0115 966 7955 Today's Opening Times 10:00 - 20:00 (BST)

Effect Of Unemployment On Crime

Published: Last Edited:

Disclaimer: This essay has been submitted by a student. This is not an example of the work written by our professional essay writers. You can view samples of our professional work here.

Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of UK Essays.

This paper follows an outline commonly known as the panel data technique to try and observe the effect or the relationship that exists between unemployment and a variety of categories of crime reported in the New York. The data set spans around sixteen regions during the period from 1984 to 1996. Fixed as well as random models are approximated to examine the prospect of a causal correlation between unemployment and crime. Hypothesis tests carried out have revealed that only the two-way fixed effects models have to be used. The most important end result of the paper is that there exists substantial proof of considerable effects of unemployment on crime, mutually for overall crime as well as for some subcategories of crime. The effect shows that crimes associated with properties are considerably discouraged by higher clear-up rates. In addition for property crime tolls, the end results point out that unemployment boosts crime. However, for violent crimes, consequence of the clear-up rate as well as unemployment is established as being insignificant.

Model Formulation

It is an ordinary observation for several countries that crime rates as well unemployment rates is positively correlated. An even more controversial issue is whether this relationship implies that unemployment lead to crime or whether crime causes unemployment or third issues cause both. Just the former of the three potential cases would imply that the effects of unemployment on crime warrant to be reckoned amongst the "non-pecuniary" costs of unemployment that must be considered during the analysis of cost-benefit of probable policies of cutting down on unemployment.

The hypothetical reinforcement of the causality perception was developed in over thirty years ago by Becker, and Ehrlich. According to Ehrlich's model, persons split their time among authorized activities along with hazardous illegitimate activities. In the event that legal earnings opportunities happen to scarce comparative to potential gains realized from crime, the model forecasts that crime will turn out to be more recurrent. A substantial rise in unemployment case might be one of the attributing factors

Various consequent empirical papers have made attempts to analyse the forecast of the Becker-Ehrlich model as well as to locate out whether the enormity of the unemployment effect is quantitatively significant. The trademark of this literature is its breakdown and the failure to attain a consensus as to whether high levels of joblessness lead to a bigger crime frequencies. In an analysis of the literature, Chapman provides information consisting of over 35 consistent studies on the topic, among which 20 were able to establish a positive connection involving unemployment rates and crime, with the rest not being able to establish such a relationship. In the framework of New Zealand, a preceding econometric study by Small and Lewis (1996), founded and utilizing the time series as well as Granger causality tests, led to a "strong support to the idea that crime and unemployment are linked in some way". (Chapman 48) Additionally, the results revealed that unemployment cases were more likely to cause crime more than vice versa.

Primarily, the purpose of this paper is to revisit the subject of whether unemployment has a contributory effect on different classes of anti-social and economic crime. For this reason, the paper shall analyse New Zealand local panel data, relapsing crime rates on unemployment rates by means of random and fixed effects models. The type of approach employed solves numerous of the trouble that have been typical of preceding empirical papers. Particularly, we ought not to reject the hypothesis that unobservable period precise effects are related with the joblessness rate. This finding proposes that time series regressions will possibly be exaggerated by absent variable bias. Still, one ought to find some piece of evidence which proposes that unemployment matters irrespective of region, time, deterrence as well as income accounting effects accounted for.

Data Sources

The data set employed throughout the paper is a panel of annual and regional level observations all based on the level of crime. The figures we obtained from the New Zealand Police for the period from 1984 to 1996 from sixteen police districts. This data was inclusive of the number of offences accounted to every police in any of the police district for seven offence set in addition to the total number of crimes, which they mutually consist of. Figures are changed into crime rates by division with the local population size, mostly estimated in terms of thousands. The described crime rate for every individual class is denoted by o1 to o7, in that order, whereas the overall crime rate is characterized by o (representing an "offence"). The various clusters of crimes used by New Zealand Police are classified as "violent offences, drug and anti-social offences, dishonesty offences, property damage offence, property abuse offences, sexual offences and administrative offences". Therefore on ought to notice that the overall rate of crime is thus a collection of very different types of crimes. Still within every category, there is a considerable degree of heterogeneity. Although it would be advantageous to bring more homogeneous groups into play, such kind of data was not available. The heterogeneity meant that every unemployment effect wants to be understood as an average effect that may be different from the result of unemployment on any constituent crime. The heterogeneity is additionally likely to steer up the outstanding variation, and in so doing the standard errors, associated with the estimates.

There exists a wide range of measure of unemployment in New Zealand. The certified measure is obtained from the Household Labour Force Survey (HLFS), which is a quarterly analysis conducted by Statistics New Zealand. The HLFS supplies the estimates which are globally analogous and are not prone to changes in the description of "being unemployed". Sorry to say, the series only comprises sub-national estimates as at 1990. The quinquennial Census of Population and Dwellings presents the most comprehensive survey of unemployment in New Zealand. On the other hand, drawbacks are the uncommon of observations presented and the actuality that varied definitions of unemployment have been practical over a long period of time.

Due to these shortcomings, the computation of unemployment chosen for this study was the figure of individuals listed as unemployed with the Department of Labour, hereby designated by UN. (Statistics New Zealand 10) yearly averages of this particular series were acquired for each of 21 employment districts for the similar stage from Statistics New Zealand's INFOS and were later matched with the 16 police districts. (Stigler 526) Unemployment rates were acquired by division with population estimation and were indicated by un. there is need to acknowledge that this is a less perfect calculation of legal employment prospects. A fraction of unemployment is associated to job search, the span of which is dependent on several factors apart from availability of jobs, as well as eligibility and benefit levels. Preferably, it could have been beneficial to include a measure involving cases of long-term unemployment. However such a statistic was is offered for the complete time period along with the necessary regions.

Additionally, other probable crime level determinants other than unemployment are also considered in the study. First of all, the clearance rate for every offence group was acquired from the New Zealand Police. This is set by the ratio of the number of crimes solved by police with respect to the overall number of crimes accounted for each region and the sub-category of the crime. The general clearance rate is symbolized p while the rates of clearance for each crime group are indicated p1-p7; anywhere the index matches the offence sub-category number.

Secondly, the piece of information regarding the average level of revenue for each area was obtained. As there is no yearly sub-national data for earnings in New Zealand, information obtained during the 1986, 1991 and 1996 Censuses on mean personal income from each police district was used. To get hold of an entire panel, the income of each district relative to the national average was considered, by means of a linear time movement to extrapolate the omitted observations. The earnings series employed in this study, designated y, is the result of this particular series and the estimate of the relative earnings of each region for the right year.

Estimation Procedure, Results and Analysis

The figure below shows the plots of the national rate of unemployment in relation to the total crime rate from 1978 to 1996.5 from the table it is visually evident that the type of series tends to move closely jointly over time. Furthermore r- the correlation coefficient - can be determined to be 0.41. Conversely, as stated in the beginning, this is by no means a clue of causality between unemployment to crime. Despite the fact that the premise of reverse causation (from crime to unemployment) seems improbable a priori, third variables are prone to have an effect on both crime and unemployment.

In a bid to assess and manage the direct potential influence of third factors, the study should employ least square multiple regression analysis. The general stratagem is to start with an extremely straightforward model and sequentially generalize. Following Myers, we make use of a log-log specification approach of the unemployment-crime relation in all instances. This provide rise to an approximate coefficient that has the explanation of elasticity. The log-log model is also reliable with Myers, who recommended a multiplicative structure for the supply-of-offences function in the event that variables are arranged according to levels. A tremendous likelihood is that unemployment is the only determinant of the crime rate and while other constraints of the model are indistinguishable, in spite of what year or region the observation is taken from (Myers 125).

This has the implication that that a collective regression can be practical to the data as shown below;

In the above equation, the subscript i signifies the area of the observation as well as t the year the observation was recorded. Denotes the residual connected with observation it. In addition εit or the error term is supposed to have mean zero. Furthermore,

As a result, the model permits for region-specific heteroskedasticity that is mainly applied while computing the variance-covariance matrix of the OLS estimator for β.

In table 2 below, the first column reports that estimate of the parameters gotten from the pooled regressions of the logarithmic rates of crime on the logarithmic rates of unemployment. These are founded on a total of eight divided regressions. The value of the total crime rate is positive and statistically different from zero and is therefore displayed in row 1. The figure 0.144 signifies that a 10% raise in the unemployment rate is related with a corresponding 1.4% increase in the crime rate.

Going all the way through the remaining coefficients of the first column divulges that for most subclasses of transgression, the approximate elasticity is considerably larger than in total. In the event of organizational offences, the elasticity reaches almost unity. The elasticity is considerably diverse from zero in six out of seven subclasses of crime. Thus, there appears to be proof for a considerable effect of unemployment on crime.

Obviously, this conclusion is provisional on the legitimacy of the model. Next, there is the need to initially examine how suitable such a regression is, provided with the data used in this study. The graphs do not reveal any relationship linking oi and uni and if it exists, the relation in question is negative, but signifies that ot and unt are optimistically correlated. With specific factors which act to bar the inability to account for a positive relation, As in Figure 2 or there are time-specific reason which produce the manifestation of a link involving unemployment and crime over time or a mix of both situations exists. Impending misspecifications of this type can be addressed by the insertion of area or even time-specific permanent effects or even both in the regression.

Officially, let

Where μi denotes a region of specific fixed effect, λto on other hand is considered as a time specific effect. On the other hand, εit characterizes a white noise error term as stated before. This is an indistinguishable requirement to the pooling case, and only the μi and λt are well thought out as the parameters to be approximated, while prior to they were a component of the error term. A probable factor that may feature in μi might be the extent of urbanisation.

The period specific effects λt capture, for instance, any change in macroeconomic conditions, such as inflation or oil-price shocks that might be expected to lead to higher levels of crime (assuming that they affect all regions equally). In addition, these effects account for changes in the propensity to report crimes over time as long as these are uniform across regions.

The expected values of the estimation results are given in columns 2-4 of table 1. Column 2 is inclusive of the time effects but have no regional effects. Under a similar capacity column 3 comprises of region effects but no time effects. On the other hand, we realize that column 4 is all inclusive with both time and region effects. As it would be expected table 1 and 2 the estimated employment elasticity is more adversely affected by the addition of period fixed effects than by the inclusion of region fixed effects. In all except one case, the degree of the unemployment elasticity plummets when time effects are incorporated and increases when the area effects are taken into account. In the most wide-ranging model, incorporating both time effects and the region, an overall fall in the elasticity is experiential, because the time effect dominates the effect of the region. The flexibility for the entire crime rate is about halved in enormity, in relation to the joint model with no fixed effects. In stipulations of arithmetical significance, barely three of the eight approximate elasticities stay put and significant - Sexual, dishonesty as well administrative offences - once region as well as time fixed effects are incorporated in the model.

The importance of the two types of fixed effects can be put into test by a employing various F-tests. The most relevant statistics are reported in table 2. In column 1, the two-way fixed effects representation is put side by side to a model with region effects only, that is the null hypothesis λt = 0 for all t. A contrast of the F-statistic with the significant value of 1.83 demonstrates that the time effects are jointly significant. A comparable conclusion may be deduced from the subsequent column of Table 2, with consideration to the region effects. From this perspective, the two-way fixed effects model is the better model. Nevertheless, as far as evaluation of the unemployment elasticity of crime is taken into consideration, efficiency gains may be realised by not including a number of of the fixed effects and modelling them as random effects as an alternative. In spite of everything, 27 degrees of freedom are nowhere to be found in the two-way fixed effects model comparative to the simple model. As a result, standard errors increase, for instance where the overall crime rate shots up by over 100%. The degrees of freedom may perhaps be conserved if the region and time effects are modelled as random effects. So as to accomplish this, one is required to assume that the unemployment rate is In order to do so, one need to assume that the unemployment rate is uncorrelated with the time and region effects, correspondingly.

Given that the assumption is valid, then the fixed effects estimator is incompetent and this calls upon the use of a random effects estimator. If not, the random effects estimator for the unemployment elasticity is inconsistent and biased. In quintessence, the resolution is whether to create inferences restricted to the effects pragmatic in the sample or unrestricted inferences with reverence to the characteristics of the population. The legitimacy of the supposition of no correlation can be tested using Hausman's (1978) test. This test entail contrasting the estimated constraint values for the unemployment elasticity in both random and fixed effects specifications. Following the null hypothesis that unemployment is uncorrelated with the random effects, the estimated coefficients using either model are reliable but just the random affects estimator is efficient. Nevertheless, in the event of a false null hypothesis, the use of random effects generates a non consistent estimator while the fixed effects estimator remains consistent.

Based on the model exclusive of time effects (Column 5) correlation involving unemployment rate and the error constituent is perceived in three out of seven cases. Given that the time effects are included, the null hypothesis is constantly acknowledged. Founded on these a choice of tests, one can confidently argue whether or not the regional effects ought to be modelled as fixed or as random. Whereas the regional effects are mutually highly significant -- as depicted in table 2 columns 2, the evidence for a correlation with unemployment is not strong. On the other hand, the two-way fixed effects model appears to be justifiable in order to hedge the results alongside this type of omitted inconsistent bias for all subclasses of transgression.

Conclusion

After, "having further muddied that already turbid waters of research" concerning the unemployment-crime relationship, facts have been established that tend to substantiate earlier conclusions made by Lewis and Small for New Zealand. End results have depicted that the overall rate of crime remains considerably affected by the unemployment rate. In particular, unemployment was established to have a major relationship to the number of treachery crimes committed. In addition, this study points to several possibilities for more research. In particular, the introduction of extra regressors that might clarify crime, for instance income inequality, may possibly lighten any remaining omitted inconsistent bias.

The unemployment-crime correlation is a mature issue. No compromise has been reached by economists for the duration of the past three decades, nor does one seem possible to emerge in the near future. Conceivably a study by Grogger on this area of study has particular significance: with the intention to "urge and achieve a certainty which simply does not exist" (Grogger 52).

Works Cited

Chapman, J.I. (1976). An economic model of crime and police. Journal of Research Crime and Delinquency, 13, 48-63.

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

Grogger, J. (1995). The effect of arrest on the employment and earnings of young men. Quarterly Journal of Economics, 110, 51-72.

Myers, S. (1983). Estimating the economic model of crime: Employment versus punishment effects. Quarterly Journal of Economics, 98, 157-166.

Statistics New Zealand (1996). New Zealand now: Crime tables. Wellington: Statistics New Zealand.

Stigler, G.J. (1970). The optimum enforcement of laws. Journal of Political Economy, 78, 526-536.


To export a reference to this article please select a referencing stye below:

Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.
Reference Copied to Clipboard.

Request Removal

If you are the original writer of this essay and no longer wish to have the essay published on the UK Essays website then please click on the link below to request removal:


More from UK Essays

We can help with your essay
Find out more