How to Evaluate Costs in the CARES Act

Executive Summary

  • The Congressional Budget Office baseline is grounded in a January economic forecast that is now historical fiction.
  • A conventional cost estimate of the CARES Act would not be valuable – but Congress should be relying heavily on the analytic capability of scorekeeping in designing CARES Act programs.
  • Congress should ensure that costs to the taxpayers are monitored on an ongoing basis, similar to the approach used for the 2008 Troubled Asset Relief Program.


There are some crises that are simply too great to approach in a conventional fashion. Such is the case with how policymakers should consider the cost of the massive relief bill – the Coronavirus Aid, Relief, and Economic Security (CARES) Act being debated on Capitol Hill. The conventional cost estimating process is unequipped to evaluate the costs of these relief efforts. Indeed, even mechanically, the Congressional Budget Office (CBO) would face significant limitations in how it could go about assessing the costs of the Act. There is too much uncertainty about future credit conditions to credibly estimate losses on the loan programs in the CARES Act. None of these deficiencies should suggest that cost assessment shouldn’t inform the CARES Act. Congress should be using the excellent analytical capability of the CBO and the Joint Committee on Tax (JCT) to use public funds to their greatest effect.

Keeping Score – How Congress Estimates Costs

When Congress considers legislation, it entrusts CBO and JCT with determining what the budgetary effects of the given legislation will be. These institutions are called “scorekeepers” and their estimates of the budget effects of legislation are called in the parlance of budget policy observers “scores.” While each institution has considerable discretion in the assumptions and methodology that underpin their analyses, they must exercise that discretion within some very rigid parameters. Federal law, rules, conventions, and traditions govern the construction of the budget baseline. The baseline is the yardstick against which the scorekeepers measure the effects of legislation. It is the budgetary counter-factual, the federal budget on auto-pilot for a given time period – generally 10 years. Budget scores are thus measurements relative to this counterfactual.

Central to the construction of this baseline is CBO’s economic forecast. This forecast is typically updated twice yearly – once in January and once in August. CBO’s current baseline is therefore keyed off of an economic forecast entirely abstracted from the effects of the pandemic. The CBO baseline is built on an assumption of real GDP growth in 2020 of 2.2 percent, an unemployment rate of 3.5 percent, and slowly rising interest rates. This was a fairly mainstream forecast for the year – and it’s now a work of historical fiction. Put another way, estimating costs against this baseline would be like flying a plane with an altimeter with the wrong sea level.

A Cost Estimate for CARES?

The CBO baseline projects budgetary flows – programmatic spending, tax collection, debt issuance – based on this economic outlook. Any cost estimate of the CARES Act would be compared to this baseline and would likely mislead policymakers. The most budgetarily consequential elements of the CARES Act would appear to be direct payments to individuals, expanded unemployment insurance (UI) coverage, and a substantial financial rescue package for American employers. The direct payments are the most easily estimated and do not present the same challenges as those policies that are somewhat contingent on forward-looking analysis. The eligibility for the payments is based off of tax data, and can be readily evaluated for cost and distributional effects. The latter analysis is more useful at this stage for advising policymakers how to direct funds. Ensuring funds go where they are intended is a good and appropriate role for congressional scorekeeping agencies.

Somewhat more difficult is assessing costs for a UI expansion. Estimating this cost requires estimating program uptake – and comparing new estimated costs to the costs of the UI projected in the baseline. CBO could certainly guess – and they’ve been forced to in the past – at what unemployment will be in a year. It’ll be a number, and it’ll be wrong. Congress should ignore the temptation to care about it.

The last, and potentially most consequential element of the CARES Act is a $500 billion package of credit assistance to American employers. When Congress considers credit programs, CBO is required to follow what is known as the Federal Credit Reform Act (FCRA) rules. Under FCRA, the basic unit of measurement is the subsidy cost, which is the net present value of all future cash inflows and outflows, discounted at term-matched Treasury rates, for a cohort of loans. These flows include the net effects of defaults, fees, and recoveries. These estimates are difficult in normal times, and any change in any of the elements of the subsidy cost – interest rates, default rates, etc. – change the cost. Every year, the subsidy costs of major ongoing credit programs are revised to reflect these changes, and they can be substantial. Again, CBO would have to estimate these elements into an impenetrable fog of uncertainty and do so against a baseline that is a hopeless anachronism.

TARP – the Model Program

Congress and CBO should follow the of the cost estimate for the Economic Stabilization Act of 2008. This was the legislation that contained the far-better-known Troubled Asset Relief Program, or TARP. When Congress considered this legislation staring into the front edge of the Great Recession, CBO stated the following in its related costs estimate: “Although it is not currently possible to quantify the net budget impact given the lack of details about how the program would be implemented, CBO has concluded that enacting the bill would likely entail some net budget cost—which would, however, be substantially smaller than $700 billion.”

The program ultimately made money for taxpayers, but that was only learned after the fact. While Congress enacted TARP without a cost estimate, it did require CBO to estimate its effects using fair value methods in its periodic baseline updates. [1] Indeed, the first estimate of the cost for TARP was in its January 2009 baseline. At the time, the CBO was estimating substantial losses under the program. Only after the economy improved (and CBO’s baseline economic forecast along with it) did these costs slowly come down as once endangered firms returned to profitability and paid the government back. For those firms receiving loans under CARES, we should hope to see the same again.


Congress shouldn’t get bogged down pretending to predict the future. CBO is asked to just that every time they produce a cost estimate. CBO is absolutely the best in the business at this, and Congress should be relying heavily on their analytical capability in designing assistance programs to do the most good.

[1] For more on fair value accounting, see: