurance systems. At that time, the CBO and the staff of the Joint Committee on Taxation (JCT) estimated the budgetary effects of the legislation to provide that information to the Congress and to update overall projections of the budget.

CBO and JCT originally estimated that the gross costs of the ACA would be more than offset by reductions in Medicare spending, increases in revenues, and other changes—such that enacting the legislation would reduce the federal budget deficit over the 2010–2019 period. This document compares projections and subsequent outcomes for health care subsidies that account for the majority of such costs and concludes the following:

• The agencies’ estimates of those subsidies proved to be close to actual amounts for 2014 and 2015 but were much too high for 2016, and
• In general, other organizations’ estimates of health care subsidies produced around the time of the ACA’s enactment were much too high for all three years.

The report does not examine the effects of other provisions of the ACA; for many of them, the actual outcomes cannot be separately identified in administrative or survey data, nor can the law’s overall effects on the deficit.

What Aspects of Their Estimates Did CBO and JCT Analyze for This Report?

The ACA significantly expanded eligibility for Medicaid; created health insurance marketplaces through which certain individuals and families receive federal subsidies; established a mandate for most legal residents of the United States to obtain health insurance or pay a penalty if they are not exempt; reduced the growth of Medicare’s payment rates for most services; imposed an excise tax on insurance plans with relatively high premiums; made changes to the federal tax code—including an increase in the Hospital Insurance payroll tax rate for high-income taxpayers, a surtax on those taxpayers’ net investment income, and annual fees imposed on health insurers; and made various other changes to Medicare, Medicaid, and other programs. Isolating the budgetary effects of the ACA (or of any complex legislation) is difficult because they are often embedded in the spending for existing programs—Medicare, for example—and in broad categories of federal tax revenues.

This report focuses mainly on CBO and JCT’s estimates of subsidies for which total amounts can be separately identified in administrative or survey data—specifically, federal spending for people made newly eligible for Medicaid by the ACA and subsidies for health insurance received through the marketplaces and the Basic Health Program. (The Basic Health Program allows states to offer subsidies to certain low-income people that are based on the subsidies available through the marketplaces.)

To examine the factors that caused CBO and JCT’s estimates of those identifiable subsidies to differ from actual amounts during the 2014–2016 period, this document compares projections and subsequent outcomes for the following groups of people:

• Those made newly eligible for Medicaid under the ACA,
• Those receiving subsidies through the marketplaces or the Basic Health Program,
• Those enrolled in nongroup insurance coverage purchased through the marketplaces or with coverage obtained through the Basic Health Program, and
• Those without coverage.

Because a decision by the Supreme Court in 2012 significantly affected eligibility for subsidies for health insurance by making the expansion of Medicaid optional for states, this document compares actual results with two sets of projections: those that CBO and JCT made in May 2013 (after the Supreme Court decision but before the coverage expansions under the ACA were implemented in 2014) and the original projections the agencies made in March 2010.

How Did CBO and JCT’s Estimates Compare With Actual Results?

In CBO and JCT’s March 2010 and May 2013 projections, the agencies’ estimates of federal spending for people made newly eligible for Medicaid by the ACA and of subsidies for health insurance received through the marketplaces and the Basic Health Program were close to the actual amounts for 2014 and 2015 but much too high for 2016. In total, for 2014 and 2015 combined, the estimate of those identifiable subsidies that CBO and JCT prepared in March 2010 was 14 percent above the actual amount. The May 2013 estimate of that total was 14 percent below the actual amount. But for 2016, the March 2010 and May 2013 estimates were above the actual amount by 43 percent and 29 percent, respectively.

To a great extent, the differences arose because actual results deviated from the agencies’ expectations about how the economy would change and how people and employers would respond to the law. To a lesser extent, the differences were caused by judicial decisions, statutory changes, and administrative actions that followed the ACA’s enactment. (For additional information, see “Notes” at the end of this document.)

How Did CBO and JCT’s Estimates Compare With Those of Other Organizations?

Compared with estimates produced in 2010 by other organizations—namely, the Office of the Actuary in the Centers for Medicare & Medicaid Services (CMS), RAND Corporation, and The Lewin Group—CBO and JCT’s March 2010 estimate for 2014 and 2015 combined was much closer to what actually happened. And CBO and JCT’s March 2010 estimate for 2016 was about the same as the average of estimates by the other organizations, though both amounts were substantially too high.

Those other organizations produced estimates in 2010 of a broader category of subsidies than those that can be separately identified in administrative data. Specifically, the broader category combines two types of subsidies:

• Identifiable subsidies for federal spending for people made newly eligible for Medicaid by the ACA and subsidies for health insurance received through the marketplaces and the Basic Health Program and
• Other changes in spending for Medicaid and the Children’s Health Insurance Program (CHIP) that stemmed from the ACA’s health insurance coverage provisions but that cannot be separately identified in administrative data.

Because the actual outcomes for those other changes cannot be identified, in this report, the estimates by other organizations (and corresponding ones by CBO and JCT) are compared with an estimated outcome for the broader category that combines actual amounts for the identifiable subsidies (90 percent of the outcome for 2016, for example) with estimates, from CBO’s March 2016 projections, of other spending on health insurance coverage provided through Medicaid and CHIP that was probably attributable to the ACA (10 percent of the outcome for 2016).

Those other changes in spending on coverage provided through Medicaid and CHIP stem primarily from higher enrollment among people eligible for those programs under rules in effect before enactment of the ACA but who were prompted to enroll by that legislation. The actual amounts of spending for those enrollees are not identifiable in administrative data because distinguishing which previously eligible enrollees were induced to enroll by the ACA and which would have enrolled in the absence of the law is impossible. So, for this analysis, CBO estimated the outcomes for those enrollees in Medicaid and CHIP using updated data and methods.

How Do CBO and JCT Analyze Subsidies for Health Insurance?

CBO and JCT used a multistep process to produce their original estimates of the effects of the ACA on health insurance coverage and federal costs. As with all major legislative proposals affecting health insurance coverage, analysts read the proposal to understand and identify its aims, consulted with outside experts, reviewed the empirical evidence from existing studies, analyzed states’ relevant behavior, and used several models to capture the complex interactions in the markets for health care and health insurance. The agencies have followed a similar process to produce subsequent baseline projections of the budget under current law, incorporating new data and methodological improvements each time.

CBO and JCT use a microsimulation model to estimate how the number of people with coverage and sources of insurance would be affected by changes in eligibility and subsidies for—and thus the net cost of—various insurance options. On the basis of administrative and survey data, that model incorporates a wide range of information about a representative sample of individuals and families, including their income, employment, health status, and health insurance coverage. The model also incorporates information from the research literature about the responsiveness of individuals and employers to price changes and the responsiveness of individuals to changes in eligibility for public coverage. The role of employers is particularly important because more than half of the U.S. population under age 65 has employment-based health insurance coverage and because offers of health insurance from employers affect eligibility for subsidized insurance in the nongroup market.

CBO and JCT regularly update that model so that it incorporates information from the most recent administrative and survey data on insurance coverage and premiums as well as information from CBO’s most recent macroeconomic forecast. In addition, they update it to incorporate the effects of relevant judicial decisions, statutory changes, and administrative actions. The agencies use the model—in combination with models to project tax revenues, models of spending and actions by states, projections of trends in early retirees’ health insurance coverage, and other available information—to inform their estimates of the number of people with various types of coverage and the associated federal budgetary costs. For example, in their March 2010 estimates, CBO and JCT projected that 92 percent of people under 65 would have some type of health insurance in 2016; in their May 2013 estimates, the agencies projected 89 percent. According to the National Health Interview Survey, 90 percent of the population had health insurance in that year. (By comparison, three years earlier, in 2013, 83 percent had health insurance.)

CBO and JCT always aim to provide estimates in the middle of the distribution of potential outcomes, and they actively seek to apply the lessons of experience. Such estimates are inherently imprecise, however, because the ways in which federal agencies, states, insurers, employers, individuals, doctors, hospitals, and other parties respond to policies affecting health care are all difficult to predict.