Pew recommendation on recidivism metric wrong for Texas

There was lots of MSM coverage this week of a new report (pdf) from the Pew Center on the States ranking states by recidivism rates, with Texas earning perhaps undeserved praise for our remarkably low rate. Doug Berman covered the report and rounded up these links:

• From CBS News here, "Breaking the cycle of repeat crime offenders"
• From CNN here, "Report promotes alternatives to prison as national recidivism rate holds steady"
• From the Dayton Daily News here, "Ohio's decline in prison recidivism among steepest in US, study says."
• From the Kansas City Star here, "Study praises efforts in Missouri, Kansas to cut prison recidivism"
• From the Statesman Journal here, "Study hails Oregon's recidivism reduction even as funding cuts threaten it"

At the Austin Statesman, Mike Ward's coverage was typical of the spin put on Texas' relatively low numbers:

The study found that 31.9 percent of Texas convicts released from prison in 2004 had returned within three years. Of those released in 2007, the year the reforms began, the recidivism rate had dropped to 24.3 percent.

Other 41 states posted an average recidivism rate of 43.3 percent in 2004, and 45 percent in 1999, according to the study. While other recent studies hint that rate has fallen slightly, Texas is still among the lowest.

Texas once had a recidivism rate of more than 55 percent.

House Corrections Committee Chairman Jerry Madden, a Richardson Republican who was an architect of the reforms, said the study confirms what Texas numbers have shown is a downward trend.

“What we did four years ago is working, and this study confirms that once again,” Madden said.

I agree with Ward's assessment that "proposals under discussions by lawmakers this spring to drastically scale back recidivism, drug-treatment and training programs would likely backfire and end up costing taxpayers billions more to build new prisons." That's certainly the lesson from states described in the report like Michigan, where reentry investments paid off by allowing the state to "shrink its inmate population by 12 percent, [and] close more than 20 correctional facilities." There's a real risk that, in a short-sighted effort to cut the budget, the Texas Legislature will eliminate treatment and reentry programming when expanding it and closing prisons would put us on a more sensible path.

But Texas' low recidivism rate predated recent probation reforms, did not result from them, and in fact is more indicative of a problem than good public policy. Simply put, the recidivism rate as calculated in the report is the percentage of those released from prison who return within three years. So the first factor that affects that statistical calculation - before we talk about any programming - is the makeup of the prisoner cohort released, which in Texas now runs more than 71,000 people per year. The more low-risk people the state incarcerates, e.g., for drug possession and increasingly low-level property offenses, the lower recidivism rates will be. So states like Texas and Oklahoma have low recidivism rates not because we're doing a great job but because we lock up too many people in the first place.

For that reason, despite the laudable wisdom of their other suggestions, it would be a mistake for Texas to adopt Pew's first recommendation: To "Define Success as Recidivism Reduction and Measure and Reward Progress." Making recidivism reduction a performance measure for budgeting would result in bad public policy, creating a perverse incentive to overincarerate low-risk offenders - exactly the most expensive outcome that nobody wants. The Pew report itself acknowledges as much in a section titled "Unpacking the Numbers."

States that send comparatively low-risk offenders to prison are likely to see lower rearrest and violation rates compared with states that concentrate prison space on more dangerous offenders. If, for example, a state incarcerates a large proportion of lower-risk offenders, then its recidivism rate might be comparatively low, because such offenders would be, by definition, less of a risk to return to prison. A state with a larger percentage of serious offenders behind bars, on the other hand, might experience higher rates of reincarceration when those offenders return to the community.

Oklahoma exemplifies the former example: “A lot of people who might be put on probation or diverted into an alternative program in another state wind up going to prison in Oklahoma,” notes Michael Connelly, administrator of evaluation and analysis in the Oklahoma Department of Corrections. “These lower level folks aren’t as likely to recidivate, so it benefits our overall numbers and makes us look like we’re doing an even better job than we’re doing.” Oklahoma’s overall recidivism rate for offenders released in 2004 was 26.4 percent, the third lowest in the country, the Pew/ASCA survey found.

Grits has described this phenomenon in the past, including a discussion with Michael Connelly, who's quoted in the Pew report. States like Texas and Oklahoma face a completely different dynamic than a place like Utah, where high recidivism rates result mostly from technical violations. Texas has about a 32% recidivism rate, with 5% of those reincarcerated sent back to prison on technical parole violations, according to a chart from Pew labeled "Exhibit 2." In Utah, by contrast, a majority of returning prisoners making up their 54% recidivism rate are technical violators, so they have more opportunities to reduce non-criminal recidivism than we do. Once a parolee commits a new crime, they've made the critical decision; for technical violators, it's the state's decisionmaking that's boosting incarceration. So in Utah, I could see making recidivism reduction, particularly among technical violators, a straight-up "performance measure." But in Texas, Oklahoma, and probably other states the metric runs counter to common sense.

If and when legislators decide to fundamentally tackle their mass-incarceration addiction, whether by choice or if forced to by the vicissitudes of history and budgeting, Texas will have little option but to lock up fewer low-risk offenders, shorten probation lengths, and focus scarce criminal justice resources on a smaller, higher-risk group of people who actually threaten public safety. When that happens, recidivism may well increase somewhat even as crime rates continue to drop, as has happened in other states. That would mark a public policy success, even if Pew's proposed benchmark runs counter to the result.