Tyler Cowen has a recent column that advocates asking a citizenship question on the 2020 census. I don’t find his argument to be persuasive, for several reasons.

Tyler does acknowledge that asking this question will make the census less accurate in estimating total population, but sees even bigger costs from not asking the question:

Not asking about citizenship seems to signify an attitude toward immigrants something like this: Get them in and across the border, their status may be mixed and their existence may be furtive, and let’s not talk too openly about what is going on, and later we will try to get all of them citizenship. Given the current disagreement between the two parties on immigration questions, that may well be the only way of getting more immigrants into the U.S., which I hold to be a desirable goal. But that is a dangerous choice of political turf, and it may not help the pro-immigration cause in the longer run.

That’s certainly not my view.  An acquaintance of mine is a legal resident who is not a citizen and doesn’t plan to ever become one.  I have no hidden agenda to get him citizenship. Regarding illegal residents, I’d be happy with an amnesty that allowed them to stay as permanent resident non-citizens.

I’m a philosophical pragmatist, so if someone could show me that there were substantial benefits to making the census slightly less accurate, I’d be willing to listen.  But exactly what are the benefits from a citizenship count?  I don’t see them. The US Constitution asks very little of the federal government.  (It already does far more than the founders thought appropriate.)  It doesn’t seem too much to ask to try to carry out one of the government’s few constitutional duties as accurately as possible.

Let’s say I’m wrong, and that the benefits from an accurate citizenship count are substantial.  Even then I’d oppose asking the citizenship question on the census form.  Instead I’d instruct the Census Bureau to use some other method to estimate the total number of US citizens.  That’s because a census question of citizenship won’t just lead to an undercount of total population, it is likely to lead to a relatively inaccurate measure of the share of the population that are non-citizens.  Thus a citizenship question on the US census is actually likely to lead to a less accurate estimate of the total number of US non-citizens than some other estimation method.

Much of Tyler’s essay glosses over the distinction between citizens and non-citizens.  Instead, the discussion either implicitly or explicitly centers on the distinction between legal residents and illegal residents.  But this sort of census question would tell us almost nothing about that issue that we do not already know.  Consider:

1. Non-citizens are composed of two groups, legal residents and non-legal residents.

2. This question would not discriminate between these two groups.

3.  Illegal residents are almost certainly much less likely to respond to this question than legal residents.

Thus suppose there are 12 million people who tell the census that they are not citizens.  What does that tell us about the number of illegal residents?  The government currently estimates that there are about 10.5 million illegal residents in America.  So would that 12 million figure imply 10.5 million illegals and 1.5 million legals?  Or would it indicate that only 2 million illegals dared respond, and the other 10 million responses were legal non-citizens?  I have no idea.

Perhaps the census could somehow produce a fairly accurate estimate of legal non-citizens, I don’t know.  But even in that case the estimate of total non-citizens (which includes legal and non-legal residents), is likely to be highly inaccurate because of illegals being undercounted.  If you asked me to produce the most accurate estimate of citizens that I could come up with, it would be the census estimate of total population (including the Census Bureau’s figure for the estimated undercount), minus the government’s estimate of legal non-citizens, minus 10.5 million.  That would likely be far more accurate than anything that comes out of a census citizenship question.

It seems to me that people who worry that there is too much immigration (a group that does not include Tyler) want to have it both ways.  They hint that we need this sort of question because illegal immigration is a big problem, and yet they don’t call for the appropriate question on the census, “Are you an illegal?” because they don’t want to be laughed at.  So instead they ask about citizenship, implying that this somehow addresses the public policy issue of illegal immigration.  It doesn’t.

I’m reasonably satisfied with the government estimate of 10.5 million illegals in America.  I’m sure it’s not exactly right, but it’s also obviously true that the number is much more than 1 or 2 million and much less than 30 or 40 million. It’s a plausible sounding number.  And no census question will get us closer to the truth.

If there are 320 million legal residents, then obviously the vast majority are citizens and a small proportion are non-citizens with green cards or visas to study or work here, etc.  I don’t feel any great urge to know the share of legal residents that are non-citizens, but if people really wish to know this information then I’d suggest using some sort of statistical study that does not make screw up one of the government’s few constitutional duties, an accurate headcount of the total population.

It’s possible that I misread Tyler’s argument.  It seems to me that he’s saying that it looks bad when the Democrats oppose this question.  It looks as if they are ashamed to admit that there are lots of illegals, and that they (the Democrats) want them to be able to stay here legally via some form of amnesty.  In fact, it looks more like some Republicans wish the Constitution did not consider total population when allocating political representation (at the federal level), rather only legal population, or citizens.  And it looks like they are trying to use this question to push the population count closer to what it would be if the Constitution were written in the way they preferred.

No doubt many conservatives who disagree with me will contest the way I characterize their motives.  But then so would the Democrats regarding Tyler’s characterization.  And again, if you ask, “Which of the two parties seems to be advocating a procedure that will produce the most accurate count of the sort that America’s founders asked for?”, I’d have to say the Democrats.  Yes, an accurate count works in their favor politically, so I’m certainly not suggesting pure motives on either side.  Rather I’m claiming that the Democratic position seems to align better with the original intent of the Constitution’s framers.  And which group is supposed to believe in original intent?

PS.  If we assume that it takes 10 seconds to respond to this question, then it will waste perhaps 20 million minutes of time.

PPS.  AFAIK, the census currently has pretty good estimates of total population.  If I recall correctly, they use various statistical techniques to estimate the undercount in specific areas, and then extrapolate to get a national undercount estimate.  I don’t believe those undercount estimates are used for re-apportionment, but they can be used by academics.  Please correct me if this is inaccurate.

PPPS.  Canada and Australia both ask this question, which seems appropriate given that rates of illegal immigration are low in both countries, and thus the population count would not be significantly affected.

PPPPS.  This post does not consider the question of whether the Trump administration decision to ask this sort of question is constitutional, or the entirely different question of whether the Supreme Court was correct to send the issue back for further clarification.  In this post I’ve also steered clear of the racial dimension to this question.  Some would argue that there is a hidden agenda to reduce the political clout of minority voters, much as the old literacy tests were more about race than literacy.  In my view, however, the argument for a citizenship question is quite weak even if you dismiss the notion that there is hidden racial bias in this decision.

PPPPPS.  Can someone check more than one box?