Economist Bryan Caplan apologizes for falling into one of the very economic biases he says the general public falls for in his fascinating new book, The Myth of the Rational Voter: Why Democracies Choose Bad Policies: The anti-foreign bias.
He had assumed, as do many Americans in this great land of immigrants, that immigrants represented a specially severe crime risk. A July 2007 working paper from the National Bureau of Economic Research set him straight. Here's how he explains it on his blog:
[D]espite their demographics, immigrants are drastically less criminal than native-born Americans. In fact, immigrants have one-fifth the incarceration rate of natives. Yes, natives are incarcerated at five times the rate of the foreign-born:
Using the 1980, 1990, and 2000 Censuses, we show that 18-40 year-old male immigrants have lower institutionalization rates than the native born in each year. The gap in these institutionalization rates widens over the decades, and by 2000 immigrants have institutionalization rates that are one-fifth of the native born.
Is this base statistical trickery? Not likely; these are raw numbers that anyone can double-check against the census. Could these results simply reflect the practice of deporting criminal aliens? Nope; our actual practice is to make immigrants serve their full sentence before expelling them. But how can we reconcile the facts with the demographics? [Study authors Kristin F.] Butcher and [Anne Morrison] Piehl show that given their demographics, we should expect immigrants to commit crimes at double the native rate. But for some reason(s), demographics yield a massive overprediction; immigrants commit crimes at one-tenth the expected rate given their demographics. Yes, if immigrants acted like otherwise similar natives, they were be ten times as criminal as they actually are.
Full text of an earlier version of the paper.