The Great Immigration Debate — Now the Great Peri-Borjas Debate

Fun to watch a public spat between two immigration economists, Giovanni Peri and George Borjas, with others chiming in on one side or the other. As Giovanni pointedly told me last year, I am not an economist. But I am a statistician (and, for what little it’s worth, was an econ minor in college). “I know where the bodies are buried,” and I claim to be able to add some value to this debate. By the way, since I know both parties to this debate, it will be more comfortable to me to refer to them by first names.

First, some background: In 1980, Fidel Castro said in essence to the U.S. president Jimmy Carter defiantly, “You want Cubans? Well, I’m going to give your Cubans!” (China’s Deng Xiaoping said to Carter, “You want Chinese immigrants? Excellent. How many millions do you want?”, knowing that Carter did NOT want so many.) So Castro opened the floodgates, including at his jails and mental health facilties, motivated at least in part by a desire to cause political trouble for Carter. Thousands of Cubans fled the country from Mariel Harbor. It also later caused political trouble for then-Arkansas governor Bill Clinton, whose resulting  defeat in the next election Hillary Clinton is said to keep foremost in her mind today.

One of the main fears, of course, was that the Marielitos would depress wages. (Immigration from Cuba has enriched the already-colorful Spanish language, another example being balseros, Cubans fleeing on balsa rafts.) Economist David Card analyzed the data in Miami, in part based on other cities he considered to be comparable to Miami. He found that there really was no appreciable effect on wages, a finding that made him one of the most famous names in immigration economics, and which played a role in his being selected for the Clark Medal, given to the best economist under age 40. By the way, the linked tribute here is by Richard Freeman, ironically George Borjas’ fellow immigration skeptic.

Giovanni Peri, my UC Davis colleague, is a relative newcomer to the field of immigration economics, and is quite stridently on the pro- side. He has become the go-to man for politicians seeking an academic imprimatur for a stance favoring expansive immigration policies. A notable example is that the Obama White House report on the projected impact of their administrative action on immigration cited Giovanni 38 times, basically citing no one else. It did mention George once — but negatively. George, on the other hand, has been cited often by the “restrictionist” side.

George did a critical analysis of the Card paper last year; Giovanni responded; George rebutted; and yesterday it was Giovanni’s turn (though the latter was not a rebuttal, in that he did not address or even mention George’s latest analysis).

Unlike some economic debates, this one can actually be followed fairly easily by noneconomists, because it basically boils down to a “he said, she said” argument as to exactly which set of workers should be studied. As I’ve said before, I’ve never liked the regional comparison method of analysis, because there are too many uncontrolled variables, leading to wild inconsistencies from one study to another; I have cited state-by-state studies on capital punishment, for instance. Yet some things are highly factual and stick out like a sore thumb.

Notable among these is George’s discovery of a major flaw in Giovanni’s analysis: In assessing the impact of the Marielitos on native high school dropouts in Miami, Giovanni treated as “dropouts” those who actually are still in school, i.e. are NOT dropouts. You don’t have to be an economist to see that this is major trouble for Giovanni’s findings.

Giovanni is a serious player in this arena. He uses sophisticated methods, and I found the approach he took in one of his papers to be especially clever (though also flawed). But my criticism of his work has often been that there are decisions he makes in his analysis process that potentially matter a lot but which he does not reveal to the reader in his subsequent writeups. I presume that he consciously chose to include the nondropouts as dropouts as described above; well, fine, but it is required academic protocol that one explain such a thing. In another one of his analyses, a regional approach to the question of whether the H-1B work visa program harms or helps native workers, he omitted from his data the most prominent H-1B using region of all — Silicon Valley! — again without disclosing that omission to the reader. The bibliographies in his papers almost never cite any work that has findings that are unfavorable to the H-1B program, when in fact one is supposed to cite all relevant prior work. These are major academic transgressions.

Indeed, Giovanni fails to cite his own relevant work. One particularly germane example is his 2012 paper, “Immigration, Labor Markets, and Productivity,” in which he finds (indeed takes it as universally agreed upon) that immigrant labor is cheap, and in fact lauds it as a boon to employers:

One common empirical finding in the literature is that immigrants are paid less than natives with similar characteristics and skills. This is in part due to the fact that many immigrants, because of less attractive outside options (such as having to go back to their home country), have lower bargaining power with the firm. In this case firms pay immigrants less than their marginal productivity, increasing the firms’ profits.

Presumably the Marielitos especially didn’t want to go back home. Low bargaining power, indeed!

Again, you don’t have to be a rocket economist to see how contradictory this is to Giovanni’s recent research finding that the Marielitos did not adversely impact native wages.

I take issue also with Giovanni’s favorite argument, that immigrants don’t reduce native wages, because immigrants and natives play complementary roles in labor markets; in plain English, the argument is that immigrants and natives do different kinds of jobs, so “no harm, no foul.” But of course, the truth frequently is that the immigrants do different kinds of work because the natives are driven out. As the complementarity claim is an especially common argument with pro-expansive immigration policy researchers, some examples are in order.

A well-known example at the low-skilled level involves the Los Angeles hotel industry, in which employers fired the unionized, largely-black hotel janitors and replaced them with immigrant Latinos. Suddenly, we were told that “Americans don’t want to do janitorial work.” A more recent, but equally egregious instance involved cleanup after Hurricane Katrina, where the African-American workers were told they were being let go, “because the Mexicans are here.”

At the high-skilled end, there is the 1989 NSF report that I often mention, which (a) advocated bringing in lots of foreign graduate students in STEM to hold down wages and (b) correctly forecast that the resulting stagnant wages would drive domestic students away from graduate study. I’ve taken even economists whom I highly admire, Bill Kerr and Jennifer Hunt, to task on this point, as their analyses fail to account for this gradual trend in the makeup of graduate programs. At any rate, you can see from this that the claim made by the industry PR people that foreign and domestic STEM workers play complementary roles (with respect to degree level) is highly misleading.

Another argument commonly made regarding the wage impact issue is “Immigrants are consumers too, thus generating more demand for goods, thus raising native wages and job opportunities.” This too is misleading, in a couple of senses.

First, consider the following “thought experiment.” Suppose immigrants were to live off the land, growing their own food etc. and engage in no consumer activity at all — but they did work for pay. To put things into even sharper focus, say they send all their earnings back home as remittances. Then once again, you don’t need to be an economist to see that they would depress wages. Of course, this is a very artificial example, but it has a real-life implication: If immigrants generate demand below a certain level, there will indeed be a decline in wages. And low-skilled workers, being poor, don’t generate much demand. It’s unclear just where the threshold is for generating “enough” demand, but it is very clear that claim “Immigrants are consumers, so they raise wages” is a gross oversimplification.

Second, the argument doesn’t account for labor market segmentation, and inelasticities in demand for labor at the high end. The number of software engineers needed by a company, say Walmart, wouldn’t increase much, if at all, if Walmart’s retail sales were to double.

Finally, a note on sample size: Some critics of George’s analysis dismiss it on the grounds that his samples are too small. This ignores the fact that his findings are indeed statistically significant. I am not a fan of significance tests, but let me point out that even very small samples can yield statistical significance. Indeed, getting a statistically significant result with  small sample is an indication of a large effect.

A little more openness to the reader, and to research done by those with contrary opinions, would make Giovanni’s work much more useful, in my humble opinion as a, ahem, noneconomist.


11 thoughts on “The Great Immigration Debate — Now the Great Peri-Borjas Debate

  1. Economists’ main claim to authority is a supposed facility with statistical analysis, so it’s bit rich if Peri dismissed your right to contribute. Your expertise exceeds that of most PhD trained economists.

    Norm, that’s a very good analysis of the culture of this debate.


    • Only some economists are econometricians, but most economists do use at least some statistics. There are some who even deride the use of graphs, referring to those who do as”chartists”.


      • No, the people economists call “chartists” are not themselves economists but rather financial industry types who employ graphs to engage in a form of entrail-reading aimed at predicting stock prices. Economists’ low opinion of these “chartists” (or “technical analysts”) has nothing to do with the use of graphs per se and everything to do with their misuse.


  2. How do the economist finding no significant impact of immigration on wages account for the existence of the minimum wage? If, for example, all additional workers were added at the legal minimum wage that would mask the impact of those additional workers on wages while the major impact would be displacing native workers. The average wage would change only if more or less workers got the minimum wage.


  3. > In another one of his analyses, a regional approach to the question of whether the H-1B work visa program harms or helps native workers, he omitted from his data the most prominent H-1B using region of all — Silicon Valley! — again without disclosing that omission to the reader.

    I’m not sure if this the case you’re referring to but Peri did omit San Jose from data provided to the Wall Street Journal for an article. I posted a description of the article at . Following is the first of three problems with the study that I listed in the conclusions:

    1) San Jose, CA and Stamford, CT are missing with no mention being made to the article’s reader. The omission of San Jose is the most disturbing since it is in the heart of Silicon Valley and has the largest share of foreign STEM workers.

    I initially emailed one of the article’s authors and asked for the data and he said that Peri had provided the data for the article’s chart. I then emailed Peri and he was kind enough to provide the spreadsheet. I found that San Jose and Stamford, CT, both of which were among the 219 metropolitan areas with data from 1990 to 2010, were missing from the spreadsheet. I asked Peri about this and he said that San Jose is an outlier and is a story of information technology, not foreign STEM.

    In fact, the second plot at shows that San Jose and Stamford both counter the trend of the other data. San Jose had average wage growth but a very high percentage change in foreign STEM workers while Stamford had a very high increase in wages but a very low increase in foreign STEM workers from 1990 to 2010. In addition, the last plot shows that San Jose had a DECREASE in the percentage of native STEM workers during this period. Hence, the omission of both cities look suspiciously like cherry-picking.

    In any event, I can see no defense for not informing the readers of this seemingly important information. This seems very similar to Peri and Yasenov’s commentary at where they critiqued Borjas’s reanalysis of Card’s study without mentioning that Borjas has posted a response to that critique a week earlier. If they knew of the response, then the commentary seems extremely dishonest. In any event, both they and the WSJ editors should have googled and/or checked with Borjas for a response. Google is your friend (unless you’re looking to hide information).


    • Interesting that Peri told you that San Jose was an “outlier.” I’ll remember that one.

      In the WSJ’s defense, the article was probably edited several days earlier, maybe even more than a week. But Peri should have given them an update.


      • Good point. I was thinking of a news story that would likely go out within a day of being written and edited. A commentary, on the other hand, might be written and edited up to a week or more before it went out. Still, I agree that Peri should have given them an update. Even a concluding sentence or two giving the location of the Borjas response and how it doesn’t affect his basic points and/or how he would address it later would have put the commentary in a whole other (and more accurate) light. As in the case of leaving San Jose out of the data, he is withholding important information from the reader.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s