The faulty empirical basis for the DOJ's withdrawal from private prisons

I have a new blog post up on NRO about The DOJ's Misguided Withdrawal from Private Prisons. Here's an excerpt:

The Department of Justice's recent decision that the federal Bureau of Prisons should wind down its private-prison contracting was apparently based on private prisons' bad record of safety and security violations relative to their public counterparts. It turns out, though, that the DOJ's understanding of private prisons' record is informed by a serious over-reading of faulty comparative studies, in particular a recent study by the Office of the Inspector General.

As a result of this over-reading, the federal government—and any states that follow the federal government's lead—may not be able to take advantage of the power of contracting to provide incentives for good behavior.

. . .

On balance . . . one might think that the IG report is fairly harmless. Yes, it compares prisons that probably aren't comparable, and some of what it measures has no obvious relationship to actual quality of confinement. No, we shouldn't take it terribly seriously as a comparative study, since there already are better studies out there. But overall it does an okay job of not overstating its results; its basic recommendation is a call for further study.

Even a harmless study can lead to harmful changes, however. The Justice Department, not sharing the IG report's caution, went ahead and read too much into its results. Where the IG's office hedged its conclusions with caveats about comparability and merely called for greater investigation, the DOJ memo made broad claims . . . .

Based on this slim reed, Deputy AG Sally Yates ordered the BOP to "either decline to renew" contracts as they expired "or substantially reduce [their] scope." It's not clear whether the DOJ was relying on studies other than the IG report, but if it was, I'd like to see those studies: All available evidence does little to suggest that private prisons are worse than public prisons overall, or that they cost more.

. . .

The far greater problem is what might have been. Prisons have begun experimenting with performance measures and performance-based contracting—which, surprisingly, is almost unprecedented. These new metrics could offer substantial possibilities for improvement in prison conditions in the future. Monetary incentives could work in the public sector—think of performance-based bonuses for public-prison wardens—but the private sector is probably best positioned to take advantage of them: If there's one thing the private sector is good at, it's trying to make money in any way it can. The DOJ could have directed the BOP to take a "mend it, don't end it" approach by truly making an effort to encourage higher quality through better contracting. The BOP could have taken the lead in developing state-of-the-art prison-contracting practices and helping to spread good performance measures nationwide. But that's unlikely to happen now.

Read the whole thing.