Plans to Update CBO’s Health Insurance Simulation Model

CBO's health insurance simulation model (HISIM) is one of the tools that the agency uses to produce baseline budget projections and cost estimates for legislation related to the Affordable Care Act (ACA). CBO is currently enhancing it through a process that is thorough and transparent and that includes extensive peer review and broad external validation.

CBO has long been dedicated to providing information that is objective, insightful, timely, and clearly presented and explained. In order to ensure independent, nonpartisan analysis, CBO must make choices about what models to use and when to use them on the basis of its best analytical assessment. Those choices are informed by knowledge and insight that CBO solicits from outside experts as part of its effort to be transparent.

CBO’s dedication to transparency applies strongly to its analyses of federal subsidies for health insurance. For instance, CBO has frequently published material about those analyses in the past and then sought input about those publications from many sources. CBO has also regularly conducted outreach and gathered feedback. The agency has intensified such transparency efforts this year.

CBO is currently following through on the general plans for HISIM that it previously described to Senator Enzi.

• CBO is obtaining systematic feedback from the research community by making presentations about various aspects of the model as they are developed.
• CBO is also publishing conceptual overviews and documentation of all parts of the model, as well as more detailed information—including computer code—about key aspects of the modeling.

The rest of this letter addresses subjects that Senator Enzi raised—specifically, the development process for the updated HISIM, documentation for it, and estimates that it can provide of health insurance coverage under the ACA.

Development Process

The update of HISIM has been very extensively reviewed. CBO is currently discussing the update with representatives from the Congress, federal agencies, states, insurers, employers, doctors, hospitals, and the general public. For example, CBO staff members met with Senate staff in October to discuss HISIM and are currently working to schedule meetings with additional Senate and House staff in November for the same purpose.

Discussion also took place during a meeting of CBO’s Panel of Health Advisers in September. The panel consists of widely recognized experts in health policy and the health care sector. Members of the panel have a variety of backgrounds, areas of expertise, and experience: They are administrators of state health programs, insurance executives, employers, doctors, hospitals, and health researchers. Congressional staff also attended the meeting.

During the same meeting, presentations were made by two members of a newly established technical review panel that provides CBO with feedback on the updates to HISIM. (That panel will provide CBO with special assistance during the next two years in the course of the current major upgrades; after that, assistance with HISIM will be provided by CBO’s regular Panel of Health Advisers.) The review panel’s members were selected for their analytic expertise and include people recommended by staff of the House and Senate Budget Committees. CBO is scheduled to have technical discussions with those reviewers on specific aspects of the model’s development and testing during the next few months. CBO also plans to obtain their feedback about preliminary baseline projections of health insurance coverage before releasing final projections in the spring.

Furthermore, CBO is conducting other outreach to experts in research organizations and industry, and the general public. For example, CBO staff engaged in discussion with a variety of stakeholders at a public event at the Bipartisan Policy Center. CBO has met with experts in microsimulation modeling at the Agency for Healthcare Research and Quality, the Heritage Foundation, the RAND Corporation, and the Urban Institute to discuss numerous technical issues. The agency also conducted a webinar with the American Academy of Actuaries that attracted about 200 participants from around the country. And CBO has worked with staff at the Congressional Research Service so that they can better communicate about CBO’s undertakings as part of their own service to the Congress. In the coming months, CBO will continue to solicit feedback from people recommended by Congressional staff and from a broad array of other stakeholders.

Documentation

During the next year, CBO plans to publish more documentation of the updated model and more detailed information, including computer code, about key aspects of that model. Taken together, those public releases will provide a comprehensive picture of the model. Specifically, before publishing its spring baseline projections, the agency plans to release the following:

• An updated slide deck that describes the model’s specifications and includes many of the key equations and parameters;
• Additional documentation describing the sources and preparation of input data; and
• Segments of computer code underlying the model’s simulations of certain decisions about insurance choices.

(CBO cannot release some aspects of the modeling publicly because the agency must maintain the confidentiality of the underlying data.)

Furthermore, CBO’s annual report on federal subsidies for health insurance will include a discussion of differences between the spring 2019 baseline projections and the previous ones.

Estimates of Coverage Under the ACA

The main points of reference for the updated HISIM come from data about health insurance coverage during the past four years. Specifically, as part of the process of developing the model, CBO is starting with data about coverage in 2015 and then using the model to estimate coverage under the ACA in the following three years. The agency will then be able to assess and report about how closely it is able to calibrate its estimates for those years with data from multiple sources about health insurance coverage. That process of checking the estimates against a large collection of external data helps validate the model.