A recent paper published in the journal PLoS ONE claims that the number of illegal immigrants currently residing in the United States is at least 50 percent greater than previously thought and likely to be twice as high. Researchers Mohammad M. Fazel-Zarandi, Jonathan S. Feinstein, and Edward H. Kaplan write that:

Our conservative estimate is 16.7 million for 2016, nearly fifty percent higher than the most prominent current estimate of 11.3 million, which is based on survey data and thus different sources and methods. The mean estimate based on our simulation analysis is 22.1 million, essentially double the current widely accepted estimate.

That PLos ONE paper levels a serious charge as virtually all demographers and researchers in think-tanks on both sides of the immigration issue and the government think that the real number of illegal immigrants lies somewhere between 11 and 12 million. 

Understandably, much of the media has run with this headline finding but have neglected to cite the substantive and convincing criticism published in PLoS One in the same issue. There are three major criticisms of the paper by Fazel-Zarandi, Feinstein, and Kaplan. The first is that their model is highly sensitive to assumptions about return migration in the 1990s. Merely replacing the authors’ assumptions with those based on Mexican return-migrant survey data brings their estimates down to the commonly accepted level. The second is that it is very difficult for millions of additional people to hide in the United States without leaving a demographic or statistical trail. Their children should show up in birth and school records, their deaths should show up in death records, and more of them should be counted in the American Community Survey or U.S. Census. The third is that they should show up in economic surveys of employment, but they do not.

Researchers, pundits, policy-makers, and members of the media should not support the PLoS ONE findings based on the quality of the criticisms. Although my doubts line up well with those of the critics cited above, there are some interesting implications if (a very big nearly-impossible if) the results of the PLoS ONE paper turn out to accurately estimate a greater number of illegal immigrants. 

The first is that the illegal immigrant crime rate would be much lower than we currently estimate. We calculate crime rates by dividing the number of convictions or incarcerations by the entire subpopulation. We then multiply the result by 100,000 to estimate the crime rate per 100,000 people. This means that a greater population of illegal immigrants would have a lower crime rate, all else being equal. The authors of the PLoS ONE study admit this. Since we know how many crimes were committed, a higher illegal immigrant population only increases the denominator which lowers their crime rate. Within a 95 percent confidence interval, the study’s most conservative finding is that there are 16.2 million illegal immigrants, the mean finding is that there are 22.1 million, and the most extreme is that there are 29.5 million illegal immigrants. 

Distributing those additional illegal immigrants across the U.S. states in proportion to where they currently live substantially lowers the illegal immigrant conviction rate for homicide in Texas. In 2016, the illegal immigrant homicide conviction rate in Texas was 1.8 per 100,000 relative to 3.2 per 100,000 for natives and 0.9 per 100,000 for legal immigrants. If there were instead 16.2 million illegal immigrants nationwide distributed in proportion to their current population, the illegal immigrant homicide conviction rate would drop to 1.2 per 100,000. For the mean estimate of 22.1 million illegal immigrants nationwide, their homicide conviction rate would be 0.9 per 100,000, equal to that of legal immigrants. In the most extreme case of 29.5 million illegal immigrants, the illegal immigrant homicide conviction rate would be 0.7 per 100,000 – the lowest of all groups that we studied.

The second is that the illegal immigrant impact on the labor market would be much smaller than researchers currently estimate. This would mean that the negative elasticities reported by Borjas and others would be less negative, and the positive ones reported by Peri and others would be less positive. There would be even less reason to debate the impact of immigrants on the wages of native-born Americans as it would be even smaller.

The third is that the positive assimilation trends observed amongst today’s immigrants, even in relation to those in the past, are occurring when the foreign-born population has been far higher than previously estimated – which would silence a lot of worriers or at least force them to move the goalposts. 

There are undoubtedly other policy debates that would be affected by the revelation that the number of illegal immigrants is far greater than it is. The PLoS ONE paper is an interesting but fatally flawed piece of research so we do not have to consider every policy area that would be affected by such a titanic shift in our assumptions. Rather than supporting restrictionist calls for additional immigration enforcement measures, a larger than expected illegal immigration population would support many of the claims made by the proponents of immigration liberalization.