On Monday the Heritage Foundation released a paper, The Fiscal Cost of Unlawful Immigrants and Amnesty to the U.S. Taxpayer. Along with Heritage’s new director Jim DeMint, the study’s senior author, Robert Rector, also published a Washington Post op-ed titled, “What amnesty for illegal immigrants will cost America.”

The key substantive point of the paper and op-ed is a $6.3 trillion number:

An exhaustive study by the Heritage Foundation has found that after amnesty, current unlawful immigrants would receive $9.4 trillion in government benefits and services and pay more than $3 trillion in taxes over their lifetimes. That leaves a net fiscal deficit (benefits minus taxes) of $6.3 trillion.

I expect making 8-9 million people here illegally into U.S. citizens would increase future deficits once these people are eligible for benefits. But there are so many problems with the Heritage study that this $6.3 trillion number is useless for making policy decisions. It might as well be plucked out of thin air.

Others have criticized the Heritage study, including a detailed critique from Cato’s Alex Nowrasteh. I’m adding to that here.

The basics of estimating government expenditures for an entitlement program are simple.

  1. Figure out the number of people who will receive subsidies;
  2. Figure out how much the government will spend, on average, per person per year;
  3. Choose your timeframe, measured in years;
  4. Multiply (1) by (2) by (3) and you’re done.

The Heritage study makes errors in each step.

1. The study purports to measure the “cost … of amnesty.” Let’s set aside whether amnesty is the appropriate word to describe what is in the Gang of Eight’s bill. The fiscal cost of “amnesty” is the incremental cost of making those here illegally eligible for government payments and services relative to what we have now. People here illegally are already imposing fiscal costs by driving on the highways, using public parks, relying on local firefighters for emergency services, and in some cases receiving uncompensated care in emergency rooms. Some children here illegally are already receiving free public education. Making those children U.S. citizens doesn’t increase the cost to the state of educating them.

There are significant incremental costs to amnesty, mostly from millions of people eventually becoming eligible for Social Security, Medicare, Medicaid, food stamps, and cash assistance.

But in addition to these incremental costs, Heritage incorporates all of the baseline costs in their $6.3 trillion estimate of “amnesty.” By measuring total rather than marginal costs, they are telling us the cost of making millions of people legal relative to the cost of somehow finding and deporting all of them. Setting aside my view that such an alternative is infeasible, it is inaccurate to describe such an estimate as the cost of amnesty. It is instead the cost of amnesty relative to some other state of the world that Heritage would prefer. That is both misleading and not particularly useful to policymakers.

The study authors acknowledge this. On page 29 of they write:

The $6.3 trillion figure represents the lifetime fiscal costs of unlawful immigrant households after amnesty. It does not represent the increased fiscal costs caused by amnesty alone. The increased lifetime costs caused by amnesty would equal $6.3 trillion minus the estimated lifetime fiscal costs of unlawful immigrant households under current law.

The killer problem comes with the way Heritage is marketing the study. $6.3 trillion is Heritage’s estimate of the total costs of illegal immigrants if “amnesty” is granted. But the op-ed is titled “What amnesty for illegal immigrants will cost America,” and the op-ed says “amnesty has a substantial price tag,” thus mistakenly presenting $6.3 trillion as the marginal cost of a proposed policy change. The study says one not very useful thing, while the op-ed markets the study’s results as something else entirely.

2. In addition to not being an estimate of the cost of “amnesty,” the $6.3 trillion number does not measure the cost of illegal immigrants, either. The study includes a number calculated from looking at 12.7 million people, then labels that the cost of “illegal immigrants.” Contained within those 12.7 million people are 4.5 million children born in the U.S. to at least one parent here illegally. These children are U.S. citizens. Heritage is inflating the population of “illegal immigrants” measured by 55% = (4.5 / (12.7 – 4.5)).

The Heritage logic is that those 4.5 million children are in the U.S. only as a result of illegal immigration, so the costs of providing them with taxpayer services are costs of illegal immigration. Heritage is correct that, had those 8.2 million adults here illegally never arrived, those 4.5 million children would not have been born in the U.S.

But you can’t unring the bell. Whatever your view on whether policy failed by granting citizenship to the U.S.-born children of an illegal immigrant, these children are now U.S. citizens and entitled to benefits (and with the requirement to eventually pay taxes). The fiscal costs of these citizen children can’t be counted as the cost of “amnesty for illegal immigrants,” since (a) they are being paid now and (b) they will continue to be paid even if their parents do not get benefits.

At best Heritage’s estimate is not the cost of “amnesty” for illegal immigrants. It is an estimate of the cost of “amnesty” for 8.7 million illegal immigrants plus the costs of benefits for 4.5 million U.S. citizen children.

While the study acknowledges that their estimate depends heavily on this unusual redefinition, the op-ed and Heritage’s subsequent promotional efforts do not. In the study the authors carefully acknowledge:

For example, any study that excludes the welfare benefits and educational services received by the minor U.S.-born children of unlawful immigrant parents from the costs assigned to unlawful immigrant households will reach very different conclusions about the fiscal consequences of illegal immigration.

The DeMint/Rector op-ed makes no such distinction and is factually incorrect because it includes the benefit spending for 4.5 million U.S. citizens (emphasis added by me):

current unlawful immigrants would receive $9.4 trillion in government benefits…

As we’ll see next, the study is also not really a complaint about the fiscal effects of illegal immigrants, or even of immigrants generally. It is instead a complaint about low income people receiving net government subsidies.

3. The logic behind the study can be summed up like this:

  • Poor U.S. citizens cost the government money because they receive more in benefits than they pay in taxes;
  • Immigrants tend, on average, to be fairly low on the income scale;
  • Illegal immigrants even more so; therefore
  • Making illegal immigrants legal (at some point) will therefore increase future deficits.

But other than the “even more so” point, this isn’t really an argument about illegal immigrants. It’s an argument about the existence of low-income Americans, a redistributive safety net and old-age subsidies, and an income tax system that exempts about 45% of Americans from paying income taxes.

Senator DeMint and Dr. Rector write:

Yet immigrants should come to our nation lawfully and should not impose additional fiscal costs on our overburdened taxpayers.

I agree with the first part, but they run into trouble on the second.  If you believe their numbers, legal immigrants also impose additional fiscal costs on our overburdened taxpayers.

The Heritage logic would apply equally to legal immigrants and to babies born to low income U.S. citizens. I know Heritage is not opposed to either of those, but the logic of this study (and the language in the op-ed) would seem to lead to such a conclusion.

4. Even more generally, the study is actually a critique of our unbalanced federal budget for everyone, and not just for the poor, much less for a subset of the poor. The federal budget is terribly out of whack, the result of promising way too many entitlement spending benefits without raising the taxes needed to pay them. (I would of course solve this by changing the spending promises, not by raising taxes.) Entitlement spending promises to seniors are the largest cause of this problem and are the drivers of unsustainable federal and state government spending trends.

Illegal immigrants are being promised far more in benefits than they will pay in taxes not just because they are on average low income, but because almost everyone is being promised old age benefits that will exceed the taxes they will pay.

In addition, Heritage’s calculations mistakenly assume that Social Security and Medicare benefits will be paid in full to newly legal immigrants for the next 50 years, even though (a) it’s obvious that neither program’s spending is sustainable in its current form for even half that time; and (b) current law includes a mandated 27% cut in Social Security benefits once the Social Security “trust fund” has a zero balance, and a 13% cut in Medicare part A benefits when the Hospital Insurance “trust fund” hits the wall. As a result, Heritage’s estimates of the direct old-age benefit costs for newly legal immigrants are too high.

5. The authors do a lot of work to group illegal immigrants by educational level (as a proxy for income) and to estimate the fiscal costs for various eligibility phase-in timeframes. But, as best I can tell, they assume that a poor, low-skilled, poorly educated illegal immigrant will remain poor, low-skill, and poorly educated, and that he will draw government subsidies his entire life. Incomes typically climb as a worker ages (including for low-skilled, low-wage workers). Some people who arrive in the U.S. illegally may initially take jobs well below their skill level because of language barriers that they later overcome. Others will get further education or build skills over time. Since government subsidies relate inversely to income, if you assume illegal immigrants will never see their education, skills, or income increase, then you’ll overestimate the government subsidies spent on their behalf and underestimate the taxes they pay. Ignoring the potential for self-improvement and economic advancement inflates the cost estimate.

Let’s not forget that those here illegally are the subset of those back home who had some combination of skills, determination, and savvy to make it to the U.S. despite significant barriers.

6. More broadly, the fear of discovery and deportation has to constrain the labor supplied by those here illegally. A talented electrical engineer here illegally might now be working in an unskilled job because green card verification is weaker for driving a cab than for working at Cisco or Intel. A spouse here illegally might choose to stay at home with the kids rather than seek paid work because the former has significantly lower risk of being discovered by immigration authorities. Eliminating these deportation risks will increase the available labor supply. Increased labor supply means a larger GDP, a larger tax base, and higher government revenues. Heritage appears to ignore all these supply-side labor effects, which is unusual from an organization that in other contexts has championed consideration of supply-side effects in fiscal estimates.

7. To get their huge numbers Heritage sums up spending over a 50-year period. They adjust their estimates of future spending and taxes for inflation but not for the time value of money.  Most official legislative cost estimates are done for a 5 or 10-year period, in which these effects are small and customarily ignored. But nobody adds up a 50-year fiscal stream without either discounting it or measuring it as a % of GDP rather than in real $. Using a 3% real long-term interest rate (CBO’s assumption), a dollar of cost 50 years from now is equivalent to a 23¢ cost today. Heritage counts this cost as a dollar rather than 23 cents. They should be showing us net present values if they want to do long-term estimates. It’s tough to say exactly, but a back-of-the-envelope calculation suggests this almost doubles their final number.

8. Finally, the DeMint/Rector op-ed purports to describe what illegal immigrants “will cost America,” when in fact they are looking only at what they will cost American governments. This is an important oversimplification and a surprising one from an institution as reliably conservative as Heritage, which I would never expect to make the mistake of equating the Government with the Nation. I happen to think the cultural and non-governmental economic benefits of a robust immigration system and a resolution of the problem of a stock of 8-9 million people here illegally far outweigh the likely impact on the federal budget, but you might make a different judgment call. Either way, it’s obvious that the federal budget is only part of the calculus, and the op-ed erred in suggesting otherwise.

To summarize, the $6.3 trillion number offered by Heritage as “the cost of unlawful immigrants and amnesty”:

  • Does not estimate the marginal cost of an amnesty program or whatever is in the Gang of Eight bill relative to anything other than a fantasy state of the world;
  • Includes costs for 4.5 million U.S. citizen children in its estimate of the costs of illegal immigrants;
  • Employs logic extensible to both legal immigrants and poor U.S. citizens, and thus is actually a critique of government tax-and-spending redistribution more than of illegal immigration;
  • Mistakenly assumes that unsustainable Social Security and Medicare benefits will be paid to newly-legal immigrants for 50 years when both programs will go bankrupt long before;
  • Ignores supply-side labor effects by mistakenly assuming that an illegal immigrant who is poor and receiving government benefits will remain poor for 50 years, that he will not learn new skills or increase his income, and that he will not find a better-paying job or take additional work after eliminating the risk of deportation;
  • Ignores the time value of money; and
  • Mistakenly labels the cost to the Government as the cost to the Nation, ignoring the cultural and non-fiscal economic benefits and costs of changing immigration policy.

There is a significant fiscal effect from making 8-9 million people into U.S. citizens and making them legal taxpayers and eventually beneficiaries eligible for the full panoply of government subsidies. But Heritage’s $6.3 trillion figure is not a good estimate of that effect and is useless for making policy decisions.

I am a fan of Heritage and respect many of their scholars and their policy work. Unfortunately this study and op-ed are not up to their usual standard. They subtract value from an informed policy debate about an important topic.