Richwine on the Net Fiscal Effect of Low-Skilled Immigrants

In Open Borders, I heavily rely on the National Academy of Sciences report on The Economic and Fiscal Consequences of Immigration to estimate the net fiscal effect of immigration.  Recently one of my graduate students pointed out this post by Jason Richwine criticizing my interpretation of the results.

Among dropouts, immigrants in the 25-64 and 65+ age categories are clearly fiscal burdens, as they cost taxpayers $225,000 and $257,000, respectively. Caplan, however, is tantalized by the age 0-24 column, which shows positive $35,000. “Even young high school dropouts more than pull their weight,” he concludes.

That conclusion is based on a misunderstanding of the table. Among immigrants in the age 25-64 and 65-plus columns, the education rows refer to the education of the immigrants themselves. However, in the age 0-24 column, education refers to the education of the immigrants’ parents. As p. 464 of the National Academies’ report explains, “If the immigrant arrives before age 25, we instead predict a future education level … based on parental education.” The reason the fiscal impact appears positive is that the model assumes that the children of high school dropouts will get more education than their parents did. In other words, most of Caplan’s “young high school dropouts” are not dropouts at all.

Richwine concludes in a gentlemanly manner:

This is an understandable mistake, as the National Academies authors should have been clearer that the age 0-24 column has a different interpretation than the other two age columns. Nevertheless, Caplan’s misinterpretation has led him far astray.

Did I indeed misread the report?  Yes.  Volume editor Francine Blau connected me with Gretchen Donehower, one of the authors of the section, and she confirmed my mistake.

Here’s the relevant NAS passage:

Because an individual’s tax payments and benefit receipts differ so much by the individual’s educational attainment, to predict future flows for an immigrant one must first predict the educational level that individual and his descendants will attain. An immigrant who arrives after age 25 is likely to maintain the education level observed on arrival, so we assume no change in educational attainment after age 25. If the immigrant arrives before age 25, we instead predict a future education level by estimating regression functions that predict offspring education based on parental education.

In hindsight, I was always a little puzzled by the NAS tables.  What does it mean, after all, to report the net fiscal impact of a 10-year-old college graduate?  I was also somewhat puzzled by how young immigrants could have such a favorable fiscal effect when taxpayers are immediately paying massive sums to educate so many of them.  But I deferred to the NAS numbers instead of double-checking the text.  I did read the whole chapter, but this qualification failed to register.  Mea culpa.

In light of Richwine’s correction, here is my revised position on the NAS report.

1. Young immigrants (ages 0-24) whose parents are high school dropouts have a positive net fiscal effect.

2. But the dropout parents themselves generally have a negative effect, even if they arrive as young adults.

3. Even 25-year-old immigrant high school dropouts have a negative net fiscal effect (-$186,000), though 25-year-old immigrant high school graduates have a positive net fiscal effect (+$72,000).

I continue to stand by several closely related controversial claims, most notably:

1. Immigrants have a much more favorable fiscal effects than matching natives.  The table showing that 25-year-old immigrant dropouts have a net fiscal effect of -$186,000 also shows that 25-year-old native dropouts have a net fiscal effect of -$388,000!

2. If you consider this an inadequate basis for restricting the reproduction of natives, it is hard to see why it is an adequate basis for restricting the migration of foreigners.

Last point: If there is a second-edition of Open Borders, I’ll definitely fix the mistake and thank Richwine for pointing out my error.

HT: Simon Laird

The post appeared first on Econlib.