Weekly Lab Report – November 5, 2025
Engagement with CBO and Policymaking Without Government Statistics
Fiscal Lab Notes is the official newsletter for the Fiscal Lab on Capitol Hill. You can check out all our work and analyses at fiscallab.org.
Meeting with CBO on Modeling and Baseline Methods
Last week, the Fiscal Lab met with staff from several divisions at the Congressional Budget Office (CBO) to trade ideas on how to model regulations and develop economic baselines.
A major topic was how to measure the effects of regulation on investment and output. CBO staff described their structural model where regulation affects firms’ “time to build” both the average and the uncertainty around it. The Fiscal Lab shared details from our complementary work using a Bayesian vector autoregression (BVAR) to measure how changes in regulatory restrictions influence economic growth. We also discussed our new structural model that simulates how compliance costs affect firms’ investment and productivity decisions.
Another thread focused on how BVARs can support analysis for baseline and alternative scenarios. BVARs can produce forecasts comparable to the model used for CBO’s baseline, but with fewer equations. That means scenarios could be estimated faster or with fewer staff, while having comparable accuracy. Both organizations were interested in using BVARs for producing forecasts and for communicating uncertainty around those forecasts.
Our exchange on baseline methods was especially valuable. The Fiscal Lab is developing a transparent, replicable baseline framework for independent policy scoring. As we begin that work, we were eager to learn from CBO’s approach to tracking market movements and legislative changes. Their insights will help us design a system that mirrors the rigor of official forecasts while offering a clearer view of the assumptions that drive them.
Altogether, it was a productive exchange. We came away with ideas to improve our models, and we hope CBO did too.
Are Policymakers ‘Flying Blind’ Without Official Government Statistics?
The current government shutdown has closed the federal statistical agencies, from the mammoth Census Bureau to the diminutive Center for Transportation Statistics. Altogether, 26 major statistical agencies and dozens more small operations throughout the federal government are closed. For example, policymakers are not getting the monthly jobs report, the unemployment rate, a reading on Gross Domestic Product (GDP), or some of the inflation measures. Indeed, some commentators argue that the federal government is flying blind on policymaking and program execution.
True, the offices of all US official statistics have gone dark, but that does not mean federal decision-makers, like the Board of Governors of the Federal Reserve System or the Office of the President of the United States, are without any data to guide them. State and municipal governments produce tons of data daily, as do the regional banks of the Federal Reserve System (remember, the Fed is financed largely through bank fees), private banks, and investment companies. Then there is the vast array of private firms from the big box retailers to the IT giants that churn out sales and customer data every day. Federal policymakers pour over these data releases nearly as intensely as they do official statistics.
There are problems, however, with these other sources of economic data. First, private firms and other governments produce data relevant to their needs and not those of the whole country, largely because they have come to rely on the official federal statistical agencies to produce data for federal policymakers. Second, federal decision-makers cannot rely on private data and even some state and municipal statistics being produced regularly, month in and out, regardless of economic conditions. Private firms especially are prone to changing or cancelling data releases depending on the needs of their companies and their financial capacity to supply publicly available data. And third, many people simply do not trust privately sourced data or even data from some states or cities.
Official federal statistics have become the dominant component of the nation’s data infrastructure. Yes, we can limp along with substitutes and temporary patches, but doing so is a lot like driving on a county road when the superhighway is closed for repairs: adequate but not the same. We expect the monthly jobs, GDP, and price reports to show us how we’re doing economically. Without them we have an incomplete and highly restricted perspective on the economy, and that does hamper good policy decisions.
We will really begin to see how much we need these stats, if the shutdown continues much longer. A shutdown end this week would allow us to have most of the jobs data for October reported on the first Friday in November. The jobs data are electronically collected from over 100,000 firms throughout the month. So, those reports are coming into the electronic collection centers but not being processed right now. It is less clear that we would have enough data to report the unemployment rate, labor force participation rates, and other labor force data. These statistics come from a survey of 60,000 households that is collected in person, over the phone, and electronically. None of that was done in October, and it is doubtful that the Bureau of Labor Statistics (BLS) would ask people to recall their labor status 30 days ago, given how much difficulty many have with accurate recall. Data on output, investment, and other components of the GDP report from the Bureau of Economic Analysis requires data from Census, BLS, and other statistical agencies, which has just not been collected over the last 30 days.
In short, the current shutdown cannot but create some missing statistics. Whether those gaps are large or small will largely depend on when the shutdown ends and how creative our patriotic statistical staffs are in generating the infrastructure we all so depend upon.