The Arizona Office of the Auditor General recently released a provocative, but flawed, performance audit of the Arizona Department of Corrections (ADC). This audit can be described simply as a mixed bag: while it does provide compelling qualitative recommendations it also relies on faulty quantitative analysis.
The quantitative weakness of the audit stems from a report issued earlier this year by the ADC. The audit relies heavily on the ADC report, especially by including calculations of comparative per capita prisoner costs. However, the auditors admittedly "did not assess the (ADC's) method for adjusting the per capita rates for comparing state-operated and private prison costs." The initial ADC report's per capita prisoner costs inaccurately suggest that Arizona's private sector correctional facilities are more expensive than public sector correctional facilities.
Public-private cost comparisons are inherently tricky. To start with, government cost accounting systems and methods are generally so opaque and convoluted that public managers cannot accurately identify the "all-in" costs of providing a service. Some costs are paid directly out of an agency's budget, while many others-such as debt service, risk management, legal, payroll, IT and other administrative functions-are paid out of some other agency's budget. If those "extra" costs aren't captured along with the direct agency spending, any cost allocation at the service level will be understated. Even within an agency budget, many costs may be borne by a central office that should actually be allocated to specific service units, facilities, etc. in a proper accounting scheme. Throw in the tendency of government analysts to look for creative ways to make their public sector colleagues look more competitive than they really are, and there's ample reason to at least question public-private cost comparison studies undertaken by government.
The state maintains the highest-security prisoners and a different population mix than the private sector, so it is difficult to make direct cost comparisons. After attempting to account for some of these differences, the ADC report concluded that it costs $3 to $8 more per day to house a medium-security inmate in a private prison, a datapoint repeated widely in the media.
Having read through the whole ADC report several times, there are few methodological issues that the ADC failed to address:
- The two summary tables (pages 3 and 5 of the report) themselves reveal the biggest flaw-the ADC report ignores correctional healthcare costs, which is a major component (and driver) of corrections spending. The report starts with an approximation of the full direct and indirect costs for both minimum and medium custody beds in public and private prisons. It then proceeds to add/subtract a series of "adjustments"-most notably, subtracting out healthcare costs-that effectively serve to wash a range of very real costs off the table as if they didn't really exist. If we're trying to make accurate judgments regarding the true costs of public and private prisons, you cannot just wish away costs that are inconvenient or tricky to normalize. This is not new-a Reason colleague pointed out a similar problem with the 2007 version of this ADC report and made a comment that holds equally today: "This type of selective accounting can only be described as 'Enron-like'-where the accounting rules are determined by the desired outcomes."
- The summary table on page 3 (minimum custody cost scenarios) shows the dramatic effect of the "adjustments." At the unadjusted level (the top line), daily per inmate costs at minimum-custody public prisons are $58.77 per inmate, compared to $52.64 in private minimum-custody facilities. Then, a murky "administrative adjustment" that's left unexplained adds $0.04 to public prison costs (now $58.80) and $2.13 to private prison costs (now $54.78)-this alone should have set off some red flags. Then comes the healthcare whammy, which drops the public sector costs by a precipitous $11.93 (now $46.87) and dropping private sector costs by $7.64 (now $47.14). Add a few more adjustments and you end up with average daily per-inmate costs of $46.81 in public sector minimum custody prisons and $47.14 in private facilities. The adjustments effectively flip the results in favor of the public sector by way of creative accounting.
- The second summary table on page 5 (medium custody cost scenarios) suffers from the same creative accounting discussed above, but it has an even deeper problem-it attempts to compare ALL of the state-run medium custody beds to ONE private prison contract. Nowhere in the table does it actually tell the reader that the 1,181 medium custody private prison beds it cites are located at just one facility (the Central Arizona Correctional Facility), nor does it explain that this particular facility houses a population of sex offenders with higher treatment requirements. Further, buried in the tables on pages 23 and 24 are data broken down by prison unit, and it becomes obvious that the state-run medium custody units have a wide range of costs. How is it fair to make a general blanket statement about cost efficiency-this table is the source of the "$3-8 more expensive per day" statement in the Republic, after all-based on a comparison of all state-run medium custody prisons (with highly variable individual costs) to ONE private prison? The short answer is that it's not a fair or meaningful comparison.
- The medium custody cost comparison is especially dubious when you consider that ADC had two other contracts for out-of-state, medium custody private prison beds in 2009-covering nearly 4,000 additional inmates-that were not included in the cost calculations! The ADC report justifies this selective omission of over three-quarters of its medium-custody, privately-held inmate population from its calculations by noting that they could "not identify the medical component of the per diem rate" for the out-of-state contracts. This is puzzling because ADC also did not attempt to calculate the medical component of their in-state public and private facilities at a facility level either, yet they nonetheless made massive adjustments to their cost data based on an average, across-the-board healthcare cost estimate-one for public facilities ($11.93 per inmate per day) and one for private facilities ($7.64 per day). This is despite the fact that different categories of inmates-even inmates in the same facility-incur vastly different levels of healthcare spending, which the ADC report smoothed out beyond recognition or meaning by using a system wide average. Even if you assume for a moment that ADC's methodology for healthcare cost adjustments is valid, then you'd still be left wondering why that same methodology should not apply equally to the out-of-state, medium custody contracts. Then there would have at least been more than one private prison facility to evaluate as a category benchmark.
- It's not clear that ADC captured other important costs to the state. Do ADC's in-house estimates cover the costs of employee pension obligations? Do they capture legal, self-insurance and other risk management costs? Do they cover debt service on state facilities financed with public debt? Since these costs are usually borne by other state agencies-and thus paid for out of someone else's budget-a true public-private cost comparison must include them if you're really trying to quantify the "all-in" costs of running a state prison. Private prisons, by contrast, capture the full range of their operational and administrative costs in their overall per diem rate, including labor, benefits, retirement, debt service, legal, risk and many other costs. Fundamentally, there's no basis for an apples-to-apples comparison in the first place if one side gets to selectively ignore some important costs.
Further, when one makes it past the summary tables in the report, they'll find some interesting data that contradict the "public sector is cheaper" storyline. Even if one takes the report's data at face value, the detailed table on page 21 reveals that private prisons had lower per diem costs than public sector prisons in the aggregate, both on an adjusted and unadjusted basis (see Table 1 below).
Table 1. Arizona Prison Cost Comparisons, 2009
Source: Arizona Dept. of Corrections, FY 2009 Operating Per Capita Cost Report, Cost Identification and Comparison of State and Private Contract Beds, February 2010, pp. 21-28.
However, since aggregate figures fail to account for differences in costs by custody level, let's assume for a moment that all state-run prisons operated at an unadjusted per diem cost of $58.80-this is the unadjusted per diem for state-run, minimum custody facilities, which is the lowest rate among the four custody categories the state lists in the report (see Table 1). Even with such a generous assumption, ADC's in-house costs would still be higher than the unadjusted per diem cost for ALL of ADC's contracted prison beds in 2009 ($58.78), which includes both minimum- and (higher cost) medium-custody inmates.
Put differently, even the lowest-cost category of public prisons is still more expensive than the average private prison, according to the state's own reported numbers. It's only when one applies creative accounting and ignores healthcare costs, out-of-state contracts and other relevancies that the lens gets distorted.
The qualitative recommendations included in the end of the report speak volumes. Mostly notably, the audit did not come out in favor of public or private prisons. If the auditors found the ADC's per capita costs to be accurate, then they would likely have recommended the ADC expand use of public facilities. Instead, the audit recommends that the ADC conduct additional analyses that accurately compare costs. This makes sense, and in such situations some form of value-for-money analysis is useful to directly show what the private sector can provide relative to the public sector. Reason has written about this extensively here, here, here, here and here.
Looking beyond the subject of privatization, the State Auditor's report also included a policy recommendation to reform criminal sentencing. They astutely noted that despite declining crime rates since the mid-1990s the incarceration rate has continued to increase. According to the Bureau of Justice Statistics Arizona had 1 crime for every 12 residents in 1995, and that figure dropped to 1 for every 22 residents in 2008. Literature that the auditors reviewed indicates that:
"The effect of incarceration on crime is limited compared to the combined effect of other factors (such as increased law enforcement, employment, and education) and diminishes as prison populations grow... although Arizona's crime rate has dropped, the State has one of the highest reported crime rates in the national despite also having one of the highest incarceration rates."
Earlier this year the Reason Foundation, in conjunction with the Howard Jarvis Taxpayers Association, called for sentencing reform to alleviate the corrections crisis in California. Read the policy summary of the study here and the full study here.
Public-private cost comparisons are admittedly complex. However, given the relevance of the subject of correctional privatization in Arizona these days, a more thorough and robust cost comparison should be prepared so the policy dialogue on corrections can move forward through factual research.
Harris Kenny is a research assistant at Reason Foundation.