How Cohorts Changed Crime Rates, 1980–2016

Abstract

Objective

Identify the effect of differences in criminal activity among birth cohorts on crime rates over time. Determine the extent to which cohort effects are responsible for nationwide crime reductions of the last thirty years.

Methods

Use a panel of state age-arrest data and frequently used economic, social, and criminal justice system covariates to estimate a proxy or characteristic function for current period effects. Combine these results with national age-arrest data to estimate nationwide age, current period, and birth cohort effects on crime rates for 1980–2016.

Results

Criminal activity steadily declined between the 1916 and 1945 birth cohorts. It increased among Baby Boomers and Generation X, then dropped rapidly among Millennials, born after 1985. The pattern was similar for all index crimes. Period effects were mostly responsible for the late 1980s crack boom and the 1990s crime drop, but age and cohort effects were primarily responsible for crime rate reductions after 2000. In general, birth cohort and current period effects are about equally important in determining crime rates.

Conclusions

Policies aimed at reducing delinquency among young children may be more effective in the long run than current policies aimed at incapacitation, deterrence, and opportunity reduction.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3

Notes

  1. 1.

    In the same vein, O’Brien (2019) supplemented his (fundamentally cohort-proxy) analysis of U.S. murder rates by assuming a unimodal age curve. These assumptions are perfectly sensible; as shown below, they are also consistent with the results presented in the text. In the absence of an alternative solution, however, they would remain unverifiable. This may be less of a problem for other dependent variables or in other fields of study.

  2. 2.

    This was demonstrated by Frisch and Waugh (1933), and forms the basis of the well-known Frisch-Waugh-Lovell theorem (Lovell 2008). A proof is provided in the Appendix.

  3. 3.

    Arrest counts were assigned to one-year bins for all age groups where BJS provides data in two- to five-year bins so as to minimize the sum of squared annual rate changes, subject to the constraint that the sum of all arrest counts remain the same as the corresponding bin in the original data. This produces almost identical results to the method of differences (Steffensmeier et al. 1992) or B-splines (McNeil et al. 1977), but preserves the sum of arrest counts. The lack of anomalies in the empirical data suggests that the minimization function is reasonable. Further details are available from the author.

  4. 4.

    Sampling decisions were guided by a “90/90 rule”: If 90% of the population were covered by law enforcement agencies reporting age-arrest statistics in 90% of years (that is, 39 of 43 years), then it could be included in the sample. A few adjustments were made in the sampling process. Delaware was not included because 27% of all arrests were attributed to the Delaware State Police, which does not cover a unique population and did not report reliably. Hawaii was not included because it was unlikely to be representative of the rest of the country. Maine was chosen instead of Rhode Island because 90% of its population was always covered. Iowa was added to improve geographic coverage. Wyoming was combined with Idaho. Thus panel size N = 15.

  5. 5.

    Feasible generalized least squares is more accurate than ordinary least squares in this context, but the standard errors are systematically too small. The bootstrap is necessary to cure this problem (Moundigbaye et al. 2018). An earlier version combining OLS with Driscoll-Kraay standard errors (which are robust to heteroskedasticity, spatial dependence, and serial correlation) produced very similar coefficients, trends, and standard errors.

  6. 6.

    Differences in twice-lagged prisons and in state government spending were positively correlated with murder, rape, and assault differences. Gubernatorial elections were valid for all crimes, but were too weak a predictor of prison rates to be useful.

  7. 7.

    To reduce the problem of capitalizing on chance, Bayesian model averaging was used in the first stage of the two-stage instrumental variables process. For all crime types, first-stage F > 10.0 with average F = 13.323; average overidentification F = 1.353 and p(F) > .10 for all crime types. In addition to demonstrating the validity of the instruments, the overidentification test also shows directly that the covariates are not significantly associated with age or cohort effects. Per Winship and Harding (2008), we could enter these covariates into an APC model in place of period effects and expect to get valid results.

  8. 8.

    It may be possible to avoid the omitted variable analysis if the covariates in Eq. (8)  account for enough of the variance in Δpt that the remainder can be chalked up to random errors. One measure of this is to compare R2 in Eq. (1) to R2 when the Δpt are replaced by their predicted values from Eq. (8). In this case the difference is significant for all crime types, indicating that some covariates have in fact been omitted. My thanks to an anonymous reviewer for the suggestion.

  9. 9.

    Specifically, the average values of κ2 for seven index crimes over the 1980–2015 time period for Australia, Canada, Denmark, Finland, Germany, Iceland, New Zealand, Norway, Sweden, and Switzerland are highly correlated with one another (average inter-country r =  + .831, Cronbach’s α =  + .980). The mean κ2 of the ten countries is correlated with U.S. mean α over the same time period with r =  + .896.

  10. 10.

    Age effects are often controlled for by including the proportion of the population in crime-prone ages (usually 18–24) as a covariate. The importance of age effects suggests that more such covariates – and perhaps more specific ones – should be used as controls.

References

  1. Aaltonen M, MacDonald JM, Martikainen P, Kivivuori J (2013) Examining the generality of the unemployment-crime association. Criminology 51:561–594

    Article  Google Scholar 

  2. Ahrnsbrak R, Bose J, Hedden SL, Lipari RN, Park-Lee E (2017) Key substance use and mental health indicators in the United States: results from the 2016 national survey on drug use and health. Substance Abuse and Mental Health Services Administration, Rockville, MD

    Google Scholar 

  3. Altonji JG, Elder TE, Taber CR (2005) Selection on observed and unobserved variables: assessing the effectiveness of Catholic schools. J Polit Econ 113:151–184

    Article  Google Scholar 

  4. Baltes PB (1968) Longitudinal and cross-sectional sequences in the study of age and generation effects. Hum Dev 11:145–171

    Article  Google Scholar 

  5. Baumer EP, Lauritsen JL (2010) Reporting crime to the police, 1973–2005: a multivariate analysis of long-term trends in the national crime survey (NCS) and the national crime victimization survey (NCVS). Criminology 48:131–185

    Article  Google Scholar 

  6. Blumstein A (2005) Disaggregating the violence trends. In: Blumstein A, Wallman J (eds) The crime drop in America. Cambridge University Press, New York, pp 13–44

    Chapter  Google Scholar 

  7. Blumstein A, Rosenfeld R (2008) Factors contributing to U.S. crime trends. In: Goldberger AS, Rosenfeld R (eds) Understanding crime trends: workshop report. National Academies Press, Washington, DC, pp 13–43

    Google Scholar 

  8. Blumstein, A., Cohen, J., and Hsieh, P. (1982). The Duration of Adult Criminal Careers. Final report to the National Institute of Justice, Carnegie-Mellon University.

  9. Blumstein A, Cohen J, Piquero AR, Visher CA (2010) Linking the crime and arrest processes to measure variations in individual risk per crime (Q). J Quant Criminol 26:533–548

    Article  Google Scholar 

  10. Bonczar TP, Hughes TA, Wilson DJ, Ditton PM (2011) National corrections reporting program: most serious offense of state prisoners, by offense, admission type, age, sex, race, and Hispanic origin. Data archive. U.S, Bureau of Justice Statistics, Washington, DC

    Google Scholar 

  11. Braga AA, Papachristos AV, Hureau DM (2012) The effects of hot spots policing on crime: an updated systematic review and meta-analysis. Justice Q 31:633–663

    Article  Google Scholar 

  12. Brush J (2007) Does income inequality lead to more crime? a comparison of cross-sectional and time-series analyses of United States counties. Econ Lett 96:264–268

    Article  Google Scholar 

  13. Bullock K, Tilley N (2003) The role of research and analysis: lessons from the crime reduction programme. Crime Prevent Stud 15:147–181

    Google Scholar 

  14. Clayton D, Schifflers E (1987a) Models for temporal variation in cancer rates. I: age-period and age-cohort models. Stat Med 6:449–467

    Article  Google Scholar 

  15. Clayton D, Schifflers E (1987b) Models for temporal variation in cancer rates. II: age-period-cohort models. Stat Med 6:469–481

    Article  Google Scholar 

  16. Cohen LE, Felson M (1979) Social change and crime rate trends: a routine activity approach. Am Sociol Rev 44:588–608

    Article  Google Scholar 

  17. Collishaw S, Maughan B, Goodman R, Pickles A (2004) Time trends in adolescent mental health. J Child Psychol Psychiatry 45:1350–1362

    Article  Google Scholar 

  18. Day, N. E. and Charnay, B. (1982). Time trends, cohort effects, and aging as influence on cancer incidence. In Magnus, K. (ed.), Trends in Cancer Incidence: Causes and Practical Implications, Hemisphere, New York.

  19. Deaton, A.S. and Paxson, C.H. (1994). Saving, growth, and aging in Taiwan. In Wise, D.A., Studies in the Economics of Aging, Chicago University Press, Chicago, IL.

  20. Donohue JJ III, Levitt SD (2001) The impact of legalized abortion on crime. Quart J Econ 116:379–420

    Article  Google Scholar 

  21. Donohue, J. and Sigelman, P. (1998). Allocating resources among prisons and social programs in the battle against crime, Faculty scholarship series, paper 51, Yale Law School.

  22. Donohue, J.J., Aneija, A., and Weber, K.D. (2017). Right-to-carry laws and violent crime: A comprehensive assessment using panel data and a state-level synthetic controls analysis. NBER working paper 23510, National Bureau of Economic Research.

  23. Durose MR, Cooper AD, Snyder HN (2014) Recidivism of prisoners released in 30 States in 2005: patterns from 2005 to 2010. U.S, Bureau of Justice Statistics, Washington, DC

    Google Scholar 

  24. Easterlin RA (1978) What will 1984 be like? socioeconomic implications of recent twists in age structure. Demography 15:397–432

    Article  Google Scholar 

  25. Farkas G (1977) Cohort, age, and period effects upon the employment of white females: evidence for 1957–1968. Demography 13:33–42

    Article  Google Scholar 

  26. Farrington DP (1986) Age and crime. Crime Justice 7:189–250

    Article  Google Scholar 

  27. Fosse E, Winship C (2019) Bounding analyses of age-period-cohort effects. Demography 56:1975–2004

    Article  Google Scholar 

  28. Frisch R, Waugh FV (1933) Partial time regressions as compared with individual trends. Econometrica 1:387–401

    Article  Google Scholar 

  29. Fu WJ (2008) A smoothing cohort model in age-period-cohort analysis with applications to homicide arrest rates and lung cancer mortality rates. Sociol Methods Res 36:327–361

    Article  Google Scholar 

  30. Fussell E (2011) The deportation theft dynamic and victimization of Latino migrants: wage theft and robbery. Sociol Q 52:593–615

    Article  Google Scholar 

  31. García-Belmonte L, Ventosa-Santalaurìa D (2011) Spurious regression and lurking variables. Statist Probab Lett 81:2004–2010

    Article  Google Scholar 

  32. Goldstein H (1990) Problem-oriented policing. Temple University Press, Philadelphia, PA

    Google Scholar 

  33. Gottfredson MR, Hirschi T (1990) A general theory of crime. Stanford University Press, Redwood City, CA

    Google Scholar 

  34. Gould ED, Weinberg BA, Mustard DB (2002) Crime rates and local labor market opportunities in the United States: 1979–1997. Rev Econ Stat 84:45–61

    Article  Google Scholar 

  35. Greenberg DF, Larkin NJ (1985) Age-cohort analysis of arrest rates. J Quant Criminol 1:227–240

    Article  Google Scholar 

  36. Grogger J (1998) Market wages and youth crime. J Law Econ 16:756–791

    Google Scholar 

  37. Harper DW Jr (2001) Comparing tourists crime victimization. Ann Tour Res 28:1053–1056

    Article  Google Scholar 

  38. Harrell FE Jr (2001) Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. Springer, New York

    Book  Google Scholar 

  39. Heckman J, Robb R (1985) Using longitudinal data to estimate age, period, and cohort effects in earnings equations. In: Mason WM, Fienberg SE (eds) Cohort analysis in social research: beyond the identification problem. Springer, New York, pp 137–150

    Chapter  Google Scholar 

  40. Hemenway D, Miller M (2000) Firearm availability and homicide rates across 26 high-income countries. J Trauma Inj Infect Crit Care 49:985–988

    Article  Google Scholar 

  41. Holford TR (1983) The estimation of age, period, and cohort effects for vital rates. Biometrics 39:311–324

    Article  Google Scholar 

  42. Hughes N, Williams WH, Chitsabesa P, Walesby RC, Mounce LTA, Clasby B (2015) The prevalence of traumatic brain injury among young offenders in custody: a systematic review. J Head Trauma Rehabilit 30:94–105

    Article  Google Scholar 

  43. Johnson R, Raphael S (2012) How much crime reduction does the marginal prisoner buy? J Law Econ 55:275–310

    Article  Google Scholar 

  44. Jones J, Schmitt J, Wilson V (2018) Fifty years after the kerner commission: african americans are better off in many ways but still disadvantaged by racial inequality. Economic Policy Institute, Washington, DC

    Google Scholar 

  45. Kazemian L, Farrington DP (2006) Exploring residual career length and residual number of offenses for two generations of repeat offenders. J Res Crime Delinq 43:89–113

    Article  Google Scholar 

  46. Kothari SP, Warner JB (2002) Evaluating mutual fund performance. J Financ 56:1985–2010

    Article  Google Scholar 

  47. Kupper LL, Janis JM, Salama IA, Yoshizawa CN, Greenberg BG, Winsborough HH (1983) Age-period-cohort analysis: An illustration of the problems in assessing interaction in one observation per cell data. Commun Stat Theory Method 12:201–217

    Article  Google Scholar 

  48. Levitt SD (1996) The effect of prison population size on crime rates: evidence from prison overcrowding legislation. Quart J Econ 111:319–351

    Article  Google Scholar 

  49. Light MT, Miller T (2018) Does undocumented immigration increase violent crime? Criminology 56:370–401

    Article  Google Scholar 

  50. Liu J, Wuerker A (2005) Biosocial bases of aggressive and violent behavior – implications for nursing studies. Int J Nurs Stud 42:229–241

    Article  Google Scholar 

  51. Lovell M (2008) A simple proof of the FWL theorem. J Econ Educ 39:88–91

    Article  Google Scholar 

  52. MacDonald J, Fagan J, Geller A (2016) The effect of local police surges on crime and arrests in New York City. PLoS ONE 11(6):e0157223

    Article  Google Scholar 

  53. Maguire M, John T (2006) Intelligence led policing, managerialism and community engagement: competing priorities and the role of the National Intelligence Model in the U.K. Polic Soc 16:67–85

    Article  Google Scholar 

  54. Marvell TB, Moody CE (1996) Specification problems, police levels, and crime rates. Criminology 34:609–646

    Article  Google Scholar 

  55. Mason WM, Fienberg SE (1985) Introduction: beyond the identification problem. In: Mason WM, Fienberg SE (eds) Cohort analysis in social research: beyond the identification problem. Springer, New York, pp 1–8

    Chapter  Google Scholar 

  56. Mason KO, Mason WM, Winsborough HH, Poole WK (1973) Some methodological issues in cohort analysis of archival data. Am Sociol Rev 38:242–258

    Article  Google Scholar 

  57. Maxim PS (1985) Cohort size and juvenile delinquency: a test of the Easterlin hypothesis. Soc Forces 63:661–681

    Article  Google Scholar 

  58. McCall PL, Land KC (2004) Trends in environmental lead exposure and troubled youth, 1960–1995: An age-period-cohort-characteristic analysis. Soc Sci Res 33:339–359

    Article  Google Scholar 

  59. McNeil DR, Trussell TJ, Turner JC (1977) Spline interpolation of demographic data. Demography 14:245–252

    Article  Google Scholar 

  60. Moundigbaye M, Rea WS, Reed WR (2018) Which panel data estimator should I use? A Corrigendum Exten Econom 12(2018–4):1–32

    Google Scholar 

  61. National Research Council (2014) The Growth of Incarceration in the United States: exploring causes and consequences. National Academies Press, Washington, DC

    Google Scholar 

  62. Nevin R (2007) Understanding international crime trends: the legacy of preschool lead exposure. Environ Res 104:315–336

    Article  Google Scholar 

  63. Nixon, R.M. 1971. Special message to the congress on drug abuse Prevention and Control, 17 June 1971.

  64. O’Brien RM (1989) Relative cohort size and age-specific crime rates: an age-period-relative-cohort-size model. Criminology 27:57–78

    Article  Google Scholar 

  65. O’Brien RM (2015) Age-period-cohort models: approaches and analyses with aggregate data. Taylor and Francis, Boca Raton, FL

    Google Scholar 

  66. O’Brien RM (2017) Mixed models, linear dependency, and identification in age-period-cohort models. Stat Med 36:2590–2600

    Article  Google Scholar 

  67. O’Brien RM (2019) Homicide arrest rate trends in the United States: the contributions of periods and cohorts (1965–2015). J Quant Criminol 35:211–236

    Article  Google Scholar 

  68. O’Brien RM, Stockard J, Isaacson L (1999) The enduring effects of cohort characteristics on age-specific homicide rates, 1960–1995. Am J Sociol 104:1061–1095

    Article  Google Scholar 

  69. Orrick EA, Piquero AR (2015) Were cell phones associated with lower crime in the 1990s and 2000s? Journal of Crime and Justice 38:222–234

    Article  Google Scholar 

  70. Osmond C, Gardner MJ (1982) Age, period, and cohort models applied to cancer mortality rates. Stat Med 1:245–259

    Article  Google Scholar 

  71. Oster E (2019) Unobservable selection and coefficient stability: Theory and validation. J Bus Econom Statist 37:187–204

    Article  Google Scholar 

  72. Peterson RD, Krivo L (2010) Divergent social worlds: neighborhood crime and the racial-spatial divide. Russell Sage, New York

    Google Scholar 

  73. Piquero AR, Farrington DP, Blumstein A (2003) The criminal career paradigm. Crime Justice 30:359–506

    Article  Google Scholar 

  74. Pullum TW (1977) Parametrizing age, period, and cohort effects: An application to U.S. delinquency rates, 1964–1973. Sociol Methodol 9:11–140

    Google Scholar 

  75. Puzzanchera, C. and Kang, W. (2017). Easy access to FBI arrest statistics, 1994–2014, Office of Juvenile Justice and Delinquency Prevention, Washington, DC. Online. Available at http://www.ojjdp.gov/ojstatbb/ezaucr/.

  76. Ranson M (2014) Crime, weather, and climate change. J Environ Econ Manag 67:274–302

    Article  Google Scholar 

  77. Raphael S, Winter-Ebmer R (2001) Identifying the effect of unemployment on crime. J Law Econ 44:259–283

    Article  Google Scholar 

  78. Rosenfeld R, Fornango R (2007) The impact of economic conditions on robbery and property crime: The role of consumer sentiment. Criminology 45:735–769

    Article  Google Scholar 

  79. Rosenfeld R, Messner SR (2009) The crime drop in comparative perspective: the impact of the economy and imprisonment on American and European burglary rates. Br J Sociol 60:445–471

    Article  Google Scholar 

  80. Savolainen J (2000) Relative cohort size and age-specific arrest rates: a conditional interpretation of the Easterlin effect. Criminology 38:117–136

    Article  Google Scholar 

  81. Shaw CR, McKay HD (1942) Juvenile delinquency and urban areas: a study of rates of delinquency in relation to different characteristics of local communities in american cities. University of Chicago Press, Chicago

    Google Scholar 

  82. Smith MD (1986) The era of increased violence in the United States: age, period, or cohort effect? Sociol Q 27:239–251

    Article  Google Scholar 

  83. Snyder, H.N., Cooper, A.D., and Mulako-Wangota, J. (2017). Arrest data analysis tool, bureau of justice statistics, Washington, DC. Online. Available at https://www.bjs.gov/index.cfm?ty=datool&surl=/arrests/index.cfm.

  84. Steffensmeier D, Streifel C, Shihadeh ES (1992) Cohort size and arrest rates over the life course: the Easterlin hypothesis reconsidered. Am Sociol Rev 57:306–314

    Article  Google Scholar 

  85. Stolzenberg L, D’Alessio SJ (2008) Co-offending and the age-crime curve. J Res Crime Delinq 45:65–86

    Article  Google Scholar 

  86. Stone DM, Simon TR, Fowler KA, Kegler SP, Yuan K, Holland KM, Ivey-Stephenson AZ, Crosby AE (2018) Trends in state suicide rates – United States, 1999–2016 and circumstances contributing to suicide – 27 states, 2015. Morb Mortal Wkly Rep 67:617–624

    Article  Google Scholar 

  87. Tseloni A, Thompson R, Grove L, Tilley N, Farrell G (2017) The effectiveness of burglary security devices. Secur J 30:646–664

    Article  Google Scholar 

  88. van Mastrigt SB, Farrington DP (2009) Co-offending, age, gender, and crime type: implications for criminal justice policy. Br J Criminol 49:552–573

    Article  Google Scholar 

  89. van Ours JC, Vollaard B (2014) The engine immobiliser: A non-Starter for auto thieves. Econom J 126:1264–1291

    Google Scholar 

  90. Weisburd D, Telep CW, Hinkle JC, Eck JE (2010) Is problem-oriented policing effective in reducing crime and disorder? Criminol Public Policy 9:139–172

    Article  Google Scholar 

  91. Werb D, Rowell G, Guyatt G, Kerr T, Montaner J, Wood E (2011) Effect of drug law enforcement on drug market violence: A systematic review. Int J Drug Policy 22:87–94

    Article  Google Scholar 

  92. White HR, Gorman DM (2000) Dynamics of the drug-crime relationship. In: LaFree G (ed) The nature of crime: continuity and change, criminal justice, vol I. Washington. DC, National Institute of Justice, pp 151–218

    Google Scholar 

  93. Wilkins, LT (1960). Delinquent Generations, Home Office Research Unit no. 7, Her Majesty’s Stationery Office, London, UK.

  94. Wilson JQ, Boland B (1978) The effect of police on crime. Law Soc Rev 12:367–390

    Article  Google Scholar 

  95. Winship C, Harding DJ (2008) A mechanism-based approach to the identification of age-period-cohort models. Sociol Methods Res 36:362–401

    Article  Google Scholar 

  96. Wright BRE, Caspi A, Moffitt TE, Miech RA, Silva PA (1999) Reconsidering the relationship between SES and delinquency: causation but not correlation. Criminology 37:175–194

    Article  Google Scholar 

  97. Yang Y, Land KC (2006) A mixed models approach to the age-period-cohort analysis of repeated cross-section surveys, with an application to data on trends in verbal test scores. Sociol Methodol 36:75–97

    Article  Google Scholar 

  98. Yang Y, Schulhofer-Wohl S, Fu WJ, Land KC (2008) The intrinsic estimator for age-period-cohort analysis: what it is and how to use it. Am J Sociol 113:1697–1736

    Article  Google Scholar 

Download references

Acknowledgements

My thanks to Jonathan Davis for suggesting the omitted variable analysis, and to two anonymous reviewers for a wide variety of insightful suggestions. This paper is much improved due to their assistance. Research support was provided by the Policy Research Institute, University of Texas at Austin.

Author information

Affiliations

Authors

Corresponding author

Correspondence to William Spelman.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix. Derivation of Level Estimates of π

Appendix. Derivation of Level Estimates of π

Estimates of π are relatively straightforward for regression in differences. The intercept is the expected change in P if there is no change in the X variables, almost a definition of a secular trend. For the age-arrest data set, there is considerable serial correlation in age, period, and cohort deviates, more or less requiring use of a differenced model. This will not always be true, however, so it is worth considering how to estimate π from a levels regression.

Let us assume that period effect Pt is a function of j causal variables Xjt, such that

$$P_{t} = \, \alpha \, + \, \Sigma_{j} \beta_{j} X_{jt} + \varepsilon_{t} .$$
(10)

Pt is presumed to be stationary about a deterministic trend, such that Pt = πt + pt. Predictor variables Xjt are likewise stationary about deterministic trends, Xjt = δj t + xjt. Pt and π are unknown, but we may reasonably assume that π is a weighted average of the predictor trends,

$${\uppi } = \, \Sigma_{j} \beta_{j} \delta_{j} .$$
(11)

Equation (A.1) cannot be estimated directly, but we can use Eq. (A.2) to estimate π if the βj can be obtained through some alternative procedure. Since pt is identified and Eq. (A-1) can be rearranged as

$$p_{t} = \, \alpha \, + \, \Sigma_{j} \beta_{j} X_{jt} {-}{\uppi }t + \, \varepsilon_{t} ,$$
(12)

we also may be able to save a step and obtain standard errors directly.

A complete proof involves j + 1 variables and requires matrix algebra, but it is not difficult to prove out Eq. (A.3) for the case of one predictor. We can generalize from there.

One Causal Predictor

Assume pt depends only on one causal variable. Then Eq. (A-3) may be estimated as

$$p_{t} = a + b \, X_{t} + ct + e_{t} .$$
(13)

Let us solve first for b, then for c.

Solve for b. As usual for regressions with two independent variables, the least squares estimate of coefficient b can be obtained from

$$b = \, \left[ {\Sigma t^{2} \Sigma Xp{-} \, \Sigma Xt\;\Sigma tp} \right] \, / \, \left[ {\Sigma X^{2} \Sigma t^{2} {-} \, \left( {\Sigma Xt} \right)^{2} } \right]$$
(14)

where, for example, ΣXt refers to the sum from 1 to T of (\(X\) t\(\stackrel{-}{X}\))(t\(\stackrel{-}{t}\)). Since p has been detrended, p and t are asymptotically uncorrelated. Thus Σtp → 0 and the second term in the numerator drops out. This leaves

$${\text{plim}}\;b = \Sigma t^{2} \Sigma Xp/\left[ {\Sigma X^{2} \Sigma t^{2} {-} \, \left( {\Sigma Xt} \right)^{2} } \right].$$

Breaking X into its component parts,

$$\begin{aligned} {\text{plim}}\;b & = & \, \Sigma t^{2} \Sigma \, \left( {\delta t + x} \right)p/\left[ {\Sigma \, \left( {\delta t + x} \right)^{2} \Sigma t^{2} {-} \, \left( {\Sigma \, \left( {\delta t + x} \right)t} \right)^{2} } \right] \\ & = & \, \Sigma t^{2} \Sigma \, \left( {\delta tp + xp} \right)/\left[ {\Sigma \, \left( {\delta^{2} t^{2} + \, 2\delta xt + x^{2} } \right) \, \Sigma t^{2} {-} \, \left( {\Sigma \, \left( {\delta t^{2} + xt} \right)} \right)^{2} } \right] \\ & = & \, \delta \, \Sigma t^{2} \Sigma tp + \, \Sigma t^{2} \Sigma xp/\left[ {\delta^{2} \Sigma t^{2} \Sigma t^{2} + \, 2\delta \, \Sigma xt\Sigma t^{2} + \, \Sigma x^{2} \Sigma t^{2} {-} \, \left( {\delta \, \Sigma t^{2} + \, \Sigma xt} \right)^{2} } \right]. \\ \end{aligned}$$

Since x and t are asymptotically uncorrelated, Σxt → 0, so this simplifies to

$$\begin{aligned} {\text{plim}}\;b & = & \, \Sigma t^{2} \Sigma xp/\left[ {\delta^{2} \left( {\Sigma t^{2} } \right)^{2} + \, \Sigma x^{2} \Sigma t^{2} {-} \, \delta^{2} \left( {\Sigma t^{2} } \right)^{2} } \right] \\ & = & \, \Sigma t^{2} \Sigma xp/ \, \Sigma x^{2} \Sigma t^{2} \\ & = & \, \Sigma xp/ \, \Sigma x^{2} . \\ \end{aligned}$$

Note that this is the formula for the coefficient on x in the simple regression, pt = a + bxt + et. Adding the trend term allows for consistent estimates of the effect of x, even if both variables are trending and only one has been detrended. Thus plim b = β and (in the two-variable case) π = bδ.

Solve for c. The least squares estimate of coefficient c can be obtained from

$$c = \, \left[ {\Sigma X^{2} \Sigma tp{-} \, \Sigma Xt\Sigma Xp} \right]/\left[ {\Sigma X^{2} \Sigma t^{2} {-} \, \left( {\Sigma Xt} \right)^{2} } \right].$$
(15)

As before, Σtp → 0, so the first term in the numerator drops out. This leaves

$${\text{plim }}c = \, {-} \, \Sigma Xt\Sigma Xp/\left[ {\Sigma X^{2} \Sigma t^{2} {-} \, \left( {\Sigma Xt} \right)^{2} } \right].$$

Again breaking X into its component parts,

$$\begin{aligned} {\text{plim}}\;c & = & \, {-} \, \Sigma \, \left( {\delta t + x} \right)t\Sigma \left( {\delta t + x} \right)p/\left[ {\Sigma \, \left( {\delta t + x} \right)^{2} \Sigma t^{2} {-} \, \left( {\Sigma \, \left( {\delta t + x} \right)t} \right)^{2} } \right] \\ & = & \, {-} \, \Sigma \, \left( {\delta t^{2} + xt} \right) \, \Sigma \left( {\delta tp + xp} \right)/\left[ {\Sigma \, \left( {\delta^{2} t^{2} + \, 2\delta xt + x^{2} } \right) \, \Sigma t^{2} {-} \, \left( {\Sigma \, \left( {\delta t^{2} + xt} \right)} \right)^{2} } \right] \\ & = & \, {-} \, \Sigma \, \left( {\delta t^{2} + xt} \right) \, \Sigma \left( {\delta tp + xp} \right)/\left[ {\Sigma \delta^{2} t^{2} \Sigma t^{2} + \, 2\delta \, \Sigma xt\Sigma t^{2} + \, \Sigma x^{2} \Sigma t^{2} {-} \, \left( {\delta \, \Sigma t^{2} + \, \Sigma xt} \right)} \right)^{2} ]. \\ \end{aligned}$$

Since Σxt → 0, this simplifies to

$$\begin{aligned} {\text{plim }}c & = & \, {-}\delta \, \Sigma t^{2} \Sigma xp/\left[ {\delta^{2} \left( {\Sigma t^{2} } \right)^{2} + \, \Sigma x^{2} \Sigma t^{2} {-} \, \delta^{2} \left( {\Sigma t^{2} } \right)^{2} } \right] \\ & = & \, {-}\delta \, \Sigma xp/ \, \Sigma x^{2} . \\ \end{aligned}$$

\(\begin{aligned} {\text{Because}}\;{\text{plim }}b & = & \, \Sigma xp/\Sigma x^{2} , \\ {\text{plim }}c & = & \, {-}b\delta \\ & = & \, {-}{\uppi }. \\ \end{aligned}\).

The coefficient on t is a consistent estimator of the trend in Pt.

Multiple Causal Predictors

If we have multiple causal variables, we estimate Eq. (A-3). This may be reframed as two-variable regression,

$$p_{t} = a + \, 1Z_{t} + ct + e_{t}$$
(16)

where, per Eq. (A-1),

$$\begin{aligned} Z_{t} = & \, \Sigma_{j} \beta_{j} X_{jt} \\ = & \, \Sigma_{j} \beta_{j} \left( {\delta_{i} t + x_{jt} } \right) \\ = & \, \Sigma_{j} \beta_{j} \delta_{j} t + \, \Sigma_{j} \beta_{j} x_{jt} \\ \end{aligned}$$

If we set θ = Σj βj δj, a weighted average of the causal variable trends, and zt = Σj βj xjt, a weighted average of the detrended causal variables, then Zt = θt + zt. From Eq. (A.7), b = 1 and (from the two-variable case) c = –bθ = –π.

Previous studies have shown that estimates b and c are Normal-distributed and converge at nominal rates (García-Belmonte and Ventosa-Santaulària, 2011), so standard errors and confidence intervals can be computed and interpreted in the usual way.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Spelman, W. How Cohorts Changed Crime Rates, 1980–2016. J Quant Criminol (2021). https://doi.org/10.1007/s10940-021-09508-7

Download citation

Keywords

  • Age-period-cohort models
  • Crime
  • Model identification
  • Birth cohorts
  • Primary prevention