The overall percentage of adults in private prisons is relatively small, but the actual impact for some States may be much greater. An article in The New Mexican, for example, suggested that New Mexico was overpaying millions of dollars to private providers that were housing more than 40 percent of the State's inmate population.
Thus, it is vital that policymakers have the best possible cost and quality information when they are making decisions regarding privatizing prisons in their jurisdiction. But what criteria should prison administrators and policymakers use when making cost and quality evaluations?
To help answer these questions, the National Institute of Justice (NIJ) assembled researchers, prison officials, private service providers, and proponents and opponents of prison privatization on March 28, 2007, to discuss this complicated and often controversial issue. At the core of the meeting was a rare occurrence: two cost and performance analyses of the same four prisons—one privately operated and three publicly operated—with different findings. The two reports are referred to in this article as the "Taft studies."
One of the Taft studies was conducted by Doug McDonald, Ph.D., principal associate with Abt Associates Inc. (referred to as the "Abt report"). The other study, funded by the Bureau of Prisons (BOP), had two components: a performance or quality analysis conducted by Scott Camp, Ph.D., a senior research analyst in BOP's Office of Research and Evaluation, and a cost analysis conducted by Julianne Nelson, Ph.D., an economist with the Center for Naval Analyses (referred to collectively as the "BOP report").
The Taft studies offer the research and public policy communities a rare opportunity to consider the different approaches that were used, why the results were different, and how this can inform not only the prison privatization debate, but in many ways, the government outsourcing, or privatization, issue in general.
Making Prison Privatization Decisions
Although every jurisdiction has its own economic and managerial idiosyncrasies, lessons learned from the Taft studies and the NIJ meeting may help administrators and public policy analysts avoid mistakes that could lead to higher taxpayer costs and possible dire consequences of poor performance. These lessons include:
- Cost comparisons are deceivingly complex, and great care should be taken when comparing the costs of privately and publicly operated prisons.
- Special care should be given to an analysis of overhead costs.
- A uniform method of comparing publicly and privately operated prisons on the basis of audits should be developed.
- Quantitative measures of prison performance, such as serious misconduct and drug use, should be incorporated in any analysis.
- Future analytical methods could allow simultaneous cost and quality comparisons.
Someone not familiar with the literature on prison privatization might assume that cost comparisons are accomplished without controversy or ambiguity. One key lesson learned from the Taft privatization studies is that comparisons are not as simple as might be presumed.
Consider, for example, per diem (or daily) costs. The chart "Average Per Diem Costs Per Inmate (in dollars) for FY 1999–2002" lists the per diem costs, in dollars, as analyzed by the Abt and BOP researchers for the three publicly operated prisons and Taft, the privately operated facility, for fiscal years 1999–2002.
|FY 1999||FY 2000||FY 2001||FY 2002|
|Publicly operated prison|
|Privately operated prison|
According to the Abt analysis, the Taft facility was cheaper to run, every year, than the three publicly operated facilities. In 2002, for example, Abt reports that the average cost of the three public facilities was 14.8 percent higher than Taft.
The BOP analysis, however, presented a much different picture. According to the BOP researchers, the average cost of the public facilities in 2002 was only 2.2 percent higher than Taft.
Why were the Abt and BOP cost analyses so dramatically different? And, importantly, what policy implications does this have?
There are two primary reasons why the cost analyses were different: (1) the way inmate population sizes were treated, and (2) what was included in overhead costs.
With respect to inmate populations, Taft had on average approximately 300 more inmates each year than the three publicly operated prisons throughout the study period. Therefore, the private service provider for Taft benefited from economies of scale that reduced average costs. To adjust for such economies of scale, the BOP researchers made adjustments to the expenditures.
Abt, in its analysis, however, did not consider economies of scale, choosing, instead, to use the actual average per diem amount that BOP paid the Taft contractor. In other words, BOP estimated what expenditures would have existed for identically sized prisons, and Abt based its analysis on actual expenditures.
McDonald, the researcher who performed the Abt analysis, argues that his approach—using actual costs that BOP paid to have a private contractor operate Taft—yields a more telling comparison. Although the BOP researchers disagree, this leads to one of the primary points of this article, which is to remind policymakers and others interested in the prison privatization issue that making cost comparisons is not a simple matter of arithmetic.
What Should Be Included in Overhead Costs?
Prison costs comprise:
- Direct operations costs, such as staff salaries, inmate food, medical care, and other services.
- Indirect (overhead) costs, such as regional and central office supervision, computer services, planning, and budget development.
With respect to overhead costs, different approaches by the two research groups led to different findings. Basing its analysis on the extent to which the government actually provided resources to support the Taft operation, Abt concluded that only a bare minimum of support was provided. Therefore, the Abt analysis reported a 100-percent savings of indirect, overhead costs for Taft during the time period in the study. BOP, on the other hand, assumed that most overhead costs (planning, auditing, and other central and regional operations) could continue to be incurred by the government, even if a private company was operating the prison. Therefore, the BOP researchers applied a 10–12 percent overhead rate (the average for BOP prisons during the 1998–2002 Taft study period), calculating privatization savings of 35 percent of overhead for that 5-year period. Here again, BOP estimated the costs that the government would have incurred by central administration, and Abt presented only what was reported.
One can anticipate that underlying assumptions regarding overhead costs will have significant implications for bottom-line estimations of costs and savings. As previously discussed, the assumptions made by Abt led to a finding of much less overhead for the Taft private provider, suggesting that the government could save a great deal of money by privatizing prisons. The assumptions underlying the BOP analysis were different, however, and led to a less sanguine conclusion. Unless policymakers are mindful of these subtleties in basic assumptions, they are not likely to delve so deeply—or even be presented with this level of detail—when considering taxpayer benefits of prison privatization.
At the March 2007 NIJ meeting, Mark Cohen, Ph.D., an economist at Vanderbilt University, presented data showing that privately operated (and sometimes privately financed) prison systems have lower costs over time than publicly operated prison systems. Although this may be true as an overall average, it is not necessarily true for a particular jurisdiction. Any prison administrator or other policymaker considering privatization would be well advised to consider the specific analytic assumptions underlying the studies.
Performance: Contract Compliance vs. Auditing
Performance is a vital part of any prison privatization discussion. In many jurisdictions, a truly accurate comparison of privately versus publicly operated prisons is hampered by different performance yardsticks. A privately operated prison, such as Taft, has a contract; performance, therefore, can be measured by compliance with specific contract terms (which, of course, can vary from contract to contract). BOP-operated prisons, on the other hand, measure performance through an auditing procedure called program review.
Because no method existed for measuring publicly and privately operated prisons on many dimensions for performance, both of the Taft studies have limitations. Until a common yardstick exists, any analysis will not be as rich as it could be. Nonetheless, it is important to make whatever performance analyses are possible—in areas such as safety, medical care, programming, and rehabilitation services—when considering prison privatization.
In the Taft comparison studies, the Abt researchers first looked at 19 functional areas—including food services, health care, safety, and security—that were specified in the Taft private-service provider contract. The contract had a scoring system, upon which possible bonuses and possible deductions would be based:
- Unsatisfactory = 0
- Marginal = 1
- Fair = 2
- Good = 3
- Excellent = 4
- Outstanding = 5
Over the first 5 years of its contract (1997–2001), the Taft private provider received a rating of 2.5; during the 2002–2004 contract period, the average rating was 2.8, which resulted in a possible award fee of nearly 50 percent of the amount allocated.
In their performance analyses, both the BOP and the Abt researchers also looked at misconduct, comparing assaults at the Taft facility to assaults at 20 publicly operated low-security prisons. Both reports found that the Taft assault rate was lower than the average of the 20 prisons; with respect to the four facilities in the Taft studies, Elkton had an assault rate similar to what would have been expected based on its inmate composition; Forrest City, Yazoo City, and Taft had lower than expected assault rates (Yazoo City was the lowest).
The researchers also considered drug use, escapes, inmate grievances, and access to medical care in their performance analyses. During the study period, Taft had a very high drug-use rate compared to the 20 BOP-operated low-security prisons. Abt noted two escapes at Taft and only two in the BOP prisons; the BOP researchers reported the same two Taft escapes, but also noted a disturbance at Taft that involved 1,000 inmates who refused to return to their cells for the 10 p.m. count.
With respect to access to medical care, the researchers found that the Taft inmates were more likely to see a physician than inmates in the 20 BOP-operated prisons.
Despite Differences, Lessons Learned
Despite differences in the approaches and assumptions used by Abt and BOP in the Taft studies, these reports represent two of the best prison privatization analyses performed so far. Administrators, policy analysts, and researchers looking at prison privatization and the larger public policy issue of government outsourcing would benefit from a closer consideration of the full reports.
Sidebar: How Did We Get the Benefit of Two Studies?
Due to the sheer expense of conducting evaluation studies, it is a rare occurrence to have competing research analyses like those discussed in this article. To understand how this happened, some historical perspective is in order.
In 1996, the U.S. House of Representatives directed the U.S. Bureau of Prisons (BOP) to perform a 5-year prison privatization demonstration project of the low- and minimum-security prisons in Taft, California. BOP awarded a 10-year contract to the Geo Group (formerly Wackenhut Corrections Corporation), which operated the facilities from 1997 to 2007. The contract was then recompeted, and a new contract to run the Taft prisons was awarded to Management and Training Corporation.
Although the U.S. Congress did not request a formal evaluation of the Taft facilities, BOP leadership decided that an evaluation of cost and quality would help them make better decisions regarding privatization. BOP funded the National Institute of Justice to secure proposals for an evaluation of Taft and similar BOP facilities. Abt Associates won that competition and conducted the study. BOP's Office of Research conducted its own independent study in order to understand how to conduct this new type of research.
About the Author
Gerry Gaes has 27 years of criminal justice experience. From 1988 to 2002, he served as director of research at the Bureau of Prisons (BOP); during that time, he worked on an interim BOP report concerning the Taft privatization study that is discussed in this article. Gaes was a visiting scientist with the National Institute of Justice from 2002 to 2007. His other research interests include the simulation of criminal justice processes, the criminogenic effect of prisons, and inmate gangs.
About This Article
This article was published as part of NIJ Journal issue number 259, released March 2008.
[note 2] Terrell, S. "Audit: N.M. Private-Prison Costs Soar," New Mexican, May 24, 2007, (accessed January 15, 2008).
[note 3] The four prisons in the Taft studies are low- and minimum-security facilities constructed by BOP during the same period. The prison operated by the private company (through a contract with BOP) is in Taft, California. The three publicly operated prisons are in Elkton, Ohio; Forrest City, Arkansas; and Yazoo City, Mississippi.
[note 4] McDonald, D.C., and K. Carlson, Contracting for Imprisonment in the Federal Prison System: Cost and Performance of the Privately Operated Taft Correctional Institution (pdf, 174 pages), final report submitted to the National Institute of Justice, November 2005 (NCJ 211990).
[note 5] Camp, S.D., and D.M. Daggett, Evaluation of the Taft Demonstration Project: Performance of a Private-Sector Prison and the BOP (pdf, 127 pages), Washington, DC: Federal Bureau of Prisons, October 2005.
[note 6] Nelson, J., Competition in Corrections: Comparing Public and Private Sector Operations (pdf, 148 pages), Alexandria, VA: The CNA Corporation, December 2005.
[note 7] In any analysis of prison misconduct, it is important to account for the composition of the inmate population (with respect to risk of misconduct) in addition to the security level of the prisons. An expected level of misconduct takes this into account; an average (actual) level does not take inmate composition into account.