How does immigration affect crime rates? The Center for Immigration Studies (CIS) has just released a thoughtful report on that question. Rather than draw any firm new conclusions, authors Steven Camarota and Jessica Vaughan question the data used up to this point that give the “conventional” answer to the question.
The conventional answer is that immigrants themselves have remarkably low crime rates compared to natives, but their children commit crimes at considerably higher rates than natives. Therefore, if we are asking whether today’s immigrants are more crime-prone than natives, the answer is “no.” But if the question is “has immigration increased the national crime rate?” then the answer is a firm “yes.”
I made this chart using Census data to illustrate the generational differences in crime and other “underclass” behaviors.
The second set of bars shows that Hispanic immigrants are institutionalized (usually that means in prison) at only about half the rate of white natives, but Hispanic natives are in institutions at about three times the white rate. As I said, this is the “conventional” view of immigration and crime supported by most academics.
Camarota and Vaughan question the validity of the Census data used to support charts like the one above. They believe we are underestimating immigrant crime and overestimating second generation crime. They make the following points:
1. The Census Bureau had to “guess” the citizenship status of over half the prisoners it interviewed for the 2000 long-form, based on other personal characteristics.
2. In other cases, administrative data from the prison were used rather than inmate interviews.
3. Inmates have a strong incentive to say they were born in the United States, because non-citizens convicted of crimes may face deportation.
4. According to the census data, the number of institutionalized immigrants has fluctuated wildly between 1990 and 2007. This is unlikely to reflect reality.
These are very important points that all researchers in this area should heed. Camarota and Vaughan are quite right that Census data on institutionalization should come with some big caveats. However, I still find the conventional view about immigration and crime to be the most convincing.
The kind of data imprecision described in points one and two would probably lead to random error, not systematic error. We can think of a worst-case scenario where the Bureau’s guesses about citizenship are so bad that they are little better than coin flips. If that were the case, then we would expect to see about an equal number of Hispanics labeled immigrant and non-immigrant. Why do the imprecise data always tend to show far fewer Hispanic immigrants than Hispanic natives? The Bureau could be systematically erring on the side of labeling someone a native, but the authors cite no evidence for that. If the error is indeed random, then the Bureau could actually be underestimating generational differences in crime.
Point four is, like one and two, quite troublesome, but I have the same reaction as above. The data are clearly unstable in the aggregate, but why do they continually show a generational difference no matter when we look? Perhaps because they still reveal an underlying truth despite their inadequacies?
Points one and two tend to undermine point three. If the Census Bureau is guessing or using other kinds of administrative data for most of the citizenship status questions, dishonesty from inmates is less of a concern.
Another reason I tend to believe the conventional view is that increased crime in the second generation is consistent with an increase in several other underclass behaviors. As the chart above indicates, labor force dropout, illegitimacy, and welfare usage are all much higher among Hispanic natives than among Hispanic immigrants. (Those data come from reliable interviews of normal people outside of prison.) It makes sense that crime would increase if all of those other underclass problems are increasing as well.
CIS has yet to change my mind on this issue, but their report is important and thought-provoking.