Who punishes whom? Bifurcation of private and public responsibilities in criminal punishment

Abstract

Who holds the legitimate right to punish criminals? While previous work has identified several factors that influence states’ decisions to delegate punishment duties to the private sector, it has not considered variation in the level of security required to implement the punishment. Delegating coercive power challenges commonly held assumptions about the appropriate locus of coercive power, and resistance is likely to be strongest when delegating highly secure services that require the greatest levels of physical coercion. Using data on American adult correctional facilities from 1990 to 2005, this article describes the current bifurcation of correctional contracting, wherein private contractors house increasing numbers of inmates in less secure correctional settings (e.g., low-security, community-based facilities) and public authorities retain near-monopoly control over inmates in highly secure settings (i.e., medium- and maximum-security prisons). Multinomial regression analyses reveal that states’ decisions to privatize highly secure facilities were associated with ideological and economic factors. However, the decision to privatize lower security facilities has become commonplace, and as a result has grown irrespective of state-level factors. These results suggest that handing over low-security services to the private sector has become a legitimate policy option, while privatizing the most secure services remains shrouded in illegitimacy.

Keywords:

• Punishment
• privatization
• legitimacy
• policy

Acknowledgements

Earlier versions of this paper were presented at the Oregon State University School of Public Policy on 11 May 2012, the Annual Meeting of the American Sociological Association on 20 August 2012, and the Annual Meeting of the Law & Society Association on May 30, 1014. Nicole Kaufman, Richard Aviles, and Monica Williams provided valuable feedback on earlier drafts of the paper. All errors are my own.

Notes

1. A notable exception was convict leasing, which allowed private entrepreneurs to lease the labor power of inmates in exchange for feeding and housing them. It was especially common in Southern states, where it provided a replacement for cheap labor following the emancipation of slaves and also minimized the potential political power of newly freed African-Americans. The practice continued in select states into the early 20th century (Hallett 2006, 43–51).

2. These figures refer to inmates held in private custody within a state, regardless of whether the inmates were sentenced by another jurisdiction.

3. The Census excludes local jails, immigration detentions facilities, privately operated facilities that are not primarily intended for state or federal inmates, juvenile facilities, military facilities, U.S. Marshals Service facilities, Bureau of Indian Affairs facilities, or hospital wings or wards dedicated to prisoners.

4. There are no private supermaximum-security facilities in the data.

5. Because states without low-security facilities were not at risk of privatizing such a facility, five state-years (Alaska in 1990; Minnesota in 2000; North Dakota in 1995; and South Dakota in 1990 and 1995) were omitted from the analysis.

6. For 1990, the crowding variable is lagged six years, to 1984, the most recent available data point for the crowding variable.

7. For brevity, the term ‘court order’ is used to refer to both court orders and consent decrees. The court order variable refers to ‘confinement’ facilities – those allowing fewer than half of inmates to leave the facility – in order to ensure data availability throughout the time series. The natural log of court orders is used because the raw distribution is highly skewed to the right. Because many states had zero court orders, and the natural log of zero is undefined, the variable is defined as ln(court orders + 1).

8. Supplemental versions of the three main models included regional dummy variables to control for unobserved regional differences within the United States. The results of these supplemental models were largely consistent with the main models. The effect of unemployment differed by type of privatization in one model; the effect of liberalism differed by type of privatization in two models; and the effect of union membership differed by type of privatization in one model, although levels of significance were lower (p < 0.10) due to multicollinearity among regional dummies and these variables.

9. Figure 3 uses estimates from Model 1. Figure 4 uses estimates from Model 2. Figure 5 uses estimates from Model 3.

10. The 2005 graph uses unemployment rates lagged to the year 2000. Overall unemployment was particularly low in 2000, which is why an unemployment rate of 5% appears to be high in the figure.