Rarely does a day pass when a politician or agency official does not make a dramatic claim that by adopting a particular policy or law, huge sums of taxpayer funds will either be expended or saved. The goal seems to be to convince the public and fellow policymakers that unless a particular action is adopted or rejected, the fiscal well-being of the state will be shaken or enhanced. This is referred to as cost-benefit analysis (CBA), and policymakers and the public are becoming increasingly dependent on--and cynical about--the science of CBA. The recent national debate on alternative public health plans highlights the uncertainty and politicalization of CBA, as dramatically different points are being presented by various interest groups regarding the same proposal.
In corrections we see a similar pattern. Advocates of efforts to reduce the nation's prison and jail populations often claim that their reforms will produce a windfall of savings that can be used to "reinvest" in more progressive and effective reforms. Others warn that if such actions are taken and that hordes of people are released from prison, crime rates will escalate to record levels, as well as the costs of crimes. These competing claims take on greater importance given the current fiscal crisis being faced by state and local governments where correctional and criminal agencies are being targeted for significant cuts. Perhaps the most dramatic example is California, where the state prison budget must be reduced by more than $1 billion in the current 2010 fiscal year.
The need to demonstrate CBA is underscored by the recent creation of CBA centers. In October, The National Conference of State Legislators (NCSL) sponsored a webinar on determining CBA for effective sentencing and corrections policy. (1) The Vera Institute was awarded a U.S. Department of Justice grant to create a Cost-Benefit Analysis Unit (CBAU) for correctional agencies. (2) Not to be outdone, the Evans School of Public Affairs, at the University of Washington, has created a Benefit Cost Analysis Center (or BCA). (3) Its definition of BCA (or CBA) is as follows:
Benefit-cost analysis (BCA), also known as cost-benefit analysis, aims to inform the decision-making process with specific types of information, namely measures in monetary terms of willingness to pay for a change by those who will benefit from it, and the willingness to accept the change by those who will lose from it. The use of monetary terms provides a common metric. Its purpose is not to price everything, but rather to order choices in a way that is informative about social choices for decision makers.
Whether it is a BCA or CBA, there is much confusion and debate about the science or perhaps more accurately the "art" of CBA. And the current fiscal crisis for state and local government has elevated the need to have more accurate CBA. This article tries to address the common methodological weaknesses and common misuses of CBA claims.
The Ambiguous Relationship Between Population Size and Costs
A major weakness in CBA is the assumption that correctional budgets change directly in relation to the number of people being supervised or incarcerated. This assumption allows the advocates of a particular program, policy or law to claim potential savings when in fact such savings or averted costs will not occur.
Between 1999 and 2004, New York state's prison population declined by about 13 percent but spending on prisons declined by just 4 percent. (4) In the city jail system, even though the jail population declined from 21,500 in 1992 to 14,000 in 2002, New York City's Independent Budget Office reported that spending on the jail system had actually increased by 16 percent.
Similarly, the budget for the Texas Department of Criminal Justice has continued to rise (see Table 1). What is particularly interesting is the projected budget for 2010 is more than $170 million greater than the 2009 budget, despite a projected decline in the prisoner population. And if one looks at the total correctional population, it grows by more than 30,000 people by 2014. The greatest growth is in locally funded probation, which is not accounted for in the TDCJ state prison and parole budget. The fact that growth in the prison population may slow at the expense of the parole and probation populations suggests that a full accounting of all costs and savings is required.
There are several reasons why agency costs rise and fall independent of the population they serve. One major reason is that a large portion of the costs associated with operating a prison are fixed and do not vary significantly by the size of the population. Correctional budgets are not designed to fluctuate based on how many people are incarcerated at a given time. It is like an airplane that has fixed costs associated with flying it from Los Angles to New York regardless of how many passengers are aboard. It is also true that because many systems have been operating beyond their design capacities, reductions in the prison population may simply allow double or triple celling to cease. Or it may be that despite reduction, local and state politicians, as well as unions, may be reluctant to lose well-paying jobs in rural or economically stressed communities.
Table 1. Prison, Parole, and Probation Populations and Texas Department of Criminal Justice Costs* Year Prison Parole Probation Population Population Population 2006 152,265 76,721 157,993 2007 153,849 76,709 159,851 2008 156,127 77,990 168,788 Projected 2009 156,928 78,462 172,808 2010 155,107 78,496 172,895 2011 155,589 80,052 175,075 2012 155,891 81,267 176,523 2013 157,831 82,602 178,216 2014 157,997 83,249 179,503 2007 % 3% 9% 12% Change Year Total Prison Parole Population Budget 2006 386,979 $2,600,119,919 2007 390,409 $2,658,263,049 2008 402,905 $2,657,156,979 2009 408,198 $2,861,034,889 2010 406,498 0,034,408,068 2011 410,716 Not Available 2012 413,681 Not Available 2013 418,649 Not Available 2014 420,749 Not Available 2007 % 8% Change * These numbers exclude some 13,000 TDCJ people housed in local jails and intermediate sanctions facilities. Texas Legislative Budget Board. 2009. Adult and juvenile correctional population projections fiscal years 2009 to 2014. State of Texas, January. Available at www.lbb.state.tx.us/PubSafetyCrimJustice/ 3_Reports/Projections_Reports_2009.pdf.
Using Fully Loaded Instead of Marginal Costs
One way around this problem is to establish marginal costs rather than "fully loaded" per capita costs. Marginal costs within a prison or jail tend to reflect immediate costs associated with adding a marginal number of prisoners to the facility count. These marginal costs often reflect food, medical, utilities and staff overtime. But even they can range from 10 percent to 50 percent of the fully loaded per prisoner cost.
For example, the Washington Institute on Public Policy uses a $22,000 "long-term" marginal estimate for its state prisoners. The California Department of Corrections and Rehabilitation (CDCR) uses a $21,900 "overcrowding" figure compared to a $49,000 per inmate per year cost to estimate how much money would be saved to depopulate crowded prisons. The state of Nevada uses a $2,200 marginal cost savings until a reduction of 500 prisoners occurs, which allows them to close the wing of a prison and thus reduce staff positions.
Averted Costs vs. Actual Cost Savings
Another common error is to claim averted costs. This involves developing a projection or forecast of the future prison or jail population and then claiming that the difference between the two is the result of whatever new policy has been introduced. The problem with this method is that it assumes the lower population is the direct result of the new policies and not some other cause. While it may be true that the policy has had some impact, to claim that it is the sole basis for the reduction is not valid. In order to make such a claim would require a rigorous evaluation that demonstrated the reforms and not other co-existing or parallel forces are producing the reduction in projected or actual growth.
An example of this type of thinking was evident when the FBI released the latest Uniform Crime Report crime rates. The dramatic decline in crime was credited to an array of individual suspects, including an aging male population, lower welfare rolls, higher numbers of immigrants, more police, more effective police, and of course higher numbers of prisoners. The point is that researchers are not able to figure out exactly which factors are the principle drivers of crime rates--only that all of these (and other) factors are associated with the crime rate decline.
In a similar manner, one can make the claim that absent a set of legislative or programmatic reforms, the prison population would have increased. But it can never really be known if the projected prison population growth would have occurred if the reforms had not been implemented. A recent example of this is from Texas, where a number of legislative reforms aimed at treatment and prison diversions were funded to ensure the state's prison population would not grow in the future. (5) The claim was based on a 2007 prison population forecast that estimated a growth from 159,492 to more than 168,000 by the year 2012. By 2009, that forecast had been substantially lowered. The lowered projection is used to claim that justice re-investment reforms averted $444 million in prison construction costs, even as the state spent $241 million for additional diversion programs and treatment capacity (see Figure 1).
However, a comparison of the 2007 and 2009 population projection publications show that the 2007 projection had assumed lower than actual parole grant rates and higher than actual new court prison admission rates--both of which were not directly related to the legislative reforms. So a significant portion of the lower 2009 projection is simply due to erroneous assumptions made in the 2007 estimate. Further, the estimates exclude about 13,000 state prisoners housed in local jails or "intermediate sanction facilities" that may or may not be increasing. To correct the problem, the 2007 projections should have been reset with the correct data on prison admissions and the parole grant rate. Also, it would have been more accurate to only claim averted construction costs in the immediate budget year and not for several years in the future as it is not certain whether the prison construction would have happened. For example, the increased prison population could have been accommodated through contracting with local jails rather than the more expensive construction route.
Another example of the long-term averted costs was published by the Washington Institute for Public Policy when it estimated that the state of Washington would avoid the need to construct up to three new prisons by the year 2030 by adopting a variety of yet-to-be-funded evidence-based treatment programs. The publication boldly stated that funding such programs at a cost of $85 million a year would avert $2 billion in taxpayer costs and reduce crime. (6) But rather than putting faith in a long-term forecast in a field in which long-term projections are notoriously inaccurate, taxpayers may feel they are better off investing their money in a less speculative investment strategy.
Jan.2007 Jan.2009 Aug. 2009 159,492 156,928 Aug. 2010 162,298 155,107 Aug. 2011 164,592 155,589 Aug. 2012 168,166 155,891 Aug. 2013 157,831 Aug. 2014 157,997 Texas Legislative Budget Board. 2009. Adult and juvenile correctional population projections fiscal years 2007 to 2012. State of Texas, January. Available at www.lbb.state.tx.us/PubSafety_CrimJustice/3_Reports/Projections_ Reports_2007.pdf. Note: Table made from line graph
Separating Costs by Classification Or Supervision Levels
Related to the fully loaded versus marginal cost estimates is the need to specify which type of correctional population is be affected by a reform. Many alternatives to incarceration target the less-expensive prisoner to house. For example, if one is proposing to divert low-risk, nonviolent offenders who have relatively short periods of incarceration, then that is very different than diverting people who would be housed in a supermax facility. Thus, it is important to estimate the custody or classification level associated with the reform.
Table 2: Estimated Overall Prison-Based Treatment System Effects All prison releases 700,000 Exposed to treatment--35% 231,000 Completed treatment--75% 173,250 Reduction in return to 17,325 prison at 10% Percentage reduction in 2.5%o 700,000 prison admissions Bed savings at 30 months 43,313 length of stay Percentage of total 1.5 2.9% million prison population Percentage of total 14 0.2% million arrests
For example, the Michigan Department of Corrections is careful to use the estimated costs of its Level I or camp programs rather than the average cost per inmate when estimating the savings associated with its Special Alternatives Intervention (SAI) program. This is done by considering that if the SAI program did not exist, the prisoners would more likely be housed in a low-security prison rather than a medium- or close-security facility. (7)
Relative to probation and parole supervision savings, the same principle applies. Efforts to divert people from prison or jail to probation or parole are fairly common. The extent to which a person poses a higher or lower risk should affect whether the person is placed on high, moderate or low supervision.
Overstating the Cost Benefits of Treatment
Finally, there is much emphasis in the reentry and treatment literature that prison populations and costs can be reduced by reducing recidivism rates. Mathematically this is correct since about 40 percent of all prisoners return to prison and about two-thirds of all prison admissions are people who have failed their probation or parole terms. However, in practical terms, it is unlikely that increasing the funding of rehabilitative services will have a dramatic impact on recidivism rates. The most comprehensive analysis of such services shows that effective programs can only reduce recidivism rates by 3 to 8 percent. (8) Given that: 1) most correctional programs have not been evaluated; and 2) most prisoners do not complete such programs while incarcerated, there simply is not a sufficient level of "treatment dosage" to have a significant effect.
Table 2 illustrates this point by showing the maximum amount of treatment services that could be reasonably delivered to the nation's prison population and the possible effects on the current 1.5 million prison population. The table assumes that one-third of all prison admissions were exposed to effective treatment services and that 75 percent successfully completed the program, reducing their recidivism rates by 10 percent. The 10 percent recidivism rate reduction means that it would require 10 people to complete the treatment before one person would be "cured" or not recidivate. This is known as the "number needed to treat," or NNT, a measure used in the medical profession as part of evidence-based practices.
With these assumptions, the treatment programs would only reduce the number of prison admissions by 17,325, which is less than 3 percent of all prison admissions. Further, the total bed savings at best would also be 3 percent.
Finally, some have used victim "pain and suffering" costs to inflate the potential savings of treatment programs. Such costs are based on estimates reported by Mark Cohen's 1988 study and a 1996 research report by Ted R. Miller and colleagues. (9) These victim costs, referred to as "pain and suffering" costs, are derived from published civil case settlements and clearly do not translate to real money from which taxpayers can benefit. At best, they are symbolic and are not even linked to the studies. Including them grossly increases the costs to victim estimates and produces very large but false savings.
As with much of the growing literature on what works and at what costs, correctional stakeholders need to be aware of misleading and sometimes purposefully erroneous CBA claims, especially when they may result in serious underfunding of core correctional services. Here are some guidelines in assessing CBA claims:
1. Are the savings based on fully loaded or marginal costs?
2. If the savings are not based on marginal costs, is there clear evidence that prison facilities will be closed and there will be a reduction in the work force?
3. Are the CBA savings "projected averted costs" or "actual cost savings"?
4. If the claims are based on projected averted costs, how sound are the projection assumptions?
5. Have the possible associated increases in probation, parole or jail populations been included in the CBA?
6. Has the CBA accounted for changes in correctional populations and costs by custody and supervision levels?
7. Does the CBA include estimated savings linked to recidivism reductions?
8. If the savings are linked to recidivism reductions, is there any evidence that such programs can be administered to a large portion of the correctional population and have the projected impact on recidivism?
9. Do any of the treatment/recidivism reduction estimates include victim "pain and suffering" costs?
10. Has the CBA been evaluated by an independent entity?
In general, CBAs based on long-term (beyond two years) averted costs should not be used because it is not possible to accurately validate future trends. Further, CBAs that are based on rehabilitative or treatment effects are equally precarious and should be heavily discounted. Even the more qualified CBA should be viewed with a healthy level of skepticism until some evidence is available that proves it is valid. For that to happen, it is necessary to take the time to see if the predictions were even close.
(1) National Conference of State Legislatures, www.ncsl.org/default.aspx?tabid=18870.
(2) Vera Institute of Justice, Cost-Benefit Analysis Unit, www.vera.org/centers/cba.
(3) University of Washington, Evans School of Public Affairs, http://evans.washington.edu/research/centers/benefit-cost-analysis/related-search.
(4) 2004 is the latest year for which prison system expenditure figures are available from the National Association of State Budget Officers.
(5) Council of State Governments Justice Center, Justice Reinvestment, www.justicereinvestment.org/states/texas/pubmaps-tx.
(6) Aos, Steve, Marna Miller and Elizabeth Drake. 2006. Evidence-based public policy options to reduce future prison construction, criminal justice costs and crime rates. Olympia, Wash.: Washington State Institute for Public Policy.
(7) Austin, James and Gabrielle Chapman. 2009. Michigan Department of Corrections special alternative to incarceration: First year process evaluation. Washington, D.C.: JFA Institute.
(8) Aos, Steve, Marna Miller and Elizabeth Drake. 2006.
(9) Cohen, Mark A. 1988. Pain, suffering, and jury awards: A study of the cost of crime to victims. Law and Society Review 22(3): 537-55; Miller, Ted R., Mark A. Cohen and Brian Wiersema. 1996. Victim costs and consequences, research report. Washington, D.C.: U.S. Department of Justice, National Institute of Justice.
James Austin, PhD., is president of JFA Institute in Washington, D.C.…