Higher new Census population estimates will affect the employment report

Over the past year, researchers have been deliberating about how to measure the size of the U.S. population. In early 2024, the Congressional Budget Office (CBO) published estimates suggesting much higher post-2020 population growth than previously estimated, largely stemming from an unanticipated surge in migration. Now the Census Bureau has revised its population estimates to be more closely aligned with the higher CBO numbers. This brief explains why the change matters and how it will affect the employment report slated to come out February 7.

The new Census estimates

In December, the U.S. Census Bureau released new estimates of the population since 2020. With this Vintage 2024 series, Census revised its estimates of the all-ages civilian non-institutionalized population (CNIP) upward for each year from 2021 through 2024 (see figure 1). The revision was more substantial than is typical, reflecting the unanticipated immigration surge over the past few years and new Census methods of accounting for that surge. As is true for every new vintage, the Census numbers also project population through the end of the following year, so Vintage 2024 includes a projection through December 2025.

The projections for population from January to December 2025 will indirectly affect official estimates of labor force and employment that will show up in the February 7 jobs report and subsequent reports throughout the year.   

Here’s why: The labor force, the number of employed people, and the number of unemployed people are all computed using the Current Population Survey (CPS), which is a monthly survey of households. The Bureau of Labor Statistics (BLS) surveys about 60,000 households each month and then extrapolates by weighting each response to produce aggregate statistics. That extrapolation method relies on estimates of the civilian non-institutionalized population ages 16+ (CNIP 16+). To arrive at those estimates, BLS matches pre-determined aggregate population levels known as population controls, which come from the annually released Census vintages. For example, the 2024 labor force and employment numbers are based on Vintage 2023 population estimates from Census released in December 2023; the newly released Vintage 2024 Census population numbers will in turn be used as population controls throughout calendar year 2025.

Importantly, the household survey will not revise CNIP 16+ back before 2025. As shown in figure 2, the published series of CNIP 16+ regularly shows an upward or downward jump in January, reflecting the new vintage projections used in each calendar year. That is, January 2024 through December 2024 will continue to be based on Vintage 2023, January 2023 through December 2023 will be based on Vintage 2022, and so on.  Economic observers generally should be careful about interpreting monthly changes between December and January, particularly changes in aggregate statistics like the size of the population or labor force.¹

We know that there will be a jump in the CNIP 16+ estimate for January 2025 stemming from the revised Census population. We don’t yet know the exact size of the jump because the Census has so far only publicly released all-ages CNIP for the latest vintage, not CNIP 16+. But we can make an educated guess using the monthly share of total CNIP that is CNIP 16+ from the Census Vintage 2023 series. We assume the January 2025 share is the same as the published Vintage 2023 December 2024 share.

We expect that the January 2025 jump will be larger than usual, reflecting the unusually large adjustment in all-ages population between Vintage 2024 and Vintage 2023. The expected jump in CNIP 16+ is shown in figure 3, with January 2025 CNIP 16+ value for “Current Population Survey” equaling our estimate of CNIP 16+ from Vintage 2024. The CNIP 16+ jump that appears in the employment report between December and January may be somewhat larger or smaller than the 3 million shown here if, in addition to revising total population, Census changes its estimated share of the population across age groups. The January 2025 jump should be interpreted as a statistical artifact, not as a large monthly change in population, labor force, or employment.

How the new Census estimates compare to other estimates

The Congressional Budget Office also publishes population estimates and estimates of annual net migration. In January 2024, CBO estimated a large flow of immigrants in 2022 and 2023, resulting in much larger population growth than they had previously projected and larger than was evident in the 2023 and 2024 CPS (based on Census Vintage 2022 and Vintage 2023 numbers). We have argued elsewhere that the case for these high immigration flows was well supported and that accounting for those flows helped resolve some labor market puzzles.

The January 2025 CBO estimates show somewhat lower immigration flows for 2024 than were previously projected, largely reflecting summer 2024 policies that significantly reduced entries of its “other foreign nationals” category at the border. It is not possible to directly compare CBO and Census net migration estimates because Census publishes annual estimates of total net immigration from July of the prior year through June of the current year, whereas CBO produces calendar year estimates. To convert Census estimates into calendar years, we rely on Census monthly population estimates and assume that the share of population growth attributable within each month to migration is the same as the annual share.

For both CBO and Census, we then apply a constant ratio of 81 percent of the CNIP 16+ population to total population based on pre-revision household data among recent immigrants. Figure 4 shows that the upward revision Census made to its net migration estimate brings it much closer to CBO’s estimates, though CBO is still somewhat higher.

Figure 5 compares estimates of the cumulative growth in CNIP 16+ for January 2010 through January 2025 from three series: CBO 2025 demographic projections, a series implied by Census Vintage 2024, and a series expected to be published by BLS with the monthly jobs report. The series from CBO is the agency’s population estimates described as “Census through 2020 plus CBO projection.” In January 2023 and January 2024, there is a significant discrepancy between CBO’s estimates and published CNIP 16+ in cumulative growth stemming from divergent immigration assumptions. In January 2025, with the release of Vintage 2024, those two series will be much closer together. As figure 3 and figure 5 show, the published estimates of CNIP 16+ will likely show an increase from January 2024 to January 2025 of roughly 5 million (whereas consistent population series would suggest 3 million) and from December 2024 to January 2025 of roughly 3 million (whereas consistent series would suggest 100,000).

Foreign-born and U.S.-born population

The Current Population Survey is also used to produce separate estimates of CNIP 16+, the labor force, and employment for the U.S.-born and foreign-born populations. The Census population controls do not directly specify the number of U.S.-born and foreign-born people; these are estimated using the CPS survey question about place of birth. The foreign-born share is indirectly affected by the Census population controls, however, because the survey weights are based on demographic factors somewhat correlated with foreign-born status, like age, sex, race, and Hispanic origin. The CPS estimates for January 2025 will be consistent with Vintage 2024 Census population by demographic group, and though the estimates with demographic detail are not publicly available yet, there could be changes that affect the estimated foreign-born share.  We anticipate that the share of the population that is foreign-born may increase slightly due to the Vintage 2024 revision, but it is unlikely to change abruptly.  

It might be helpful to think of the U.S. population as a pie. Survey data show that the U.S.-born share of the pie has been getting smaller over the past few years with high levels of immigration and low fertility rates. At the same time, the previous Census population estimates suggested the pie was only getting slightly larger over that period. A smaller slice and only slightly bigger pie meant that the size of the U.S.-born population was estimated to have declined from January 2023 to January 2024. The apparent decline was at odds with other evidence that there were more U.S.-born people turning 16 than there were deaths. Now, with the new Census revision, we see the overall pie is growing bigger over time, more than Census previously estimated. We expect, therefore, that the U.S.-born CNIP 16+ will rise sharply from December 2024 to January 2025 in published data.

The red line in figure 6 shows our estimate of the cumulative increase in U.S. born CNIP 16+ since 2020 as implied by Vintage 2024. To compute U.S.-born CNIP 16+, we estimate total CNIP 16+ as in figure 5, and we apply the share of the 16+ population that is U.S. born from the published monthly CPS data, using the December 2024 CPS share for January 2025.  We expect the published data to show a jump in the U.S.-born civilian noninstitutionalized population of 3.2 million between January 2024 and January 2025.

Figure 6 plots annual estimates in order to include estimates implied by CBO’s series “Census through 2020 plus CBO projection.” We can compute monthly estimates for Vintage 2024 and published CPS data. For the change in the U.S.-born CNIP 16+ population between December 2024 and January 2025, we estimate that the consistent Vintage 2024 series implies an increase of 95,000 while the published data will show a much larger 2.4 million.

Similarly, the Vintage 2024 estimates will imply a higher foreign-born population in January 2025, as shown in figure 7 (estimated as above using the December 2024 CPS foreign-born share for January 2025). For the change in the foreign-born CNIP 16+ between December 2024 and January 2025, we estimate that Vintage 2024 implies an increase of 21,000 while the published data will show 530,000.

Our conclusion is that, because they are based on meaningfully different Census Vintages, extreme caution should be exercised when using the published CPS data to measure month-to-month changes in the size of the population, labor force, or employment since 2021. In particular, the monthly changes in those series from December 2024 to January 2025 reflect the shift to a new Vintage rather than actual developments over that period.

Alternatively, there are two time-consistent measures of population: First there are the CBO estimates published in their 2025 demographic projections, titled “Census through 2020 plus CBO projection.” Those are available for CNIP 16+ and available on an annual basis. Second there are the Vintage 2024 estimates published by Census. Those are currently only available for CNIP all ages but are available monthly. The Vintage 2024 estimates incorporate less net migration between 2021 and 2024 than the CBO estimates. While in previous work we argued that the Census Vintage 2023 estimates were clearly too low, our assessment is that both CBO and the revised Census Vintage 2024 series are in the range of reasonable estimates. Differences between Vintage 2024 and CBO are small relative to the remaining uncertainty around immigration flows.

Using a time-consistent population measure, it is then reasonable to apply the labor force participation rates, the employment-to-population ratio, and other shares published in the Current Population Survey to compute the size of the labor force, employment, and other measures. We appreciate that those CPS-derived shares may be somewhat sensitive to changing demographic distributions in the population controls. Nonetheless, they are more comparable across changing vintages than the population levels.

What the new Census estimates imply for labor force growth and employment growth

We next turn to what the most recent population estimates imply for growth in the labor force. We use our estimates of CNIP 16+ from figure 5 and the monthly published estimates for the labor force participation rate, assuming that this rate stays constant from December 2024 to January 2025. Figure 8 shows that the published CPS data will likely show a significant increase in the size of the labor force from January 2024 to January 2025, as the published data for January 2025 align with the Vintage 2024 data, while that for January 2024 aligns with the Vintage 2023 data. In contrast, if the Vintage 2024 data were applied to both January 2024 and 2025 estimates, it would likely imply a more modest increase, as would the CBO time-consistent population series.

For example, the published CPS will likely show labor force growth between December 2024 and January 2025—a single month—of roughly 1.8 million because of the change between Vintage 2023 (December) and Vintage 2024 (January). If both months were consistent with Vintage 2024, the growth in the labor force over that one month would likely be closer to 70,000. The labor force growth between January 2024 and January 2025 will be approximately 3 million, but the labor force growth implied by a time-consistent series would be something like 1.5 to 2 million.

The new published CPS estimates will likely also imply faster growth in recent months in employment, both overall and separately for U.S.-born and foreign-born adults. We apply our estimates of CNIP 16+ to the monthly published estimates for the employment-to-population ratio for both groups, assuming that this ratio stays constant from December 2024 to January 2025.

Figure 9 shows employment in January from 2020 to 2025 for U.S.-born CNIP 16+. Using our preferred population measure from Vintage 2024 or CBO throughout, rather than the switch in vintages used in the soon-to-be-published CPS estimates, would imply more modest growth from January 2024 to January 2025 (and faster growth during 2023). For the change in the U.S.-born CNIP 16+ employment between December 2024 and January 2025, we estimate that Vintage 2024 implies an increase of 55,000 while the published data will show a much larger 1.4 million.

Figure 10 shows the same estimates for the foreign-born population. Here, the growth in the published data should be much smaller because only about 18 percent of the total increase in the population is attributed to the foreign-born population (in line with the recent foreign-born share measured in the CPS) when the Vintage 2024 estimates are incorporated in the published CPS data. For the change in the foreign-born CNIP 16+ employment between December 2024 and January 2025, we estimate that Vintage 2024 implies an increase of 13,000 while the published data will show an increase of 340,000.

Recent net migration affected how fast U.S. employment can grow

Consistent with the approach we used in earlier analyses, we estimate that recent net migration had a significant effect on labor force and employment growth. Figure 11 uses data from the American Community Survey to show labor force participation for all recent immigrants age 16 and up. For each year since 2016, about half of new arrivals report being in the labor force in the year they arrive. Each cohort shows increasing connection to the labor force over time. Using these trends to estimate labor force participation for those in the recent immigration surge, we find the surge significantly boosts the pace that employment could grow each month without spurring unwanted inflation. CBO’s latest net migration estimates suggest that the pace in potential employment growth was boosted by roughly 100,000 a month in 2023 and 75,000 a month in 2024, while the latest Census Vintage 2024 estimates suggest the boost was about 70,000 and 55,000 monthly in 2023 and 2024, respectively.²

The Vintage 2024 estimates will affect population estimates over 2025

The new administration has entered office with a clear desire to reduce the number of immigrants and has already taken action to reduce border flows, ramp up deportations, and make legal migration harder. We anticipate a significant slowdown in net migration over the coming year. It is possible net migration could even turn negative if the administration’s policies are sufficiently extreme. Because immigrants make up the bulk of population growth in the U.S., the anticipated slowdown will drive notably slower population growth, and therefore labor force and employment growth, over 2025. In other words, more modest job growth than in recent years will be consistent with a healthy labor market as well as low and stable inflation.

The slowdown also may complicate interpreting data on labor force and employment levels during 2025. Census has already produced its (Vintage 2024) projections that will be used to determine CPS population controls throughout 2025. Assuming the share of the population change attributable to net migration in the coming year is equal to the average share from July 2023 to June 2024, we estimate that those projections are consistent with net migration of around 1.3 million non-institutionalized adults over the course of 2025, lower than we have seen in the past few years but higher than historical norms and higher than we anticipate under the Trump administration. Indeed, in the scenario we estimate with 430,000 interior removals, 900,000 adjudications of parole or asylum leading to removals, and 220,000 people voluntarily outmigrating in response to stricter enforcement, we estimate negative total net migration of 650,000 in 2025 (the corresponding decrease of 530,000 in CNIP net migration is shown by the gray bar in figure 12.) In this scenario, CPS data published during 2025 will likely overstate growth in the true level of population, labor supply, and employment throughout the year.

What do the new Vintage 2024 estimates mean for the gap between employment in CPS and employment according to the Establishment Survey?

In our previous work, we explored the differences between measures of employment in the CPS and in the Establishment Survey, which reflects what firms report. After adjusting the CPS measure so that it reflects a similar concept as the Establishment Survey, the published estimates of employment in the CPS have been lower by increasing amounts since 2022. In December 2024, the gap between employment in the Establishment Survey and the CPS was a remarkably large 4.3 million.

The incorporation of Vintage 2024 will help to shrink that gap. Our rough estimate is that using the Vintage 2024 estimates boosts the level of the CNIP 16+ population in January 2025 by about 2.9 million. That means it boosts the estimate measure of employment in the CPS by about 1.7 million.

To add another wrinkle, the employment report on February 7 will also incorporate into the Establishment Survey new data from the Quarterly Census of Employment and Wages. As economist Jed Kolko explains, incorporating that new data will revise down employment in the Establishment Survey—perhaps by roughly 800,000.

Taken together, the gap between employment in the two surveys—after adjusting employment measures to be comparable—will be a smaller 1.9 million. Notably, if instead we use population and employment implied by the time-consistent CBO series instead of CPS, the gap with employment in the Establishment Survey is roughly 500,000, which is much less remarkable relative to the historical gap.

Conclusion

Rapid and unexpected changes in population make the estimation of official population measures incredibly challenging. Recent improvements in the Census population estimates make those official counts align more closely with outside estimates, including those from the Congressional Budget Office. Nonetheless, extreme caution should be exercised when using the published CPS data to measure month-to-month changes in the size of the population, labor force, or employment since 2021. In particular, the monthly changes in those series from December 2024 to January 2025 mainly reflect the shift to a new vintage rather than actual economic developments over that period.