A re-examination of the opportunity and motivation effects underlying criminal activity

Abstract

The opportunity and motivation effects underlying criminal activity presented in Cantor and Land (1985) are revisited. It is argued that a full appreciation of these effects requires examination of the evolution of criminal activity over the business cycle rather than in relation to simple measures of unemployment, as has been considered often in the literature. Using the derived cyclical components of a number of socio-economic indicators, the empirical evidence in support of the proposed theoretical relationships is examined. It is found that in contrast to the results obtained when using unemployment, consideration of alternative cyclical indicators provides very strong support for the presence of opportunity and motivation effects of the form predicted by Cantor and Land (1985). An interesting distinction between results for property and violent crimes is noted and discussed.

Keywords

Crime economic conditions motivation effects opportunity effects unemployment

Introduction

The analysis of the potential relationships between crime and unemployment has resulted in the emergence of a huge literature spanning a range of academic disciplines (see, inter alia, Becker, 1968; Cantor and Land, 1985; Cohen and Felson, 1979; Ehrlich, 1973; Marvell and Moody, 2001; Smith, 1997), drawing upon a variety of theories including rational-choice theory (Becker, 1968; Cornish and Clarke, 1987; Ehrlich, 1973) and strain theory (see Agnew, 1992, 2001; Arvanites and DeFina, 2006; Cloward and Ohlin, 1960; Merton, 1938). In summary, a review of the development of the theoretical literature shows an emergence in the early literature of typically agreed conclusion of a positive relationship between unemployment and crime.¹ In contrast, the empirical research portrays an inconsistency in the results obtained, providing a general conclusion of inconclusiveness (Box, 1987; Freeman, 1980; Long and Witte, 1981; Tarling, 1982) which Chiricos (1987: 188) refers to as a ‘consensus of doubt’. When considering the resulting response to this uncertainty, a reading of the empirical literature gives the impression that many practitioners have sought to consider the empirical methods employed rather than focus upon the theoretical foundations underlying their analysis. For example, alternative modelling techniques have been employed by some (Carmichael and Ward, 2001; Mocan and Bali, 2010), while others have pondered upon the accuracy of data sources (MacDonald, 2000). However, a marked departure in this literature is provided by Cantor and Land (1985), hereafter referred to as CL.² The key development introduced by CL concerns the decoupling of the influence of unemployment upon crime into opportunity and motivation effects. As a result of the countervailing nature of these effects, unemployment is not viewed as having a single directional impact upon criminal activity, thereby providing a means for understanding or re-evaluating previous studies.³ More precisely, it is argued that while the opportunity effect relates to crime increasing in good times (i.e. a positive relationship between crime and economic conditions), the motivation effect relates to crime increasing in bad times (i.e. a negative relationship between crime and economic conditions). As a consequence, opportunity is viewed as pro-cyclical, while motivation is counter-cyclical.

The above pro-cyclicality and counter-cyclicality of opportunity and motivation effects prompt the current analysis.⁴ More precisely, it is argued that as unemployment has been long recognized to be a poor or lagging indicator of economic conditions, the analysis of opportunity and motivation effects needs to move beyond consideration of this variable. As a result, the current study considers the relationships between the cyclical components of alternative socio-economic indicators and crime to explore the evolution of crime over the business cycle directly. To achieve this, bivariate correlations between the created cyclical components of socio-economic indicators and alternative crime classifications are derived. It is examined then whether the opportunity and motivation effects identified from this analysis (1) exhibit the directional effects predicted by CL via the use of sign testing and (2) are significant statistically. While a similar motivation underlies the recent work of Phillips and Land (2012), the method and underlying approach to the present study differ in two crucial respects. First, Phillips and Land (2012) do not consider cyclical components. Second, in contrast to the current analysis, Phillips and Land (2012) consider unemployment alone as a socio-economic indicator. Consequently, the focus of the current analysis (the link between criminal activity and cyclical economic conditions) is not considered by Phillips and Land (2012).

Opportunity and Motivation: The Predictions of Cantor and Land (1985)

When categorizing opportunity effects, the changes in socio-economic conditions can be seen to have two complementary effects. First, as an economy booms in an economic upturn there will be associated increases in household income and consumption. This will provoke both an increase in more lucrative criminal opportunities and a rise in the number of available ‘crime targets’. Second, as working hours and the affordability of leisure activities increase, criminal opportunities will become more accessible and levels of guardianship will fall. Clearly both of these outcomes will have a positive impact upon crime. As noted by CL, theories relating to opportunity effects are well recognized in the literature requiring a conjunction of the availability of suitable targets and the absence of guardianship. This is apparent in studies including Arvanites and DeFina (2006), Cohen and Felson (1979), Felson (1994), Finkelhor and Asdigian (1996), Land and Felson (1976) and Wilcox et al. (2003). The resulting prediction is therefore that ‘opportunity’ is pro-cyclical, having a tendency to increase crime in good times and reduce it in bad times.

In contrast to the pro-cyclical nature of ‘opportunity’, ‘motivation’ is deemed to be counter-cyclical. In short, the pressure of ‘bad times’ is predicted to lead to an increase in criminal activity. As with ‘opportunity’, theories supporting or attempting to explain ‘motivation’ abound. Arguably, the most obvious candidate here is strain theory as provided in the works of Agnew (1992, 2001), Cloward and Ohlin (1960), Cohen (1955) and Merton (1938).⁵ A recurrent feature underlying this research is the gulf between the actual and expected state of affairs for individuals, with society playing a significant role in determining the latter. The argument for a negative relationship between crime and economic conditions, under the heading of ‘motivation’, originates in the desire to achieve goals or, in the words of more recent research of Messner and Rosenfeld (2001), to live ‘the American Dream’. Beyond strain theory, other theories supporting ‘motivation’ include rational choice theory (see, for example, Becker, 1968) and Marxist theories (see, for example, Hughes and Carter, 1981).

The predictions of CL can therefore be expressed as opportunity and motivation having positive and negative effects, respectively, upon crime. It is the direct examination of whether these positive and negative effects exist that prompts the current study. Importantly, the present study considers this issue within the framework specified by CL, by conducting its empirical analysis using the underlying cyclical components of socio-economic indicators. Using the approach outlined by CL, the extent to which the predicted effects are observed is considered via the calculation of bivariate correlations between these cyclical indicators and alternative classifications of crime. At this point, the distinction between correlation and causation should be noted, with the former obviously not providing evidence of the latter. However, use of correlation analysis in the present analysis is supported in two main ways. First, research such as that by Winsberg (1993) and Young (1993) has employed bivariate correlation analysis previously to explore the properties of crime, considering its similarity across regions and with unemployment respectively. Second, the series considered are changes in crime and cyclical components of various economic indicators. These series are stationary and hence not subject to the misleading or spurious association or correlation discussed in Hoover (2003), thus ensuring an increased robustness of the results obtained.⁶

While the recent research of Phillips and Land (2012) has considered a similar issue to the present article, it has done so in a fundamentally different manner. In particular, Phillips and Land (2012) consider the sign of opportunity and motivation effects via use of measures of unemployment. As such, the research adopts the exact approach avoided in the current study with the present focus on cyclical measures of broader socio-economic indicators. An extension of analysis beyond unemployment is not new in the literature, as Arvanites and DeFina (2006) and Rosenfeld and Fornago (2007) consider gross state product and consumer sentiment respectively, albeit in a different manner and in the context of property crime and robbery only. Given the identified importance of cyclical conditions in this work and the employment of cyclical components, the present analysis and that of Phillips and Land (2012) differ fundamentally and crucially in nature also.

The particular approach adopted in the current study has not been followed although Arvanites and DeFina (2006) do employ detrended series in the analysis of property crime and Cook and Zarkin (1985: 124) note that ‘secular movements dominate the cyclical movements’. However, it should be noted that Cook and Zarkin (1985) attempt to overcome this using an averaging procedure, their examination of correlation is undertaken using actual series rather than their cyclical components. As a consequence of this, the present article complements and extends previous analyses via the analysis of cyclical correlations.

Data and Method

Data

The following analysis considers a range of data on crime and economic and social indicators. All series are measured in annual observations over the period 1970 to 2009. The crime data are obtained from the Federal Bureau of Investigation via the Uniform Crime Reporting Statistics website. The particular series considered are four classifications of violent crime (murder, rape, robbery and aggravated assault) and three classifications of property crime (burglary, larceny and motor vehicle theft).⁷ The socio-economic indicators considered are unemployment (as utilized by CL and Phillips and Land, 2012) along with per capita measures of real personal disposable income, real GDP and real consumers’ expenditure.⁸

The present study employs aggregate, or national-level data. The prime motivating factor underlying the use of national-level data is the desire to replicate the level of analysis employed by CL. In this respect, the views of O’Brien (1999) and Paternoster and Bushway (2001) are shared, in that any attempt to re-examine CL should take place within their framework, rather than their approach being re-specified and criticized.⁹ Beyond this, national data are employed as many theories relate to national-level concerns (see, inter alia, Cohen and Felson, 1979; LaFree, 1998; Messner and Rosenfeld, 2001) and many prior studies have employed national data (see, inter alia, LaFree, 2005; Sutton, 2004).¹⁰ Further to this, recent research into the presence of a ‘national trend’ in crime has suggested that the similarities found in geographically disaggregated data mitigates any potential gains resulting from their use in place of national-level analysis (see Cook and Winfield, 2013; McDowall and Loftin, 2009).¹¹

Method

CL operationalize the distinction between opportunity and motivation by arguing that the former occurs within a restricted period, while the latter operates with a lag or delay. Using unemployment as a measure of socio-economic conditions, it is therefore proposed that the link in a particular period between the change in crime and the level of unemployment is a measure of opportunity. In a similar fashion, the link between the change in crime and the change in unemployment is taken as a measure of the impact of motivation upon crime. Formally, this means that the change in crime ( $Δ c r_{t}$ ) is assumed to be driven by the opportunity associated with the level of unemployment ( $u_{t}$ ) and the motivation captured by the change in unemployment ( $Δ u_{t}$ ) which can be expressed as:

Δ c r_{t} = f (u_{t}, Δ u_{t})

(1)

where $f (\cdot)$ denotes a general functional relationship.

However, given the arguments above that unemployment served as a proxy for the business cycle, ( $u_{t}$ , $Δ u_{t}$ ) are replaced by the level and change in the cyclical components of alternative socio-economic indicators (aggregate and disaggregate measures of unemployment, income, GDP and disposable income) in the present empirical analysis. These can be denoted as $c y c_{t}$ and $Δ c y c_{t}$ respectively. To derive the cyclical components of the economic indicators, the Hodrick-Prescott filter (Hodrick and Prescott, 1997) is applied.¹² The filter provides a relatively straightforward method of isolating the trend component of a series of interest. Denoting the variable under examination as $y_{t}$ and its underlying (to be derived) trend as $τ_{t}$ , the Hodrick-Prescott filter derives the trend as the solution to the following quadratic minimization problem:

m i n (\sum_{t = 1}^{T} {(y_{t} - τ_{t})}^{2} + λ \sum_{t = 2}^{T - 1} {(Δ^{2} τ_{t + 1})}^{2})

(2)

Equation (2) is therefore a two-part expression where the relative importance placed upon the two terms is determined by the smoothing parameter ( $λ$ . )The first part of the expression can be seen to relate to the goodness of fit of the proposed trend, as given by the sum of its squared deviations about the series of interest. The second part of the expression relates the smoothness of the proposed trend. As this second term is defined as the squared value of the change in the change in the trend (i.e. the second difference of the trend), it is minimized when the underlying trend is specified as a linear trend.¹³ Consideration of the extreme cases of $λ = 0$ and $λ = \infty$ illustrates the nature of the two terms. With $λ = 0$ , all weight is placed upon the first term in (2) which is then minimized by setting the trend equal to the actual series. Alternatively, with $λ = \infty$ , the resulting trend is a linear process for the reasons outlined above. In practice, neither of these extreme values will be employed as a relatively smooth trend reflecting the series is desired. In an examination of the optimal value of $λ$ . employ, Ravn and Uhlig (2002) propose a selection rule based upon the frequency of the data under examination. For the annual data under investigation here, adoption of this rule results in a value of $λ = 6.25$ . With the filter applied, the underlying cyclical components of the socio-economic indicators ( $c y c_{t}$ ) can be derived as the difference between the natural logarithmic values of the series and their Hodrick-Prescott trends.

As noted above, the focus of the present analysis is to examine whether the directional effects of opportunity and motivation are as predicted by CL. To do this, the signs of the bivariate correlations between { $Δ c r_{t} c y c_{t}$ } and { $Δ c r_{t} Δ c y c_{t}$ } are examined. In the case of unemployment-based cyclical indicators, opportunity and motivation effects are predicted to produce negative and positive correlations respectively. Conversely, these correlations are predicted to be positive and negative respectively when considering the ‘non-unemployment-based’ cyclical indicators.¹⁴

Results

The results from the empirical analysis are presented in Table 1. The analysis can be considered to contain two components. First, the results can be viewed in terms of the extent to which theoretically predicted signs for the relationships are observed in practice. Second, it can be considered in terms of the information provided on statistically significant opportunity and motivation effects.

Table 1.

Assessing opportunity and motivation effects.

Indicator	Effect	Crime classification							Sign p-value
Indicator	Effect	Murder	Rape	Robbery	Assault	Burglary	Larceny	Motor theft	Sign p-value
Unemployment	Opportunity	−0.301	−0.203	−0.327	−0.180	−0.212	−0.144	−0.259	0.09
	Opportunity	(0.06)	(0.22)	(0.04)	(0.27)	(0.19)	(0.38)	(0.11)
	Motivation	−0.009	−0.117	0.188	−0.022	0.249	0.256	−0.154
	Motivation	(0.95)	(0.48)	(0.25)	(0.89)	(0.13)	(0.12)	(0.35)
Real income	Opportunity	0.322	0.348	0.185	0.271	0.146	0.035	0.291	0.02
	Opportunity	(0.05)	(0.03)	(0.26)	(0.09)	(0.37)	(0.83)	(0.07)
	Motivation	−0.047	0.135	−0.206	0.055	−0.278	−0.319	0.084
	Motivation	(77.71)	(0.41)	(0.21)	(0.74)	(0.09)	(0.05)	(0.61)
Real GDP	Opportunity	0.380	0.335	0.294	0.245	0.188	0.035	0.292	0.01
	Opportunity	(0.02)	(0.04)	(0.07)	(0.13)	(0.25)	(0.83)	(0.07)
	Motivation	−0.064	0.174	−0.267	−0.004	−0.334	−0.389	0.070
	Motivation	(0.70)	(0.29)	(0.10)	(0.98)	(0.04)	(0.01)	(0.67)
Real consumption	Opportunity	0.369	0.328	0.157	0.267	0.059	−0.076	0.316	0.01
	Opportunity	(0.02)	(0.04)	(0.34)	(0.10)	(0.72)	(0.64)	(0.05)
	Motivation	−0.138	0.030	−0.409	−0.031	−0.514	−0.495	−0.004
	Motivation	(0.40)	(0.86)	(0.01)	(0.85)	(0.00)	(0.00)	(0.98)

Notes: The tabulated figures are calculated bivariate correlation coefficients between the series under consideration. The accompanying figures in parentheses are the associated p-values (expressed in percentage terms) for the test of whether the coefficient equals zero. Shaded cells indicate significance at the 10 per cent level. Shaded cells with bold text indicate significance at the 5 per cent level.

From inspection of the signs of the correlation coefficients for the various combinations of crime and socio-economic conditions presented in Table 1, it is clear that the theoretical predictions of CL are overwhelmingly supported. With regard to opportunity effects, the predicted sign is detected in 27 of the 28 cases considered.¹⁵ The results for motivation effects are not as compelling (18 out of 28 cases producing predicted signs), but this is in part due to the poor performance of unemployment. Given this is a noted poor indicator of economic conditions, primarily as a result of being a lagging indicator with other factors in the labour market responding to economic conditions ahead of unemployment levels altering, this is perhaps not surprising. While the other indicators generally provide more support for the predicted influence of motivation effects, it is consumption which is most supportive of CL’s predictions. To assess formally the support for CL’s prediction, the final column of the Table 1 provided two-sided probabilities for binomial testing of the coefficients’ signs.¹⁶ The presented p-values show some support for CL’s predictions when using unemployment as a socio-economic indicator (the presented p-value indicates marginal significance at the 10 per cent level), but overwhelming support using the alternative indicators, particularly GDP and consumption.

Turning to the significance of the derived correlation coefficients, results similar to those above are noted with unemployment again being the least supportive indicator of CL’s predictions while consumption is the most supportive. Furthermore, it is apparent when considering the results for the non-unemployment indicators that the most prominent supportive results for opportunity effects arise in connection with crimes that have relate to violence (murder, rape, assault), while for motivation effects the relevant crimes are those with a financial basis (robbery, burglary, larceny). The results for motor vehicle theft are interesting as despite being classified as a property crime, the evidence of opportunity effects is in keeping with violent crimes. However, that motor vehicle theft should exhibit results similar to violent crimes is not a novel finding (see Cook and Cook, 2011). In addition, previous research has highlighted underlying components in motor vehicle theft relating to financial gain and joyriding, which may have counterbalancing effects thus making interpretation of results for this classification less straightforward (see Paternoster and Bushway, 2001).

Discussion and Concluding Remarks

In this article, Cantor and Land’s (1985) theoretically predicted effects of opportunity and motivation upon the evolution of crime have been examined. The crucial feature of the present analysis was to note that opportunity and motivation effects were dependent upon economic conditions as given by the business cycle. This led to the derivation of the cyclical components of alternative socio-economic indicators including, and extending beyond, unemployment. The results obtained provided clear support for the presence of opportunity and motivation effects of the form predicted by Cantor and Land (1985). Interestingly, the most supportive results were obtained for the socio-economic indicators other than unemployment, with the use of consumption providing particularly compelling results. That unemployment should not lead to the detection of significant results is perhaps unsurprising given its acknowledged paucity and limitations as an indicator of economic conditions.

As stated, the results of the analysis provided support for the theoretical predictions of Cantor and Land (1985). An interesting feature of the findings was the distinction between the results for violent and property crimes. While the former displayed evidence of significant opportunity effects, the latter provided evidence of significant motivation effects. Considering the latter results, previous studies have presented findings of motivation effects underlying property crime (see, for example, Arvanites and DeFina, 2006) and this has a clear theoretical underpinning via strain theory. However, while Arvanites and DeFina (2006), for example, present significant results for motivation effects, the findings of the current study concerning opportunity effects in violent crime are new. However, the increased social interaction, reductions in guardianship and increased levels of alcohol consumption during upswings in the economy (or ‘good times’) are well documented. Considering the final issue in particular, numerous studies support the pro-cyclical nature of alcohol consumption (see Freeman, 1998) while others document its positive correlation with violent crime (see Cnossen, 2007; Galanter, 2002 and studies therein). As such, the changed behaviour and activities associated with upswings, in contrast to downturns, in the economy provide clear mechanisms for the increased violent crime detected.

Clearly, the present analysis has implications for policy. In particular, while the results obtained using unemployment provide only marginal evidence in support of a link between economic conditions and opportunity and motivation effects, the introduction of more direct, or ‘better’, indicators provided clear evidence of such relationships. In particular, the results obtained using the cyclical components of real consumption provided significant relationships for all classifications of crime considered, thus illustrating the importance of opportunity and motivation effects over the course of the business cycle. As a result, a role for policy is apparent with noted opportunity and motivation effects indicating a potential for addressing increases in crime during differing phases of the business cycle. More precisely, policy would need to address the witnessed counter-cyclical effects present for property crimes and the pro-cyclical nature of violent crimes. However, while the results provided demonstrate a clear link between the economic cycle and alternative classifications of crime, the analysis could be extended to provide analysis for alternative economies or differing time periods as a check of the robustness of the findings.

Footnotes

Acknowledgements

This research has benefited from funding received from the British Academy (ref: SG100765). We are grateful to the editors and anonymous referees for numerous comments which have substantially improved the presentation and content of this article.

Notes

Author biographies

Steve Cook holds a DPhil from the University of Oxford and has an interest in the empirical analysis of hypotheses spanning a range of disciplines including economics, finance and criminology. Recently, he has received funding from the British Academy to explore the linkages between economic conditions and criminal activity.

Duncan Watson holds a PhD from Loughborough University and has interdisciplinary research interests which have resulted in the publication of articles in journals covering a range of areas of research. Primarily empirical in nature, his research has sought to explore and employ alternative or plularist approaches.

References

Agnew

(1992) Foundation for a general strain theory. Criminology 30: 47–87.

Agnew

(2001) Types of strain most likely to lead to crime and delinquency. Journal of Research in Crime and Delinquency 38: 319–361.

Arvanites

DeFina

(2006) Business cycles and street crime. Criminology 44: 139–164.

Becker

(1968) Crime and punishment: An economic approach. Journal of Political Economy 76: 169–217.

Box

(1987) Recession, Crime and Punishment. Basingstoke: Macmillan.

Cantor

Land

(1985) Unemployment and crime rates in the post World War II United States: A theoretical and empirical analysis. American Sociological Review 50: 317–332.

Carmichael

Ward

(2001) Male unemployment and crime in England and Wales. Economics Letters 73: 111–115.

Chiricos

(1987) Rates of crime and unemployment: An analysis of aggregate research evidence. Social Problems 34: 187–212.

Cornish

Clarke

(1987) Understanding crime-displacement: An application of rational choice theory. Criminology 25: 933–947.

10.

Cloward

Ohlin

(1960) Delinquency and Opportunity. New York: Free Press.

11.

Cnossen

(2007) Alcohol taxation and regulation in the European Union. International Taxation and Public Finance 14: 600–732.

12.

Cohen

(1955) Delinquent Boys: The Culture of the Gang. Glencoe, IL: Free Press.

13.

Cohen

Felson

(1979) Social change and crime rate trends: A routine activities approach. American Sociological Review 44: 588–607.

14.

Cook

(2011) Are US crime rates really unit root processes? Journal of Quantitative Criminology 27: 299–314.

15.

Cook

Zarkin

(1985) Crime and the business cycle. Journal of Legal Studies 14: 115–128.

16.

Cook

Winfield

(2013) Crime across the states: Are U.S. crime rates converging. Urban Studies (forthcoming).

17.

Ehrlich

(1973) Participation in illegitimate activities: A theoretical and empirical investigation. Journal of Political Economy 81: 521–565.

18.

Felson

(1994) Crime and Everyday Life. Thousand Oaks, CA: Pine Forge Press.

19.

Finkelhor

Asdigian

(1996) Risk factors for youth victimization: Beyond a lifestyles/routines activities theory approach. Violence and Victims 11: 3–20.

20.

Freeman

(1998) A note on economic conditions and alcohol problems. Journal of Health Economics 18: 661–670.

21.

Freeman

(1980) Crime and unemployment. In: Wilson

(ed.) Crime and Public Policy. San Francisco, CA: ICS Press, pp. 89–106.

22.

Galanter

(2002) Recent Developments in Alcoholism: Alcoholism and Violence. New York: Kluwer.

23.

Greenberg

(2001) Time series analysis of crime rates. Journal of Quantitative Criminology 17: 291–327.

24.

Hale

Sabbagh

(1991) Testing the relationship between unemployment and crime: A methodological comment and empirical analysis using time series data from England and Wales. Journal of Research in Crime and Delinquency 28: 400–429.

25.

Harding

(1976) Can Theories Be Refuted? Essays on the Duhem-Quine Thesis. Dordrecht: Kluwer.

26.

Hodrick

Prescott

(1997) Postwar U.S. business cycles: An empirical investigation. Journal of Money, Credit, and Banking 29: 1–16.

27.

Hoover

(2003) Nonstationary time series, cointegration, and the principle of the common cause. British Journal of the Philosophy of Science 54: 527–551.

28.

Hughes

Carter

(1981) A declining economy and sociological theories of crime: Predictions and explanations. In: Wright

(ed.) Crime and Criminal Justice in a Declining Economy. Cambridge: Gunn and Hain, pp. 5–26.

29.

Pesaran

Shin

(2003) Testing for unit roots in heterogeneous panels. Journal of Econometrics 115: 53–74.

30.

LaFree

(1998) Losing Legitimacy: Street Crime and the Decline of Social Institutions in America. Boulder, CO: Westview.

31.

LaFree

(2005) Evidence for elite convergence in cross-national homicide victimization trends, 1956 to 2000. Sociological Quarterly 46: 191–211.

32.

Land

Felson

(1976) A general framework for building macro social indicator models: Including an analysis of changes in crime rates and policing expenditures. American Journal of Sociology 82: 565–604.

33.

Levin

Lin

Chu

(2002) Unit root tests in panel data: Asymptotic and finite-sample properties. Journal of Econometrics 108: 1–24.

34.

Long

Witte

(1981) Current economic trends: Implications for crime and justice. In: Wright

(ed.) Crime and Criminal Justice in a Declining Economy. Cambridge: Gunn and Hain, pp. 69–143.

35.

MacDonald

(2000) The impact of under-reporting on the relationship between unemployment and property crime. Applied Economics Letters 7: 659–663.

36.

Marvell

Moody

(2001) The lethal effects of three-strikes laws. Journal of Legal Studies 30: 89–106.

37.

McDowall

Loftin

(2009) Do US city crime rates follow a national trend? The influence of nationwide conditions on local crime patterns. Journal of Quantitative Criminology 25: 307–324.

38.

Merton

(1938) Social structure and anomie. American Sociological Review 3: 672–682.

39.

Messner

Rosenfeld

(2001) Crime and the American Dream. Belmont, CA: Wadsworth.

40.

Mocan

Bali

(2010) Asymmetric crime cycles. Review of Economics and Statistics 92: 899–911.

41.

O’Brien

(1999) Measuring the convergence/divergence of ‘serious crime’, arrest rates for males and females: 1960–1995. Journal of Quantitative Criminology 15: 97–114.

42.

Paternoster

Bushway

(2001) Theoretical and empirical work on the relationship between unemployment and crime. Journal of Quantitative Criminology 17: 391–407.

43.

Phillips

Land

(2012) The link between unemployment and crime rate fluctuations: An analysis at the county, state and national levels. Social Science Research 41: 681–694.

44.

Ravn

Uhlig

(2002) On adjusting the Hodrick-Prescott filter for the frequency of observations. Review of Economics and Statistics 84: 371–375.

45.

Rosenfeld

Fornago

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

46.

Smith

(1997) Explaining variation in state-level homicide rates: Does crime policy pay? Journal of Politics 59: 350–367.

47.

Sutton

(2004) The political economy of imprisonment in affluent western democracies, 1965–1990. American Sociological Review 69: 170–189.

48.

Tarling

(1982) Unemployment and crime. Home Office Research Bulletin 14: 28–33.

49.

Wilcox

Land

Hunt

(2003) Criminal Circumstance: A Dynamic Multi-Contextual Criminal Opportunity Theory. Chicago, IL: Aldine De Gruyter.

50.

Winsberg

(1993) Are crime waves in the United States regional or national? Journal of Criminal Justice 21: 517–520.

51.

Young

(1993) Unemployment and crime: Not a simple relationship. American Journal of Economics and Sociology 52: 413–415.