Evaluating the quality of research is laborious. Unless you re-do the whole paper yourself, how do you know the authors were not only truthful, but careful? Faced with this quandary, one of my favorite heuristics is to ask: Did the authors want to find this result? If the answer is No, I put a lot more credence into the results. In research as in the law, statements contrary to interest count more.
For example, when I learned that most economists find little effect of national education on national GDP, I was simultaneously surprised and convinced. Surprised, because I know that most economists want to find a big effect of national education on national GDP. Convinced, because if legions of smart people fail to find the result they want, the best explanation is that what they want to believe isn’t true.
Moving along… Many people debate the effect of immigration on redistribution. Libertarians (like me) want to find that immigration reduces redistribution, because we like immigration but dislike redistribution. As a result, non-libertarians sensibly discount libertarian claims that this is so. Sensible libertarians would do the same. People who agree with you are still flawed human beings, right?
So how can we approach the truth despite these impediments? Well, most of the researchers who study the effect of immigration on redistribution are, in fact, left-wing. They like immigration and redistribution, so they want to find that immigration increases redistribution.
In practice, however, they usually don’t. And it disturbs them! Here are two notable examples.
First, Alesina, Miano, and Stantcheva’s recent NBER working paper:
Support for redistribution is strongly correlated with the perceived composition of immigrants — their origin and economic contribution — rather than with the perceived share of immigrants per se. Given the very negative baseline views that respondents have of immigrants, simply making them think about immigration in a randomized manner makes them support less redistribution, including actual donations to charities. We also experimentally show respondents information about the true i) number, ii) origin, and iii) “hard work” of immigrants in their country. On its own, information on the “hard work” of immigrants generates more support for redistribution. However, if people are also prompted to think in detail about immigrants’ characteristics, then none of these favorable information treatments manages to counteract their negative priors that generate lower support for redistribution.
While Alesina et al. maintain a scholarly tone, their sympathies are pretty clear:
Anti-immigration parties have an incentive to maintain and even foster the extent of misinformation. Because information is endogenous, a vicious cycle of disinformation may arise. The more natives are misinformed, the more they become averse to immigrants and redistribution, and the more they may look for confirmation of their views in the media.
Second, Soroka, Banting, and Johnston’s chapter in Globalization and Egalitarian Redistribution (Princeton University Press and Russell Sage Foundation, 2006):
International migration does seem to matter for the size of the welfare state. Although no welfare state has actually shrunk in the face of the accelerating international movement of people, its rate of growth is smaller the more open a society is to immigration. To the extent that spending growth is inescapable, mandated by the aging of populations in industrial societies, specific parts of the welfare states—especially the parts that redistribute from rich to poor or from the old to the young—may truly have shrunk in the face of migratory pressures. Whatever the details, the typical industrial society might spend 16 or 17% more than it now does on social services had it kept its foreign-born percentage where it was in 1970.
Their sympathies, too, are also pretty clear:
What do the propositions imply? They do seem to vindicate Miller’s (1995) worries about threats to the national basis of the welfare state… The attitudinal problem is more among natives than newcomers and reflects more the apprehension of cultural threat than the fact of threat. But those apprehensions, combined with often realistic appraisals of cost and benefit, mean that the human component of globalization may represent a constraint on the expansion of welfare states that seemed fully consolidated two decades ago.
If you want to go meta, you could naturally object, “Libertarians want to believe that non-libertarian researchers support their desired conclusions. So why should I trust Bryan’s summary of the research?” Yes, it’s a problem, but it’s not insuperable. Try reading Alesina et al.’s literature review and see for yourself if I’m being unfair.