The Shallowness of Garett Jones’ Immigration Research

Garett Jones’ recent speech at the Intellectual Freedom of Navigation Conference highlights his strengths as a writer and his weaknesses as a researcher. Garett contributed no original empirical work to analyzing how immigrants affect the countries where they settle. My 12,000-word two-part review of Garett’s The Culture Transplant contains more original empirical work than his book. He has offered no original theoretical insight into how immigrants affect the societies where they settle. He is either unfamiliar with the vast peer-reviewed literature on how immigrants affect their new societies, misunderstands much of it, or attempts to cover his mistakes through repetitive criticisms of others.

There is a robust empirical and theoretical literature that examines how immigrants affect institutions and other factors that contribute to economic growth in the countries where they settle. Garett did not contribute to that literature. He largely ignored it. His avoidance of this literature may reflect that it contradicts his thesis, but some omissions appear to be simple oversight. A paper published well before his book in a prominent economics journal, where Jones has also published, examined whether immigrants to Argentina had damaged that country’s economic institutions. This matters because Garett makes much hay out of the idea that immigrants ruined Argentina’s institutions, but somehow ignored the only peer-reviewed empirical paper on the topic. Coincidentally, I’m sure, the paper’s authors also find that immigrants aren’t responsible for Argentina’s institutional decline. The working paper was out long before Jones sent his book to the publisher. There’s no excuse for that.

His misunderstanding of academic research persists in his treatment of the so-called trust literature. Garett and I have disagreed about immigration, and much of our debate concerns the so-called trust literature, a dubious empirical project by some economists that purports to have found that general social trust, as measured by a specific survey question from the World Values Survey (WVS), GSS, and elsewhere, is positively causally related to income. My peer-reviewed paper in Kyklos (with Andrew C. Forrester) identifies five serious empirical and theoretical problems that undermine the findings of the trust literature. Briefly, they are 1) There is a concerning paucity of instances of trust being incorporated in formal growth models, 2) Responses to the trust question itself are internally invalid, 3) Responses to the trust question do not generally predict trusting behavior in real-world micro-level experiments or in trust games, with some exceptions, 4) Many of the major papers in the trust literature are contaminated by various types of sample bias, and 5) Omitted variable bias is a serious concern. Andrew and I attempted to resolve these problems, ran a simple regression using the best commonly employed methods from the trust literature, and found no statistically significant relationship between economic output and generalized social trust at the regional level in the United States during the 1972-2018 period.

Garett claims that our Kyklos paper systemically misquotes, misunderstands, and miscites much of the peer-reviewed “academic literature on interpersonal trust [emphasis added].” Garett is confused because our paper is about generalized social trust, not interpersonal trust. The relevant economics literature concerns generalized social trust, which is the measure economists use to study the link between trust and output. Interpersonal trust refers to the trust you have in people personally known to you. Generalized social trust is a broader measure from the WVS, GSS, and other surveys mentioned below that asks about people in general with the question: “Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people?” This might just be an innocent error by Garett. Regardless, I took a screenshot of Garett’s written comments because I don’t trust him not to stealth edit his error.

Garett claims that Andrew and I mischaracterized the findings of the economics literature on generalized social trust. It would certainly be news to the editors of Kyklos, the peer reviewers who thoroughly vetted our paper, and other economists who work on cultural economics and agree. Garett should take his concerns about our supposed scholarly incompetence or dishonesty to the editors of Kyklos and ask for a retraction, but that would require him to be specific for a change. It would also require him to know the difference between interpersonal trust and generalized social trust.

Are there other reasons why Garett doesn’t raise complaints with the editors of Kyklos and instead just relies on unspecific vibes-based critiques that signal his unwillingness to accurately criticize his opponents? I’ll let the readers here develop their own theories to explain the gap between Garett’s griping and his actions. Although there is no peer-reviewed academic literature on this topic, I will refer to that as the Garett Gripe Gap (GGG).

Elsewhere, Garett tweeted that I misquoted or misunderstood many papers about generalized social trust. However, you can’t check his claims because he deleted his tweets after I responded to them in a twitter thread. With screenshots of the disputed papers, I demonstrated that Garett misread every one. This is a pedantic complaint, but it can explain the persistence of the GGG. My thread includes citations and quotes for anyone interested, but Garett deleted his tweets, so you can’t verify his specific accusations against my corrections. All you have are the latter. Best as I can tell from my existing responses, the following explains some of the disagreement.

Garett claimed that “Measuring People’s Trust,” by Ermisch et al. (2009), found that responses to the trust question predicted trust outcomes in experiments, contrary to my claim that this paper found that responses to the survey didn’t predict experimental outcomes. Here’s what the authors wrote: “The study also asks the typical survey question that aims to measure trust, showing that it does not predict ‘trust’ as measured in the experiment” (749). Point to me.

Additionally, Garett relies on “Trust and Growth,” by Algan and Cahuc (2013), but they rely on “Using Experimental Economics to Measure Social Capital and Predict Financial Decisions,” by Karlan (2005) that finds, “In other words, the GSS [trust] questions predict default, or trustworthy actions, but fail to predict, savings, or trusting actions” (1697-1698). Another point for me.

Garett similarly takes issue with my related description of what “Trust, Voluntary Cooperation, and Socio-economic Background: Survey and Experimental Evidence,” by Gächter et al. (2004), finds. Here’s from their paper: “Trust as measured by the popular GSS trust question is not significantly correlated with cooperative behavior [in the experiments]” (523). Three points for me. The list goes on, and you can read the Twitter thread for yourself, but they all end in the same way.

However, I did make at least one error that is worth mentioning: not taking screenshots of Garett’s tweets from years ago because I didn’t think that he would delete them after being shown he was wrong. I’m skeptical of how trust, as currently measured, affects economic output, but my debate with Garett did convince me that trust matters greatly in scholarly debate. I do not trust Garett or his work, and neither should you. “Put up or shut up,” is crude but apt. Garett has had ample time to provide specific criticisms of my scholarly work. Since he hasn’t, it is well past time for him to shut up.