Sorry, Heritage, your number is still wrong

I’d like to thank the Heritage Foundation for taking the time and effort to respond to my post, “Eight problems with the Heritage immigration cost estimate.” Heritage’s response, written by Derrick Morgan, is professional and takes a constructive tone. I still disagree with their conclusion, but I appreciate their efforts to make this a discussion rather than the usual screaming matches one finds online.

I’m not sure it adds a lot of value for me to go point-by-point rebutting each of Mr. Morgan’s responses to my original technical points. He is right that we agree in some areas, even though he doesn’t highlight all areas of our agreement. I am sorry to say that where we still disagree his responses have not convinced me to change my original critiques.

While I therefore stand by each point of my original critique, I think it’s most helpful if I highlight a few points on which we agree, and then explain again only the most important reasons why I continue to recommend policymakers ignore this study and its headline number.

Where Heritage and I agree

Heritage highlighted a few important points of agreement, as well as a few other smaller ones.  We agree on the following.

  • Making legal those now here illegally would increase government spending and deficits. This is because much of this population is low income, and the benefits they would receive from government exceed the taxes they would pay.
  • Therefore, as Heritage frames it, “amnesty will cost taxpayers.”
  • It is highly unlikely that the additional economic growth resulting from making these folks legal would offset these costs, especially since governments would capture only a small fraction of that additional growth.

If you read Mr. Morgan’s response carefully you can see that Heritage agrees with several other points I have made.

As best I can tell, Heritage and I agree on the following.

  • Heritage’s $6.3 trillion headline number does not represent their estimate of what they label as “amnesty,” but instead something different.
  • This number includes the costs of government services now being used by 8-9 million people here illegally.
  • This number includes current and ongoing costs of 4.5 million U.S. citizen children. These costs would remain if you could somehow find and deport everyone now here illegally.
  • While this number includes a one-time +5% bump in income (I missed this), it ignores future skill and income growth of newly legal U.S. citizens, and the effects of that growth on the budget.
  • This number excludes the supply-side effect of increased labor from newly-legal Americans who would have more ability to work and for some take higher-skilled jobs that better match their abilities and training.
  • This number ignores the time value of money and treats an inflation-adjusted dollar of costs 50 years from now as equivalent to a dollar of costs today.

The baseline & citizen kids disagreements

Heritage and I disagree on what numbers are relevant for policymakers. I argue that the relevant number for policymakers is the incremental cost of what is now being debated in Congress, eventually making legal those now here illegally. (Heritage calls this “amnesty,” I shy away from that term.) This increment, I argue, should be measured relative to current policy: the reality that 8-9 million people are in the U.S. today, using government services and imposing fiscal costs on governments.

Heritage argues policymakers should also know the cost of the proposed policy relative to a state of the world in which illegal immigration had never occurred. How can that possibly be relevant to any policy decision now being debated? That illegal immigration did occur, and those people are here now. Even if you somehow had a way to find every illegal immigrant, deport him or her, and perfectly prevent all new illegal immigration, what are you going to do with 4.5 million U.S. citizen children? Deport them too? That’s what the Heritage study tries to estimate.

While Heritage or some policymakers might prefer that state of the world, it’s impossible to get there from here, both because we don’t know how to find and deport everyone, and because you can’t undo the existence of 4.5 million U.S. citizens. It is therefore an irrelevant basis for comparison.

Some may argue, “How can it hurt to know this additional number, even if it’s kind of silly and not relevant to the current choice being considered by Congress?” It hurts because reporters, policymakers, and Heritage shorthand things, and policymakers mistakenly substitute the useless number for the useful one. That is definitely the case here. Unless you are scrupulously reading the report’s footnotes (or my blog), you’d never know that Heritage most of the time does not actually claim that $6.3 trillion is the cost of the “amnesty” provisions in the Gang of Eight bill.

This confusion, which always exists in real-world legislative processes, makes a misleading number harmful. If this confusion is intentional, then it’s deception and highly irresponsible.

The costs of more poor Americans

I think Heritage and I are pulling in roughly the same direction on concerns about our enormous and unsustainable entitlement state. The Heritage work on illegal immigration focuses on the additional fiscal costs of having more poor Americans. I am willing to bear some (I said some) additional fiscal costs of making legal those here legally if it’s part of a long-term solution that dramatically reduces the flow of future illegal immigration, and if that solution significantly expands our high-skill immigration.

At the same time, I think parts of our low-income support structure are seriously fouled up. I’d highlight three big concerns: the explosion of food stamp eligibility in recent years; the explosion of those claiming disability insurance; and the insanely high effective marginal tax rates on some low-income workers as their income climbs and they lose eligibility for government benefits. I think all three problems should be high priorities for policymakers to fix.

I place an even higher priority on dialing back quite dramatically the taxpayer subsidies for the broadly defined “middle class” in the form of old-age entitlements and now new health insurance subsidies. Because they are bigger and apply to so many more people, these unsustainable middle class benefit promises are even more fiscally damaging than the low-income subsidies so often highlighted by Heritage.

And I think I differ from Heritage in that I see both of these solely as problems of our entitlement state, not as costs of illegal immigration. Let’s fix the whole immigration system and offer new, better structured, fiscally sustainable entitlement promises to the middle class and the poor. Let’s not use fouled-up entitlement spending structures as an excuse not to improve immigration policy.

The fatal flaw:  No discounting

For me the fatal flaw in the Rector-Richwine paper is not accounting for the time value of money. I applaud Heritage for trying to estimate the long-term costs of immigration policy, and I am a proponent of long-term cost estimates (especially for old-age entitlement programs where they are most relevant). Heritage repeatedly refers to its “exhaustive” study, a “100 page paper” that contains “a complex analysis of 34 categories of taxes and 73 expenditure categories.” But if your study tries to examine costs over a 50-year timeframe and you’re not doing a net present value calculation, everything else is moot. You are overestimating lifetime costs by nearly a factor of two simply by not discounting future costs and benefits. You undercut the rest of your hard work by ignoring this elementary principle.

Sorry, Heritage, it’s still not a useful number

Estimating the costs of making legal those here illegally is difficult. Well-intentioned estimators will take different approaches to the same problem. But when you estimate the costs of the wrong population, compare it to a fanciful/impossible baseline, and ignore the time value of money, your headline number adds no useful information to the immigration policy debate. Each of these problems leads to an overstatement of the cost, as does each other smaller concern I highlighted with the paper and op-ed. When those flawed conclusions are then further misrepresented as the “Cost of Amnesty,” real harm is done to the policy debate on an important national issue.

I continue to recommend immigration policymakers ignore the Heritage study and especially its $6.3 trillion headline number. Ask CBO for an estimate instead.

8 responses

  1. I agree that discounting is important, but it is convention in government budget discussions to use raw numbers in projections (maybe started by the CBO?)

  2. Agree with most of your post. But NPV isn’t useful in this case because the question isn’t “what is the most we should pay for this program now,” which is the question that NPV answers. But as Jacob mentioned, government budgeting is also based on nominal values — I think the idea is that when governments determine budgets, they don’t do so on a risk or inflation-adjusted basis but rather on a nominal basis.

    • CBO scoring is based on a 10 year window compared to presently enacted policy. Over that short a time-frame, inflation and discount rate are less of an issue. The Heritage paper covers a 50-year period and compares to a non-existent hypothetical situation where illegal immigration never occurred in the first place.

      • C200,

        “The Heritage paper covers a 50-year period and compares to a non-existent hypothetical situation where illegal immigration never occurred in the first place.”

        That partially true. However, Heritage also argues that Amnesty will make the current batch of illegals much more expensive by enabling them to consume more welfare (of all forms).

        Quote from Keith Hennessey (not Heritage)

        “Making legal those now here illegally would increase government spending and deficits. This is because much of this population is low income, and the benefits they would receive from government exceed the taxes they would pay.”

        Keep in mind that the current illegals could be sent home by taking away the jobs and handouts. In other words, America could save $6.3 trillion by sending the illegals home vs. giving them Amnesty.

      • c200,

        If you read the Heritage update, they have now broken down the cost into Amnesty vs. the status quo. The status quo will cost $1 trillion. Amnesty will cost $6.3 trillion. That makes the incremental cost of Amnesty $5.3 trillion.

  3. If you are going to discount, an obvious question is what discount rate to use. Apparently, Heritage used zero. Keith Hennessy proposes to use 3%. The problem is… Their is no evidence that a 3% discount rate has any validity over long periods of time. Right now the 30 year TIPS rate is 0.639%. That’s much closer to Heritage than what Keith Hennessy is proposing.

    Of course, you can argue that real rates of return are depressed of late. However, TIPS have essentially never returned 3%. Rates in the range of 1-2% were the norm before the crash of 2008.

    T-Bills provide another metric. The average real rate of return on T-Bills from 1928 to 2013 appears to be well under 0.5%. A quick check online shows that real returns on T-Bills have always been quite low. Another source gives a 0.6% real rate of return from 1926-2005.

  4. “This number includes current and ongoing costs of 4.5 million U.S. citizen children. These costs would remain if you could somehow find and deport everyone now here illegally.”

    Really. Immigrants and citizens leave the U.S. all the time. In the great majority of cases, they take their children with them. No law prevents them from doing so and no effort is made to stop them.


Get every new post delivered to your Inbox.

Join 6,545 other followers