More impatient people are more likely to commit crime

Gary Becker's famous model of rational crime suggests that criminals weigh up the costs and benefits of crime (and engage in a criminal act if the benefits outweigh the costs). Time preferences matter in this model, because the benefits of a criminal act are usually realised immediately, whereas the greatest costs (including the penalties of being caught occur in the future. So, someone with a higher discount rate (a greater preference for the present over the future) will be more likely to commit crime, because the costs will be more heavily discounted. In other words, people who are more impatient (and therefore have a greater preference for the present) will be more likely to commit crime.

Is there evidence to support this idea that more impatient people are more likely to commit crime? This new article by Stefania Basiglio (Universitá degli Studi di Bari), Alessandra Foresta (University of Southampton), and Gilberto Turati (Universitá Cattolica del Sacro Cuore), published in the Journal of Economic Psychology (ungated earlier version here), provides some supporting evidence. They make use of data from the National Longitudinal Survey of Youth 1997 (NLSY97), using data from the 2008-2011 survey waves (with a sample of nearly 6000 observations), when the cohort was aged 24 to 31 years old. Their dependent variables are self-reported measures of whether the survey respondent engaged in property crimes, violent crimes, or drug crimes, in the previous twelve months.

One of the interesting aspects of this study is how Basiglio et al. chose to measure impatience. Because the NLSY doesn't include a survey measure of time preference, they instead use a variety of variables that are expected to be correlated with impatience. As they explain:

...we consider several similar observed variables 𝑋, available in the NLSY97, which represent individual behaviors for which impatience plays a role: the saving rate 𝑋1 is defined as the ratio between total savings and income; smoking 𝑋2 and drinking 𝑋3 are measured by the average number of cigarettes smoked in the past month and the average number of drinks consumed in the past month, respectively; obesity 𝑋4 is defined by a dummy which is equal to one if the Body Mass Index is equal or higher than 30; risky sexual behavior 𝑋5 is measured by the number of sexual partners that the individual had in the previous 12 months; 𝑋6 is a dummy for using hard drugs (like cocaine or meth) in the previous 12 months; 𝑋7 is a dummy for participating to a worship service at least once a month; finally, we define a dummy about marital status 𝑋8, which takes value one if the individual is married.

Basiglio et al. then use factor analysis to extract a single factor (a single variable) that best summarises the information contained in all eight of those variables. This is a useful approach to reducing the dimensionality of data, but also a handy way to proxying for a latent variable like impatience. The proxy variable seems to pick up the right correlations with each of the variables, being the same correlation that we would expect with impatience:

Factor loadings take up the expected signs: we find a negative correlation between being married, obesity, and having attended a worship service and the extracted common factor 𝐹1, while we find a positive association for all the other proxies. Consistent with our expectations, the strongest positive linkages are with drinking (0.260), smoking (0.204), and hard drug use (0.203); the strongest negative linkages are with attending a worship service (−0.267) and being married (−0.210).

Using this proxy variable as their measure of impatience, and controlling for age, ethnicity, education. occupation, and whether the individual had been jailed in the previous year, they find that:

The marginal effect for impatience is positive and significant for all types of crimes... The result suggests that being more patient is associated with a lower probability of committing crimes. The correlation of our proxy for impatience is stronger for drug crimes.

One of the main problems with the proxy variable is that it doesn't have a natural interpretation in terms of the size of the coefficients. However, taking the results as given in Table 4 of the article, a one-standard-deviation increase in the impatience measure is associated with a 5.4 percentage point higher probability of having committed any crime, a 1.8 percentage point higher probability of having committed a violent crime, a 3.1 percentage point higher probability of having committed a property crime, and a 6.4 percentage point higher probability of having committed a drug crime. Those are substantial effects, given that the baseline probability of committing those crimes are 6 percent for any crime, 2 percent for violent crime, 4 percent for property crime, and 6 percent for drug crime.

Basiglio et al. then look at differences by demographic group, and find no differences between men and women, or between people whose parents have college education compared to those whose parents have no college education. They do find some evidence that the effects of impatience are larger for non-Black/non-Hispanic men than for other men, for total crimes and drug crimes only. It is difficult to see what we can take away from the demographic analyses though - we would need some theory as to why impatience would affect different groups' crime decision-making differently.

Basiglio et al. also find that the results remain after controlling for risk preferences, which is an important robustness check, since people who are more impatient may also be those that are willing to take on more risk. Now, the results are not causal, but they do suggest that impatient people are more likely to commit crime.

If we accept these results, what are the policy implications? Basiglio et al. suggest that education may be a solution to reducing crime, to the extent that it both increases the opportunity costs of crime and makes people more patient. However, I think that there is a more immediate solution, which is to make the punishment of crime more immediate and more certain. If people who heavily discount the future are more likely to commit crimes, then the costs of committing crime (and being caught) have to be more severe, or (and this may be more effective overall) the punishment needs to come more quickly after the crime is committed. Either way, that probably means more resources devoted to policing and the criminal justice system.