The GSS and the Political Externalities of Immigration: A Guest Post by Sam Wilson

I cannot reject the null hypothesis that there are no significant long-term political externalities to open immigration.

... screening for communicable disease and criminal record as part of the entrance process, but that's hardly controversial.

On this blog it IS controversial. I say that because we have people here who implicitly claim a certain moral superiority because they favor, on principle, open immigration. But having restrictions is not, on principle, open. Once you allow screening on those grounds you do not have a principled objection to using other grounds to screen.

What about controls for country of origin and multi generation effects of Hispanic groups? I believe their voting behavior in the second generation is to the economic left of the public at large. Moreover, their support for affirmative action, etc. is above average, their 2nd gen crime rates are higher, their academic achievement is lower.

Do I have to spell this out? Abortion.

I think pro-choicers are worried that Mexican immigrants are mostly Catholic and therefore oppose abortion rights. All this stuff about favoring infrastructure and military presence etc. is not on most people's radars when they discuss immigration.

While I think this analysis is valuable I don't find it convincing. Setting aside all of the difficulties regarding survey accuracy and reliability (i.e. surveyees ad-hoc forming opinions, surveyees acting in contradiction to their stated opinions, etc.) what I think matters most with regard to political externalities is net voter behavior. If, on net, immigrants tend to favor a particular party at the voting booth then regardless of the immigrants views on specific issues it is entirely reasonable to be concerned about the policies enacted by the elected representatives of said political party.

You should really run models without controlling for party-ID. To a somewhat lesser extent the same goes for education and income.

It is good to disaggregate this type of analysis though, so you can speak about differences among immigrants. If its truly open-borders we're interested in evaluating the impact of, then we should ask precisely who the immigrants are that are likely to come (education, income, race, national origin, etc.)

And what evidence is there that the GSS sample is representative of illegal immigrants?

More diversity means more people eligible for zero sum racial/ethnic preferences that my family is not eligible for. It means more pushing for expansion of who gets the goodies from the government, such as the current push to make Arabs a preferred group:

http://isteve.blogspot.com/2012/06/feds-considering-declaring-arabs.html

The analysis controls for education, race, income, and political party. It asks whether second or third generation Mexican immigrant lower-income high school dropouts are politically different than 4th or 5th generation Mexican low-education high school dropouts in the same party. It asks whether 2nd-generation German immigrants differ from the descendants of the earlier German immigrants.

This is like studies that find that parental IQ has no effect on child IQ, after you control for number of books in the home, parental income, and numerous other proxies for parental IQ.

The major political externalities of open borders seem to be intermediated via the "control variables" following, all of which seem to be excluded from consideration in the thesis:

1) A dumber population because of a relative increase of unskilled labor from countries whose immigrants and their descendants have had persistently lower IQ, income and education (which Bryan's work, or that of Jason Richwine). This is the biggie. Garrett Jones' work, among others suggests that the IQ externalities are huge. These could be nullified by some kind of guest-worker system with strict enforcement and no birthright citizenship, which would be great economically for workers and locals, but is not really on offer.

2)A poorer population which would vote for redistribution.

3) Increased racial diversity, which would reduce solidarity and social trust, per the work of Robert Putnam ("Bowling Alone"), increase the scale of affirmative action and similar transfers (it has so far, new immigrants automatically benefit from such), and affect many political issues where there are strong differences among racial groups.

4) Shifting the partisan balance, and making the ideological distribution in America more closely resemble one party's than the current American distribution. Basically, Republicans losing.

I would add to BK's excellent list, the increasing racialization of American politics, as can be seen in the 2012 campaign, where Romney and the Republicans are constantly derided as the white party while Obama attempts to assemble a coalition of minorities.

Thanks for the comments so far.

Ken B: Perhaps you're right, though I feel it's appropriate to acknowledge that incremental libertarianism is better than no liberty at all.

ajb: Unfortunately, the GSS does not appear to track country of origin (though if someone finds such a variable, I'd be happy to include it and re-run my regressions)

Coupon Clipper: perhaps I know different anti-immigration folks than you do, but the arguments I usually encounter are about job theft. The folks I know who are ardently pro-choice are often the ones who are (marginally) more pro-immigration.

Jason: I agree for the most part. I am often leery about using survey data for the very reasons you mention. I am a little bit more comfortable in this sort of setting because the act of responding to a survey isn't all that different from voting. Again, to reference Arnold Kling, voting and survey response is voice. Tying surveys to consumer behavior is confusing voice with action, a categorical error.

Michael Bishop: I did run the simple regressions. The results were consistent with or without controls.

Steve Sailer: It seems to me that arguments about group preference should be arguments that push back at the scope of the state's authority to grant rent privileges, not so much to bar people from coming to the US to work and support their families.

BK: Yes, in a nation of immigrants, we're measuring one generation against another. One way to interpret these results is to note that assimilation happens as quickly as the second generation, which seems to diffuse at least a few objections to open borders.

My claim isn't that allowing immigrants will produce a more classically liberal society, it's more that it won't necessarily produce a more statist society.

@Sam Wilson:
I like your phrase incremental libertatrianism. I think the describes my positions well. I for instnace favor more immigration, but not any settled principle that we must have more or that we cannot care about the kind of immigrant. (To be deliberately provocative I will meantion that Canada in particular has been benefitting from well-educated Jews fleeing France for french Quebec. Give us more!)

I make a sharp distinctioon between incremental and (tendentious!) the doctrinaire libertarianism characteristic of so many here.

>My claim isn't that allowing immigrants will produce a more classically liberal society, it's more that it won't necessarily produce a more statist society.

Sam, your results don't show that. Amnesties of current illegal immigrants will increase the share of low-education low-IQ Hispanic-Americans. There is data showing that for at least 4 generations Mexican immigrants retain low education, low IQ levels, and significant political differences with the median voter. Your analysis just hides the ball about the actual consequences of the policy.

It's worth noting that recent immigration status hasn't made a big difference for political attitudes (although controlling for political party is pretty bogus), but you should not go around claiming to have found that amnesty and reduced enforcement against illegal immigration will not make the USA more statist.

For true open borders, there would be a massive shift towards the global poor, who differ persistently in race, education, income, IQ, etc after immigration. See the depopulation of Puerto Rico after Puerto Ricans gained the right to enter the continental US, a movement staunched only after Puerto Ricans in Puerto Rico became eligible for large transfers from American taxpayers.

Also, in the open borders case assimilation would probably be attenuated, as super-numerous recent immigrants would have relatively less interaction with the existing population.

[Comment removed pending confirmation of email address and for crude language. Email the webmaster@econlib.org to request restoring your comment privileges. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

BK: As I mentioned above, I ran the regressions both with and without the political party controls with similar results. You have a very good point about extrapolation--once you push past the boundaries of the sample, the results may no longer hold.

As far as political party prediction, I just ran the strict generation variables (thirdgenonly, secondgen) on my aggregated party dummies (rep, dem, ind). Secondgen predicts democratic party membership. Thirdgenonly is statistically insignificant for all three.

As far as assimilation, the best I think I can do for now is point to other instances of mass immigration to the US. Irish, Polish and German immigrants seemed to assimilate pretty well. If enclaves persist over generations (as with, say, Hacidic Jews or Mennonites), I think it's worth investigating to what extent groups like that end up influencing national policy.

What about controls for country of origin and multi generation effects of Hispanic groups? I believe their voting behavior in the second generation is to the economic left of the public at large. Moreover, their support for affirmative action, etc. is above average, their 2nd gen crime rates are higher, their academic achievement is lower.

ps, in my comment (which as of now is still in the spam filter) i missed that you're already working w Jones.

[Gabriel: Please see your email. If you'd like to edit and resubmit the comment we'll be happy to publish it.--Econlib Ed.]

Have you looked at the work of Tino such as http://super-economy.blogspot.com/2011/09/more-on-bryan-caplan-ethnic-diversity.html ?

[broken link fixed--Econlib Ed.]

"As I mentioned above, I ran the regressions both with and without the political party controls with similar results"

Thanks, I slipped over that after reading it in the list of controlled variables.

"As far as assimilation, the best I think I can do for now is point to other instances of mass immigration to the US. Irish, Polish and German immigrants seemed to assimilate pretty well. "

Those groups also did well in their home countries (IQ, education, income, etc) and new immigrants from those countries still do well.

The Mexican data already exist, and there hasn't been economic or political convergence after 4 generations. I don't see any measure of Hispanic ethnicity (and racial composition therein, Cuban Spaniard vs mestizo Mexican American, vs Mexican native) in your study either, by the way.

People learn the language and such quickly, but differences in IQ that track country of origin, race, and ethnicity persist. And as Bryan has shown, those IQ differences affect political views.

ajb: I'd love to be able to parse those very issues, but I didn't find those variables in the GSS. When I head back in to gather some of the economic-related variables, I'll be sure to look closer. If you don't hear back from me here, I'll be sure to post something on my home blog http://euvoluntaryexchange.blogspot.com/

I will say that my initial concern about any relatively low performance of second-generation Hispanic immigrants may plausibly be endogenous to policy. I will hedge this by saying I don't want to make any definitive claims before examining the data firsthand, so consider this response something of a teaser.

Gabriel: not officially, not yet (though I am certainly open to that option. Garett Jones is in the top 5 of my favorite living macroeconomists).

Sam wrote:

Coupon Clipper: perhaps I know different anti-immigration folks than you do, but the arguments I usually encounter are about job theft. The folks I know who are ardently pro-choice are often the ones who are (marginally) more pro-immigration.

BK, sourcreamus: Thanks for the links. I will be sure to include them as part of my continuing investigation. The assimilation studies in particular should provide some good context for the GSS data.

It's interesting that this post is immediately followed by Arnold Kling's excellent one on opening/closing minds.

You describe the arguments of those who disagree with you as not reasons but "excuses", and write things like "(that is) the closest thing to a reasonable objection one might have". Do you think this kind of rhetoric serves to open minds, or close them?

A hearty second to Michael's call for you to not over-control the model and to evaluate the counterfactual on the prospective margin not the current margin. Your model assumes that we're assessing effects of immigration holding constant human capital, age, race, and (really?) party id. This is interesting in an analytical sense but as policy analysis you're confusing mechanisms with controls. From the perspective of a native who fears a shifting political consensus, does it really matter if migrants would shift the median vote because they are migrants per se rather than if the effect is mediated through control variables? You can then add on to this that the policy counterfactual is not more immigration with migrants who match natives on other observables (as you've modeled it), nor even migrants with their actual vector of observables for our current migrant stream (as would be implied by the bivariate), but more immigration with migrants with a new set of observables appropriate to the expanded margin.

(Yes, I know you said it's robust to bivariate but it's too important an issue to footnote, really more of a show us kind of thing).

Salem: point well taken. I shouldn't have presumed that the audience here was either familiar with or concurred with the pro-immigration arguments forwarded in the economics profession. Assuming economic literacy is an unfortunately common error among economists in general and economics students in particular.

Callowness is no excuse, however. Tone is important, and I think it's good advice to pay more heed to how I sound to others when I post. Thank you for highlighting my gaffe.

What BK said. Bluntly put, some people believe that Haiti, for example, is a mess it is because it is full of Haitians (some if it is IQ, the rest cultural, like south of Italy vs north). When we admit Haitian immigrants, we are making the US a bit more like Haiti. It is highly doubtful if the self-selection of immigrants can overcome the differences with respect to native population (IQ diff. of over 1 SD). I understand that Haitians are suffering horribly from living in a broken society, and we need to take that into account, but open borders...

I would like Brian (and Sam) to try to address this type of concern; if they agree with the above assessment, perhaps in the current pc climate they are not able to publicize their views, but then "of which we cannot speak we must remain silent".

"I would like Brian (and Sam) to try to address this type of concern"

It's been pointed out on here for years but is never addressed in a direct manner. That should tell you something. On the positive, at least they allow the arguments to be made and don't overmoderate them just to save face.

Searching for articles on immigration shows that your new colleague Noel Johnson has a 2008 paper on immigrants and voting behavior (Immigrant Participation in U.S. Elections) showing that their behavior does not converge to that of the native population and that this is a product of different cohort effects (i.e. the low end Hispanic boom post 1960s). This they tie to lack of assimilation in general although it is not explored in detail. Nonetheless BK's objections seem quite relevant and totally sidestepped by your work. It's quite relevant because the authors note that controlling for race/origin when studying assimilation tends to make immigrants look "the same" yet Johnson and LaFountain note they clearly are not, especially in recent decades.

If one is concerned about the country's IQ, perhaps we should allow in all immigrants that pass an IQ test above the US average score.

This is a bit off topic (not much), but take a look at the "Genius" types that current immigration law allows in.

http://in.reuters.com/article/2012/06/29/entertainment-us-usa-visa-extraordinary-idINBRE85S09A20120629

This blog post starts out claiming that most US voters want stricter immigration restrictions. That sounds like most voters want to allow fewer immigrants into the country. I am not sure if that is the case. I for one would like to see more immigration, but better screening of entrants. Many good reasons for such screening given in previous comments. This is the tall fence, wide gate approach to immigration. The problem with immigration is not the number of people, rather how they enter the country.

Mr. Econotarian: An interesting idea that would never happen- though still better than full open immigration. Still, full open immigration might get a better sample of the Mexican population than the present situation, where poor and less-than-bright Amerindians who don't even have Spanish as a first language are the ones with the most to gain by emmigrating.

Conceptually, you lack a distinction between immigration and citizenship. Citizens may vote, non-citizen immigrants may not.

This is not to disparage the content of your post. The results are good, as far as they go.

But if you mean to respond to people worried about political externalities, it makes sense to distinguish citizens from other immigrants. If the descendants of immigrants, who presumably will be born in the states and get citizenship by birth, are indistinguishable from other voters, then maybe they should be the focus.

Furthermore, the distinction has implications for federalism. If states have more control over immigration, but not citizenship, then it becomes much more likely that some states will experiment with looser borders.

Mr. Econotarian writes:

If one is concerned about the country's IQ, perhaps we should allow in all immigrants that pass an IQ test above the US average score.

In 21st Century America, there is simply no reason to have high levels of immigration, as we currently do. Immigration—from all parts of the world—should be greatly curtailed. And the few immigrants we do let in should be highly select.