The Budget Lab recently released a report titled “Lower Immigration Means Lower Productivity Growth”. It is based on the following logic: lower immigration means fewer working-age people; fewer working-age people means fewer potential entrepreneurs; fewer entrepreneurs means fewer new firms; and fewer new firms means lower productivity.
(The Budget Lab describes itself as “a non-partisan policy research center dedicated to providing in-depth analysis of federal policy proposals for the American economy” and is affiliated with Yale University.)
Unfortunately, each link in the chain depends on assumptions that are far more contestable than the report implies.
Let’s start with the working-age population. Lower immigration means fewer people, but it does not tell us whether the country is less productive or wealthy per capita. Much of the argument for immigration as a demographic necessity quietly conflates population decline with population aging. The real concern is usually aging, since that means fewer workers supporting more retirees.
The Budget Lab reports the change in the absolute number of working-age people under reduced immigration, but it does not show that immigration substantially changes the working-age share of the population. CIS has shown in the past that immigration has only a modest effect on population aging. The difference in the working-age share of the U.S. population by 2060 between maintaining 2019 levels of net migration and having zero net migration is only about two percentage points. Maintaining the 2019 working-age share would require quintupling net migration levels. At that point, immigration is no longer a solution to population aging; it would overwhelm the country before it “fixed” the age structure with all the labor market, housing, education, and political problems that would follow.
Once its assumptions are questioned, the precision of the model becomes much less impressive. We are left with a familiar problem in immigration economics: a sophisticated estimate of one side of the ledger, presented as though it tells us much more than it really does.
Even then, age structure is not the same as productive capacity. A 35-year-old is not automatically an economic asset because he is 35. The immigrant’s skills, earnings, fiscal contribution, language ability, and social behavior all matter. Models of this kind implicitly treat age as the key difference between natives and immigrants, while pushing human capital and fiscal contribution into the background. Given that immigrant households, both legal and illegal, use substantially more welfare than native households, and given the less-than-impressive educational profile of much of current immigration, that assumption is questionable.
The Budget Lab also leans heavily on the idea that immigrants are more likely to be entrepreneurs than natives. But “immigrant” is a broad legal category covering arrivals from all over the world with very different characteristics from each other. Some immigrant groups are heavily overrepresented in innovation, while others are not. Some legal immigrants are exceptional scientists, engineers, or entrepreneurs. Others instead arrive through family, humanitarian, student, temporary-worker, or status-adjustment channels and are unlikely to be pioneers in new and disruptive technologies or ideas. The Budget Lab report is not estimating the loss from excluding a handpicked group of exceptional founders, but instead the effect of reduced immigration flows in general.
In reality, there is a large discrepancy between the sources of our most innovative immigrants and the sources of much current immigration, as the two usually come from different regions of the world. If the “missing” immigrants are not the ones disproportionately responsible for high-value entrepreneurship, then the model has imported the prestige of exceptional immigrant founders into a much broader and less accomplished population.
Then there is the assumption that fewer firms means lower productivity. What kind of firms? The Budget Lab’s own appendix shows that much of the missing immigrants are concentrated in sectors such as construction, professional and administrative services, accommodations, and retail. These are not exactly the commanding heights of technological progress. A missing landscaping contractor, retail worker, or construction subcontractor is not the same thing as a missing semiconductor firm.
In fact, if employers cannot rely on endless foreign labor, they may have to use labor more efficiently. Historically, reduced access to immigrant labor encouraged mechanization and native reallocation. Recent evidence from Denmark also suggests that access to non-Western migrant labor can reduce incentives for robot adoption.
The broader issue with the Budget Lab’s model is that it essentially treats the economy as if lower immigration is simply the existing economy minus several million people. But that is not how economies adjust. When a disruption occurs, natives relocate in response to economic signals. Employers recruit differently to attract prospective workers. Firms learn to replace the lost labor with machinery, software, and other productivity-enhancing investments. Some younger natives move into opportunity-rich regions they might otherwise have been priced out of. Some low-productivity businesses never form, while more efficient firms expand. These are not minor or trivial details. They are the counterfactual.
The Budget Lab’s report does not prove that lower immigration makes America less productive in the real world. It merely establishes that if one assumes the missing immigrants are sufficiently entrepreneurial, that their firms would have been productivity-enhancing, that natives and capital do not adequately replace them, and that the other costs of immigration are outside the frame, then lower immigration produces a modest negative result. Once those assumptions are questioned, the precision of the model becomes much less impressive. We are left with a familiar problem in immigration economics: a sophisticated estimate of one side of the ledger, presented as though it tells us much more than it really does.