Social Science Research 45 (2014) 33–45

Contents lists available at ScienceDirect

Social Science Research
journal homepage: www.elsevier.com/locate/ssresearch

Link between unemployment and crime in the US:
A Markov-Switching approach q
Firouz Fallahi a,⇑, Gabriel Rodríguez b
a
Department of Economics, University of Tabriz, Tabriz, Iran
b
Department of Economics, Pontificia Universidad Católica del Perú, Peru

a r t i c l e i n f o a b s t r a c t

Article history: This study has two goals. The first is to use Markov Switching models to identify and ana-
Received 5 February 2013 lyze the cycles in the unemployment rate and four different types of property-related crim-
Revised 14 December 2013 inal activities in the US. The second is to apply the nonparametric concordance index of
Accepted 22 December 2013
Harding and Pagan (2006) to determine the correlation between the cycles of unemploy-
Available online 2 January 2014
ment rate and property crimes. Findings show that there is a positive but insignificant rela-
tionship between the unemployment rate, burglary, larceny, and robbery. However, the
Keywords:
unemployment rate has a significant and negative (i.e., a counter-cyclical) relationship
Markov-Switching models
Business cycles
with motor-vehicle theft. Therefore, more motor-vehicle thefts occur during economic
Unemployment expansions relative to contractions. Next, we divide the sample into three different subs-
Crime amples to examine the consistency of the findings. The results show that the co-move-
Concordance ments between the unemployment rate and property crimes during recession periods
are much weaker, when compared with that of the normal periods of the US economy.
Ó 2013 Elsevier Inc. All rights reserved.

1. Introduction

Although economic theory anticipates the existence of a positive relationship between unemployment and crime, empir-
ical works have found mixed results. Chiricos (1987) reviewed 68 studies about the relationship between crime and unem-
ployment rate, and found that only less than half of these studies have found positive significant effects of unemployment on
crime rates. That is, most of these studies have shown a negative or no relationship between crime and unemployment.
Cantor and Land (1985) and Cook and Zarkin (1985) are among those who have tried to explain these mixed findings of the
empirical studies. Cook and Zarkin (1985) stated that during the recession, the opportunity cost of attempting to crime and
also the legitimate opportunities of earning income deteriorate. Therefore, the number of property and violent crimes esca-
late; they called this the motivation effect.1 In addition, recession might cause a reduction in the policing due to lower tax rev-
enues of governments. The results of this change might be a lower chance of crime detection and punishment; hence, it might
increase the crime rate. On the other hand, during recession, the number and quality of criminal opportunities decline (Cook,
2010), because more people stay at home, which increases the guardianship of properties, and hence, the number of burglary
victims would decrease. The same could be said for the robbery, because people carry less cash, and as Cook (2010) stated, they
have more tendencies to defend what they have. Cantor and Land (1985) called this the opportunity effect. Motivational and
opportunity effects would produce upward and downward fluctuations in crime rates; consequently, the net effect of unemploy-
ment, as an indicator of bad economic conditions, on crime is ambiguous. In other words, when the opportunity effect is larger

q
We thank the editor, two anonymous referees, Lynda Khalaf, Marcel Voia, Gamal Atallah, and David Greenberg for their extremely thoughtful and
thorough comments, which substantially improved the quality and presentation of the paper.
⇑ Corresponding author. Fax: +98 411 335 6013.
E-mail address: ffallahi@tabrizu.ac.ir (F. Fallahi).
1
In fact, motivational effect, originally, has been introduced by Becker (1968).

0049-089X/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.ssresearch.2013.12.007

that is. However. as stated by Freeman et al. mo- tor-vehicle theft.3 As mentioned earlier. linear models may provide a weak fit. i. Andersen (2012) used neighborhood-level panel data to examine the association between crime and unemployment. the effect of a business cycle. 1985. Consequently. whereas Witt et al.. proxied by unemployment. the estimated significant relationship between unemployment and crime does not necessarily show the causality between these variables. Cantor and Land (1987).g. MS models allow having multiple endogenously determined structural breaks in the data. This study has two goals. therefore. they also differ in their findings. while the latter is more similar to dynamic approaches. Greenberg (2001) and Allen (1996) revealed the same result for robbery.7 2 The effect of business cycles on crime could be very complex. there would be a negative (positive) link between unemployment rate and crime. the association between the unemployment rate and crime has been investigated extensively in different studies. Greenberg. On the other hand. Rodríguez / Social Science Research 45 (2014) 33–45 (smaller) than the motivation effect.. Some scholars argue that it is the lagged unemployment. 1991. one needs to use nonlinear models. Allen (1996) showed that lagged unemployment has a negative effect on larceny. Some of these studies have employed national aggregate data (Kollias and Paleologou. larceny. Our sampling period covers 1977:2 until 2004:4 and four types of property crimes6 are analyzed: burglary. such as the Mar- kov Switching (MS) models. hence. Allen.e. 1996). Fallahi. the results show the link over time. (2012) and Phillips and Land (2012) came to a conclusion that the rate of burglary is significantly lower during recessions.4 An important step forward in understanding the relationship between crime and unemployment was introducing the ef- fect of unemployment lag to the literature. hence. which makes the prediction of these effects really difficult (Cook. (1999) highlighted the importance of gender on crime. The first goal is to examine the behavior of different types of property-related felonies and the second goal deals with the potential relationship between the unemployment rate and property crime rates in the US econ- omy. more auto-theft occurs during economic expansions. the occurrence timing of these two effects could be different. The results about the link between larceny and unemployment are mixed and contradictory. 1985. such as those by Entorf and Spengler (2000). on crime might go in either direction or even it could be null. Most of these studies were carried out using multiple regression models. it would be better to study that along with other property crimes. and robbery. state. However. although robbery is considered a violent offence. national-levels). Consequently. as Cantor and Land (1985) argued. These studies not only differ in terms of aggregation level of the data used for analysis (e. 2012. Several studies have shown a positive relationship between unemployment and property crime. the study by Hale and Sabbagh (1991) is among those that concluded that unemployment has no significant influence on auto-theft. 1991. 1996. Therefore. In fact. the results show the relationship between unemployment and crime at one point of time. Can- tor and Land (1987) showed that lagged unemployment positively affects larceny and burglary. Phillips and Land (2012) are among those who analyzed this subject at the county. Yet. Therefore. For example. Fallahi et al. state-level vs. 2006). Yet. it is motivated primarily by economic need. therefore. (1999). Larceny is another property crime that has gained attention from research community. vector autoregression or error correction mod- els. The same is true for crime as well (see Freeman et al. 4 As one of the referees of the present paper points out. based their investigation on subnational. and auto-theft. one reason for these contradictory results might be the difference in the data type used to study the link. 1999 for a survey). Cook and Zarkin (1985). Britt (1997) and Lee and Holoviak (2006) examined age-specific data on crime. Hale and Sabbagh. .5 MS models have been used widely in the literature and have proven their ability to explain the changes in the pattern of time series.e. in fact. whereas the motivation would respond with a lag (Arvanites and Defina. but the motivation effect needs time. Allen (1996). but Phillips and Land (2012) finding a negative relationship. These models are able to capture changes in the behavior of time-series by allowing the switch- ing between regimes or states. with Cantor and Land (1987) and Hale and Sabbagh (1991) docu- menting a positive link.. Taking this point into consideration. rather than current unemployment. 7 As stated by Raphael and Winter-Ebmer (2001) and Grant and Martinez (1997). while others have found a negative relationship. The majority of studies on robbery found a positive significant effect of unemployment on robbery (Cook and Zarkin. hence. a negative rela- tionship between business cycles and burglary. 6 These four types of crime cover 92% of the total crime in the US in 2004. 2012). they might cancel each another. Moreover. Allen. in fact. 1987. they try to distinguish the motivational and opportunity ef- fects. that has a potential influence on crime. 2001. and Edmark (2005). others have used time-series data. i. opportunity effect would appear instantaneously. However. in another study. and Edmark (2005) identi- fied a positive co-movement between the unemployment rate and rate of burglary. the opportunity effect is instantaneous. Cook and Zarkin. Edmark (2005) and Fallahi et al.2 Besides. while others. but Edmark (2005) claimed that there is no significant link between unemployment and robbery.. 2010). (2012) showed that unemployment and auto-theft co-vary significantly. and national levels to shed more light on the impact of data aggregation level on findings. which means that the rate of auto-theft is higher during economic con- traction. the former can be regarded as a static approach. The same is true for burglary. Some of these papers have used cross-section data. All of these methods assume a stable behavior of the variable under examination. Hale and Sabbagh. Levitt (2001). studies by Cook and Zarkin (1985). these two effects have opposite signs and. However. it is lagged. Cantor and Land. burglary.. i. G. state.e.34 F. Allen (1996). and Phillips and Land (2012) suggested that the unemployment rate and rate of auto-theft move in opposite directions. no unequivocal support for the link between unemployment and crime has been found. hence. it is better to be cautious in comparing the results of cross-sectional papers with that of the time- series ones. 5 In other words.or county-level data. the unemployment rate is directly related to business cycles and has a cyclical pattern. 3 It is worth mentioning that. Fallahi et al.

the unemployment rate has a significant relationship with motor-vehicle theft. the period from 1990 to 2004) and the third subsample when the US economy was at a normal state. It is denoted as MSM(2)-AR(k). The findings of sub- sample studies show that the links between unemployment and property crimes are higher for the first subsample (i. the link is much weaker than the links found for the whole sample and the other two subsamples.’’ These changes (breaks) could happen because of various reasons. ð1Þ where l shows the mean of the variable yt . Take weather conditions as an example. Using this approach. Then. yt2 ... and the variable yt moves between these states. variance. for example. The contemporaneous relationship between the rate of unemployment and motor-vehicle theft is negative (which supports the existence of oppor- tunity effect).. two regimes. j 2 f1.. In this model. as stated by Hamilton (1994). the intercept is considered as regime-dependent to obtain an 8 In the first-order Markov switching approach. The second section describes the methodology. and robbery. yt1 . knowing that it is sunny today. p2M 7 6 21 7 P¼6 7. Following Hamilton (1989). st is an unobserved variable indicating the state of the variable yt . . when the crimes in the US are known to have declined. to check the relationship between the cycles of the unemployment rate and crime variables. . economic cri- ses. we divide the sample into three subsamples. . To shed more light on the stability of the link between unemployment and property crimes. we have M different possible states or regimes for the variable. To take these changes into account. we can discuss about the state of the variable at the next period. Mg. Markov switching approach provides us with a solution.. If the light is green now.. we found that during the period from 1990 to 2004. It is possible to estimate a model for the period before break and another model for the periods after the break. In some cases. and regulation changes. The second subsample locates recession phases in the US and the third subsample consists of the observations coinciding with the period of normal state in the US economy. The results from the correlations between the lagged cycles of unemployment and crime indicate that as time goes by. we have different approaches to follow. and red. .. the fifth section gives the conclusion. knowing the present light or state reveals the next state with certainty. Suppose it could be rainy. and an autoregression of order k. The remaining portion of this paper is organized as follows. during the recession periods (the second subsample). wars. however. They can create different regimes or states for the variable. ¼ Pr½st ¼ jjst1 ¼ i ¼ pij . Lastly. yellow.. The third sec- tion shows the results obtained. We only can assign probabilities on the potential conditions for tomorrow’s weather. . we know that it will turn red. . F.. In addition. One way to calculate these probabilities is the Markov switching (chains) approach that is a branch of probability theory. having information about the present period. . instead of the mean. which indicates a Mar- kov switching autoregressive model (MS-AR) with regime-dependent mean. 2. and if it is yellow now. the exact date of break is not known most of the time.The situation gets complicated when the variable is a stochastic process. larceny. the probability that it rains the following day is 3%. 2. . burglary. Methodology Let yt denote the variable of interest in period t. Rodríguez / Social Science Research 45 (2014) 33–45 35 The Markov-Switching method is used to examine the behavior of the unemployment rate and property crime rates. . there was a positive and significant relationship between the unemployment rate and robbery. or sunny. G. Fallahi. only the value of the mean is regime-dependent. The first subsample covers data from the first quarter in 1990 through the last quarter of 2004. and the fourth section deals with synchronization of cycles using the nonparametric Con- cordance Index of Harding and Pagan (2006). we can estimate a different model for each regime without a priori knowl- edge about the date of breaks or regime changes. Different methods can be used to study the behavior of a series. . we may say that the probability that it is sunny the following day is 80%. and the probability of having a cloudy day is 17%.. the state of the variable at time t þ 1 depends on the state of the variable at time t. Sometimes the variable is deterministic. which implies that less motor-vehicle theft has been reported during recessions when the unemployment rate is higher. PMM where pij shows the probability of moving from regime i to the regime j. 5 pM1 pM2 . . and t  i:i:d. . p1M 6p p22 . PM with j¼1 pij ¼ 1. . Therefore. A full description of the dynamics of yt could be obtained if we have a probabilistic description of how the variable changes from one regime to another. In the model (1). On the other hand. a model where the mean. such as the traffic light that has three lights or states: Green..e. r2 Þ. 8i. the relative size of motivation effect and opportunity effect changes. ... st2 ¼ j. 4 . and autoregressive coefficients are regime dependent is denoted by MSMAH(m)-AR(k). Other specifications are available. However. Such simplest model is a Markov chain. we have a matrix of transition probabilities 2 3 p11 p12 . The findings show that there is a positive but insignificant association between the unemployment rate. Nð0. However.8 which is a model where Pr½st ¼ jjst1 ¼ i. we cannot be certain about tomorrow’s weather condition. . . . ‘‘[m]any variables undergo episodes in which the behavior of the series seems to change quite dramatically. we know that it will turn yellow. cloudy. Therefore. For example. we may propose the following model for the variable yt : yt  lst ¼ /1 ðyt1  lst1 Þ      /k ðytk  lst4 Þ þ t . where m indicates the number of states (regimes) and k reflects the order of the autoregression. such as financial panics. Therefore.

Larceny. The smoothed probabilities and the filtered probabilities of being at each regime at each point of time are presented in Fig. Concerning the standard deviations. the Schwarz information criterion (SIC). while several studies have used annual aggregate data. the rate of burglary was declining. Based on the information criteria and the LR (likelihood ratio) sta- tistic. the results should not be used to judge about long-run relationships. The data were seasonally nonadjusted. and the feasible point optimal statistic suggested by Elliott et al. 3. indicating that all crime time series are I(1) processes. Using the selected lag structure. The null hypothesis of a unit root was not rejected for any of the crime series. the MSIAH(2)-AR(4) was selected as the best model. which showed that for this time series. The models where the mean and intercept are regime-dependent are not equivalent. 1987:2–1988:3.9 The data were obtained from the Uniform Crime Reports of the Federal Bureau of Investigation (FBI) and they are national aggre- gates. we selected the best fitting MS models. and the Hannan–Quinn information criterion (HQ). 13 This finding is in line with the fact that the index crimes in the US have been declining since the 1990s.004. Empirical results of the MS-AR models Quarterly data of four property crime series (Burglary. The p-value of the Davies’s statistic for the null hypothesis of linearity was 0. Therefore. the unconditional probabilities of being at the lower and higher burglary rate regimes are 79% and 21%. this rate increased. and 1993:2–2004:4. I(0). The lag length was determined using information criteria such as the Akaike information criterion (AIC). Larceny All information criteria indicated that the preferred model is the MSIA(2)-AR(1).10 The data for the quarterly unemployment rates were taken from the Bureau of Labor Statistics. This indicates the presence of asymmetries in the duration of the regimes of lower and higher burglary rates. whereas the second regime has the duration of 1. 12 Using LR test and Davies’ (1987) bounded LR test. indicating that between the second quarter of 1977 and the end of 2004. not those reported in crime victimization surveys. Based on the transition probabilities. The presence of unit roots was formally tested using the GLS-based Augmented Dickey-Fuller (ADF GLS ) statistic of Elliott et al. We would like to thank an anonymous referee for pointing this out. allowing for changes in the inter- cept (I). (1996). The results showed that the first regime has a negative intercept. 10 Notice that the rates are reported property crime rates based on crimes reported to the police. based on the information of the whole sample. The same figure also shows the observations since 1993:2 until the end of the sample form the first regime (decreasing bur- glary rates). In addition.13 3. 2005). and show the number of each crime per 100.36 F. and Table 1 shows the results for this model. the results showed that the sec- ond regime was more volatile than the first one (standard deviation of 7. Table 1 presents the results for this mod- el. The null hypothesis of linearity was rejected for all selected models. a linear model was specified. the property crime series were modeled in first differences11 and the unemployment rate entered in levels. the null of linearity should be rejected in favor of the MS model. Given the previous results of the unit root tests.59 quarters. A generalized version of the model suggested by Hamilton (1989) was estimated. The results also showed that 79% of the observations for burglary were categorized in the first regime. 1980:3–1980:4. the results showed that in the first regime. a decrease in the rate of burglary was present. All esti- mations were carried out with the MSVAR class for Ox (see Krolzig.1. in 88 quarters. The transition probabilities revealed that the duration of the first regime is 6. . the first regime stands for lower and second regime stands for higher burglary rates in the US. when compared with 5. intend- ing that the periods of reduction in the rate of burglary are much longer than those with positive change at this rate.77. 1989:2–1991:3. while the intercept of the second regime is positive and 9 It should be mentioned that this study used quarterly aggregate data. and Robbery) for the US were used. Motor-Vehicle Theft.12 3. We used the TRAMO/SEATS procedure of Gómez and Maravall (1992) to remove the seasonality and detect the presence of outliers in the series. In fact.000 of population. 1981:3–1984:3. Burglary For this time series. MS models with two states were estimated. Once the coefficients of the model are estimated and a transition matrix is calculated. and the corresponding series of probabilities is known as ‘‘filtered probabilities’’. and in the second regime. Overall. 1. 1985:1–1985:4. In addition. (1996). 1992:2–1992:3. On the other hand. 11 It is worth noting that as we are using the first difference of the property crime rates and as differenced data show the short-term fluctuations only. the unem- ployment rate appeared to be stationary.2. 1986:2–1986:4. it is clear that the first regime is persistent with a probability of 84%. They imply different dynamics of adjustment of the variables after a change in regime. Rodríguez / Social Science Research 45 (2014) 33–45 MSIAH(m)-AR(k). Before estimating the MS models. variance (H). one can calculate the probability of being at state j at each time only based on the information up to that date (not the whole sample). This shows that the first regime (lower burglary rates) was formed by the periods 1977:2–1979:3. one can calculate the probability of being in state j at each period. G. Fallahi. the model allowed for different variances for each regime. and autoregressive parameters (A). respectively. This series of probabilities is known as ‘‘smoothed probabilities’’. These data cover the period from 1977:2 to 2004:4. that is.50).09 quarters.

836 0.02 Likelihood ratio linearity test’s probability Davies 0. the first regime stands for the periods with decreasing rates of lar- ceny and the second regime shows the periods with increasing rates of larceny.737) 0.909 3.974 p22 0.738) 0. 1.687 (5.201 (2.224) Standard errors r1 5.484) 4. 1977 (2) .11 12.352 (2. G.065) /42 0.922 (4.279 Transition probabilities p11 0.5 1980 1985 1990 1995 2000 2005 Fig. respectively.31 Regime 2 1.409 (3.558) 0.504 1.758) /41 0.112 (2.279 r2 7.046 (0.56 13. F.2004 (4) DBurglary 25 0 1980 1985 1990 1995 2000 2005 Probabilities of Regime 1 1.083 (0. According to these results.207) /12 0. The durations of the regimes were 12 and 11.0 filtered smoothed 0.253) Autoregressive parameters /11 0.221 (1.248 (2.03 38.944) /21 0.917 0.055 (0.016 0.859) 0. both of them are significant. smoothed.920) 2.197 (1.909 3. Rodríguez / Social Science Research 45 (2014) 33–45 37 Table 1 Results of univariate MS-AR models for crimes.028 t-Statistics are in parentheses.050 (0.212 (2.923 Durations Regime 1 6. and filtered probabilities of Burglary.534 (3.882 0. Burglary Robbery Larceny Motor-Vehicle theft MSIAH(2)-AR(4) MSI(2)-AR(3) MSIA(2)-AR(1) MSI(2)-AR(1) Intercepts l1 2.349) 1. the first regime accounts for 88% of the observations (98 quarters).281) l2 5.774 1.49 1. From 111 observations included in the estimation.714 (4.09 22.360 0. while the second regime accounts for 13 quarters.231 (2.092 (5.143 (1.251) 1.818) 1.619) /31 0. MSIAH(2)-AR(4) model.59 8.418 14. The .418 14. The series.589) /22 1.002 0. Fallahi.5 filtered smoothed 1980 1985 1990 1995 2000 2005 Probabilities of Regime 2 1.0 0.857) 34.56 quarters.004 0.765) /32 0.056 (0.955 0.370 0.853 (3.644) 0.016 (0.607) 0. MSIAH(2)-AR(4).

4.. At the same time. 3.e. MSIA(2)-AR(1). 2. 3. 2 shows the smoothed and filtered probabilities of larceny in the two regimes. The results also indicated that 32% of the observations are classified in the higher rate regime. 1989:1– 1989:1. when compared with 38 for the first regime. 1986:1–1986:1.5 1980 1985 1990 1995 2000 2005 Probabilities of Regime 2 1. the rate of motor-vehicle theft in the US had been decreasing during these periods. this model was selected as the best model. respectively. Therefore. The expected duration of this regime was 13 quarters. Robbery The SIC suggests the MSI(2)-AR(3) as the best model that captures the movements of robbery in the US The results from the estimation of this model are shown in Table 1.. the second regime (increasing rates of larceny) consists of 1979:3–1979:4.0 filtered smoothed 0. the SIC suggested an MSI(2)-AR(1) as the best model. G. the larceny in the US was decreasing.e. for the first and second regimes. The intercepts are negative and positive.0 filtered smoothed 0. According to these results. a close investi- gation of this model revealed that it does not fit the data well and it cannot capture the movements of this time series. and 1994:1–1994:1. and the second regime shows all the periods with positive changes in this rate (i. This means that if there is a decreasing rate of larceny at any period. Motor-vehicle theft For this variable. However. and there is a 3% probability of moving from the first (lower rates of motor-vehicle theft) to the second regime (higher rates of motor-vehicle theft). The first regime stands for all the periods with negative changes in the rate of robbery (i. Table 1 shows the results for this model. 1991:4–1992:1.e. It can also be seen from Fig. and filtered probabilities of Larceny. MSI(2)-AR(1) model.. Fallahi. which matches the data and its movement well. The series. respectively. Fig.3. Both the regimes are very persistent.5 1980 1985 1990 1995 2000 2005 Fig. the changes in motor-vehicle theft had been positive and the rate of motor-vehicle theft had been increasing. It shows that during 1980:1–1984:2 and from 1990:2 to the end of the sample. The opposite direction has an 8% probability. and the second regime stands for the periods with positive changes of this rate. when compared with 36%. lower robbery rates).38 F. higher rates of robbery). For the rest of the sample. smoothed. with a probability of 92%. i. 1987:1–1987:1. The first regime shows the periods with negative changes at the rate of motor-vehicle theft. 2 that from 1994 onwards. the AIC and the LR test suggested that the best model is an MSIAH(2)-AR(1). 1977 (2) . Fig. Rodríguez / Social Science Research 45 (2014) 33–45 transition probabilities indicated that the first regime is more persistent than the second regime. Notice that most of these periods last only for two quarters.2004 (4) 100 DLarceny 50 0 -50 1980 1985 1990 1995 2000 2005 Probabilities of Regime 1 1. 1989:4–1989:4. 1980:2–1980:2. 3 shows the smoothed and filtered probabil- ities of being at each regime. it will be dif- ficult to return to a regime with an increasing rate. . the variable was at the first regime.

Therefore.. The expected durations were 22 and 8. Fallahi.119 and 0. Regime 1 shows all the periods with low unemployment rate in the US. The only periods with high unemployment rates are 1979:4–1984:3. there are different estimations for the coefficients of lagged variables in regimes. the regime with higher unemployment rates is more volatile than the first regime. smoothed. the first and second regimes show the ‘‘normal’’ and ‘‘recession’’ periods in the US economy. then there is a high probability for a negative change (positive change) in the next period. respectively. the unemployment rate in the US was at a lower regime. The longest period with a decreasing rate of robbery is 1992:1–2004:4. Unemployment rate Although the information criteria selected the MSI(2)-AR(3) as the best model. it cannot capture the complete character- istics and movements of the series. Fig. Fig. for regime 1 and 2. During the period of study. Table 2 shows the results of estimation of this model. 1990:3–1992:1. 3. and 2001:1–2001:4. These periods almost perfectly match the recession periods in US. and the second regime stands for all the periods with higher unemployment rate. one can test whether the timing of the cycles of the unemployment rate and . To answer this question. for the first and second regimes. Synchronization of cycles So far. almost 13 years. The lower robbery rates regime included 70% of the observations (77. with probabilities of 94% and 84%.0 filtered smoothed 0. to remain at the same state in the next period.5 1980 1985 1990 1995 2000 2005 Fig. Transition probabilities indicated that both the regimes were per- sistent. i. 1977 (2) . indicating that if there is a negative change (positive change) in this rate in any period. The LR linearity test strongly supported the nonlinearity in this variable. This model allows for regime-dependent autoregressive parameters and variance. The mean. MSI(2)-AR(1) model. During the rest of the study period. respectively. respectively. and asymmetries. and filtered probabilities of Motor-Vehicle Theft.0 filtered smoothed 0.5.e. respectively. One of the questions of interest in this study is the existence of any relationship between the unemployment rate (as an indicator of the position of the economy) and the prop- erty crime variables. The model that can fit the data well and describe the changes in the unemployment rate in the US is the MSIAH(2)-AR(3). 4 shows the smoothed and filtered probabilities for both the regimes.3 quarters).5 quarters. tran- sition probabilities.5 1980 1985 1990 1995 2000 2005 Probabilities of Regime 2 1. F. Rodríguez / Social Science Research 45 (2014) 33–45 39 MSI(2)-AR(1). 4. the MS models for each single time series were estimated and their properties. The expected duration of low unemployment is almost three times longer than that of higher rates of unemployment. 79 quarters form the regime of low unemployment rate and the rest (32 quarters) are classified as the second regime (higher unemployment). G. 3. The rest of the study period oscillates between the regimes. with probabilities equal to 95% and 88%. hence. The first regime is more persistent than the second regime based on the tran- sition probabilities.285. such as expected durations. 5 shows the smoothed and filtered probabilities for both the regimes. Therefore.2004 (4) 20 DMotor Mean(DMotor) 10 0 -10 1980 1985 1990 1995 2000 2005 Probabilities of Regime 1 1. have been investigated individually. The estimated stan- dard deviations are 0.

221) /3 0.085) 0.939 0.613 (8. to see whether higher unemployment rate periods are correlated with higher property crime rates regimes or not.500) 0. Rodríguez / Social Science Research 45 (2014) 33–45 MSI(2)-AR(3).5 1980 1985 1990 1995 2000 2005 Fig. .630) /2 0. The mean. The CI is a nonparametric statistic that shows the proportion of time at which two series are in the same regime.056 (0. . proposed originally by Harding and Pagan (2006).189 (9. MSI(2)-AR(3) model.48 6.192) Standard errors 0. Fallahi. Table 2 Results of MSIAH(2)–AR(3) for the unemployment rate. and filtered probabilities of Robbery. 4. property crimes under study in this research is similar or not.163 0. G. T. 1977 (2) . For two series xt and yt for t ¼ 1.285 Duration 16. the CI for xt and yt is given by the following expression: ( ) 1 XT X T CI ¼ T Sxt Syt þ ð1  Sxt Þð1  Syt Þ : ð3Þ t¼1 t¼1 . smoothed. that is.302) 0.730 (2.607 (2.5 1980 1985 1990 1995 2000 2005 Probabilities of Regime 2 1. . the Concordance Index (CI). In other words.2004 (4) 5 DRobbery Mean(DRobbery) 0 1980 1985 1990 1995 2000 2005 Probabilities of Regime 1 1. These cycles will be compared with those of the property crime series to test for common cycles.837 t-Statistics are in parentheses.166 (1. has been used. Then. The unemployment rate will be considered as the reference variable and its cycles as the reference cycle.036 (0.271) /1 1. let Sxt ðSyt Þ be a dummy variable that takes the value of unity when xt ðyt Þ is in the first re- gime and the value of zero when it is not in the first regime.14 Likelihood ratio linearity test’s probability Davies 0.0 filtered smoothed 0. 2.000 Transition probabilities Regime 1 Regime 2 Regime 1 0. For this purpose. one can check for synchronization of cycles.119 0.757) 1.139 (1. .061 Regime 2 0.0 filtered smoothed 0.40 F. Regime 1 Regime 2 l 0.

The series. and they are unsynchronized when qS ¼ 0.. F. therefore. However. the mean corrected concordance index (Artis et al. For this purpose.5 5. If the former is larger than the latter. smoothed. assume that the means are 0. a value of 0. . the mean values of the series will be higher than 0.5. They are counter cyclical if CI ¼ 0. Therefore.5. For example. their mean. The CI index has a value of unity when Sxt ¼ Syt and a value of zero when Syt ¼ ð1  Sxt Þ. Suppose that the mean of Sxt and Syt is 0. Now. 14 For example. It is natural to state that having a high concordance index means high common cycle. the value of unity for this index corresponds to qS ¼ 1 and value of zero to qS ¼ 0. then rSx rSy ¼ r2Sx . However. the question is how high should it be to interpret that as pro-cyclical? Even for two unrelated series. the expected value of the concordance index may be 0.68. Now. Another problem that exists in using this index is that it depends on the expected values of Sxt and Syt . the expected value of the concor- dance index will be 0.0 0. MSIAH(2)-AR(3). Fallahi.14 The above-mentioned formula for concordance index can be written in a different way as follows: X T CI ¼ 1 þ 2T 1 Sxt Syt  lSx  lSy ¼ 1 þ 2qS rSx rSy þ 2lSx lSy  lSx  lSy .5 filtered smoothed 1980 1985 1990 1995 2000 2005 Probabilities of Regime 2 1.5 and these two series are unsynchronized. this index is bounded between zero and unity. which is higher than 0. which confirms the assumption that they are unrelated.5 or higher. If the former is smaller than the latter. Assuming that the two series are statistically uncorrelated (qS ¼ 0Þ. Rodríguez / Social Science Research 45 (2014) 33–45 41 10. ð4Þ t¼1 where qS is the estimated correlation coefficient between Sxt and Syt . consider tossing two fair coins.8 for this index means that two series (xt and yt ) are in the same regime 80% of the time. one can conclude that these series are counter-cyclical. qS ¼ 1 (qS ¼ 1) shows that two cycles are perfectly positively (negatively) synchronized. and filtered probabilities of the Unemployment Rate. hence. that is. G. this expected value can be compared with the calculated value from the series.5. and one may think that these two series have common cycle although they are not related. As CI is defined as the proportion of time that two series are in the same state.0 1980 1985 1990 1995 2000 2005 Probabilities of Regime 1 1.0 0. If Sxt ¼ Syt or Syt ¼ ð1  Sxt Þ. the mean value of the series has to be taken into account. 2004) is considered.5. These two series are called pro-cyclical if CI ¼ 1. This indicates that the number of periods where the series are in the same state is higher than that if they were uncorrelated. one can state that there is a link between the cycles. However. 5. if the regime that takes the value of one has higher duration than the other. Therefore. then.0 Unemployment 7. the probability that both coins are in the same state (either heads or tails) is 0.8. the significance of this result has to be checked to see whether the ratio of these two is statisti- cally different from one or not. the expected value of this index will be: EðCIÞ ¼ 1 þ 2lSx lSy  lSx  lSy : ð5Þ The expected value of being at each regime can be measured by dividing the number of periods at that regime by T. the expected concordance index will be 0.5 1980 1985 1990 1995 2000 2005 Fig.

and robbery is pro-cyclical. However. The findings were the same as those earlier for the relationship between the first and second lags of unemployment rate cycles. as these cor- relations were insignificant even at 10%. In addition. only the results for motor- vehicle theft is statistically significant at 5%. Then. Nevertheless.595 E(CI) 0. 1987). there is a negative contemporaneous relationship between the unemploy- ment rate and motor-vehicle theft in the US that implies that less motor-vehicle theft has been reported during recessions.015 0. we may conclude that there is no significant link between unemployment and rob- bery.016 PT Let Sx ¼ T 1 t¼1 Sxt indicate the estimated probability of being at the first regime.670 0. as time goes by. the relative size of motivation effect and opportunity effect changes.123 0. 16 As we have used quarterly data.16 For example. only one of the property crimes is significantly contemporaneously con- cordant with the unemployment rate cycle. Fallahi.685 0. Rodríguez / Social Science Research 45 (2014) 33–45 Table 3 Concordance Index (CI) between the unemployment rate and property crimes.658 0. stated CI is monotonic in qS . hence. the mean corrected concordance index will be X T CIcorr ¼ 2T 1 ðSxt  Sx ÞðSyt  Sy Þ: ð6Þ t¼1 As mentioned earlier.712 for the burglary and unemployment rate indicates that burglary and unemployment are in the same state or regime (i. However. However.053 0. the results show that unemployment and burglary rates are linked to each other.039 t-ratio 1.2% of time. Also. when the unemployment rate is higher. which means that burglary increases when the economy is in recession.178 0. when the null hypothesis is true. larceny. Table 3 shows the calculated CI. To do so.15 The previous findings were based on the assumption that the relationship between crimes and unemployment rate. larceny. the sign of the correlation between the lagged cycles of unemployment and each one of the property crime types changed. if there is any. and robbery is positive. hence. however. . EðCIÞ is lower than unity. is contemporaneous. Unemployment Unemployment Unemployment rate & Unemployment rate & rate & Burglary rate & Larceny Motor-Vehicle theft Robbery CI 0. as the ratio of CI to E(CI) is higher than 1.. According to the robust t-ratios reported in Table 3. Although these correlations were insignificant. In doing so. therefore. the unemployment rate in the US and burglary are pro-cyclical. and the correlation between the cycles of unemployment rate and those of robbery showed a negative relationship.474 2. we assumed that changes in the regime of unemployment rate at t  i may be related to changes in the regime of crime series at time t. and q for each pair of variables using the regime classifications based on the MS models in Section 3. while motivation effect was high- 15 This could be interpreted as evidence that supports the existence of the opportunity effect. what we are talking about here is the relative contemporaneity of a crime opportunity mechanism (Cantor and Land. which shows that the unemployment rate and motor-vehicle theft are counter-cyclical.582 0. The estimated concordance of 0. However. We next calculated the concordance index of the lagged cycles of the unemployment rate and current cycle of the property crimes to examine the presence of the motivation effect. The same is true for CI larceny and robbery.8% of time. This shows that the relationship between the unem- ployment rate and burglary. larceny. the estimated correlation between the unemployment rate and burglary.055 0. this relationship is counter-cyclical between the unemployment rate and motor-vehicle theft. However. and motor-vehicle theft (see Table 4).293 0. as introduced by Cantor and Land (1985) and Phillips and Land (2012). G. and robbery. the results changed for robbery. In this regression. Syt is strongly serially correlated. where i shows the number of lagged periods.459 0. they showed that. In addition. then they had to be in the same regime only at 65.218 0.42 F.578 q ^ 0. robust estimated standard errors have to be used (such as HAC Newey-West method). Harding and Pagan (2006) suggested that one could use a regression model to deal with this problem. the error term inherits the serial correlation properties of Syt .231 CIcorr 0. E(CI).712 0. the following regression can be used: r1 1 sy Syt ¼ a1 þ qS rSx Sxt þ ut : ð7Þ Now. In summary. one of the shortcomings of the concordance index is that it does not allow for a statistical testing of the result. burglary.545 0. Thus.e. it is negative between the unemployment rate and motor-vehicle theft. When more lags were introduced to the model. the hypothesis that qS ¼ 0 can be tested using the t-ratio of the coefficient of r1 Sx Sxt . as Harding and Pagan (2006). we can shift our attention from CI to the qS obtained from the estima- tion of (7). if they were independent. for motor-vehicle theft. we found no contemporaneous significant relationship between the unemployment rate and the following crimes: burglary. rising or declining together) in 71. In addition. larceny.

030 0.344 t-Statistics are shown in parenthesis.800 0.235⁄ 0.219 (0.169 0.219 0.031 0. Consequently.245 8 lag 0. however. it was lower than the opportunity effect.274 0.112 t-ratio 1.183 0.067 0. F. during recessions.911 0. when the unemployment rate was low. the results with no lag (contemporaneous results) are reported in Table 3.094 0.076 Notice that. G.205 0.024 0.254 2 lag 0.095 0.805 5 lag 0.817 0. was negative.077 0.520 1.468 0.981) 0.086) 0. Unemployment Unemployment Unemployment rate & Unemployment rate & Burglary rate & Larceny Motor-Vehicle theft rate & Robbery CI 0. That is.193 0.183 0.696 The 2nd Regime of Unemployment CI 0.297 0.494 0. Conclusions Although economic theory anticipates the existence of a positive relationship between unemployment rate and crime rates. 5. During this period.017 0. The same was true for motor-vehicle theft.585 4 lag 0.038 t-ratio 1.316 0. er for larceny up to three lags of unemployment cycle.592 7 lag 0. In addition.732 q 0. when compared with the whole sample.130 0.024 0.027 0.052 0.347 0.029 0.171 0. as denoted by regime 1 in Section 3. For example. the state of the economy does affect the rela- tionship between the variables.513 2.497 0. Therefore.062 (0. the findings showed that the motivation effect existed for motor-vehi- cle theft. However.178 1.163 1. which is the sum of the motivation and opportunity effects. the opportunity effect was higher than the motivation effect for lags high- er than three.311 3 lag 0. the results showed a significant and positive association between unemployment and robbery.022 0. and the researchers who want to study the economics of crime must take the state of the economy into consideration.103 t-ratio 1. when the crimes in the US are known to be declining.287 (1.086 0. the calculated CIs in the period corresponding to the normal state of the US economy. ⁄ Significant correlations at 10% or lower. The second subsample located recession phases in the US.5 and referred to as the second regime of the unemployment rate.053 0. this link was weaker than the relation- ships found for the whole sample and the other two subsamples. most of these studies have shown a negative or no relationship be- tween crime and unemployment.106 0.705 6 lag 0.125 0.004 0. and found that only less than half of these studies have found positive signif- icant effects of unemployment on crime rates. The third subsample consisted of the observations coinciding with the period of normal state in the US economy.5.145 0.293 0. Rodríguez / Social Science Research 45 (2014) 33–45 43 Table 4 The estimated correlations between crime and the lagged unemployment rate. showed a much higher concordance between unemployment and property crimes. the second regime of the unemployment rate. The first subsample covered data from the first quarter in 1990 through the last quarter of 2004. To shed more light on the pattern and stability of the relationship between unemployment rate and property crimes.944) 0.286⁄ 0.020 0. The 1st and 2nd Regimes of Unemployment corresponds to the normal conditions and recession in the US economy. Chiricos (1987) reviewed 68 studies about the relationship between crime and the unemployment rate.181 0.911 0. Unemployment Unemployment Unemployment rate & Unemployment rate & Burglary rate & Larceny Motor-Vehicle theft rate & Robbery 1 lag 0.814 0.785 0. respectively.019 0.994 0. Based on the results from these three subs- amples (reported in Table 5). as identified in Section 3.878 0.608 0.077 t-ratio 0.883 Subsample from 1990:1 to 2004:4 E(CI) 0.574 (2. the link between the unemployment and property crimes were higher for the period 1990:1– 2004:4.112 0. the contemporaneous effect.246 0. Table 5 Synchronizations between the unemployment rate and property crimes in different subsamples.042 t-ratio 1.833 0. Fallahi.114 0.094 t-ratio 1.041 t-ratio 0.176 0.774 0. .152) The 1st Regime of Unemployment CI 0. empirical works have found mixed results. we divided the sample into three subsamples and checked the consistency of the results obtained across the subsamples.013 t-ratio 0.806 0.

Grant II. Int. 20 (1). 1987. Soc. C.L. 1985. One reason for this finding might be the fact that during recession.. The estimated models for the data captured the movements in these variables. bur- glary. Gómez. 2006. Furthermore.44 F. Forces 75. and the contemporaneous relationship was negative for motor-vehicle theft.. 769–799. 1994. J.. K. unemployment volatility. Rodríguez. J. Soc. 1997. Pap.. 50 (3). Unpublished paper.. the findings show that the opportunity effect is larger than the motivation effect for motor-vehicle theft. P.M... For example.). In other words. References Allen. K. R.D. 1969–1985. Synchronization of cycles. Unemployment and crime: is there a connection? Scand. 693–697. 169–217. Criminol. Delinq. (2) the second subsample covering the recession periods. Besides. 813–836. for burglary. Harding. 17 (4). We applied Markov-Switch- ing models to the unemployment rate and to the first difference of the four types of property crime to identify downturns and upturns for each of these variables. Toror.. Chiricos. Princeton. were used to test the existence of any relationships between the unemployment rate and property crimes in the US economy. we used quarterly data from the US economy over the period of 1977:2–2004:4 to examine the link between the unemployment rate. Reconsidering the unemployment and crime relationship: variation by age group and historical period. larceny. Biometrika 47. Oxford Econ. the crime rates dropped substantially in the US. Freeman. Econ.. C. Quant. it is worth noting that as we used the first difference of the property crime rates. 1996. 357–384. Soc. Rev. 440–448. P.. the results of these MS-AR models. Res. 2006.J. it is known that since the early 1990s. In other words. G. 14 (1).. 187–212. our results only explain the short-term relationship between the fluctuation or changes in the property crime rates and movements in the unemployment as an indicator of economic condition. 2001. D. 33–43..A.A. Soc. Rodríguez / Social Science Research 45 (2014) 33–45 In this study. D. 75–106. J. Criminol. Property crime-yes. Legal Stud. Britt. the results showed a procyclical but insignificant link between the unemployment conditions and these three property crimes. Am. Unemployment and crime rates in the post-World War II United States: a theoretical and empirical analysis. ‘‘Comment on Lauritsen and Heimer’’. Time Series Regression with Arima Noise and Missing Observations – Program Tram. O. G. Efficient tests for an autoregressive unit root. Don S. J. Princeton University Press. 1968.C. and Land.. G. and rob- bery. 56 (1). 34. Stock. Arvanites. The European business cycle. Econ.. Time Series Analysis. 1989. 2000. 76. 1996. Spengler. the motor-vehicle theft rate is low.B. jointly with concordance index.M. Crime and punishment: an economic approach.17 Lastly. we divided the sample into three subsamples: (1) the first subsample covering a period from 1990:1 to 2004:4. (Eds.. Sociol. 317–332. the unemployment rate was found to have a significant relationship with motor-vehi- cle theft. Econ. 139–164. violence-no. Polit. To check the stability of the relationship between the unemployment rate and property crime. Crime and the restructuring of the US economy: a reconsideration of the class linkages. H. Public Policy 9 (4).B. H. Zarkin. with the differenced data show- ing the short-term fluctuations only. Moreover. The findings showed that there is a positive but insignificant relationship between the unemployment rate. A. Business cycles and street crime. Florence. The unemployment rate. 41 (6). J.. and the estimated models for the property crimes have captured this fact correctly.. 1985. 2004. Econometrica 64 (4). 3529–3571. the co-movements between the unem- ployment rate and property crime deteriorated. Handbook of Labor Economics.. Card. J. H. 13 (4). Hamilton.J. Unemployment and crime: a neighborhood level panel data approach. and robbery.. 1999. Hypothesis testing when a nuisance parameter is only identified under the alternative. Criminology 44.. Fallahi. D. Time series analysis of crime rate. Ramiro. G. Sci. Sabbagh.. Pagan. larceny. J.. Socioeconomic conditions and property crime: a comprehensive review and test of the professional literature. 107 (2). T. Hale. 3... as an indicator of economic condition. These findings are in agreement with those of Phillips and Land (2012) who used annual crime data for the US. the results for the larceny must be used cautiously. Econ. A. Hamilton. more people stay at home and also they guard their properties. Maravall. when the unemployment rate is low. and (3) the last subsample covering the observations from the normal state of the economy. J. European University Institute. Land. the sign of the total effect changed over the time for all the property crime types.. Cook. Defina. Cantor. Davies. Elliott. Probl. Entorf. D. Criminol. the relationships found between unemployment and property crimes only showed that these property crimes and unemployment rate are following similar co- movements and that it does not necessarily show causality relationships. The counter-cyclical link between unemployment and motor-vehicle theft means that during recession.. T. Quant. 1987. The economics of crime. Martinez Jr. M... 2010.H. 1–44. R. 2012. J. The results showed that the link between unemployment and property crimes increased when we studied the first and third subsamples. 1992. vol.J. Econometrica 57 (2). Artis. Rates of crime and unemployment: an analysis of aggregate research evidence. C. Finally. 400–417.. Am. Rev. 291–327. J.. and four types of property crime.. D.. 1997. Becker. Andersen.. 405–428. Int. and it showed that the size of opportunity and motivation effects do change by time. J. H. Cook. 551. Testing the relationship between unemployment and crime: a methodological comment and empirical analysis using time series data from England and Wales. the results cannot be used to judge the long-term relationship between unemployment and property crimes. the size of motivation effect was found to be larger than the size of the opportunity effect. when the unem- ployment rate is high. 28 (4).. Next. These findings could be interpreted as a support for the opportunity effect. However. J. Sociol. G. 115–128. 1987. Res. Krolzig. as argued by Cantor and Land (1985) and Phillips and Land (2012). Cantor. J. 39 (6). Socioeconomic and demographic factors of crime in Germany – evidence from panel data of the German states. 2012. 132 (1). during the recession period. K. V. Crime and the business cycle. In: Ashenfelter. D. M. Fallahi. pp.S. Rothenberg. therefore. . Working Paper 92/81. 1615–1628.H. Unemployment and crime rates: A bivariate seasonal ARIMA analysis of monthly data. Econom. Law Econ... A new approach to the economic analysis of nonstationary time series and the business cycle. 59–79... T. F.D. 353–373. 293– 308. Pourtaghi. as larceny has a fairly low rate of reporting. Edmark.. and crime. 17 We are grateful to David Greenberg for pointing this out to us. However. 1991. R. R. Greenberg. 2005.

J. Soc.M. Int.J. 681–694. Soc. F.. Raphael. Criminol. 44 (1). 391–400. <http://www. 259–283. Lee.31k for Ox 3.uk/economics/staff/hmk/index. J. 2005.. MSVAR 1. state... R.kent. Witt. Phillips. G. 39 (3). S. Levitt. and national levels.ac. 983–1000.html>. Criminol.. Paleologou... S. 2001. C.. Sci. S. Alternative strategies for identifying the link between unemployment and crime.. AppliedEconomics Letters 13. 2012. A..C.. 2006. Labour market conditions and crime in Greece.. 377–390. H. R. Holoviak. 805–810. Land. Crime and economic activity: a panel data approach. Identifying the effect of unemployment on crime.M. J. Nigel. J. 2012. Clarke. Krolzig. Br. D. S. The link between unemployment and crime rate fluctuations: an analysis at the county. . 2001. Quant. 41 (3). Fallahi. F.. 39 (12). Unemployment and crime: an empirical investigation. K. Law Econ. Winter-Ebmer.4. 1999. J. Res. Econ. Rodríguez / Social Science Research 45 (2014) 33–45 45 Kollias.D. 17 (4).