Three Critical Questions in Law Enforcement In this section, I overview what I consider the three most important questions in current research on the economics of crime and punishment. A. The Efficacy of Deterrence The previous section discussed some potential policy tools that are availTABLE to the government to restrict crime. In principle, the government might attempt to limit the benefits to crime or raise the legal wage. However, historically the most important weapons against crime have been the direct tools of arrest and punishment.
Before the 1 9th century, serious crimes were nearly punished with death or other forms of severe corporal punishment (see Faculty, 1995). Lesser crimes were punished with public humiliation and lesser forms of corporal punishment. In some societies, fines were also used (even occasionally for murder). After 1800, prisons became the standard form of punishment for almost all crimes, throughout Europe and the U. S.. There are several possible explanations for the evolution. Some theorists credit the increasing distaste of the public for painful, public physical punishment.
Alternatively, other theorists have argued that the use of prisons was part of the common event to restrict the power of the state in imposing punishment. This vein of thought emphasizes that the restrictions on cruel and unusual punishment were part of the Bill of Rights, which are generally meant to restrict the powers of government. A final theory is that incarceration makes more sense as professional criminals replace amateurs. When crimes are committed as random acts by generally productive persons, it is very costly to engage in incarceration.
However, as crime became more professional, incarceration leads to social gains as criminals are incapacitated. The modern economics literature on the efficacy of forms of punishment began in the asses. Exemplars of this literature include the early work of Isaac Earlier (1 973, 1975), which generally found quite significant deterrence elasticity. The critical parameter being estimated is the elasticity of the quantity of crime with respect to expected size of punishment. Earlier (1973) estimates this elasticity at -. Using ordinary least squares. Matheson and Passel (1976) examined sub-city geography and found a particularly large elasticity of -3. This literature represented an advance over previous empirical ark, which generally used case studies and performed empirical exercises like comparing neighboring states which had differences in the level of punishment. The new economic literature used both cross-sectional and time series regressions and controlled for a rich variety of control variTABLEs, on top Of deterrence variTABLEs.
However, as early as the mid-asses, Taylor (1978) was focusing on the problem that punishment levels are endogenous with respect to the level of crime. There is a particularly rich literature from this period on the impact of capital punishment. Earlier (1975) gives results suggesting a major impact of capital punishment using variation from U. S. States. Archer and Gardner (1984) examine cross national evidence and find no negative impact of capital punishment on murders. Given the nonentity problem, we are unlikely to have a compelling finding on the impact of capital punishment any time soon.
All of these estimates suffer from a basic identification problem and recent empirical work on deterrence has focused on solving this identification problem. The problem is that the level of punishment is rarely determined randomly. Instead, governments choose the level of policing and the level of enmeshment in response to local factors such as the amount of crime. Thus, for example there is a very high positive correlation between the number of police per capita and the amount of crime per capita. Should we infer from this that police cause crime? Probably not.
A better interpretation is that localities vote more on police when they have a higher crime rate. This problem exists with almost all deterrence variTABLEs. As arrest rates and punishment levels should be a function Of the level Of crime, all coefficients are expected to be biased. In the mid-1 9905, Steven Levity has revolutionized the field of deterrence elasticity estimates by writing a series of papers on the impact of deterrence. Levity uses clever, exogenous sources of variation that should determine the level of policing, but should not themselves be a function of the level of crime.
For example, Levity (1996) uses UCLA cases against prison overcrowding to look at the effects of releasing criminals from prisons. These UCLA cases are filed against prisons and if they succeed, they generally lead to a significant release of criminals. In this paper, he finds that increasing flow of criminals out of prisons due to these cases significantly increases the crime rates. Levity 1997) uses the fact that politicians tend to spend more on policing during election years.
He uses the variation created by electoral cycles across cities (correcting for general time effects and city effects) to assess the impact of more police on the levels of crime. He fields modest, but quite significant impacts of policing on the level of crime. Elasticity estimates generally range between . 1 and . 5, which are in line with the findings of Groggier (1991). Ares and Levity (1997) examine the introduction of Lack across cities. Lack is a technology that emits electronic signals from stolen cars and enTABLEs police o catch auto thieves.
Again there is a substantial amount of crime reduction created by the introduction of Lack. These papers demonstrate both the basic approach and the difficulties in doing work on the elasticity of crime with respect to deterrence. In principle, the best forms of identification can come with the aid of government. If governments can actually be convinced to randomly allocate policing resources across areas of a country, then very convincing estimates of the effects of policing can be estimated. While it will certainly be useful to estimate elasticity in the classic manners discussed above (I . Regressions run across space of the level of crime on the level of policing), the results will be stronger if exogenous sources of variation can be found. Furthermore, given the nonentity problem it is crucial that ordinary least squares estimates be treated skeptically. B. Endogenous Law Enforcement and Punishment The principal behind the identification problem is that law enforcement responds to the level of crime. This is one example of the many reasons why we must stop viewing law enforcement as a black box that just serves the needs of the people, or maximizes some social welfare function that we have n mind.
Increasing, research is focusing on the behavior of the various actors in law enforcement, which can subvert even the most wellbeing policies. This is not to say that these law enforcement officials are in any sense evil, but they respond to their own incentives which are not necessarily the incentives that benevolent social planners would want them to have. If this issue is severe within the United States, it is far more so in developing countries. Many police forces in developing countries are either corrupt or committed to their own, occasionally brutal policies.
Anecdotes (and survey evidence) documenting extremely high levels of police extortion abound. Judges are often unreliTABLE in developing countries. In many of these countries, it would do great justice to the truth to just assume that the police force and judiciary will simply obey the policy dictates handed down from on high. The U. S. Economics research on the objective functions of police and the judiciary is much more limited than the research on the efficacy Of deterrence. Posses (1994) wrote a pioneering article attempting to understand what it is that judges maximize.
Kessler and Pill (1997) examined the extent to which juries and judges undo mandatory sentence guidelines (a typical on-high policy rule) by changing the charges the criminals face. Glasses, Kessler and Pill (1997) show the Federal prosecutors, who have more authority than state prosecutors, prosecute criminals with more income and higher levels of human capital. We argue that this relates to the career concerns faced by prosecutors who want to distinguish themselves by prosecuting high profile criminals. There is also a literature on the determinants of punishment.
Friedman (forthcoming) argues that one reason why punishments are not more severe (recall that Becker, 1968, suggested that extremely high penalties would be optimal) is that policemen can’t be trusted with too much power. High penalties enTABLE the police to extract high bribes and terrorize the populace. As a result, less severe penalties are optimal when the behavior of police are taken into account. Glasses and Sacerdotal (1999) examine the factors that are associated with higher levels of punishment for murderers in the United States. We find that people who kill African- Americans get shorter sentences.
Murderers of women get more severe punishments. This is true even of people who are guilty of vehicular homicide where the victim is presumably random. The tendency to punish people who kill African-Americans less is lower in those counties with demographics that are associated with more pro-shack attitudes. These behavioral patterns presumably illustrate the tastes of Judges who make most of these sentencing decisions. The total U. S. Research in this area is newer, less comprehensive and less convincing than the research on deterrence.
However, this is clearly a particularly hot area for new research and hopefully the World Bank’s search on developing countries will particularly focus on this area. C. Social Explanations Of Crime The Becker model of crime suggests that the opportunity cost of time will directly effect the propensity to engage in criminal activity. As such, we expect there to be an increase in the amount of crime as the opportunity cost of time declines. There is a long literature on the connection between either unemployment or wages and the amount of crime.
Early work on this theme includes White (1980) which finds that crime declines as unemployment rises Freeman (1986) looks at the National Longitudinal Survey of Youth and examines the factors that determine the probability of going to jail. He finds that youths in poverty are more likely to end up being arrested or going to jail than youths who are wealthier. Surprisingly, he finds that church attendance is a particularly strong predictor Of the level Of crime. Indeed, this result echoes a basic finding of the crime literature.
Measures of the opportunity cost of time do predict crime, but not as effectively as other more sociological variTABLEs. Even with the basic opportunity cost of crime variTABLEs, there is always a problem in interpreting the coefficients. There may be a negative connection between an individual’s wages and his proclivity to crime, because higher wages mean a higher opportunity cost of time. Alternatively, more successful people might just have a lower proclivity towards crime because they are healthier emotionally or are better-integrated into society.
At the community level, higher wages may create less crime but on the other hand higher crime may deter economic development and lower wages. As such there is an nonentity problem with the level of wages. Gould and Weinberg (1999) address this problem and show that even dealing with this causality problem here is still a negative connection between wages and crime. However, increasingly there is a body of work documenting the importance of social factors in explaining the level of crime. As mentioned earlier, Freeman (1986) shows a strong connection between church attendance and crime for disadvantaged youths.
Glasses and Sacerdotal (Bibb) show that the most important variTABLE which explains the variance of crime rates across cities within the United States is the percent of single parent families in the population. This survives instrument for this variTABLE with welfare payments in the states. Case and Katz (1991) examine criminal activity (among other things) in a survey of disadvantaged Boston inner city youth. They find that individuals’ tendencies to engage in crime increase substantially when peers are also engaging in crime. This survives instrument for the crime rates of peers using exogenous characteristics of those peers.
Glasses, Sacerdotal and Chainman (1996) address the issue of the high variance of crime rates across space. We argue that these crime rates are evidence for the existence of social interactions between criminals. In other words, if one person’s rimming activities increases the benefits (or decreases the costs) of his neighbor engaging in crime then We should expect to find a high variance Of crime rates over space. We present a methodology for using the variance of crime rates over space to estimate the extent to which there are social interactions in crime. IV.
Other Important Questions This section briefly surveys some of the other important issues in the economics of crime literature. A. Measuring the Costs of Crime For policy purposes, it is frequently important to understand the costs Of crime. Generally, there are two approaches. One approach is an accounting exercise where researchers just try to add up all of the losses from crime. Donahue and Salesman (1998) is an example of this sort of exercise. The largest loss category is usually the values of lives lost due to murder. Other straightforward costs include the amount that is spent on crime prevention.
These costs would include the amount that is spent by the government on police and the judiciary. Crime cost estimates have generally been unTABLE to quantify the social costs that accrue from underinvestment in the legal sector because returns are appropriated by criminals. There is a debate about whether or not the dollar value of goods lost should be included in the social costs of crime. If crime meant that the value of the goods lost exactly equaled the value of the goods received by the criminal, then there is, in fact, no social cost from the redistribution that crime creates.
Of course, generally the goods are valued less by the criminals who take them than by the people who lose them so there are social losses. The value of goods taken should in equilibrium by equal to the opportunity cost of the criminals time and the time of the criminals that is spent on crime instead of gal activities does represent a real social welfare loss. For this reason, it may be correct to include the value of goods lost as a social loss. Another approach to estimating the social costs of crime is to examine housing price hedonism.
Housing prices reflect the willingness of individuals to live with community level characteristics. Thus the difference in prices between a high crime and a low crime community will reflect the willingness of consumers to live in a high crime area. A pioneering paper in this area was Thales (1975). Naturally, this type of research relies upon good housing price tat and that unobservTABLE neighborhood and housing characteristics are not strongly correlated with the neighborhood level of crime. One advantage of the housing price data is that it can actually tell us about the value of marginal reductions in the level of crime.
Most reasonTABLE policy interventions will certainly never lead to a complete eradication of the level of crime. Therefore these accounting exercises don’t always make much sense, if they are trying to estimate the total costs Of crime. The goal of this type of research is to estimate the gains in social welfare from a marginal reduction n the crime rate. A final important issue is the extent to which criminal activities harm the career prospects of young men. Freeman (1986) investigates the general problems of crime and the career paths of young men. Groggier (1995) presents the strongest evidence on the career costs of going to prison b.
The Function of Prisons: Deterrence, Incapacitation and Rehabilitation In principle, prisons can serve three functions for crime prevention. The literature on the economics of crime has focused on the role of prisons as a deterrent. However, prisons will also serve to incapacitate criminals and lastly orisons, at least in principle have the capability of rehabilitating criminals. Rehabilitation can only be understood as serving to change the costs and benefits of crime so that crime becomes less likely in the future. The literature on deterrence has generally not found much support for the idea of rehabilitation.
Needles (1996) presents a particularly detailed survey of the amount of recidivism and shows strikingly high levels of continued criminal activity for released convicts. This is in line with previous findings which almost always show that the propensity for crime does not fall with prison. Indeed, many others have suggested that prisons act to train criminals and reduce their suitability for legal employment which would suggest that prison sentences create criminals rather than rehabilitate them. The only cases where we know that something like rehabilitation can occur is when criminals get extremely long sentences.
As it is well known that the propensity towards crime falls sharply among older men, very long sentences can in some sense rehabilitate criminals. The relative importance of deterrence and incapacitation depends in theory on the heterogeneity among criminals. In a oral of homogeneous citizens, then imprisoning a group of criminals will have little effect on the level of crime. New criminals will just replace the old ones. When there is a great deal of heterogeneity then locking away the particularly crime prone may indeed act to reduce the level of crime. Levity (1998) tests the relevance of deterrence vs.. Incapacitation by examining the impact of arrest rates in different crimes. A higher arrest rate in one crime (say auto theft) would tend to increase the crime rate in robberies if deterrence is important. The idea is that criminals will tend to substitute out f auto theft into the now cheaper form of crime: robbery. Conversely, if incapacitation is important then higher arrest rates in auto theft will lead to lower levels of robbery as long as some auto thieves also engage in robberies. Levity finds much stronger evidence for incapacitation than for deterrence. C.
Gangs and Drugs Gangs appear to be associated with a significant amount of crime in the United States. The organization of young men into organizations that exist to protect territory, engage in various crimes and sell drugs is a major feature of many American cities. While no one has produced a definitive paper linking the level of gang activity with the level of crime, two recent papers have come forth attempting to understand the level of gang activity. Croaker and Yelled (1994) follow the ideas of sociologist Morris Jackson’s and produce a formal model of gang behavior.
They suggest that gangs provide services for their local communities and are protected by their communities from the police. They exist in part to fill holes in services that might alternatively be produced by the government. For examples, residents of some communities do not rust the police and gangs are seen as more reliTABLE providers of public safety. Levity and Venerates (1999) have acquired the finances of a particular gang’s activities. They find that there are large revenues acquired by the leadership of the gang but regular members of the gangs only earn modest amounts.
Furthermore, they find that gangs pay more for soldiers during periods of gang wars, supporting further the economic concept of compensating differentials. Most importantly they have provided an invaluTABLE guide into the inner working of a gang. Acquiring such information outside of the U. S. Loud be particularly useful. Research on gangs, and on drug trafficking, is really in its infancy within the economics of crime literature in the LLC. S. However, given the importance of gangs and drug trafficking in many developing countries this seems like a particularly promising area for future research.
At the least, work in this area should attempt to produce research along the lines of Levity and Venerates (1998 that gives us some idea of how these gangs operate. D. Guns and Weaponry A major controversy is the extent to which the cheap availability of guns increases or decreases the levels of crime. While relatively systematic research on this topic had been done prior to the 1 sass, since 1 996 there has been a veriTABLE explosion of research on different gun laws. In a particularly controversial study, John Lott presented evidence supporting the idea that a greater availability of guns reduced rather than increased the level of crime.
The bulk of his work, which has seen been published in a volume More Guns, Less Crime, used laws which are passed requiring counties to allow individuals to buy guns. Lott finds that the level of violent crime appears to decrease in counties that have made guns more availTABLE. The basic economics of the controversy rests on the idea that guns will increase crime when they are made availTABLE to criminals. However, as guns are made availTABLE to libidinal citizens, the law of crime may decline as criminals are afraid to prey upon armed victims.
Perhaps the most convincing piece of evidence in Loot’s study is the relative prevalence of ‘hot burglaries” in Europe relative to the U. S. “Hot” burglaries are defined as burglaries that occur when residents are at home. Lott suggests that “hot” burglaries are rare in the U. S. Because burglars fear armed homeowners more than they fear the police. Indeed, in this case anecdotal evidence (interview’џ with criminals) strongly supports the importance of this idea. However, there are many other cases where the presence of weaponry appears to lead to more bloodshed.
It is hard to imagine the deadliness of American inner city gangs without the weaponry that they have recently acquired. Levity and Donahue (1998) present an elegant model where the increased presence of weaponry leads to more conflict. While this have never been shown, it seems reasonTABLE that weapons will act to increase crime if they are made particularly availTABLE to crime-prone citizens. Weapons may crease crime if they are carefully made availTABLE only to those who are unlikely to be criminals. Of course, such differential policies become tricky to enforce as long as resale of weapons is a possibility.
Glasses and Gleeson (1998) suggest that the presence of guns in a culture changes far more than just the level of crime. They suggest that areas of the U. S. Where guns are widespread have developed a tradition of private justice substituting for State justice. This may be good or ill, but policies towards weaponry must face the fact that when weapons are common the entire structure of the society often changes. . The Returns to Crime A final area of research is to actually estimate the determinants of the returns of crime. There is very little research in this area.
Glasses and Sacerdotal (AAA) attempt to understand whether big cities increase the returns to crime. We argue that proximity to the rich may make increase the returns to crime. The evidence of Carrel and House (1990) shows the presence of strangers (who are presumably targets for crime) increases the amount of criminal activity. Indeed, there is a substantial connection between city size and the average value of goods stolen in a crime. This fact appears in both the Uniform Crime Reports when the value of goods stolen is regressed on average city size and in the National Crime Visualization Survey.
This research is limited because it only examines the role of city size and many other factors might also influence the returns to crime. Furthermore, the equilibrium average level of goods stolen is a dubious measure of the actual returns to crime. The average level of goods stolen is an equilibrium quantity. There are certainly cases where an overall increase in the availability of victims means that more and more marginal crimes are omitted. Thus the average quantity of goods stolen may in fact decline with the availability of victims.
Alternatively, in cities where crime is particularly hard, only the most professional criminals might engage in crime. As such, in those cities the average value of goods stolen might be very high. In principle, research in this area might be better if it focused on variTABLEs which clearly predicted a greater availability of increased targets and examined the effects of these targets on the amount of crime (and of the value of goods stolen). However, it is hard to think of exogenous factors which would differ across space and would increase the returns to crime. As such, this must remain an area of research that is truly in its infancy.
V. Overview of Data Sources In this section, I review the basic data sources that are used in empirical work on crime. The first type of data set is the visualization study. This is unquestionably the primary workhorse of measuring crime. To the extent that a hot spots index will be created, at least 50 percent Of that hot spots index should be based on mini-visualization studies. To be really useful, this sort of study must have geographic identifiers which enTABLE the research to ink the individual to the community in which he lives and was robbed. Then the researcher also needs to know attributes of that community.
These attributes may be drawn from broader national surveys or national census. The second type of data set relies on official crime statistics. This data set is useful but relies on the quality of arrest and reporting data. With this sort of data at the locality level, one might run regressions connecting city level characteristics with the level of reported crime. However, reporting (and recording) problems in many of the countries that we are discussing is so aerobatic that this type of sunset becomes difficult to use. The third type of data set involves homicide (or intentional injury) data from hospitals and morticians.
This type of data is generally more reliTABLE than official crime statistics. It is also less sensitive to changing definitions of crimes across cultures. This should also be part of the hot sports index. In principle, data on the victim (and the perpetrator when the case is brought to trial) can also provide a rich set of stylized facts about the natural of homicide in the country in question. A final data source is the offender survey. This type Of survey can both be oaken through traditional survey methods where respondents are asked if they have been arrested (or less reliably if they have committed a crime).
