What the National Academies Found on Immigration's Fiscal Costs

Ask a partisan whether immigration pays for itself and you will get a slogan before you get a number. One side says immigrants are a gold mine. The other says they are a sinkhole. Both sides are quoting a report that says neither.

That report is The Economic and Fiscal Consequences of Immigration, published in 2016 by the National Academies of Sciences, Engineering, and Medicine. It is the most careful single attempt to answer the question, it was written by a committee of economists and demographers, and it is a book-length study of "it depends." Let's find out what it depends on, because the answer is the best argument in the whole debate for choosing immigrants rather than merely counting them.

What the report is, and what it is not

Start with definitions, because this subject dies of vague words. A fiscal impact is the difference between what a person or group pays to governments (income, payroll, sales and property taxes, fees) and what governments spend on them (schools, health programs, cash and food assistance, roads, courts, police). It is an accounting concept. It is not a measure of anyone's worth, and it says nothing about the moral standing of a person who is a net recipient. Every American child is a net fiscal cost for the first eighteen years of life, and nobody concludes that children are a mistake.

An economic impact is a different thing: what happens to output, wages, prices and the incomes of native-born workers. The Academies report covered both, in separate chapters, and the two verdicts are different. On the economy, the committee found that immigration raises overall output and that the gain to the native-born is real but modest. On wages, it found that the long-run effect on the average native-born worker's pay is small, with the most likely losers being earlier immigrants and native-born workers without a high school diploma. We will leave the wage fight to Two Economists, One Boatlift, and the Fight Over Immigration and Wages and stay with the ledger here.

The report is a study of a population. It is not a verdict on any family. The average is a statement about a very large group of very different people, and the whole point of this post is that the spread inside the group is bigger than the argument over its mean.

Finding one: the first generation costs state and local governments money

Here is the headline that restrictionists remember. The committee found that first-generation immigrants, as a group, cost state and local governments more than the native-born do, per person. The driver is not a mystery and it is not a scandal. It is children. A household with school-age children costs a school district money, and the first generation has, on average, more school-age children per adult than the native-born do. K-12 education is the single largest line in a state and local budget, and it is paid mostly from state and local taxes.

Notice what the claim is and is not. It does not say immigrants use more welfare than natives in some dramatic way. The report's accounting shows that a large share of the state and local gap is education spending on the children of immigrants, and those children are, as the next section shows, the payoff. A fair reader says so. A fair reader also says the other half out loud: the bill arrives at the level of government that cannot print money, cannot run a deficit in most states, and did not choose the policy.

That last point is the one with a constitutional edge. Immigration is a federal power. The Constitution gives Congress the power "To establish an uniform Rule of Naturalization" (Article I, Section 8, Clause 4). The states do not vote on who comes. They educate, treat and police whoever shows up. If you design a system in which one level of government decides and another level of government pays, do not be surprised when the payer complains. We will return to that.

Finding two: the second generation is a fiscal star

Now the half of the report that restrictionists skip. The committee found that the adult children of immigrants, the second generation, are among the strongest fiscal contributors in the whole population. They are, on average, better educated than their parents, they earn more than their parents did, and they pay more in taxes than they cost. In the committee's accounting the second generation looks better than the third-and-higher generation of native-born Americans.

I want conservatives to sit with that, because it is the strongest evidence for an old conservative claim: that America assimilates. A country that turns the children of laborers into engineers, nurses and owners of small businesses is a country whose institutions work. The schools are doing their job. The family is doing its job. The labor market is rewarding effort. The second generation's performance is not an accident of genes or a gift of statistics. It is what a functioning civil society produces, and it is the reason the cost in the first generation is better described as an investment than as a loss. It is the argument of Lincoln's Electric Cord and the Case for a Civic Nation Anyone Can Join showing up in a spreadsheet.

But an honest conservative also has to say that an investment is only as good as the thing you invest in. If the children of a newcomer who arrives with little schooling and little English perform much like the second generation of the report, fine. If the arrangement fails to produce that second-generation lift, the first generation's cost does not get repaid. The report does not claim the lift is guaranteed. It reports the average as it was in the historical data it had.

Finding three: education is the dial

This is the section that matters most for policy. When the committee projected the lifetime fiscal impact of an immigrant by educational attainment, the ordering was exactly what any parent would predict. Immigrants with more schooling were far more positive for government budgets than those with less. The reason is simple. Income taxes are progressive, so a high earner pays a lot, and means-tested benefits fall as income rises. A person who earns more pays more in and draws less out.

The report's projections showed that the net fiscal contribution of immigrants with a graduate or professional degree was strongly positive, with the federal side the most favorable. Immigrants without a high school diploma were, on the committee's accounting, net costs. And the same pattern holds for the native-born. A native-born adult with a doctorate is a fiscal asset; a native-born adult who never finished high school is, on average, a fiscal cost. The committee did not discover that immigrants are special. It discovered that education and earnings determine a person's fiscal balance, and immigrants are no exception.

That is a point against both slogans. "Immigrants are a gold mine" fails because a large share of the immigrant flow has little schooling, and those arrivals are net costs in the first generation. "Immigrants are a sinkhole" fails because a large share has a great deal of schooling, and those arrivals are net contributors from the day they start working. The average is a blend, and the blend can be changed by changing the mix.

Finding four: the federal and state ledgers point in different directions

The report found that at the federal level the fiscal picture of immigration is much more favorable than at the state and local level. Federal revenue is dominated by income and payroll taxes. A working-age immigrant who arrives ready to work pays into Social Security and Medicare for decades, and the federal government does not have to educate the children; the states do. In the 75-year projections the committee ran, the federal side generally looked better than the state and local side, though the results rest on assumptions. State and local budgets were the weak side, particularly in the first generation.

Put the two together and you have an uncomfortable fact for everyone. The same immigrant can be a federal net contributor and a state net cost at the same time. A policy fight over whether "immigrants pay their way" is partly a fight over which ledger you consult. A Washington think tank that looks at federal taxes finds a surplus; a school board looking at its own budget finds a bill. Both are looking at the same data.

There is a second wrinkle that the committee was careful about, and I will be too. Some federal spending, such as national defense and interest on the debt, does not rise when a person arrives. How you treat that spending changes the answer. If the new arrival is charged a share of the debt interest and the defense budget, the fiscal picture looks worse; if you treat those costs as fixed and unaffected by population, it looks better. The committee showed results under different treatments. That is not evasion. That is the honest way to present a number whose assumptions are a matter of dispute. It is also why a politician who cites a single dollar figure from this report, in either direction, is telling you less than the report did.

The arithmetic of why who comes matters more than how many

Now let's build the argument from the arithmetic. Everything below is a labelled hypothetical with round numbers chosen to show the logic. These are not figures from the report, and the report's own estimates are the ones to check.

Suppose a group of immigrants has a lifetime net fiscal impact per person, and consider two simplified types.

• Type A: arrives young, has a college degree, works for four decades. Suppose the lifetime net contribution to all levels of government is plus $200,000.
• Type B: arrives with little schooling and little English, works at low wages and has children in public schools. Suppose the lifetime net impact is minus $100,000.

Now compare two policies, each admitting 1,000,000 people.

1. Policy X admits 1,000,000 people, of whom 20 percent are Type A and 80 percent are Type B. The total is 200,000 × $200,000 = $40 billion in, and 800,000 × $100,000 = $80 billion out. Net: minus $40 billion.
2. Policy Y admits 1,000,000 people, of whom 60 percent are Type A and 40 percent are Type B. The total is 600,000 × $200,000 = $120 billion in, and 400,000 × $100,000 = $40 billion out. Net: plus $80 billion.

Same number of people. A swing of $120 billion, entirely from the mix. Now try the opposite experiment: keep Policy X's mix but cut the number in half, to 500,000. The net is minus $20 billion. You improved the ledger by $20 billion by cutting the count in half, and you still lost money. Then double Policy Y to 2,000,000 people, same mix. The net is plus $160 billion. Doubling the volume of a good mix helped; halving the volume of a bad mix merely lost less.

That is the lesson. Volume is a multiplier. It multiplies whatever the mix is. Whether the multiplier is good or bad news depends on the sign of what it is multiplying, and the sign is set by selection: by age, by skills, by schooling, by English, by whether the arrival is entering the labor market or the safety net. A policy debate that argues only about the multiplier, "more" versus "fewer," and never about the thing being multiplied, is arguing about the wrong variable.

Let me add one more layer, again a hypothetical. Suppose Type B's children become Type A on average, as the second-generation finding suggests. Then Type B is not a permanent minus $100,000; it is a minus $100,000 in the first generation and a plus in the next. The sign of the family may be positive over two generations. Now you have a real, defensible argument for admitting a lower-skill flow, one the committee's second-generation result supports. A conservative who wants a skills-based system has to answer that argument, not ignore it. The answer is that the uncertainty in the second-generation lift is large, the cost arrives now and the payoff arrives in a generation, and a prudent country chooses the arrangement whose payoff does not depend on the assumption. Select for the skills. Let the second generation be a bonus, not a bet.

The other side's best case

Here is the strongest version of the opposing view, and I want its defenders to recognize it.

The case runs: first, the committee found that immigration is positive for overall growth and that native-born workers, on average, are barely affected in the long run. Second, the federal ledger is favorable, and Social Security and Medicare in an aging country need young workers paying in. Third, the second generation is an economic success and a fiscal contributor. Fourth, the state and local cost is mostly schooling for children, which is an investment in future taxpayers, so counting it as a "cost" is like counting a factory's construction as a loss. Fifth, the sensible response to the state-and-local mismatch is for the federal government to share the cost, not for the country to shut the door.

That case is not stupid, and several of its premises are supported by the report. The fiscal benefit to the federal government is real. The second generation really does outperform. The construction analogy is a serious point.

The answer, point by point

One. Premise one is true on average and silent on the mix. "Immigration is good for growth" is a statement about the aggregate. It does not tell you which immigrants, and the report's own education tables show that the sign of the fiscal effect is different for different groups. An argument that stops at the aggregate is an argument that cannot tell Policy X from Policy Y, and I just showed that they differ by $120 billion.

Two. The federal ledger is favorable because the federal government does not educate the children. That is a quirk of who pays for what, and it is not evidence that the arrival is a net benefit to the country. The country is the sum of its governments. A cost shifted from Washington to a school district in Texas has not disappeared. It has been moved to a taxpayer who had no vote.

Three. The second-generation result is genuinely good news, and it is the reason for optimism about assimilation. But it is an average built from past arrivals, in a particular economy, with particular schools. Prudence says to select on what you can observe now. Nobody should stake the budget on a lift that depends on institutions we are not maintaining.

Four. The "school spending is an investment" analogy is right that an investment is not a loss. It is wrong that all investments pay. The right test of an investment is the return, and the return on educating a child depends on what the child becomes. The committee's own education tables are the evidence that the return differs by group.

Five. Sharing the cost federally is a reasonable idea, and I will say so. It also makes the case for choosing carefully stronger, not weaker. If the federal government is going to pay the state-and-local bill, then the taxpayers of every state have a stake in the selection rules that Congress writes. Nobody should want to pay more for a system whose mix is set by accident.

The Founding and the question of who decides

Return to Article I, Section 8, Clause 4. The Framers gave the power over naturalization to Congress and gave no state a veto. In Federalist No. 42, Madison explained why a uniform rule was needed: a patchwork of state rules would let one state's generosity become every state's problem. The Founders understood the mismatch. They did not live in an age of school districts and Medicaid, but the principle is old: a power exercised in one place that creates costs in another is a power that has to be exercised responsibly.

And the early Congresses did select. The first naturalization law, in 1790, applied to "free white persons" of "good character" after two years of residence. We reject the racial limit and should. The principle of a character requirement, and the long line of laws that followed, including the public-charge bars traced in Self-Sufficiency Was the First Immigration Rule, the Public Charge Story, shows that a country has always been allowed to ask what a newcomer will contribute. Selection is not a modern cruelty invented by restrictionists. It is a standing feature of how a self-governing people decides who joins it.

There is also a modern way to write the selection. A points system that rewards age, English, schooling and a job offer, the kind discussed in What an American Points System Would Look Like Written Into Law, aims the dial at exactly the variable the Academies found to matter most. It does not say that anyone's worth depends on a diploma. It says the immigration system is a policy with a budget, and a policy with a budget should be designed on purpose. That argument connects to the larger fight over The Case for Nuclear-Family Immigration Over Extended Family Chains, where the same question of selection runs through the family-preference categories.

What a fair reader should conclude

Let me be plain about the verdict, since "it depends" is a place to start, not a place to stop.

The Academies report does not show that immigration is a net fiscal disaster. It does not show that immigration is a free lunch. It shows that the fiscal value of an immigrant is driven by education, age at arrival and earnings, that the first generation costs state and local governments while the federal ledger looks better, and that the second generation tends to be strongly positive.

From that, the conservative conclusion is not "less immigration." The conclusion is better-selected immigration. If you want the fiscal benefits and you want to limit the fiscal costs, you do it by admitting people on the basis of what predicts a positive contribution, and by designing the system so the cost does not land on whichever state happens to be the destination. A country that admits a million people by lottery or by chain, without regard to skills, is leaving $120 billion in my hypothetical on the table. A country that admits a million by points is collecting it. The humane thing and the prudent thing point in the same direction here, because a system that works for the taxpayer is a system that can stay open to more people for longer.

The honest limit of all this is the limit of the report: it is a model with assumptions, built on past data. Read the report's summary chapter itself, and read the tables, before you trust anyone's two-sentence version of it, including mine.

The case in brief

1. Fiscal impact is a ledger of taxes paid and services used, and it is not a measure of any person's worth.
2. The 2016 National Academies report found first-generation immigrants cost state and local governments more than natives do, mostly through schooling, while the federal ledger is more favorable.
3. The second generation is among the strongest fiscal contributors, which is the best evidence that America assimilates.
4. Education and earnings drive a person's fiscal balance, so the mix of who arrives matters more than the headcount, as the labelled arithmetic above shows.
5. The sound policy response is selection by skills, age and English, with the costs of a federal decision shared fairly rather than dumped on a few states.

What you can do: find the summary chapter of the 2016 report on the National Academies website, read the education tables yourself, and then ask your members of Congress, through the page for your representatives, whether the immigration rules they vote on were written with those tables in view.