Skip to main content

Advertisement

Log in

The Distribution of Police Protection

  • Original Paper
  • Published:
Journal of Quantitative Criminology Aims and scope Submit manuscript

Abstract

This paper investigates the distribution of police protection in the United States by race and class. By examining police employment and demographic data for every general-service police jurisdiction in the US, I find that poor and heavily-nonwhite jurisdictions employ far fewer officers per crime than wealthy and white jurisdictions do. That finding contrasts with an older body of literature on the distribution of police protection, which examined the distribution of police resources across neighborhoods within individual cities and found little inequality. I also find that inequality in police protection has grown since 1970—a finding that contrasts with the increasingly equal distribution of resources for education, the other major claim on local government revenues—largely because criminal victimization became more concentrated in disadvantaged communities. (In the process, I find that contrary to widespread impressions, the crime rate fell very little in the most disadvantaged jurisdictions from 1980 to 2000, and violent crime actually increased). Finally, by examining data about federal grant programs, I find that the rise of federal contributions to local policing in the 1990s slowed the growth of inequality somewhat, suggesting that revenue-sharing has a real but modest role to play in reducing inequality in police protection. Together these findings highlight a neglected aspect of equality in criminal justice.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. For example, blacks have substantially less confidence than whites in the ability of police to protect them from violence (Jones 2005), and they are more likely than whites to say that we spend “too little” on the crime problem (Maguire 2005: 135). The same is true for low-income respondents.

  2. Some studies analyzing the determinants of police strength have also investigated variation across jurisdictions (Maguire 2001 provides a thorough, and generally critical, review), but that literature takes a different form from the educational inequality literature: Instead of examining inequalities in the level of policing directly, this literature aims to develop a comprehensive explanation for variations in police strength. In any case, these studies mostly focus on a small number of cities or even a single jurisdiction (some aggregate the number of officers to the national level and attempt to explain variation in police strength over time). None examines the distribution of policing nationwide.

  3. William Fischel writes: “Given a stock of numerous and independent districts…local public education was converted into an essentially private good. To get the benefits of the schools, you had to buy a home in the community whose property taxes covered the cost of education… Local schools, though nominally in the public sector, became more like a private good. Tuition for education quality differentials was extracted in the housing market, not at the schoolhouse door” (Fischel 2009: 222).

  4. Moreover, the immediate beneficiaries of education are children, and even in Tiebout’s story it is not children but their parents who choose to trade off educational resources against low taxes or better public services in other realms. Perhaps we demand equal opportunities for education because we are unwilling to allow parents to trade this aspect of their children’s life chances against their own tax bill. If so, a similar logic applies to policing. Recent data from the National Crime Victimization Survey shows that minors are the victims in almost half of all violent crimes, and these minors do not choose to live in underpoliced neighborhoods any more than they choose to live in communities with poor schools.

  5. The Tieboutian literature has identified many preconditions for this conclusion. Most important, jurisdictions must have some way of forcing residents to consume a minimum level of housing or Tiebout sorting will collapse into a game of “musical suburbs”, in which the poor chase the rich to free-ride on the services financed by their taxes. In practice, zoning rules serve this purpose (Hamilton 1975).

  6. I tried to exclude any county sheriffs that do not provide general-purpose police services. (These agencies typically operate a county jail and provide security for courts and other county buildings.) Most of them can be easily identified in the UCR dataset because the FBI codes their jurisdiction’s population as zero. Occasionally, however, the UCR has erroneously assigned nonzero population to these agencies. When my own algorithm for identifying the population policed by county sheriffs assigned an agency zero population, I reviewed its website to verify that it did not in fact provide general purpose law-enforcement services anywhere in the county. Of course, even sheriffs that do provide general-purpose police services may spend more time than municipal police on other tasks such as operating a jail, providing court security, or serving civil processes (Hickman and Reaves 2006), and these additional duties make their personnel counts hard to compare with a municipal agency’s. The FBI asks agencies to leave employees who mainly perform ancillary functions out of their employee counts, but some may do so imperfectly. In the end, there is unavoidable variation in the functions that different police agencies perform, and my analysis suffers from this problem to the same extent as other studies that rely on the UCR (e.g. most studies of the effect of police on crime). Nevertheless, the inequalities I will identify are large enough that this problem is unlikely to change my qualitative findings. Maltz (1999) provides a detailed critique of the UCR, and Maguire et al. (1998) provides a useful analysis focused on police employment data.

  7. In a few states such as Pennsylvania, state police rather than county sheriffs serve unincorporated areas. Unfortunately, there is usually no way to determine how many state police officers are assigned to an area, so my analysis excludes these areas; together they represent about 1.5% of the US population. Practical considerations also made it necessary to eliminate Indian reservations. Some tribal police are included in the UCR dataset, but many of these areas also receive police services from the Bureau of Indian Affairs, which is not; these agencies police less than 0.1% of the US population.

  8. In cases where a locality crosses county lines, the necessary data appear in the summary level 55 (place-county level) Census files. In cases where a New York or Ohio village crosses town or township lines, the data appear in the summary level 070 (place-remainder level) files. In addition to the difficulties with county sheriffs discussed in the text, several other complications make it inappropriate to rely exclusively on the crosswalk FIPS codes. For example, hundreds of localities across the country—most of them very small—contract with their neighbors for police protection, and dozens more have banded together to form regional departments. Many of these arrangements are identified in the UCR dataset (I tried to identify the rest by contacting relevant state officials, such as the Governor’s Center for Local Government Services in Pennsylvania and the California Contract Cities Association), but the crosswalk file often does not provide the FIPS code for the jurisdictions involved in them; instead it assigns each regional department the FIPS code for one of its participating localities or for the entire county that contains it. Details about these and other matching complexities are available from the author.

  9. The UCR file presents the FBI’s own estimate of the total population served by each agency—though not, crucially, the economic or racial composition of that population. The correlation between my population estimates and the FBI’s for all agencies was at least 0.99 for every year in my analysis. The correlation for county sheriffs was 0.96 for 1970, 0.95 for 1980, 0.93 for 1990, and 0.95 for 2000. In many cases where my estimate diverges from the FBI’s, my estimate appears to be more accurate (see e.g. footnote 6). My algorithm does, however, suffer from one systematic error: In years when a municipal agency does not appear in the UCR dataset at all, my algorithm will not subtract its population from the total county population when estimating the population served by the county sheriff. Since agencies serving over 97% of the population reported personnel data from 1975–2000, this problem is unlikely to be large, but in 1970 missing data is more common. To check whether these gaps affected my conclusions, I analyzed how my results would change if I replaced my own population estimates with the FBI’s. To do that, I first estimated the composition of the county sheriff’s jurisdiction using my usual algorithm, and then I scaled down each demographic datum (e.g., the number of blacks in the jurisdiction) according to my best estimate of the total population living in nonreporting jurisdictions in the county. (Specifically, I multiplied each datum by the ratio of (1) the FBI’s estimate of the total population served by the sheriff to (2) my own estimate of that population.) For 1980, 1990, and 2000, this adjustment had almost no effect on my findings, and even in 1970 it had only a modest effect. In all cases my qualitative conclusions persisted.

  10. The results I will discuss in this paper measure how “wealthy” a jurisdiction is using its per-capita income. Household income is arguably a better measure, but the necessary data were not available for 1970. To check the robustness of my results, I reanalyzed my findings for 1980–2000 using average household income in place of per-capita income, and I reanalyzed my findings for 1970 using average family income in place of per-capita income. None of these re-analyses changed my results substantially. I also reanalyzed the findings about race using the fraction of each jurisdiction’s population that is black rather than the fraction of the population that is nonwhite. Again, my basic findings persisted.

  11. Some school finance research has used the Gini coefficient and related measures like the Thiel index (Murray et al. 1998), but these measures were developed to study income inequality, not the distribution of other goods. The Gini coefficient can be misleading in other contexts because it quantifies inequality in resources regardless of who benefits from it. (For example, a distribution of teachers that favored the poor could yield the same gini coefficient as a distribution that favored the rich.) Since inequality in the distribution of any social good matters mainly when it exacerbates the inequalities that accumulate in other spheres of life (Walzer 1973), the important question is not whether the distribution of policing is unequal but whether that distribution favors whites and the wealthy. The CI provides the information needed to answer that question, but the Gini does not. The same criticism applies to Coulter’s (1980) Coefficient of Inequity, discussed in the policing literature.

  12. One source of uncertainty about this and other findings from my analysis inheres in all analyses of UCR data: The underreporting of crime to the police. Homicide is usually considered the best-reported crime, so it might seem useful to examine trends in police protection per homicide, but unfortunately more than three-quarters of all jurisdictions (representing more than one-third of the U.S. population) reported no homicides in 2000. Vehicle theft, however, is much more widespread and also very well-reported: The most recent NCVS found that over 90% of vehicle thefts were reported to the police, and reporting rates did not vary much by race or income (U.S. Dept. of Justice 2010: 97–100); the same was true in the 1970s (e.g. U.S. Dept. of Justice 1977: 76–77). While vehicle theft suffers from its own problems as a proxy for the overall crime rate—in particular, rates of car ownership vary across cities, and carless households are particularly common in big cities like New York and Boston—it may be worth noting that the trends reported in the text for police per index crime also hold for police per vehicle theft: The income 20-20 for police per vehicle theft grew from 0.50 in 1970 to 2.03 in 2000, while the racial 20-20 for police per vehicle theft grew from 2.70 to 3.83. This trend makes it harder to see how changes in reporting rates could account for the trends I discuss in the text.

  13. The claim in the text is purely compositional, not causal: It says that the “police per crime” ratio for poor jurisdictions would have tracked the ratio for rich jurisdictions if its denominator had not grown so much. (One way to make this counterfactual precise is to imagine a world in which police per capita remained at their actual 2000 levels but the average violent crime rate in the poorest quintile fell at the same rate as the richest quintile—i.e., it fell 21% from 1970 to 2000 rather than growing 95%. On these assumptions, the bottom quintile would have had 283 police per 100 index crimes rather than 114, yielding a 20-20 of 0.96 rather than the actual value of 2.37.)

    Of course, if crime had actually fallen in the poorest jurisdictions, the decline might have had a variety of consequences, and those consequences could provide the basis for alternative counterfactual scenarios. In particular, the poorest jurisdictions might not have hired so many police if their crime rates had fallen. That thought suggests an alternative counterfactual: If the level of police per capita in the poorest jurisdictions had tracked level of police in the richest (i.e. if it had grown only 44% rather than 86% from 1970 to 2000), then the bottom quintile would have had 219 police per violent crime in 2000, yielding a 20-20 of 1.24; thus even in this conservative counterfactual, the lion’s share of the inequality in police protection would vanish. Of course, this counterfactual suffers from its own difficulties, and not just because of uncertainties about how many police the bottom quintile would have hired: Large changes in crime would presumably affect other factors as well, such as the population and median income of these jurisdictions. I leave exploration of such possibilities for future research, which would require different data, appropriate instrumental variables, and a more comprehensive framework for understanding the relationships among crime, residential composition, and police hiring than I can muster here; for now rigorous description and interpretation may be a more appropriate goal (cf. Freedman 1991; Berk 1991).

  14. Unfortunately, the literature on revenue-sharing has produced no consensus about the extent to which federal funds supplant local expenditures, though the current conventional wisdom suggests that supplantation is less extensive than economic theory would predict (Hines and Thaler 1995).

  15. It is worth noting, however, that the need-based standard has at least two different justifications, and only one of them is commonly appreciated. The simplest justification rests on the claim that each individual should receive benefits proportional to her level of need—for example, that a needier student should receive more attention from her teacher because it will take more effort to help her reach a given level of competence. A different justification derives from the collective nature of many public services. In the case of a collective service like police protection, an individual who lives in a high-need area will typically find that her own access to public services gets crowded out by the heavy demands made by her neighbors. For example, if I live in a jurisdiction with a high burglary rate, and my neighbor just across city lines lives in a low-burglary jurisdiction, then I will receive less protection from burglary even if the two jurisdictions have the same number of police per capita. If my neighbor’s house is burglarized her police department will be able to devote considerable time and effort to investigating her case, but my case will receive little attention because my department has so many other burglaries to investigate. The problem is not that I have an above-average need for services that police cannot meet; I am not like the disadvantaged pupil who needs more attention from his teacher to succeed as well as his better-prepared peers. My police department simply has fewer investigative resources to devote to me when I do need its help. The same logic applies to community problem-solving, order maintenance, and preventative patrol.

  16. Alternatively, future research on the distribution of police protection might investigate other analyses of local policy choices. The most influential alternative to the dominant Tieboutian tradition focuses on the internal political dynamics of individual jurisdictions rather than competition among local governments for residents (Stone 2004), and although it is difficult to study those dynamics in large-n quantitative studies like this one, case study research could analyze how regional patterns in the distribution of police protection arise out of the internal politics of individual jurisdictions (including the way each unit of government mimics, reacts to, and exchanges ideas with its neighbors [Leicht and Jenkins 2007: 71–72]).

References

  • Anderson E (1990) Streetwise. University Chicago Press, Chicago

    Google Scholar 

  • Anderson E (1999a) Code of the street. W.W. Norton, New York

    Google Scholar 

  • Anderson ES (1999b) What is the point of equality? Ethics 109:287–337

    Article  Google Scholar 

  • Anderson ES (2007) Fair opportunity in education: a democratic equality perspective. Ethics 117:595–622

    Article  Google Scholar 

  • Banzhaf HS, Randall W (2008) Do people vote with their feet? An empirical test of Tiebout. Am Econ Rev 98:843–863

    Article  Google Scholar 

  • Basolo V (2000) City spending on economic development vs. affordable housing. J Urban Aff 22:317–332

    Article  Google Scholar 

  • Bayley D, Shearing C (2001). The new structure of policing. National Institute of Justice Research Report, NCJ 187083

  • Berk R (1991) Toward a methodology for mere mortals. Sociol Methodol 21:315–324

    Article  Google Scholar 

  • Cottrol R, Diamond R (1991) The second amendment: toward an afro-americanist reconsideration. Georgetown Law J 80:309–361

    Google Scholar 

  • Coulter P (1980) Measuring the inequity of urban public services. Policy Stud J 8:683–698

    Article  Google Scholar 

  • Dowding K, John P, Biggs S (1994) Tiebout: a survey of the empirical literature. Urban Stud 31:767–797

    Article  Google Scholar 

  • Dworkin R (2000) Sovereign virtue. Harvard University Press, Cambridge

    Google Scholar 

  • Fischel W (1989) Did Serrano cause proposition 13? Nat Tax J 42:465–473

    Google Scholar 

  • Fischel W (2009) Making the grade: the economic evolution of American school districts. University Chicago Press, Chicago

    Google Scholar 

  • Freedman D (1991) Statistical models and shoe leather. Sociol Methodol 21:291–313

    Article  Google Scholar 

  • Frug G (1998) City services. NY Univ Law Rev 73:23

    Google Scholar 

  • Hamilton B (1975) Zoning and property taxation in a system of local governments. Urban Stud 12:205–211

    Article  Google Scholar 

  • Harding D (2009) Violence, older peers, and the socialization of adolescent boys in disadvantaged neighborhoods. Am Sociol Rev 74:445–464

    Article  Google Scholar 

  • Hickman M, Reaves B (2006) Sheriff’s Offices, 2003, Bureau of Justice Statistics Report NCJ 211361

  • Hines J, Thaler R (1995) The flypaper effect. J Econ Perspect 9:217–226

    Google Scholar 

  • Hochschild J (1995) Facing up to the American dream. Princeton University Press, Princeton

    Google Scholar 

  • Hochschild J, Scovronick N (2003) The American dream and the public schools. Oxford University Press, New York

    Google Scholar 

  • Holcombe R, Asghar Z (1981) The determinants of federal grants. South Econ J 48:393–399

    Article  Google Scholar 

  • Hoxby C (2000) Does competition among public schools benefit students and taxpayers? Am Econ Rev 90:1209–1238

    Article  Google Scholar 

  • Hughes K (2006) Justice expenditure and employment in the United States, 2003, U.S. Department of Justice, Bureau of Justice Statistics Bulletin. (NCJ 212260)

  • Jones J (2005) Confidence in Local Police Drops to 10-Year Low, Gallup News Services. http://www.gallup.com/poll/19783/confidence-local-police-drops-10year-low.aspx (retrieved 9/22/2010)

  • Kakwani N, Wagstaff A, van Doorslaer E (1997) Socioeconomic inequalities in health: measurement, computation, and statistical inference. J Econometr 17:87–103

    Article  Google Scholar 

  • Kennedy D (1994) Can we keep guns away from kids. Am Prospect 5:74–80

    Google Scholar 

  • Kennedy R (1997) Race, crime, and the law. Vintage, New York

    Google Scholar 

  • Kleiman M (1992) Against excess: drug policy for results. Basic, New York

    Google Scholar 

  • Levitt S (1999) The changing relationship between income and crime victimization. Federal Reserve Bank NY Econ Policy Rev 5:87–98

    Google Scholar 

  • Levy F, Wildavsky A, Meltsner A (1974) Urban Outcomes. University of California Press, Berkeley

    Google Scholar 

  • Lindgren S, Zawitz M (2001). Linking uniform crime reporting data to other datasets, Bureau of Justice Statistics Technical Report, NCJ 185233

  • Lineberry R (1977) Equality and urban policy: the distribution of municipal public services. Sage, Beverly Hills

    Google Scholar 

  • Lobao L, Hooks G, Tickamyer A (2007) The sociology of spatial inequality. SUNY Press, Albany

    Google Scholar 

  • Maguire E (2001) Research evidence on the factors influencing police strength in the United States. In: Christopher K, Ed Maguire, Gretchen M (eds) A national study of police hiring and retention issues in police agencies. Report submitted to the National Institute of Justice

  • Maguire E, Snipes JB, Uchida C, Townsend M (1998) Counting cops: estimating the number of police officers and police agencies in the United States. Policing Int J Police Strat Manag 21(1):97–120

    Article  Google Scholar 

  • Maguire K et al (2005) Sourcebook of criminal justice statistics, 2003. USGPO, Washington

    Google Scholar 

  • Maltz M (1999) Bridging gaps in police crime data, a discussion paper from the BJS Fellows Program, Bureau of Justice Statistics report NCJ 176365

  • Mladenka K, Hill K (1978) The distribution of urban police services. J Polit 40:13–133

    Article  Google Scholar 

  • Moore M (1992) Problem-solving and community policing. Crime Justice 15:99–158

    Article  Google Scholar 

  • Murray S, Evans W, Schwab R (1998) Education finance reform and the distribution of education resources. Am Econ Rev 88:789–812

    Google Scholar 

  • National Advisory Commission on Civil Disorders (1968) Report of the national advisory commission on civil disorders. Bantam Books, New York

    Google Scholar 

  • Ostrom E (1983) Equity and police services. In: Whitaker G, Phillips CD (eds) Evaluating the performance of criminal justice agencies. Sage, Thousand Oaks

    Google Scholar 

  • Ostrom E, Whitaker G (1973) Does local community control of police make a difference? Some preliminary findings. Am J Polit Sci 17:48–76

    Article  Google Scholar 

  • Peterson P (1981) City limits. University of Chicago Press, Chicago

    Google Scholar 

  • Rae D (1981) Equalities. Harvard University Press, Cambridge

    Google Scholar 

  • Rich R (1982) Analyzing urban service distributions. Lexington Books, Lexington

    Google Scholar 

  • Roth J et al (2000) National evaluation of the COPS program. Urban Institute, Washington

    Google Scholar 

  • Rothstein J (2007) Does competition among public schools benefit students and taxpayers? Am Econ Rev 97:2026–2037

    Article  Google Scholar 

  • Sklansky D (1999) The private police. UCLA Law Rev 46:1165–1287

    Google Scholar 

  • Skogan W, Fridell L (2004) Fairness and effectiveness in policing. National Academies Press, Washington

    Google Scholar 

  • Smith S (2007) Lone pursuit: distrust and defensive individualism among the black poor. Russell Sage Foundation, New York

    Google Scholar 

  • Snyder T (2008) Digest of education statistics 2007. National Center for Education Statistics, Washington

    Google Scholar 

  • Stark K, Zasloff J (2002) Tiebout and tax revolts. UCLA Law Rev 50:801–858

    Google Scholar 

  • Stone CN (2004) It’s more than the economy after all: continuing the debate about urban regimes. J Urban Aff 26:1–19

    Google Scholar 

  • Thacher D (2004) The Rich Get Richer and the Poor Get Robbed: inequality in US criminal. Victimization, 1974–2000. J Quantitat Criminol 20:89–116

    Article  Google Scholar 

  • Thacher D (2005) The local role in homeland security. Law Soc Rev 39:635–676

    Article  Google Scholar 

  • Tiebout C (1956) A pure theory of local expenditures. J Polit Econ 64:416–424

    Article  Google Scholar 

  • U.S. Dept of Justice (1977) Criminal victimization in the U.S., 1975. USGPO, Washington

    Google Scholar 

  • U.S. Dept of Justice (2010) Criminal victimization in the U.S., 2007. USGPO, Washington

    Google Scholar 

  • Walzer M (1973) In defense of equality. Dissent 20:399–408

    Google Scholar 

  • Weicher J (1971) The allocation of police protection by income class. Urban Stud 8:207–220

    Article  Google Scholar 

  • Zald M (1991) Sociology as a discipline: quasi-science, quasi-humanities. Am Sociol 22:165–187

    Article  Google Scholar 

Download references

Acknowledgments

I would like to thank Frank Levy for helpful discussion; the editors and two anonymous reviewers at JQC for very helpful comments; and Terry Adams, Gary Bauer, Galo Falchettore, Justin McCrary, and Grace York for providing data.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to David Thacher.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Thacher, D. The Distribution of Police Protection. J Quant Criminol 27, 275–298 (2011). https://doi.org/10.1007/s10940-010-9125-3

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10940-010-9125-3

Keywords

Navigation