The 43 reference contexts in paper Giovanni Cerulli, Maria Ventura, Christopher F Baum (2018) “The Economic Determinants of Crime: an Approach through Responsiveness Scores” / RePEc:boc:bocoec:948

  1. Start
    2130
    Prefix
    There is no single cause identifying the different levels of crimes over time, as a number of determinants, often interacting, contribute to their variations. These may range from social to geographical and historical causes, and events whose effect is only indirect, but equally strong. For instance,
    Exact
    Levitt & Dubner (2005)
    Suffix
    argue that the legalization of abortion throughout the country in 1973 has been critical in reducing crime rates in the following generation, and attribute this to the decrease in the birth rates among the most disadvantaged or unstable social categories.
    (check this in PDF content)

  2. Start
    2856
    Prefix
    The issue with this type of setting is that most of the previous literature in the field have mainly analysed each driver individually, without necessarily providing a global account of the phenomenon. For instance,
    Exact
    Lochner (2004)
    Suffix
    focused on the role of education, while Fowles & Merva (1996) focus on the effect of changes in the distribution of wage income. Although an intensive knowledge of each single factor can definitely offer some critical insight, we believe that gathering the main determinants and observing how those interact in an inclusive context should result in more realistic scenarios that would refle
    (check this in PDF content)

  3. Start
    2920
    Prefix
    The issue with this type of setting is that most of the previous literature in the field have mainly analysed each driver individually, without necessarily providing a global account of the phenomenon. For instance, Lochner (2004) focused on the role of education, while
    Exact
    Fowles & Merva (1996)
    Suffix
    focus on the effect of changes in the distribution of wage income. Although an intensive knowledge of each single factor can definitely offer some critical insight, we believe that gathering the main determinants and observing how those interact in an inclusive context should result in more realistic scenarios that would reflect society in all its aspects.
    (check this in PDF content)

  4. Start
    3936
    Prefix
    standard approach contains the obvious limitation of assuming that coefficients are identical across units and time, thus ignoring potential heterogeneity in the impact of the factors, whose effects are likely to vary according to the geographical and historical framework. Therefore, we choose instead to adopt the method of responsiveness scores, whose successful application by
    Exact
    Cerulli (2014)
    Suffix
    in the field of technological innovation has raised the possibility for its use in different fields. Responsiveness scores are based on an iterated Random Coefficient Regression model (Wooldridge (2010)) for each unit and for each factor.
    (check this in PDF content)

  5. Start
    4144
    Prefix
    Therefore, we choose instead to adopt the method of responsiveness scores, whose successful application by Cerulli (2014) in the field of technological innovation has raised the possibility for its use in different fields. Responsiveness scores are based on an iterated Random Coefficient Regression model
    Exact
    (Wooldridge (2010))
    Suffix
    for each unit and for each factor. A more detailed and technical description of the underlying 2 econometrics follows. The main advantage of this technique is that it relaxes the classic assumption that each observation of the population has the same slope, thus allowing for idiosyncratic responses.
    (check this in PDF content)

  6. Start
    5552
    Prefix
    and provides a rationale, based on past research, for the obtained outcomes while Section 5 concludes. 2 Theoretical background of the determinants of crime The causes of crime have been widely studied by social sciences, with economic determinants acquiring greater relevance during the last decades. Although the analysis of the effect of income on delinquency was not new to the field (e.g.,
    Exact
    Fleisher (1966)),
    Suffix
    this stream of modern literature was pioneered by Becker (1968). In his seminal paper, he presents the criminal’s choice, intended as the main trigger of the “supply of offenses,” as a standard microeconomic problem of expected utility: the individual chooses whether to commit a crime by comparing its expected benefits with its costs, which also can include the loss of an outside option, usuall
    (check this in PDF content)

  7. Start
    5619
    Prefix
    while Section 5 concludes. 2 Theoretical background of the determinants of crime The causes of crime have been widely studied by social sciences, with economic determinants acquiring greater relevance during the last decades. Although the analysis of the effect of income on delinquency was not new to the field (e.g., Fleisher (1966)), this stream of modern literature was pioneered by
    Exact
    Becker (1968).
    Suffix
    In his seminal paper, he presents the criminal’s choice, intended as the main trigger of the “supply of offenses,” as a standard microeconomic problem of expected utility: the individual chooses whether to commit a crime by comparing its expected benefits with its costs, which also can include the loss of an outside option, usually represented by income from a legal, and less risky, activi
    (check this in PDF content)

  8. Start
    6200
    Prefix
    : the individual chooses whether to commit a crime by comparing its expected benefits with its costs, which also can include the loss of an outside option, usually represented by income from a legal, and less risky, activity. He also introduces the theme of punishment of crime, which enters the problem both in the forms of probability of being caught and magnitude of the penalty.
    Exact
    Ehrlich (1973)
    Suffix
    refines and extends this model, and gives a greater twist to the discussion of the responsiveness of individuals to economic incentives and to their interaction. In this context, many other factors besides individual income can be included in the analysis of the determinants of crime, as they modify people opportunities in legal activities, and therefore their expected returns from engaging in o
    (check this in PDF content)

  9. Start
    7426
    Prefix
    Through income effects, education increases expected wages, and therefore the returns from legitimate work, increasing the opportunity cost of crime. Following this idea, most scholars would predict a decrease in criminal activity for an increase in education, although
    Exact
    Levitt & Lochner (2001)
    Suffix
    find that certain mechanical knowledge and related disciplines can improve chances in the criminal world. Resources allocated to education also create time constraints that should work in keeping teenagers away from committing offenses.
    (check this in PDF content)

  10. Start
    7689
    Prefix
    Following this idea, most scholars would predict a decrease in criminal activity for an increase in education, although Levitt & Lochner (2001) find that certain mechanical knowledge and related disciplines can improve chances in the criminal world. Resources allocated to education also create time constraints that should work in keeping teenagers away from committing offenses.
    Exact
    Witte & Tauchen (1994)
    Suffix
    find that greater time spent in school is correlated with a lower probability of criminal activity. Moreover, a stream of literature also associates greater education with higher life satisfaction, as in Oreopoulos (2007) and Lochner (2004) finding that higher levels of education increase patience and risk aversion, thus lowering crime.1Usher (1997) considers a fourth channel, a civic externality
    (check this in PDF content)

  11. Start
    7917
    Prefix
    Witte & Tauchen (1994) find that greater time spent in school is correlated with a lower probability of criminal activity. Moreover, a stream of literature also associates greater education with higher life satisfaction, as in
    Exact
    Oreopoulos (2007) and Lochner (2004)
    Suffix
    finding that higher levels of education increase patience and risk aversion, thus lowering crime.1Usher (1997) considers a fourth channel, a civic externality of education, which is assumed to affect one’s willingness to commit an offense.
    (check this in PDF content)

  12. Start
    8050
    Prefix
    Moreover, a stream of literature also associates greater education with higher life satisfaction, as in Oreopoulos (2007) and Lochner (2004) finding that higher levels of education increase patience and risk aversion, thus lowering crime.1
    Exact
    Usher (1997)
    Suffix
    considers a fourth channel, a civic externality of education, which is assumed to affect one’s willingness to commit an offense. In general, we would therefore expect a negative relation between educational attainment and crime.
    (check this in PDF content)

  13. Start
    8473
    Prefix
    In general, we would therefore expect a negative relation between educational attainment and crime. Labor market conditions have also been widely used to motivate different incidence of crime. Although employment status has been prevalent in the early literature, with
    Exact
    Fleisher (1963)
    Suffix
    establishing a positive effect of unemployment on delinquency, modern works have rather focused on the pecuniary aspect of labor market. Gould et al. (2002) remark that, as employment is highly cyclical, it can hardly explain a long-term trend in crime, as observed in the United States.
    (check this in PDF content)

  14. Start
    8624
    Prefix
    Although employment status has been prevalent in the early literature, with Fleisher (1963) establishing a positive effect of unemployment on delinquency, modern works have rather focused on the pecuniary aspect of labor market.
    Exact
    Gould et al. (2002)
    Suffix
    remark that, as employment is highly cyclical, it can hardly explain a long-term trend in crime, as observed in the United States. Moreover, Gumus (2003) distinguishes between short-term and long-term effects of unemployment, and concludes that while 1The attitude toward risk is a factor that Becker (1968) also takes into consideration in his model. 4 after being unemployed for a short period p
    (check this in PDF content)

  15. Start
    8786
    Prefix
    Although employment status has been prevalent in the early literature, with Fleisher (1963) establishing a positive effect of unemployment on delinquency, modern works have rather focused on the pecuniary aspect of labor market. Gould et al. (2002) remark that, as employment is highly cyclical, it can hardly explain a long-term trend in crime, as observed in the United States. Moreover,
    Exact
    Gumus (2003)
    Suffix
    distinguishes between short-term and long-term effects of unemployment, and concludes that while 1The attitude toward risk is a factor that Becker (1968) also takes into consideration in his model. 4 after being unemployed for a short period people tend to look for another job, a long spell of unemployment increases the likelihood of criminal activity.
    (check this in PDF content)

  16. Start
    8937
    Prefix
    Gould et al. (2002) remark that, as employment is highly cyclical, it can hardly explain a long-term trend in crime, as observed in the United States. Moreover, Gumus (2003) distinguishes between short-term and long-term effects of unemployment, and concludes that while 1The attitude toward risk is a factor that
    Exact
    Becker (1968)
    Suffix
    also takes into consideration in his model. 4 after being unemployed for a short period people tend to look for another job, a long spell of unemployment increases the likelihood of criminal activity. The wage from legal activities matters both as a component of income, and as the opportunity cost of criminal actions.
    (check this in PDF content)

  17. Start
    9297
    Prefix
    toward risk is a factor that Becker (1968) also takes into consideration in his model. 4 after being unemployed for a short period people tend to look for another job, a long spell of unemployment increases the likelihood of criminal activity. The wage from legal activities matters both as a component of income, and as the opportunity cost of criminal actions. Concerning the first aspect,
    Exact
    Buonanno (2003)
    Suffix
    highlights that both the income of the offender and that of the victim represent relevant factors, as the first is a cost while the second an incentive to commit crimes, thus leading to expect opposite signs of their effects.
    (check this in PDF content)

  18. Start
    9652
    Prefix
    Concerning the first aspect, Buonanno (2003) highlights that both the income of the offender and that of the victim represent relevant factors, as the first is a cost while the second an incentive to commit crimes, thus leading to expect opposite signs of their effects. Different studies have also led to believe that income inequality plays a critical role in determining crime levels.
    Exact
    Buonanno (2003)
    Suffix
    highlights that income inequality can be thought as a measure of the differential between legal and illegal payoffs and ̇Imrohoroˆglu et al. (2006) identify it as one of the variables having the greatest effect on the crime rate.
    (check this in PDF content)

  19. Start
    9800
    Prefix
    Different studies have also led to believe that income inequality plays a critical role in determining crime levels. Buonanno (2003) highlights that income inequality can be thought as a measure of the differential between legal and illegal payoffs and ̇
    Exact
    Imrohoroˆglu et al. (2006)
    Suffix
    identify it as one of the variables having the greatest effect on the crime rate. As explained by Kelly (2000), the direct effect of inequality is to juxtapose those with low returns from their legal activities and people with considerably higher wealth.
    (check this in PDF content)

  20. Start
    9925
    Prefix
    Buonanno (2003) highlights that income inequality can be thought as a measure of the differential between legal and illegal payoffs and ̇Imrohoroˆglu et al. (2006) identify it as one of the variables having the greatest effect on the crime rate. As explained by
    Exact
    Kelly (2000),
    Suffix
    the direct effect of inequality is to juxtapose those with low returns from their legal activities and people with considerably higher wealth. A second consequence of inequality is explicated by the strain theory of Merton (1938), who argues that because of the same juxtaposition of social classes, frustration of the poorest individuals could represent a triggering motive
    (check this in PDF content)

  21. Start
    10166
    Prefix
    As explained by Kelly (2000), the direct effect of inequality is to juxtapose those with low returns from their legal activities and people with considerably higher wealth. A second consequence of inequality is explicated by the strain theory of
    Exact
    Merton (1938),
    Suffix
    who argues that because of the same juxtaposition of social classes, frustration of the poorest individuals could represent a triggering motive for criminal activity. In both cases, the impact of inequality should then lead to higher crime rates.
    (check this in PDF content)

  22. Start
    10581
    Prefix
    In both cases, the impact of inequality should then lead to higher crime rates. Police presence and public expenditure in protection has also claimed a relevant impact on crime. The topic is extensively discussed by
    Exact
    Levitt & Dubner (2005)
    Suffix
    : although recognizing that there may be simultaneity bias, as police presence action affects crime trends, but also a higher crime rate will call for higher police protection. They provide some interesting evidence of an inverse effect of police on crime by using exogenous changes in the number of police around the time of elections to find that additional police can lower the crime
    (check this in PDF content)

  23. Start
    11182
    Prefix
    some interesting evidence of an inverse effect of police on crime by using exogenous changes in the number of police around the time of elections to find that additional police can lower the crime rate. On the other hand, by looking at the significant crime decline of the 1990s in the United States, they also show that innovative police strategies had a more limited effect.
    Exact
    Draca & Machin (2015)
    Suffix
    note that the effect of policing is oriented towards a deterrence mechanism, which enters the framework outlined in Becker (1968) by decreasing the expected returns from crime. Di Tella & Schargrodsky (2002) partly confirm this suspected negative effect by using a natural experiment to find the effect of larger police forces on car theft.
    (check this in PDF content)

  24. Start
    11322
    Prefix
    On the other hand, by looking at the significant crime decline of the 1990s in the United States, they also show that innovative police strategies had a more limited effect. Draca & Machin (2015) note that the effect of policing is oriented towards a deterrence mechanism, which enters the framework outlined in
    Exact
    Becker (1968)
    Suffix
    by decreasing the expected returns from crime. Di Tella & Schargrodsky (2002) partly confirm this suspected negative effect by using a natural experiment to find the effect of larger police forces on car theft.
    (check this in PDF content)

  25. Start
    11390
    Prefix
    Draca & Machin (2015) note that the effect of policing is oriented towards a deterrence mechanism, which enters the framework outlined in Becker (1968) by decreasing the expected returns from crime.
    Exact
    Di Tella & Schargrodsky (2002)
    Suffix
    partly confirm this suspected negative effect by using a natural experiment to find the effect of larger police forces on car theft. Finally, Buonanno (2003) highlights the importance of social interactions and social networks among the determinants of crime.
    (check this in PDF content)

  26. Start
    11562
    Prefix
    Draca & Machin (2015) note that the effect of policing is oriented towards a deterrence mechanism, which enters the framework outlined in Becker (1968) by decreasing the expected returns from crime. Di Tella & Schargrodsky (2002) partly confirm this suspected negative effect by using a natural experiment to find the effect of larger police forces on car theft. Finally,
    Exact
    Buonanno (2003)
    Suffix
    highlights the importance of social interactions and social networks among the determinants of crime. This raises the issue of immigration, whose link with criminality has always been rather controversial.
    (check this in PDF content)

  27. Start
    12399
    Prefix
    Due to all these challenges, the literature on migration and crime is not as extensive as on the other determinants. Nevertheless, at least in the United States, racial inequality is still a dominant feature, and it has been widening with the Great Recession
    Exact
    (Kochhar & Fry (2014)).
    Suffix
    This suggests an intrinsic disadvantage of being “different” that, again as in Merton (1938), might be manifested as a higher propensity to engage in criminal activities. 3 Data and methodology 3.1 Data and variables description The dataset is a panel constructed for 50 US states2for the period 2000–2012.
    (check this in PDF content)

  28. Start
    12501
    Prefix
    Nevertheless, at least in the United States, racial inequality is still a dominant feature, and it has been widening with the Great Recession (Kochhar & Fry (2014)). This suggests an intrinsic disadvantage of being “different” that, again as in
    Exact
    Merton (1938),
    Suffix
    might be manifested as a higher propensity to engage in criminal activities. 3 Data and methodology 3.1 Data and variables description The dataset is a panel constructed for 50 US states2for the period 2000–2012.
    (check this in PDF content)

  29. Start
    15185
    Prefix
    Moreover, an indicator for the political orientation of the state (red or blue) is used as a control variable. Later in the analysis, measures of total household income, and a diversity index constructed as in
    Exact
    Ottaviano & Peri (2006)
    Suffix
    5are also used in order to rank the states and investigate the aforementioned effects for given categories. 3.2 The responsiveness scores approach As explained above, the investigation of the socioeconomic determinants of crime is not new to the literature.
    (check this in PDF content)

  30. Start
    15586
    Prefix
    & Peri (2006)5are also used in order to rank the states and investigate the aforementioned effects for given categories. 3.2 The responsiveness scores approach As explained above, the investigation of the socioeconomic determinants of crime is not new to the literature. In this paper, however, we propose a new type of analysis based on the Responsiveness Scores technique as proposed by
    Exact
    Cerulli (2014, 2017).
    Suffix
    The main advantage of this approach is that it allows for heterogeneous responses of the macro units of observation to the factor variables. In what follows, we set out a concise methodological account of this method. 5The index is constructed as the probability that two individuals that are randomly drawn from the population of the state are born in the same country:DIst= 1− ∑M i=1 ( CoBist T Ps
    (check this in PDF content)

  31. Start
    16676
    Prefix
    Q) changes, conditional on the other (Q−1) factors: x−j= [x1,...,xj−1,xj+1,...,xQ].(1) Algebraically, it is the derivative ofyonxj, givenx−j, when one allows for each observation to have its own responsiveness score. We assume thatx−jis a vector of all exogenous variables. RS are obtained by an iterated random–coefficient regression (RCR), whose basic econometrics can be found in
    Exact
    (Wooldridge 2010,
    Suffix
    pp. 141–145). The calculation of RS follows this simple protocol: 1. Definey, the outcome (or response) variable. 2. Define a set ofQfactors thought of as affectingy, and indicate the generic factor withxj. 3.
    (check this in PDF content)

  32. Start
    17366
    Prefix
    Finally, aggregate by unit (row) and/or by factor (column) the E(bij|xi,−j) thus getting synthetic unit and factor responsiveness measures. Analytically, a RS is the “partial effect” of a factorxin a RCR
    Exact
    (Wooldridge, 1997;
    Suffix
    2003; 2004). Indeed, for eachj= 1,...,Q, define a RCR model of this kind:      yi=aij+bijxij+ei aij=γ0+xi,−jγ+uij bij=δ0+xi,−jδ+vij (2) whereei,uijandvijare freely correlated error terms with: E(ei|xi,−j;xij) =E(uij|xi,−j;xij) =E(vij|xi,−j;xij) = 0(3) It is easy to see that the regression parameters,aijandbij, are both non-constant as they depend on all the other inputsxexceptxjdue to the
    (check this in PDF content)

  33. Start
    20408
    Prefix
    In particular, fixed–effect estimation, by allowing for arbitrary correlation betweenαiandηit, can mitigate a potential endogeneity bias due to misspecification of previous equation and measurement errors in the variables considered in the model
    Exact
    (Wooldridge 2010,
    Suffix
    pp. 281–315). As such, a panel dataset may allow for more reliable estimates of the responsiveness scores than OLS estimates on a cross-section. If the variables are standardized, eq. (9) becomes: yit=γ0+xi,−j,tγ+δ0xijt+xijt·xi,−j,tδ+αi+ηit(10) which simplifies the formula.
    (check this in PDF content)

  34. Start
    24189
    Prefix
    We can also assess the volatility of this effect by looking at the standard deviation. We find thatEducationhas, on average, a weak but clearly positive effect. This at 12 first might seem counterintuitive, although it is consistent with some of the literature.
    Exact
    Buonanno & Leonida (2005a),
    Suffix
    while studying the impact of education on crime in Italian regions, conclude that there are non-linearities in the effect of education on crime. That is, crime decreases with the increase of education when the latter is low, but rises when it’s already high.
    (check this in PDF content)

  35. Start
    25355
    Prefix
    Again, despite disagreeing with the earlier literature, this would still seem consistent, as mentioned in Section 2, with the line of research claiming little impact of unemployment on crime. Moreover, empirical evidence suggests that the job finding rate is particularly high in the US, and explains much of the cyclicality in unemployment
    Exact
    (Shimer (2005)).
    Suffix
    If spells of unemployment are short, it is likely that shocks in this variable may not have a significant effect on crime. ThePolicefactor shows one of the most relevant scores, with no ambiguity in its sign.
    (check this in PDF content)

  36. Start
    27140
    Prefix
    At the same time, in certain situations a higher wage income could reflect a higher 13 20012004200720102013 EducationEmployment PoliceInequality IncomeForeign population Figure 2: Timepaths of the responsiveness scores for the period 2000–2012. household income, or income per capita. In this case, as explained by
    Exact
    Buonanno (2003),
    Suffix
    there could be a higher potential gain (the victims’ income) from certain kinds of crime, and specifically property crime (Fleisher (1966)). Finally, crime has a predominantly positive responsiveness to the share ofForeign born.
    (check this in PDF content)

  37. Start
    27289
    Prefix
    13 20012004200720102013 EducationEmployment PoliceInequality IncomeForeign population Figure 2: Timepaths of the responsiveness scores for the period 2000–2012. household income, or income per capita. In this case, as explained by Buonanno (2003), there could be a higher potential gain (the victims’ income) from certain kinds of crime, and specifically property crime
    Exact
    (Fleisher (1966)).
    Suffix
    Finally, crime has a predominantly positive responsiveness to the share ofForeign born. Although this is true on average, and for most of the observations, for a small part of them the opposite is true.
    (check this in PDF content)

  38. Start
    29021
    Prefix
    This would suggest that an increase in the average level of education, when this is relatively low, brings a less than proportional increase in crime, while implying a decrease in the crime rate for very high levels of education. With reference to the above mentioned work by
    Exact
    Buonanno & Leonida (2005b),
    Suffix
    this suggests a further nonlinearity in the responsiveness to the education level. What is even more striking is the result achieved from plotting the responsiveness toEducationand a measure of cultural diversity (Ottaviano & Peri (2006)).
    (check this in PDF content)

  39. Start
    29262
    Prefix
    With reference to the above mentioned work by Buonanno & Leonida (2005b), this suggests a further nonlinearity in the responsiveness to the education level. What is even more striking is the result achieved from plotting the responsiveness toEducationand a measure of cultural diversity
    Exact
    (Ottaviano & Peri (2006)). To
    Suffix
    investigate this, we check whether 15 different levels of diversity interact with the responsiveness to education. The relation is unambiguously negative, with the effect of education on crime decreasing and soon becoming negative for higher levels of the diversity index.
    (check this in PDF content)

  40. Start
    32555
    Prefix
    Interestingly, this is not the case for the lowest level of expenditure where returns are increasing in magnitude, therefore suggesting that the presence of police is still so low that even a slight increment has a strong negative effect on crime. Overall, this seem to be a reassuring result. As pointed out by
    Exact
    Levitt & Dubner (2005),
    Suffix
    higher expenditures on police protection can lower criminality, but if this is not paired with other complementary measures, this turns not to be necessarily the most efficient way to foster social order.
    (check this in PDF content)

  41. Start
    35817
    Prefix
    75% of the states in the Midwest are ranked below average when units are ordered according to their average Gini coefficient, i.e. income tends to be more fairly distributed. On the basis of the literature, a possible hypothesis would therefore be that, suffering less from inequalities, crimes that are mainly due to resentment and social tensions
    Exact
    (Merton (1938)),
    Suffix
    as violent crimes, are less common, with possibly more property crime. According to Buonanno & Leonida (2005b) and Abdullah et al. (2015) education has the effect of reducing income inequality. Thus, a lower Gini index could be signalling for higher education level, which in turn points to a greater likelihood of committing property crimes.
    (check this in PDF content)

  42. Start
    35921
    Prefix
    On the basis of the literature, a possible hypothesis would therefore be that, suffering less from inequalities, crimes that are mainly due to resentment and social tensions (Merton (1938)), as violent crimes, are less common, with possibly more property crime. According to
    Exact
    Buonanno & Leonida (2005b) and Abdullah et al. (2015)
    Suffix
    education has the effect of reducing income inequality. Thus, a lower Gini index could be signalling for higher education level, which in turn points to a greater likelihood of committing property crimes.
    (check this in PDF content)

  43. Start
    45426
    Prefix
    discrepancies from the analysis performed at the aggregate level. 5 Conclusions This paper investigates how the level of engagement in illegal activities responds to six of the main socioeconomic factors identified in previous literature in 50 US states. The analysis of the state level idiosyncratic effects of these considered determinants is made possible by the use of responsiveness scores
    Exact
    (Cerulli (2017)).
    Suffix
    We assess the magnitude of their effect of these six factors in each state for each year. Moreover, we have distinguished between two relevant categories of crime, and found differential behaviors in trends and factor returns for their responsiveness to the considered factors.
    (check this in PDF content)