Interview with Tracey Kyckelhahn of the Bureau of Justice Statistics
Tracey Kyckelhahn is a statistician with the Bureau of Justice Statistics (BJS), part of the Office of Justice Programs in the U.S. Department of Justice. She is the author of the report State Corrections Expenditures, 1982-2010, which was published in December. We asked Kyckelhahn about BJS’s sources for justice system costs and some challenges associated with collecting and analyzing them. These comments represent Kyckelhahn’s views and not necessarily those of BJS.
If you’d help us set the stage here, what information does BJS collect on criminal justice system expenditures?
- Expenditure and employment extracts, collected by the U.S. Census Bureau through its government finance survey;
- Operating expenditures by law enforcement agencies, through short-form and long-form surveys in the Law Enforcement Management and Administrative Statistics collection;
- Expenditures by prosecutors’ offices in state court systems, collected by the National Census of State Court Prosecutors through this survey;
- Operating and capital expenditures through surveys by the Census of Jail Facilities of jurisdictions and facilities; and
- Information about fees paid to a facility for housing another agency’s inmates, through this survey by the Census of State and Federal Adult Correctional Facilities.
What are some general issues that arise in interpreting justice-system spending patterns and trends?
One problem is that in some states, not all costs related to justice system spending can be disaggregated from total state expenditures. For example, some employee benefits are in a fund for all state employees, and therefore those for corrections or police protection or court employees cannot always be separated.
What are some examples of how people have interpreted or used BJS’s statistics well—and examples in which the statistics have not been used so well?
I have seen a couple of state-specific studies that have used the government finance data as part of a comprehensive analysis of the costs and benefits of state justice systems. In these studies the expenditure data for the criminal justice functions was combined with other expenditure data from the states, as well as arrest, conviction, and inmate data. The expenditure data can be very useful when combined with other information.
One caution I would give in using the expenditure data in tandem with other data is to be sure to know what is included in each expenditure category and to know what levels of government to include. For example, I have seen the state expenditure data for judicial and legal services divided by the number of criminal convictions in a state to calculate cost per conviction for a state. However, the judicial and legal services category also includes civil courts, and it is therefore inappropriate to use that to calculate cost of conviction. Also, judicial and legal services are often paid for by localities, and to get the total cost for judicial and legal services one would need to combine state and local expenditures. The Census Bureau has a classification manual that specifies which types of spending are in which categories, and how that has changed over time for its entire government finance survey.
Another way I have seen expenditure data on corrections used inappropriately is related to average vs. marginal cost. I’ve seen reports state that a correctional agency could save X amount of dollars if a certain program were used to release prisoners, and to calculate the savings, they multiplied the average cost to house an inmate by the number of inmates released. However, the marginal cost should be used in such a situation. Unless very large numbers of inmates can be released, only non-fixed costs such as food can be saved. This is especially true in correctional systems that are crowded. Average cost should be used for things like determining the total amount needed to fund a correctional system, not the savings generated by the release of a relatively small number of inmates.
Another problem is the difficulty of making appropriate state-by-state comparisons. Aside from cost-of-living differences among states, often people assume that either more or less spending is better. So a state that spends less on corrections must doing something right (or wrong) compared to a state spending more—or a locality spending more on police protection (or education or health care, etc.) must be more proactive than one spending less. However, there needs to be an examination of the entire justice system to determine how to most efficiently produce the best results, as opposed to simply looking at the spending of one segment. It can be very educational to do comparative analyses; it’s just that those comparisons should be more than surface deep.
What’s a particularly interesting point that comes out of the recent State Corrections Expenditures, 1982-2010 report?
One thing I found interesting is the regional similarity in trends in per capita spending. Wisconsin and Minnesota have the same pattern, the Mid-Atlantic states share a pattern; New Hampshire and Vermont share a pattern; Pennsylvania, Ohio, and Michigan are grouped together, and so on. I think that could be an area for further analysis.
Something I think the general audience will find interesting is that in the nation, overall spending on corrections is actually a relatively small component of total state expenditures—generally about 3 percent or less—and this has been true for decades. Also, over the past 10 years, the amount spent on corrections has declined, while it has increased in categories such as education and public welfare. Both of these categories have always been a much larger share of state expenditures than corrections.
Given the interest and need for marginal cost information—for prisons in particular—what resources would you recommend?
The Washington State Institute for Public Policy (WSIPP) has done some interesting work using marginal costs. The tables for their report on crime-reduction programs is helpful in its discussion of how to calculate marginal costs—and the report itself addresses how to approach quantifying savings from those programs. I’d also recommend a WSIPP report that discusses an investment tool for studying sentencing and corrections policy options. I also think your CBKB page on marginal costs is great. And the Bureau of Labor Statistics has various reports examining issues related to marginal cost and marginal value, including one about the Employment Cost Index.
Some researchers use regression analysis to calculate marginal costs by examining the relationship between changes in workload and changes in spending. What data would you need to estimate state-specific marginal costs of incarceration using regression analysis, and why?
To do this for the nation we would need a considerable amount of information that is beyond our resources. However, if we could do such a study, what we would need would depend on how detailed we wanted our analyses to be. At the very least, if we had information on how much a prison spends per inmate on items such as food or medical costs, we could begin to have some idea of the marginal cost, as those are items that are probably among the most likely to change with the addition or removal of one inmate. But it would be interesting to have such cost information broken down by the characteristics of the prisoner. Ideally, one would be able to answer questions related to how much certain types of prisoners cost and how much certain numbers of prisoners cost. For example, food costs may not vary much by type of prisoner and so a certain amount will be saved each time a prisoner is released, whereas other costs, such as medication costs, will vary a lot, depending on the physical and mental health of the individual. It would be nice to incorporate this information to the point where we could at least group types of inmates and say what the costs were for older inmates, mentally ill inmates, and so on.
Some fixed costs, such as staffing and utilities, may be dependent on the type of security level—for example, violent inmates might cost more to supervise. But the costs will also depend on the number of inmates being supervised. For supervision costs, we would need to know how much the inmate population needs to decline before these fixed costs can be reduced. Then there are fixed costs that are unlikely to see much reduction without large population declines, such as those for central administration. Of course, all of this varies by state and over time.
What kinds of questions would BJS like to answer about criminal justice system expenditures?
We are examining the feasibility of producing a report on local corrections expenditures. There are a few more challenges with the local data than the state data that we are trying to resolve. Also, in the future we might examine the shifting responsibilities for justice expenditures among the federal, state, and local levels.