Fifteen Ways to Improve CBO & Scorekeeping

Fifteen Ways to Improve CBO & Scorekeeping
April 11, 2018
By Demian Brady

As the budget referee helping determine the fate of legislation, the Congressional Budget Office (CBO) often finds itself in the crosshairs of criticism. Last year’s health care debates brought to light several concerns about the agency’s transparency and the assumptions it plugs into its models. This controversy triggered the first comprehensive oversight hearings in Congress on the scoring agency since it was established in the 1970s. Members and panelists addressed institutional and cultural problems such as improving openness and increasing the flow of information between the agency and legislators since ultimately CBO works under the rules that Congress imposes.

With the release of CBO’s updated budget forecast, the time is ripe for considering the best ways to reform the agency to help it to better execute its important mission. Here are 15 of the best reforms to improve CBO and the quality of budget data available to lawmakers and the public, in no particular order.

  1. Include a score using a realistic policy baseline. CBO scores legislation using a current-law baseline, which generally takes the written text at face-value: programs or tax provisions that are enacted or further extended on a temporary basis are assumed to terminate when their authorizations expire. Knowing this, legislators can pack their bills with budgetary gimmicks in order to generate a favorable score. Including a score based on a realistic policy baseline would employ more honest assumptions to uncover the true fiscal impact of proposals.  Congress has required that CBO report a point estimate in its scores, as opposed to a range. While CBO states that it strives to report the middle ground of possible outcomes, modest changes in assumptions (or in legislative provisions) can often result in significantly different projections. Better explanation of sensitivity in scoring would illustrate potential swings in outcomes over the long-term and highlight the significance of assumptions used in a score.
  1. Include current spending levels in estimates for reauthorizations. As the five- or ten-year budget window in a cost estimate is compared against CBO’s baseline for the next decade, cost estimates should also provide data on the existing funding levels of the programs reauthorized by the bill. Currently, CBO simply reports amounts relative to a baseline, leading to situations where consistent increases in agency funding can be characterized as a “cut” in expenditures if the increase is slightly less steep than CBO’s baseline assumed. Including current spending levels would help provide much-needed context. This would especially help to provide clarity on CBO’s scores of bills to reauthorize farm programs.
  1. Apply a “confidence index.” Some analyses produced by CBO are more easily modeled, and thus more likely to be accurate, than others. Yet the agency produces only a single estimate for a given piece of legislation, including no context in the budget tables as to whether their confidence in the analysis is high or low based on the reliability of data inputs. If it utilized a “confidence index” assessing the likelihood of the outcomes presented in its analysis, or even a range of possible scores, the agency could better inform lawmakers about legislation and the difficulties inherent in scoring it. Gauging the potential uncertainty of outcomes would be useful for flagging gimmicky fake “savings” used to get a favorable CBO score.
  1. Make behavioral assumptions clear in cost analyses. In the Congressional oversight hearings this year, several Members complained that they had to have multiple rounds of discussions directly with CBO staff in order to learn the details of the assumptions that were used to score legislation. The basis of estimates in CBO’s legislative scores should, to the extent practicable, clearly describe the key behavioral assumptions applied to the model.
  1. Congress should hold regular oversight hearings. Communication between Members of Congress and CBO was cited as a recurring problem in this year’s series of hearings. Regular annual oversight hearings would promote consistent communication between the entities while also providing a review of the agency’s work. While this runs some risk of providing non-substantive grandstanding opportunities for some in Congress, it would also provide an outlet for CBO to discuss challenges it faces in scoring and tools and cooperation it might need from Congress to make its job easier.
  1. Make better use of private sector models and fair-value accounting. Last year, CBO provided the House Committee on Financial Services with a supplemental report about the soundness of the National Flood Insurance Program. For this analysis, CBO utilized a private sector model and found a $1.4 billion shortfall between the program’s claims paid out and premiums taken in - a larger gap than was foreseen with the actuarial model that is normally used to project NFIP’s fiscal outlook. CBO also occasionally re-evaluates federal credit and loan programs using fair-value accounting instead of the procedures stipulated in the Federal Credit Reform Act of 1990. Under fair-value accounting, subsidy estimates are calculated based on market values instead of rates on U.S. Treasury securities. Through this corrective lens, the true liabilities of programs like the Export-Import Bank are exposed: instead of generating savings of $14 billion, it actually produced a loss of $2 billion. CBO should survey for and utilize commercially-developed models and fair-value accounting on a more formal and regular basis to assess the scale of the liabilities and risks imposed on taxpayers through federal credit and loan programs.
  1. Improve modeling of competition. Yuval Levin points out that CBO has a problem accounting for competition in its models: CBO tends to “model competition as having minimal effect on costs while modeling price controls to be efficient and effective. ... Competition is obviously much more difficult to model than mandates and price controls, but the agency’s experience with Medicare Advantage and the Medicare prescription-drug benefit suggests that it tends to significantly understate the effects of competition — which obviously has consequences for its scoring of reforms intended to increase the market orientation of the health-care system.” CBO should embark on an internal effort to reevaluate its approach to modeling effects of competition.
  1. Standardize formatting of scores and improve capability for providing bulk data. NTUF joined with a coalition in an open letter to CBO identifying improvements in transparency and data formatting. The letter noted that tables included in cost estimates do not “appear to be available as a structured data format that can be systematically gathered.” Standardizing the presentation of budget information would make it easier for outside individuals or organizations to access CBO’s scoring data and perform additional research.
  1. Provide scores for junior members. Given its current workload, CBO typically only scores legislation that is supported by committee chairmen or leadership. This means that numerous bills introduced by more junior Members of Congress remain unscored by CBO and, as a result, languish in Congress—despite that fact that they could save taxpayers billions. As a result, innovative ideas to shrink government are held back, while the full implications of schemes to expand government only come to light long after they have gained traction among influential lawmakers.
  1. Lower the threshold for requiring dynamic scores. Current rules only require CBO to include an estimate of macroeconomic effects, known as “dynamic scoring,” when the direct spending or revenue proposal under consideration in a bill has a budgetary impact greater than 0.25 percent of the GDP. Compared to a static analysis, dynamic analysis more accurately portrays the economic effects of revenue legislation. The agency should provide dynamic analysis on more legislation, and Congress should ensure it has sufficient capacity to do so.
  1. Provide a qualitative assessment of how important the baseline was in developing its cost estimates. Economist and former White House Director of Budget Policy Paul Winfree notes, “Some cost estimates are entirely driven by baseline assumptions. Others are less constrained. For instance, CBO often acknowledges that the policy world has changed when new information becomes available. But cost estimates for new legislation are often provided against an outdated baseline to remain consistent in evaluating all proposals over the course of a legislative session. The difference is important to policymakers.” Better explaining how tethered an estimate is to baseline assumptions would be helpful in evaluating its usefulness.
  1. Improve transparency. During the hearings, CBO’s Director said that scoring is more than just modeling: projections rest upon the assumptions that are plugged into the model. Both should be made more transparent. The Show Your Work Act, introduced as S. 1746 and H.R. 3822, would require CBO to publish its data and models. Additional steps to improve transparency were laid out in our coalition letter, including improved searchability for scores and standardization of web URLs.
  1. Improve reporting requirements of unfunded mandates. The Unfunded Mandates Reform Act (UMRA) requires CBO to report the unfunded cost burdens imposed on the private sector or state, local, and tribal governments through legislation reported by authorizing committees. UMRA does not cover review of unfunded mandates in legislation that goes straight to the floor, in amendments, or in appropriations. A point of order can be raised against legislation with unfunded mandates above the thresholds. In 2017, those thresholds were $78 million for intergovernmental mandates and $156 million for private-sector mandates (adjusted annually for inflation). Of all the mandate statements CBO issued in the 2010-2017 timespan, only 4 percent of 252 intergovernmental mandates and 15 percent of 451 private sector mandates met the scoring threshold. Even then, one former CBO Director told NTUF        that the methodology employed to estimate impacts of federal mandates on lower-level governments does not generate reliable, useful results. CBO should more consistently review unfunded mandates, and Congress should ensure they have the capacity to do so.
  1. Don’t count savings before they are advanced. Fuzzy budget math for the Center for Medicare and Medicaid Innovation (CMMI) has been used to understate the cost of Medicare by as much as $45 billion. CMMI was enacted in the Affordable Care Act to reduce costs through demonstration programs that experiment with different payment and service delivery models. With those assumptions, CBO assumes that CMMI yields savings that amount to 0.1 percent of Medicare expenditures. However, CBO is unable to account for any actual savings that should have resulted from CMMI. Similarly, CBO scored the Independent Payment Advisory Board – another program established in the ACA to reduce its CBO score – as achieving significant savings in Medicare despite huge uncertainty surrounding when or even whether its authority to enact automatic cuts would ever be triggered. Budgetary savings should not be scored as a matter of faith but should be established on the determination of realistic assumptions and demonstrable outcomes. If they are unmeasurable after the fact, then the assumptions should no longer be used going forward.
  1. Provide scores for program integrity. Many pieces of legislation authorize additional spending to monitor fraud, waste or abuse within a government program. These “program integrity” reforms offer the potential for billions of dollars of savings for taxpayers when applied to massive entitlement programs. Current budget rules do not allow savings generated later in the budget window to be counted as an offset against the initial spending required. In other words, later savings potentially greater than the original expenditure do not “pay for” the initial increased activities. While there is significant risk that cynical Members of Congress might exploit such a system to disguise spending hikes as efforts to save money, additional analysis from CBO could at least provide a rational foundation for discussing significant reforms that could reduce burdens on taxpayers but don’t meet current scoring rules.

The reports and analyses produced by CBO play a pivotal in the development of legislation in Congress, thus it is crucial that the figures at the heart of policy debates are accurate and clear. There are obvious budgetary and knowledge resource challenges that constrain CBO’s ability to meet the growing demand for legislative scores and analytical studies across a broad range of policy areas. There is also a roll that external organizations, such as NTUF’s Taxpayers’ Budget Office project, can fulfill to help hold the CBO accountable and fill in gaps in its analyses.