Wednesday, April 2, 2008

The Undocumented Commit 21 Percent of Crimes?

by David Wilson
Co-Author, The Politics of Immigration: Questions and Answers

Until recently almost all studies have indicated that immigrants are far less likely to commit crimes than the general population. But a Feb. 27, 2008 article in Time magazine, "Immigration: No Correlation With Crime," notes that Prof. Eric Rasmusen, an economist at Indiana University, has come to a strikingly different conclusion. In postings to his blog in 2007 (accessed on March 17, 2008), he announced he had "run the numbers" and discovered that undocumented immigrants have a crime rate six times that of other US residents, causing about 21 percent of crimes; their criminal activities cost the US $84 billion a year, he says.

The reason for this startling difference from the other studies, Prof. Rasmusen tells us, is that the other studies "are about legal immigrants, not illegal immigrants. Legal immigrants are by definition unusually law-abiding, and include Indian doctors, Korean grocers, and Mexican grandmothers who are unlikely to be committing many murders and rapes."

So how does Prof. Rasmusen establish the crime rate for undocumented immigrants?

As he points out, data on out-of-status inmates "isn't readily available from the government." There are, however, some interesting suggestions in a 2005 report by the Government Accounting Office (GAO), "Information on Criminal Aliens Incarcerated in Federal and State Prisons and Local Jails" (GAO‑05‑337R). The report bases its numbers on the Department of Justice's State Criminal Alien Assistance Program (SCAAP), which partially reimburses state and local jurisdictions each year for the cost of incarcerating undocumented immigrants "who were convicted of a felony or two misdemeanors and incarcerated for a minimum of four days," according to the GAO.

Numbers based on SCAAP are questionable, as Prof. Rasmusen notes. Some jurisdictions don't apply for the reimbursements, and some probably underreport because of poor record-keeping. Jurisdictions may also overreport, and it's worth noting that overreporting results in higher payments from the federal government. Still, the SCAAP numbers may well be the best we have for now.

Starting from this GAO report, Prof. Rasmusen says there were about 49,000 undocumented inmates in federal prisons at the end of calendar 2004, and some 77,000 in state prisons and 138,000 in local jails in fiscal 2002 (the fiscal 2003 figures were incomplete). This would make a total of about 264,000 undocumented immigrants in federal, state, and local prisons in or near the year 2004. Prof. Rasmusen then takes the US prison population in 2005 as 1,496,000, and concludes that the undocumented inmates make up 21.4 percent of this prison population and therefore account for 21.4 percent of the estimated $400 billion that he says crime costs the US each year.

The problem with all this is that Prof. Rasmusen's numbers make no sense.

1. Nowhere does the GAO report say that there were 49,000 undocumented inmates in federal prisons at the end of calendar 2004. The 49,000 number is for "criminal aliens," which includes the entire immigrant federal prison population, both documented and undocumented, excluding immigrants held in immigration detention. (The 49,000 number is close to the Bureau of Prison's estimate of 26 percent to 27 percent for noncitizens in the federal prisons; but note that the Department of Justice's midyear report on the prison population for 2005 gives the total number of noncitizens held in federal prisons as 35,385, far less than the GAO's number. The reason for the discrepancy is not clear.)

2. Since the GAO report's figures for state and local prisons are based on reimbursements under SCAAP for the entire fiscal year, these numbers include prisoners that were incarcerated or released during the year along with the prisoners who remained in the system throughout the year. But figures for prison populations are usually based on the number of inmates in the system on a certain date--December 31, for instance--not on the number of inmates that passed through the system during the year. So these figures need to be adjusted before we make comparisons with the total prison population.

For example, the GAO report shows 77,000 undocumented inmates in state prisons in fiscal 2002, but the Justice Department's midyear 2005 report on prisoners gives a total of just 54,804 noncitizen inmates, documented and undocumented, in the state systems on June 30, 2002 (the total had grown to 57,393 by June 30, 2004). Clearly the number of undocumented prisoners must be lower than the total number of noncitizen prisoners. To use the SCAAP reimbursements to estimate a figure for the undocumented state inmates at a given time, we'd need to know the inmates' average stays in state prisons. (The reports on noncitizen inmates, which are only issued for the middle of the year, may undercount, since some large states, like Virginia, don't report. But they may also overcount, since other large states, like New York, include naturalized citizens in their figures, and other states include inmates in local jails.)

The GAO report is more informative for the local jails, where sentences are generally much shorter than in the state systems and inmate turnover is therefore much higher. The GAO says that in fiscal 2003 SCAAP reimbursed local systems for 147,000 prisoners, who "spent a total of about 8.5 million days in jail." This works out to a little less than two months for each prisoner, so probably about one sixth of these prisoners--24,500--were in jail on any given day.

So in fact the GAO report doesn't give us a way to estimate the number of undocumented prisoners except in the local jails. But for the sake of argument, let's take all the noncitizen prisoners in both the federal and state systems in 2004 as undocumented immigrants; if we then add in our estimated number of undocumented immigrants in local jails at any point during fiscal 2003, we come out with a total of about 131,000, less than half Prof. Rasmusen's number.

3. Prof. Rasmusen postulates a US prison population of 1,496,000 in 2005, and then unaccountably subtracts his figure for undocumented prisoners from the total to come up with a percentage of 21.4 percent (it would actually be about 17.6 percent if he'd left the undocumented prisoners in the total). But where does he get this 1,496,000? He cites a government URL, but the government has an annoying habit of moving its websites around, and that URL has disappeared. Wherever he got his number, it's off by more than 600,000. The actual prison population at the end of 2004 was 2,135,335, according to the Justice Department--177,600 federal, 1,243,745 state and 713,990 local.

If we follow Prof. Rasmusen's methodology, but with corrected numbers, we get an undocumented prison population of 131,000 out of a total prison population of 2,135,335; the undocumented would then make up about 6.1 percent of the prison population. While probably still way too high--since it overstates the number of undocumented in the federal and state systems--this percentage is still less than a third of Prof. Rasmusen's number.

By this calculation the undocumented would have a somewhat higher crime rate than the population as a whole, since undocumented immigrants are about four percent of the general population. But undocumented immigrants are mostly young adults and are disproportionately male; if we compare them to a native-born population with similar characteristics, the immigrants show a much lower crime rate--which is exactly what most studies have indicated. Prof. Rasmusen's numbers change exactly nothing.

[Please read Prof. Rasmusen's May 1, 2008 comment, posted below.]


Eric Rasmusen said...

I agree with the 6.1% figure in this post. See:

Dave said...

Bravo for correcting Rasmusen's errors although whether they were errors or intentional misstatements one can only ponder.