Testing Theory and the Analysis of Time Series Data

Although the relationship between unemployment and crime has been a longstanding interest in criminology, there is little agreement about appropriate models for analyzing this relationship. David Greenberg's (this issue) discussion highlights two issues that raise questions about recent research on the unemployment–crime relationship. First, he extends the work of Hale and Sabbagh (1991) and argues that cointegration methods should be used instead of first-differenced regression models to analyze unemployment and crime time series data. Second, he argues that previous attempts to test his strain theory linking unemployment to the age distribution of crime rely on flawed hypotheses, inappropriate data, and faulty measurement. In this paper, I address both of Greenberg's claims. I begin with a discussion of the relative utility of cointegration analysis and of first-differenced regression models for the analysis of the unemployment–crime relationship, focusing on the link between theory and statistical model. I then discuss the possibility of ever testing and falsifying Greenberg's strain theory.     

Theory, Operationalization, Identification, and the Interpretation of Different Differences in Time Series Models

The principals in this exchange, David Cantor, David Greenberg, and Kenneth Land, have each contributed outstanding methodological and substantive scholarship to criminology and to the social sciences more generally. In exchanges such as this we are engaged in a collective process of refining and advancing our knowledge in the areas of both criminology and statistical methods. The issues raised by Greenberg provide an opportunity to reflect on time series modeling and to revisit some issues concerning operationalization and model specification. I do not expect any of us to agree fully on these issues, but we may be able to shed some light on these issues, and collectively this may help the larger community of criminologists and social scientists.     

Unemployment and Crime Rate Fluctuations: A Comment on Greenberg

Several ways in which the specification of the Cantor and Land (1985) conceptual model of transient relationships between aggregate unemployment and crime rate fluctuations differs from that of Greenberg (2001) are noted. It follows that we do not accept Greenberg's Eq. (1) as a valid theoretical representation of the processes of interest. We briefly review the substantive context from which our investigation began in the mid-1980s. We also review the time series properties of our model and of the aggregate unemployment and crime rates used in its estimation. We note how the time series behavior of various crime rates determines which parts of the Cantor and Land model are and are not likely to be estimated as statistically significant for those series. We conclude with some comments on the limitations of aggregate time series research designs for testing the behavioral hypotheses used to generate expected relationships between aggregate unemployment and crime rates and suggest some alternative research designs.     

Alternative Strategies for Identifying the Link Between Unemployment and Crime

National-level time series data are a crude tool for distinguishing between two alternative behavioral explanations for a link between unemployment and crime. Consequently, inferences drawn from aggregate time series estimates are likely to be misleading. A more fruitful approach to learning about the link between unemployment and crime would be to utilize a menagerie of different methodological approaches such as cross-section and panel data analysis of less geographically aggregated areas, natural experiments, international data, individual-level data, and ethnography.     

Theoretical and Empirical Work on the Relationship Between Unemployment and Crime

Part of the ongoing debate between Cantor and Land and Greenberg centers on differing opinions about the question of interest in Cantor and Land (1985). We begin this article with our opinion that Cantor and Land's theory relates changes in the business cycle to changes in the aggregate rate of crime. We then question whether year-to-year changes adequately reflect changes in the business cycle, which last on average 4 years, and we refer to an article by Cook and Zarkin (1985) which presents an alternative method of measuring business cycle changes. We also discuss how Greenberg's use of cointegration provides an alternative way of addressing the difficult statistical problem of nonstationarity without resorting to first differences. Our final contribution involves noticing that opportunity and motivational theories of crime can be structurally identified by focusing on different types of crime rather than temporal lags. We demonstrate this idea by splitting car theft into joyriding and theft for profit. We show that joyriding appears to be driven by opportunity, while the causal structure of theft for profit is less clear.     

Poisson-Based Regression Analysis of Aggregate Crime Rates

This article introduces the use of regression models based on the Poissondistribution as a tool for resolving common problems in analyzing aggregatecrime rates. When the population size of an aggregate unit is small relativeto the offense rate, crime rates must be computed from a small number ofoffenses. Such data are ill-suited to least-squares analysis. Poisson-basedregression models of counts of offenses are preferable because they arebuilt on assumptions about error distributions that are consistent withthe nature of event counts. A simple elaboration transforms the Poissonmodel of offense counts to a model of per capita offense rates. Todemonstrate the use and advantages of this method, this article presentsanalyses of juvenile arrest rates for robbery in 264 nonmetropolitancounties in four states. The negative binomial variant of Poisson regressioneffectively resolved difficulties that arise in ordinary least-squaresanalyses.     

Prison commitments,crime, and unemployment: A theoretical and empirical specification for the United States, 1933–1985

This paper examines sources for the changing commitment rates to U.S. state prisons (PCR) from 1933 to 1985 using a variety of time-series techniques. Theoretically, it resolves ambiguous interpretations of how crime, unemployment, and imprisonment are related. Hypotheses that crime and punishment are in equilibrium are rejected. Our final specification supports theories integrating institutionally endogenous and socially exogenous causes of prison use and includes feedback effects between crime and punishment. We reached several general conclusions. (1) Changes in PCR are due partly to changes in the levels of unemployment, age composition of the population, and military active-duty rates. (2) Effects of the criminal justice system, captured as autoregressive institutional drift, account for approximately half of the year-to-year fluctuations in the PCR. The contemporaneous prison discharge rate also influences the rate of prison commitments. (3) Neither the specified nor the unspecified institutional effects mediate the effects of other exogenous variables. (4) Under most simultaneous-equation specifications, the crime rate is moderately influenced by the contemporaneous unemployment rate and strongly influenced by prior levels of prison commitments. The preferred simultaneous causal model estimates a modest positive coefficient for the unemployment-crime causal path and a substantial positive coefficient for the unemployment-prison commitment causal path.     

Revisiting the Relationship Between the Economy and Crime: The Role of the Shadow Economy

Despite decades of research exploring the relationship between the economy and crime, there is a lack of clarity in this literature. Questions remain, particularly with respect to how the economy is measured and whether the relationship between the economy and crime is the same across contexts. The literature to date has overlooked what is called the “shadow” economy, which represents unreported economic activity. We examine the relationship between traditional measures of the legitimate economy (e.g. unemployment) and crime as well as whether the shadow economy moderates this relationship using a panel of U.S. states from 1997 to 2008. Our results show that the shadow economy reduces the strength of the relationship between the legitimate economy and crime, and the effect of the economy on crime is conditional on the size of the shadow economy.     

Unemployment, Economic Theory, and Property Crime: A Note on Measurement

Drawing on rational choice theory, this study considers how best to measureunemployment within the context of the unemployment–property crimerelationship. Specifically, we use ARIMA techniques to examine the relativeefficacy of using the conventional Bureau of Labor Statistics (BLS)unemployment rate and two alternative measures of the demand for labor aspredictors of monthly counts of U.S. property offenses for the years 1982through 1996. The bivariate time series analyses indicate that while theBLS unemployment rate exhibits null effects, the number of individualsunemployed for 15 weeks or more and the capacity utilization ratesignificantly affect the level of property crime. The implications ofthese results are discussed.     

Crime, Prison, and Female Labor Supply

This paper introduces the vector autoregressive (VAR) methodology into themainstream of applied criminological research as an alternative to theestimation of structural models of crime. The paper presents cointegrationtest statistics for a second-order VAR of crime, prison population, femalelabor force participation, and durable consumption. Our results for theimpact of imprisonment on the crime rate are similar to Levitt's (1996)and substantially larger than previous estimates (e.g., Marvell and Moody,1994). We find that the short-run response of the crime rate to increasedlabor force participation of women is larger than the long-run effect. Theimplication is that major social changes, like the increased labor supply ofwomen, may have surprising impacts and that social institutions may takesome substantial period of time to adjust to such major changes.     

On Theory, Models, Model-Testing, and Estimation

I reply to comments and criticisms of my article raised in this Special Issue by Cantor and Land, Britt, O'Brien, Levitt, and Paternoster and Bushway.     

Unemployment and Crime: Toward Resolving the Paradox

While official crime statistics from many countries show that unemployed people have high crime rates and that communities with a lot of unemployment experience a lot of crime, this cross-sectional relationship is very often not found in time-series studies of unemployment and crime. In Australia there have been no individual-level or cross-sectional studies of unemployment and adult crime which have failed to find a positive relationship and no time-series which have supported a positive relationship. Consistent with this pattern, a time series of homicide from 1921 to 1987 in Australia reveals no significant unemployment effect. A theoretical resolution of this apparent paradox is advanced in terms of the effect of female employment on crime in a partriarchal society. Crime is posited as a function of both total unemployment and female employment. When female employment is added to the model, it has a strong positive effect on homicide, and unemployment also assumes a strong positive effect.     

Reconsidering the unemployment and crime relationship: Variation by age group and historical period

Although considerable research has examined the direct effects of age and economic conditions on crime, relatively little work has investigated the joint influence that age and unemployment may have on rates of criminal behavior. This study extends prior research on the unemployment-crime relationship by testing simultaneously for (1) variation in the unemployment-crime relationship by age group and (2) variation in the unemployment-crime relationship over time. Age-specific arrest and unemployment time-series data for the United States from 1958 to 1995 are used to test these hypotheses. The two main findings indicate that (1) unemployment has a greater motivational effect on property crime among youth and young adults and (2) the unemployment-crime relationship varies over time, but in a way that appears to be more random than systematic. The implications for future research are discussed.     

国际犯罪与涉外犯罪、跨国犯罪关系辨析——以基础概念解析为进路

国际犯罪与涉外犯罪、跨国犯罪之间虽然在起源上具有不可磨灭的历史渊源,但它们在内容上并不具有同一或者包容关系,而是一种即相互交叉又彼此保持一定独立性的关系。只有当涉外犯罪、跨国犯罪具有"国际性"时,即它们的危害性达到国际刑法所要求的严重程度时,才能上升为国际犯罪。而在国际犯罪中,除了具有国际性的涉外犯罪和跨国犯罪外,还包括具有国际性的既不属于涉外犯罪又不属于跨国犯罪的单纯的国内犯罪。     

Violent Crime in the United States of America: A Time-Series Analysis Between 1960–2000

This paper empirically assesses the effects of socio-economic and demographic variables on violent crime in the United States. Using national-level time-series data over the period 1960–2000, an unrestricted vector autoregressive (VAR) model was estimated for overall violent crime, murder, rape and assault. The results indicate that there is no long-run relationship among the examined variables, but significant short-run relationships hold. Imprisonment growth, income inequality, alcohol consumption, and racial composition of the male youth population are shown to influence the short-run behaviour of violent crime.     

Structural Covariates of Violent Crime Rates in Germany: Exploratory Spatial Analyses of Kreise

This research applies the techniques of exploratory spatial data analysis (ESDA) and spatial regression modeling to explain variation in robbery and assault rates across 413 districts or “Kreise” in Germany. The findings from ESDA reveal a distinct clustering of low rates in southern Germany, especially for robbery. The results of the spatial regression analyses indicate that for the nation at large, Kreise with high robbery and assault rates tend to be those with comparatively high levels of socioeconomic deprivation and a more urbanized environment, findings consistent with those commonly reported with data for areal units in the USA. We also observe net regional effects for the south in the regression models that pose puzzles for further inquiry into the German case, and “null effects” of the eastern region that have implications for more general debates of the potentially criminogenic consequences of the transition to market economies.     

论独立教唆犯之犯罪预备性——以我国刑法第29条第2款的规定为视角

独立教唆犯是预备犯,这是因为,从教唆行为的本质上看,教唆行为是教唆者所教唆之罪的犯罪预备行为,且这种犯罪预备行为的本身在法律上不具有正当性。为了尽早阻断教唆行为对法益的破坏作用,有的国家在刑法总则中对独立教唆犯予以原则性地处罚规定,采取的是非独立预备犯的立法模式,将犯罪构成要件的可罚性前置化。采取独立预备犯立法技术的国家,在刑法分则中为一些独立教唆行为规定了特定的犯罪构成要件,使一些教唆行为成为其相应的犯罪构成要件的实行行为,但不是教唆者所教唆之罪的犯罪实行行为。     

犯罪中止时间性构成条件探微

根据现行刑法规定,犯罪中止成立的必备条件之一是必须发生在“犯罪过程中”,即人们所谓的时间性构成条件从刑法的谦抑性思想和主客观相统一原则的角度而言,“犯罪过程”应始于着手犯罪,而非始于犯罪预备行为;为了贯彻鼓励行为人主动放弃犯罪,有效地保护社会合法权益的立法精神,应将“犯罪过程”的终点确定为发生行为人追求的危害结果。     

Economic and Sociological Correlates of Suicides: Multilevel Analysis of the Time Series Data in the United Kingdom

For the effects of social integration on suicides, there have been different and even contradictive conclusions. In this study, the selected economic and social risks of suicide for different age groups and genders in the United Kingdom were identified and the effects were estimated by the multilevel time series analyses. To our knowledge, there exist no previous studies that estimated a dynamic model of suicides on the time series data together with multilevel analysis and autoregressive distributed lags. The investigation indicated that unemployment rate, inflation rate, and divorce rate are all significantly and positively related to the national suicide rates in the United Kingdom from 1981 to 2011. Furthermore, the suicide rates of almost all groups above 40 years are significantly associated with the risk factors of unemployment and inflation rate, in comparison with the younger groups.