Long-term unemployment and violent crime


This study reinvestigates the relationship between unemployment and crime, but is the first to focus explicitly on the effects of long-term unemployment on crime. A unique finding is that long-term unemployment shows a strong association with violent crime, an effect which is greater than that of total unemployment on property crime in this and most previous studies. Long-term unemployment thus identifies a marginal group for committing crime (particularly violent crime) better than total unemployment, with the duration of unemployment plausibly increasing the strain that fosters violent behaviour.

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

Fig. 1
Fig. 2
Fig. 3
Fig. 4


  1. 1.

    In the short term, Cantor and Land (1985) assumes that improved economic conditions cause a negative opportunity effect; when economic conditions improve, it increases the criminal opportunities. However, in economics this mechanism is explained as higher returns to crime and the effect is assumed to be captured by the income level and not the unemployment level.

  2. 2.

    Fougère et al. (2009) estimate effects of long-term unemployment simultaneously with effects of youth and adult unemployment and report no effect of long-term unemployment.

  3. 3.

    To the best of our knowledge, no one has used aggregated post-2000 data and only Grönqvist (2011) and Rege et al. (2009) have used post-2000 individual register data.

  4. 4.

    Saridakis (2004) and Greenberg (2001) use national time-series data to analyse the relationship between duration of unemployment and crime.

  5. 5.

    At the moment there are 290 municipalities in Sweden. The municipalities of Nykvarn and Knivsta were created in 1999 and 2003, respectively, and are therefore excluded. They are also very small.

  6. 6.

    Most missing values are in the covariates, but there are also 17 missing values in the long-term unemployment variable.

  7. 7.

    Excluding the share participating in labour market programmes does not affect the results in this study.

  8. 8.

    The unemployment measures are therefore taken from different sources, but a measure from SCB could have been used for regular unemployment too. The main reason for not using this measure is that we want a measure for total employment that is similar to that used in previous studies, and the SCB measure differs slightly in its construction. Moreover, the effect of total unemployment on property crime estimated with this measure is smaller than with the measure from AMS.

  9. 9.

    Because the standardised measures are computed from unemployment at municipal levels, we also weight with municipality population size.

  10. 10.

    We have unemployment data for 1997 and can therefore construct a first difference variable for 1998.

  11. 11.

    When calculating the within variation for the standardised measures with xtsum in STATA, the variation is 1.26-fold larger for long-term unemployment than total unemployment. With our method of calculating variation, a larger difference shows that the variation is mainly larger between years than to the mean.

  12. 12.

    Weapon theft is also excluded, because weapons may be used when committing a violent crime. However, such thefts are very few, so excluding them does not change the results.

  13. 13.

    Robbery is categorised by NCCP as both a violent and a property crime. However, robbery is included in the Appendix when specific crime categories are analysed separately.

  14. 14.

    Another reason to add the first difference is that municipality income levels are nonstationary.

  15. 15.

    Different age groups are also targets and victims of different crimes (NCCP 2008).

  16. 16.

    Thus, the short-term measure is the average yearly total unemployment rate minus the share long-term unemployed during some part of the year. The measures are therefore not totally mutually exclusive because they are measured differently, but we are confident that this does not pose problems for the analysis. That is, the constructed short-term variable should capture short-term unemployment fluctuations that do not turn into long-term unemployment fluctuations during the year.

  17. 17.

    In regressions also including indicators of migration patterns (inflow and outflow of individuals) and the divorce rate, the results are not changed.

  18. 18.

    When adding both linear and quadratic municipality-specific time trends, long-term unemployment loses its significance for violent crimes. However, the effect stays large, 1.54. This specification (adding 2 times 288 variables) probably removes too much of the variation in violent crimes to identify a significant effect.

  19. 19.

    We lack data at the municipal level for murder and violence against officials. Nevertheless, since the number of such crimes is very low, the variable has too many zeros and too little variation. Murder and manslaughter cases are also too few as the number of reported murders and manslaughter cases during most of our period has typically varied between 2 and 3 per 100,000 inhabitants (NCCP 2008).

  20. 20.

    In 2008, almost 50 % of the social benefits recipients in Sweden were registered job-seekers at the National Labour Market Board (Mörk 2011).

  21. 21.

    High IV estimates could also be caused by a weak instrument (Murray 2006), but this does not seem to be the case in, e.g., Lin (2008) and Raphael and Winter-Ebmer (2001).

  22. 22.

    Further information on data sources and the description of the construction of the instruments are provided in the Appendix.

  23. 23.

    See Gould et al. (2002) for a more profound discussion of the instrument.

  24. 24.

    For the purpose of comparison, the results of the baseline OLS regression, with the same period as in our 2SLS regression, are presented in Table 8. As can be seen, the result is similar to the baseline regression in Table 1.

  25. 25.

    By including a squared variable, we have more instrumental variables than endogenous variables. Yet, performing an overidentification test is pointless, since it is not two instruments, but rather one instrument modelled with two variables.

  26. 26.

    For a discussion on the social costs of crime, see for example, Cohen et al. (2004); Cohen 1998); Miller et al. 1996) and Boardman et al. (2011).


  1. Agell J, Öster A (2007) Crime and unemployment in turbulent times. J Eur Econ Assoc 5(4):752–775

    Article  Google Scholar 

  2. Agnew R (1992) Foundation for a general strain theory of crime and delinquency. Criminology 30:47–87

    Article  Google Scholar 

  3. Boardman AE, Greenberg DH, Vining AR, Weimer DL (2011) Cost-benefit analysis; concepts and practice, 4th edn. Pearson Education Inc, New York

    Google Scholar 

  4. Buonanno P, Montolio D, Vanin P (2009) Does social capital reduce crime? J Law Econ 52:145–170

    Article  Google Scholar 

  5. Cantor D, Land KC (1985) Unemployment and crime rates in the post World War II United States: a theoretical and empirical analysis. Am Soc Rev 50:317–332

    Article  Google Scholar 

  6. Card D, Dahl GB (2011) Family violence and football: the effect of unexpected emotional cues on violent behavior. Quart J Econ 126:103–143

    Article  Google Scholar 

  7. Chiricos TG (1987) Rates of crime and unemployment: an analysis of aggregate research evidence. Soc Probl 34:187–212

    Article  Google Scholar 

  8. Cohen MA (1998) The monetary value of saving a high risk youth. J Quant Criminol 14:5–33

    Article  Google Scholar 

  9. Cohen MA, Rust RT, Steen S, Tidd ST (2004) Willingness-to-pay for crime control programs. Criminology 42:89–108

    Article  Google Scholar 

  10. Cullen JB, Levitt SD (1999) Crime, urban flight, and the consequences for cities. Rev Econ Stat 81:159–169

    Article  Google Scholar 

  11. Edmark K (2005) Unemployment and crime: is there a connection? Scand J Econ 107:353–373

    Article  Google Scholar 

  12. Ehrlich I (1973) Participation in illegitimate activities: a theoretical and empirical investigation. J Pol Econ 81:521–565

    Article  Google Scholar 

  13. Eide E, Rubin PH, Shepherd JM (2006) Economics of crime. Found Trends Microecon 2:205–279

    Article  Google Scholar 

  14. Felson M (1998) Crime and everyday life. Pine Forge Press, Thousand Oaks

    Google Scholar 

  15. Fougère D, Kramarz F, Pouget J (2009) Youth unemployment and crime in France. J Eur Econ Ass 7:909–938

    Article  Google Scholar 

  16. Fryer R, Heaton P, Levitt S, Murphy K (2005) Measuring crack cocaine and its impact. Econ Inq 51:1651–1658

    Article  Google Scholar 

  17. Gould E, Weinberg B, Mustard D (2002) Crime rates and local labor market opportunities in the United States: 1979–1997. Rev Econ Stat 84:45–61

    Article  Google Scholar 

  18. Greenberg D (2001) Time series analysis of crime rates. J Quant Criminol 17:291–327

    Article  Google Scholar 

  19. Grogger F (1998) Market wage and youth crime. J Lab Econ 16:756–791

    Article  Google Scholar 

  20. Grönqvist H (2011) Youth unemployment and crime: new lessons exploring longitudinal register data. Working Paper 7/2011, (SOFI) Stockholm University

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

    Article  Google Scholar 

  22. Levitt SD (1997) Using electoral cycles in police hiring to estimate the effect of police on crime. Am Econ Rev 87:270–290

    Google Scholar 

  23. Levitt SD (2001) Alternative strategies for identifying the link between unemployment and crime. J Quant Criminol 17:377–390

    Article  Google Scholar 

  24. Levitt SD (2004) Understanding why crime fell in the 1990s: four factors that explain the decline and six that do not. J Econ Perspect 18:163–190

    Article  Google Scholar 

  25. Lin M-J (2008) Does unemployment increase crime? evidence from U.S. data 1974–2000. J Hum Resour 43:413–436

    Article  Google Scholar 

  26. Machin S, Meghir C (2004) Crime and economic incentives. J Hum Resour 39:958–979

    Article  Google Scholar 

  27. Miller TR, Cohen MA, Wiersema B (1996) Victim costs and consequences: a new look. NCJ 155282, National Institute of Justice, Washington DC

  28. Murray M (2006) Avoiding invalid instruments and coping with weak instruments. J Econ Perspect 20(4):111–132

    Article  Google Scholar 

  29. Mustard D (2010) How do labor markets affect crime? new evidence on an old puzzle. In: Benson B, Zimmerman P (eds) Handbook on the economics of crime. Edward Elgar Publishing, Cheltenham, pp 342–358

    Google Scholar 

  30. Mörk E (2011) Från försörjningsstöd till arbete—Hur kan vägen underlättas? Report 2011:6, Institute for Evaluation of Labour Market and Education Policies (IFAU), Uppsala

  31. NCCP (2005) Brottsligheten bland personer födda i Sverige och i utlandet. Report 2005:17, The National Council for Crime Prevention, Stockholm

  32. NCCP (2008) Brottsutvecklingen i Sverige fram till år, 2007 Brottsutvecklingen i Sverige fram till år, (2007) Report 2008:23. The National Council for Crime Prevention, Stockholm

  33. Papps K, Winkelmann R (2000) Unemployment and crime: new evidence for an old question. New Zealand Econ Pap 34:53–72

    Article  Google Scholar 

  34. Rege M, Skardhamar T, Telle K, Votruba M (2009) The effect of plant closure on crime. Statistics Norway, Research Department, Discussion Papers No 593

  35. Raphael S, Winter-Ebmer R (2001) Identifying the effect of unemployment on crime. J Law Econ 41:259–283

    Article  Google Scholar 

  36. Saridakis G (2004) Violent crime in the United States of America: a time-series analysis between 1960–2000. Eur J Law Econ 18:203–221

    Article  Google Scholar 

  37. Stock J, Wright J, Yogo M (2002) A survey of weak instruments and weak identification in generalized method of moments. J Bus Econ Stat 20:518–529

    Article  Google Scholar 

  38. Willis M (1997) The relationship between crime and jobs. Working paper, University of California-Santa Barbara

Download references


We would like to thank Roberto Galbiati, Hans Grönqvist and participants in the seminar at ESPE (2012) for helpful comments and suggestions. Research grants from the Swedish Council for Working Life and Social Research and the Health Economics Program (HEP), Lund University are gratefully acknowledged.

Author information



Corresponding author

Correspondence to Martin Nordin.



This section describes the instrument used, which was constructed by connecting the industry (or business) sectoral composition of employment at municipal level with the national growth trends in these sectors. It thus measures the expected variation (predicted by the national trend) in labour demand in the municipalities. We begin by constructing the national growth rate in employment in industry j between time \(t -1\) and time t :

$$\begin{aligned} g_{j }= (E_{jt} - E_{jt-1}) / E_{jt-1} \end{aligned}$$

where \(E_{jt }\)is the number of employed workers in industry j at time t at the national level. Then, we multiply the national growth rate, \(g_{j}\), by the municipality-specific sector composition of employment, lagged one period, \(E_{ijt-1}\):

$$\begin{aligned} \mathrm{{Predicted\,Employment}}_{it}={\sum ^{16}_{j=1}} E_{ijt-1} \times (1 + g_{j}) \end{aligned}$$

The employment data at municipal level collected by SCB are only available for the period 2000–2010. Here, we use employment data differentiated into 16 different industry sectors (Tables 7, 8, 9, 10).

Table 7 Dependent variables used in the model and descriptive statistics
Table 8 Effect of total unemployment and long-term unemployment on crime rates in Sweden, 2001–2010
Table 9 Effect of total employment and long-term unemployment on specific crime categories in Sweden, 1998–2010. \(N=3696\)
Table 10 Effect of total unemployment on specific crime categories in Sweden, 1998–2010, when municipality-specific linear time trends are added. \(N=3696\)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Nordin, M., Almén, D. Long-term unemployment and violent crime. Empir Econ 52, 1–29 (2017). https://doi.org/10.1007/s00181-016-1068-6

Download citation


  • Crime
  • Violent crime
  • Unemployment
  • Long-term unemployment

JEL Classification

  • J2
  • K14
  • K42