Nova workboard

a blog from young economists at Nova SBE

Education externalities, the effect on crimes


In class we have discussed the reasons that may push the demand for education, in particular among the others factors, the externalities that education can produce. We focused our attention on the “functioning effect”, or better, the effect of education (at a primary level in this case) to facilitate social and economic interaction necessary to conduct a normal life (let’s say basic calculus for example). This topic generated  a lot of interest in class and gave me thoughtful insights to explore it better. In my opinion one of the most interesting externality generated by education is the effect on crime: a criminal behavior in the related literature has been connected to the “inner structure” as a deviation from a psychological and mental health or connected to impact of social circumstances, it’s clear that the effect of education can be easily included in both channels. The first question that naturally arises in me regards the direction in which this relation operates. The easiest way to think a manner by which education affects crime is an income effect that is due to better education: it directs an individual to more social acceptable goals and to respect of law, reducing her propensity to crime, it increases the income from legitimate work and raise the opportunity cost of criminal activity. Moreover increasing the wage it makes more costly the time spent in prison and outside the labor market. Time spent in school reduces the time available to spent in criminal activities (Tauchen et al.). Other effects involve the psychic ambit, in fact some evidence (Becker and Mulligan) show impact on patience and risk aversion (more patience means low discount rate and so a higher valuation of future earnings, people that drop out from school instead, reveal, according to Oreopulos, a high focus on immediate cost of schooling). If we think in that way we expect the relation between the two variables to be negative, but going deeper in literature it comes out that there is a possible channel that can drive the relation in a positive way, in fact more education may increase the self-protection against punishment by law (we can think that the best way to evade the law is to know the law) also the skills learnt in school could be used in criminal activities and so increase the income from those. I decided to construct my own dataset to try to figure out if a correlation between education and crime was relevant and in which direction goes. Data on crime are taken from UN’s dataset in crimes and the proxy I used is “Intentional homicide” defined here as unlawful death purposefully inflicted on a person by another person in 2012. Instead data for education are provided by OECD and the proxy I used is the percentage of people that attained only primary school with an age in the range 25-64 years as a measure of low level of education. Then I matched the two proxies for each country available and what I obtained is a sample composed by 36 observations. It’s shown below a subsample:


Attained only primary school 25-64 years

Crime rate

















Czech republic

















As we can see from the graph there seems to be a positive correlation between an high level of people who attained only the primary school in a country and the rate of homicide in the same country. A simple calculation of the Pearson correlation coefficient shows a value of 0,313 to confirm my expectation. But as we can imagine this is only a correlation and not  a causation (the STATA regression gives a coefficient on crimeRATE of 0,13 but is not consistent as there are a lot of variables that can cause the relation that are missing, for example income), in fact I used a broad measure of both the measures, for a more accurate analysis is better to work with more microeconomic data.

 Angelo Saponara


On the relation between education and crime – Isaac Ehrlich

The Crime Reducing Effect of Education –  Stephen Machin, Olivier Marie, Sunčica Vujić

Education as a Deterrent to Crime – Dan Usher

Work and Crime: An Exploration Using Panel Data – Ann Dryden Witte, Helen Tauchen

The endogenous determination of time preference – Gary S. Becker and Casey B. Mulligan


Author: studentnovasbe

Master student in Nova Sbe

2 thoughts on “Education externalities, the effect on crimes

  1. In a recent post, Angelo Saponara describes the correlation existing between education and criminality. This question – behind which we can see the debate about the link between poverty and criminality – is highly sensible.

    First of all, on an empirical point of view, Angelo correctly stated that correlation does not mean causality. Indeed, “a number of empirical issues make it difficult to interpret the (above) described education-crime relationships as causal” (Hjalmarsson et al., 2011). The correlation could be due to unobserved characteristics. For instance, a low ability or a lack of risk aversion could explain both criminality and low education. They also point out the phenomenon of reverse causality: the potential effect of the criminality of an individual and the education results.

    Next to that, I found it quite surprising to use the intentional homicides rate to derive the correlation, as the irrational character of a homicide excludes any thoughts about opportunity costs and future incomes. To strengthen his intuitions, Angelo would better have used an indicator of youth and/or little criminality, as it should (I insist on that) be more “economic-driven”.

    Finally, the policy implications of this debate (a longer compulsory education) are dubious. Indeed, a policy must always tackle the drivers of a problem. And in that case, I don’t think that a lack of education would be the main driver of criminality. Consequently, and despite all the respect I have for the outstanding work made by teachers in helping all the students (particularly those in troubles), I don’t think that a longer education would really tackle the criminality problems (particularly seeing the fact that the job of the teacher and the education in general are more and more denigrated in our society). As for me, the roots of violent crimes are more to see in irrationality and the problem of youth criminality is more to find in the education given (or not) by parents that could or not be combined to bad influences of peers (another peer effect?). Hence, instead of stretching the compulsory education, it would be much more efficient to re-make the parents aware of their responsibilities. However, it does not seem to be the path taken…

    Hjalmarsson, R., Holmlund, H. & Linquist, M. (2011). Can education policy be used to fight crime?. Source: (consulted on 03/12/2013)

  2. It is true that the demand for education is based on the expected impact that education has in the society, the externalities it can produce. The public demand for education depends on the externalities.
    I agree that there are two directions in the relation between crime and education: a positive, where education gives better tools to use in crimes (so crime and education vary in the same direction), and a negative, that states that education, as it originates higher opportunity costs of committing crime (income, social acceptance…) varies negatively with crime. As you focused on the negative direction of this relation, you are assuming that that one is the strongest. For a more complete analysis the test for correlation in the case of the positive relation between education and crime could be done.
    Is it known that people with high levels of education have the ability and the incentive to commit crimes as tax evasion, misuse of funds, fraud, inside trading or money laundering. In fact, several cases are known such as Madoff, a massive ponzi scheme in 2008, Parmalat, a case of falsified accounting documents in 2003, or Royal Ahold because of inflating promotional allowances, also in 2003.
    The big focus of the blog post was the negative relation between crime and education and it was perfect the use of the income effect concept applied to this case, as it highlighted the fact that propensity to crime is lower when the level of education is higher once income raises the opportunity cost of criminal activity. Also, time spent in school reduces the time available to spend in criminal activities plus crime is not socially acceptable, and high educated individuals have more respect for the law.
    As said above, the hypothesis of the positive effect could be tested, or even if it is stronger than the negative one.
    The variables chosen may not be the most appropriated. For crime the proxy the choice was “Intentional homicide”, which is not the best when we are talking about education and crime. If the individual has low education, has low income or even is unemployed he might very well steal without killing. Probably, if you had chosen for instance burglaries or robberies the positive correlation could even be stronger, because the individuals’ objective is not to murder but to get money.
    The proxy used for education, the percentage of people that attained only primary school with an age in the range 25-64 years, is a good choice. However, it would be interesting to see the effect by level of education.
    From the graph presented it is observable a positive correlation between the level of people who attained only the primary school (as a measure of low level of education) and the rate of homicide in the same country, but still it is small.
    A fundamental aspect is that correlation does not imply causality, so even if the correlation between high education levels and crime is negative (we see in the graph that when the level of education is lower crime level is higher i.e., when the level of education is higher crime level is lower as well) we cannot show empirically with the presented data that because an individual has a low education level he will commit more crimes (homicides).
    This is an interesting and important topic, especially when we can see that burglary crimes show an increasing trend from 2007-2010 in EU-27. Compared to 2007, 7 % more cases of domestic burglary were reported in 2010. Also, robbery begins to denote a non decreasing trend. These are worrying facts.