As Congress readies legislation to repeal and replace the Affordable Care Act (ACA), Congressional Budget Office (CBO) estimates will play an important and respected role as they did in the passage of the law in 2010. We now know that many of CBO’s projections of important aspects of the ACA have significantly differed from actual outcomes. In this piece, I highlight CBO’s key past errors in projecting effects of the ACA. They can largely be grouped into two categories. First, CBO projected that the exchanges would be stable by now with more than twice as many enrollees as they currently have, rather than suffering from severe adverse selection in most states as they now are. Second, CBO projected that the ACA Medicaid expansion would be much smaller and less expensive than it has turned out to be.

These errors were caused by two primary mistakes in CBO’s model and assumptions. First, CBO significantly overestimated the degree to which the individual mandate would induce relatively healthy people with middle class income to buy coverage in the exchanges. Second, CBO failed to anticipate that states would respond to the federal government’s elevated reimbursement rate for the Medicaid expansion by maximizing enrollment and paying insurance companies extremely high payment rates for this population. CBO has not yet explained if or how it has corrected its models for these past mistakes, but it should do so if it wants to improve confidence in its estimates of repeal and replace legislation.

Exchange Enrollment Much Less Than CBO Projected

When the ACA passed in 2010, CBO projected 21 million people would be enrolled in the exchanges in 2016. After the Supreme Court ruled that the Medicaid expansion was optional for states and not compulsory, CBO increased its projection of 2016 exchange enrollment to 22 million as some people who would otherwise have been enrolled in Medicaid in non-expansion states were then expected to enroll in the exchanges instead. Exchanges plans have proved much less attractive than expected as enrollment will average only about 10 million people this year. This means that CBO’s last projection of exchange enrollment before the exchanges opened overshot actual 2016 enrollment by 120 percent.

In a briefing to congressional staff on the effects of repealing the ACA that I attended as a Senate staffer in June 2015, I asked CBO about their then-projection that exchange enrollment would nearly double between 2015 and 2016. They responded that this was largely because of both an increase in awareness of the individual mandate and the increase in the associated tax penalty. In stark contrast to the projections of CBO and others, enrollment barely increased between 2015 and 2016.

CBO’s model has consistently and significantly overestimated the effect of the individual mandate in inducing people to enroll in the exchanges. Since higher income people were supposed to be more affected by the mandate (the penalty increases with income), a less effective than expected mandate means that exchange enrollees are also much poorer than expected. If CBO has not adequately adjusted its model for its mistake about the effectiveness of the mandate, its estimates of bills to repeal and replace the ACA—which will almost certainly eliminate the individual mandate—will continue to incorporate this source of inaccuracy.

Insurers Performance Much Worse Than CBO Projected

In February 2014, one day before a congressional oversight hearing examining the ACA’s risk corridor program, CBO released an estimate that the risk corridor program would net the federal government $8 billion over three years. CBO’s estimate assumed that insurers would be quite profitable selling exchange plans since insurers with “excess” profits contributed to the program while those with “excess” losses received payments from the program.

It turns out that CBO was significantly mistaken about insurer profitability. Insurers have incurred large losses on ACA plans, losses which have grown over time. These losses resulted in a $2.5 billion risk corridor deficit in 2014 and resulted in a $5.8 billion deficit in 2015. The large losses have also driven many insurers from the exchanges.

Reinsurance Program Subsidies Much More Generous Than CBO Projected

Insurers’ losses in 2014 and 2015 would have even been larger if not for receiving larger per enrollee payments through the reinsurance program than expected. The reinsurance program compensated insurers for a large share of the cost of their most expensive enrollees. CBO projected that reinsurance payments lowered premiums by about 10% in 2014. In an April Mercatus study I coauthored, we found that net reinsurance payments equaled about 20% of premiums—double what CBO expected. CBO’s estimates of insurer profitability look even worse since the agency significantly underestimated reinsurance payments as a percentage of premiums.

Medicaid Expansion Enrollment Much Greater Than CBO Projected 

Since the exchanges have enrolled so few relative to expectations, the vast majority of the newly insured are enrolled in Medicaid. The figure below shows CBO’s most recent estimate of Medicaid expansion enrollment along with CBO’s 2010, 2014, and 2015 estimates. The figure—as well as the one on spending—adjusts CBO’s previous year estimates for its current assumptions about state adoption of the expansion. (Currently 31 states have adopted the expansion.) This adjustment allows for a better intertemporal comparison because it holds constant CBO’s assumptions about the percentage of the newly eligible Medicaid population residing in expansionary states.

Medicaid expansion enrollment is much higher than CBO expected when the ACA passed in 2010, and it is also significantly higher, particularly in 2017 and beyond, than estimated in both CBO’s 2014 and 2015 reports. Essentially this means that far more people—roughly 50% more—have enrolled in Medicaid in the states that expanded than expected by CBO.

Medicaid Expansion Spending Much Greater Than CBO Projected

In addition to higher-than-expected enrollment, spending per newly eligible Medicaid enrollee is much greater than CBO expected. As I wrote in July when the Obama administration released the 2015 Medicaid actuarial report, government spending on newly eligible enrollees equaled about $6,366 in 2015—an amount 49% higher than CMS’s projection of $4,281 from just one year earlier. In April 2014, CBO projected the Medicaid expansion enrollee average cost would be approximately $4,200 in 2015, a number very close to the erroneous CMS projection.

Both higher-than-expected enrollment and spending per enrollee has resulted in the Medicaid expansion being much more costly than projected. For example, in April 2014, CBO projected that the Medicaid expansion would cost $42 billion in 2015. The actual cost was approximately $68 billion, about 62% higher. The figure below shows CBO’s past projections of federal spending on the Medicaid expansion, again adjusting CBO’s previous year estimates for its current assumptions about state adoption of the expansion.

Medicaid expansion is proving much more expensive than CBO expected, largely because the agency failed to anticipate how states would respond to the elevated reimbursement rate for ACA Medicaid expansion enrollees. Many states have set very high payment rates to insurers for the expansion population with the cost dispersed to federal taxpayers.

More New Medicaid Enrollees Were Already Eligible Than CBO Projected  

Recent research suggests that only between 30% and 40% of new Medicaid enrollees in 2014 were made eligible for the program by the ACA. In contrast, CBO’s most recent estimate projected that 13 million people would be added to Medicaid in 2016 because of the ACA—11 million, or 85%, as newly eligible and 2 million, or 15%, as previously eligible. This large discrepancy has significant implications for the proper share of federal and state spending as well as the practical implications of repeal.

Previously eligible but unenrolled Medicaid recipients could generally enroll at any time including at the point of needing medical care, meaning that the practical effect of repeal on coverage loss may be significantly overstated. To put it simply, many of the newly covered Medicaid recipients will remain eligible even if the ACA is repealed without a replacement.

Economic Growth After Obamacare Much Lower Than CBO Projected

In January 2010, CBO projected that growth in real gross domestic product (GDP) would average 3.2% from 2010 to 2016. By way of comparison, the annual GDP growth rate after the first six years of another severe recession (1981-82) averaged 4.6%.

Economic growth after the Great Recession has been anemic by historical standards and relative to expectations. As the figure below shows, annual real GDP has increased just 2.1%, 50% below the average growth rate predicted by CBO and less than half the growth rate during the Reagan recovery. The weak economic recovery has produced lower health care spending, and thus lower health insurance premiums, than would have resulted with a stronger recovery.

Conclusion

Projecting the economic impact of major pieces of legislation is a difficult task with substantial amounts of uncertainty. To its credit, CBO acknowledges this, even as there is a tendency elsewhere to treat CBO estimates as gospel. Since CBO estimates will once again play a prominent role in the coming ACA repeal and replace debate, it is important to appreciate the large degree of uncertainty and to understand CBO’s key mistakes estimating the ACA. In particular, it would be good to know the steps CBO has taken to correct for its two biggest mistakes—overestimating the effect of the individual mandate and failing to anticipate how states would respond to the elevated reimbursement rate for the Medicaid expansion population. CBO should then inform lawmakers how it has adjusted its model and assumptions. And irrespective of CBO’s methodological changes, lawmakers should proceed with full awareness of the limits of CBO’s projection capability.