Here I draw from an essay by Jed Kolko, posted on Mathew Yglesias’ Substack newsletter, Slow Boring:

Kolko observes that in late 2023 and early 2024, government agencies released wildly different estimates of 2023 net immigration: 1.1 million people, said the Census Bureau, and 3.3 million, said the Congressional Budget Office (CBO). They create a range in estimates of net migration under the Biden administration between 5 and 9 million.

He notes that government agencies, such as the Census Bureau and the Congressional Budget Office (CBO), use varying methodologies, leading to significant discrepancies in immigration estimates. The Census Bureau relies on survey data from the American Community Survey (ACS) and the Current Population Survey (CPS), which use lagged data. This approach can underestimate recent immigration trends, especially during periods of rapid change. Kolko does not explicitly recognize that the surge in humanitarian program migration (asylum, Humanitarian Parole, and Temporary Protected Status) is very largely Latin American and Caribbean in origin.  The ACS, for example, provides data on foreign-born residents who lived abroad a year ago, but this lag can be problematic when immigration patterns shift quickly.

On the other hand, the CBO incorporates more immediate administrative data, such as visa issuances and border encounters.  This would better capture the humanitarian surge, and possibly also an administrated high figure for undetected migration.  But this method may also overstate immigration due to repeated encounters at the border.

These methodological differences, Kolko concludes, have led to a large gap between Census and CBO estimates, particularly in 2022 and 2023. The Census estimates tend to be lower due to their reliance on lagged data, while CBO estimates are higher, reflecting more recent administrative data.