Do Government Shutdowns Affect Subsequent Data Revisions?

Nov. 12, 2025, marked the end of the longest U.S. government shutdown to date. During the shutdown, some agencies within the government stopped collecting data. For example, the U.S. Bureau of Labor Statistics will not release several reports for October 2025, including the jobs report and consumer price index.

Government shutdowns are not new phenomena. Since 1980, there have been 17 instances in which the U.S. government has experienced a lapse in funding.We include the funding lapse that occurred Feb. 9, 2018, in our analysis. Some do not classify this event as a shutdown due to its brevity. Many of these shutdowns were short, with only four of the 17 lasting longer than a week. When the government reopens and data finally become available, are data revisions that follow larger than normal?

Data revisions are not limited to the periods following government shutdowns.Past St. Louis Fed publications provide background on economic data revisions; for example, see “Data Revisions with FRED” and “How Important Have Employment Revisions Been the Past Two Years?” Some economic data are estimated from surveys and a limited panel of firms, only to be later revised when more information is available. For gross domestic product (GDP) specifically, there are three primary releases: The advance, preliminary and final estimates of GDP are released during the first, second and third months, respectively, after the end of the quarter. The difference between the advance and preliminary numbers is the first revision to GDP. Similarly, the difference between the preliminary and final numbers is the second revision to GDP.

GDP Revisions after Government Shutdowns since 1980

The table below compares average absolute revisions to GDP growth in normal quarters and in quarters during and following a government shutdown. It also compares data for normal quarters with those for all shutdowns and for the three longest shutdowns since 1980. (Note, however, that we exclude the shutdowns in 2025 and 2026.)We exclude the 2025 shutdown, which lasted from Oct. 1 to Nov. 12, and the 2026 shutdown, which lasted from Jan. 31 to Feb. 3, because we are not yet three quarters removed from them. Given this, the three longest shutdowns we examine occurred in 1995-96, 2013 and 2018-19. The two columns on the right show the first and second revisions to GDP, respectively. The first row shows the average absolute revision during normal quarters. The two panels that follow show the average revision for all shutdowns and for the three shutdowns before 2025 that lasted more than one week, respectively. In each panel, the four rows contain the average revision for:

• The quarter prior to the shutdown, data for which might be collected or analyzed during the shutdown
• The quarter that includes the shutdown
• The quarter immediately following the shutdown
• The second quarter following the shutdown

Based on this cursory analysis, three preliminary conclusions emerge. First, the quality of the data for the quarters before and containing the shutdown is broadly comparable to any other quarter, regardless of the length of the shutdown. The difference between the average absolute revision observed during these periods and the average absolute revision in normal quarters is similar. Second, shutdowns do appear to affect the size of revisions for the next two quarters, with the subsequent quarter having higher-than-average revisions and the second quarter after a shutdown generally having lower-than-average revisions. Third, the length of the shutdown does not seem to be important, only that it occurred.

Can we take these results as a sign that a shutdown lowers the quality of the subsequent data? Well, not so fast. As we noted above, the first few releases of the past quarter’s output data are estimated using a limited set of data. Karen Dynan and Douglas Elmendorf’s 2001 paper argues that these provisional estimates of GDP tend not to properly account for turning points (PDF). Thus, it is possible that the shutdown itself induces firms to push off hiring or investment and/or consumers to push off purchases in a way that is not accounted for in the estimates.

Notes

1. We include the funding lapse that occurred Feb. 9, 2018, in our analysis. Some do not classify this event as a shutdown due to its brevity.
2. Past St. Louis Fed publications provide background on economic data revisions; for example, see “Data Revisions with FRED” and “How Important Have Employment Revisions Been the Past Two Years?”
3. We exclude the 2025 shutdown, which lasted from Oct. 1 to Nov. 12, and the 2026 shutdown, which lasted from Jan. 31 to Feb. 3, because we are not yet three quarters removed from them. Given this, the three longest shutdowns we examine occurred in 1995-96, 2013 and 2018-19.