There’s a new paper about the minimum wage and its effects on crime. I wrote a paper (with Amanda Agan) about the minimum wage and crime (here’s a slightly older ungated version). I have received several requests to comment on the new paper because, based on the abstracts, our papers appear to generate conflicting results. Spoiler alert: they don’t. Sorry to disappoint those who came looking for an academic blood bath.
I am happy to talk about the new paper, by Fone, Sabia, and Cesur (FSC), but let’s get the big part out of the way. Our paper on the minimum wage looks at criminal recidivism, defined as a return to prison, for those who have been released from prison. These are people whose conviction resulted in them being in incarcerated in a prison (not jail) who, on average, served nearly 2 years and were subsequently released at age 35. The FSC paper uses arrest data. Their principal observation regards property crime arrests committed by 16-24 year olds.
Our two papers identify fundamentally different results about fundamentally different populations that, in my opinion, hinge on completely different mechanisms.
Our paper is old news, so I won’t belabor the point. Succinctly, we found that minimum wage increase of $0.50 reduced the probability an individual returns to prison within 3 years by 2.15%. The availability of state EITCs also reduced recidivism, but only for women.
The FSC paper use’s Uniform Crime Report data to look at arrests. Here’s the figures and tables that I’ll focus on for our discussion:
FSC find that property crime arrests increase for 16-24 year olds in an event study estimate, where an increase in the minimum wage of at least $1 serves as an “event”:
Property crime arrests in their diff-in-diff estimate reaffirm this estimate. They also, however, observe negative effects on property crime arrests on 35-49 year olds, though the coefficient is too noisy to be statistically significant. These results are similar to ours, though because we were looking at individual recidivism we had the benefit of estimating over ~6 million observations (vs the 45 thousand county-years of FSC).
When FSC dig into the crime categories further, there is no effect on burglary, robbery, or auto theft. The property crime effect is entirely in larceny. Let’s also note the positive effect of the minimum wage on vandalism.
Here’s an important tidbit: UCR data does not distinguish between misdemeanor petty (petit) larceny and felony larceny. One last result: employment is noisily declining for 16-24 year-olds who have not yet completed high school.
Let’s add it all up: when a state increases the minimum wage by at least $1, we observe an increase in larceny and vandalism arrests of 16-24 year-olds, without any effect on robbery, burglary, auto theft, or violent crime, all while reducing the employment of 16-24 olds who have not yet completed high school. Can you see where I’m going with this?
Shoplifting. When states significantly increase the minimum wage, employers stop hiring teenagers. Those teenagers, laden with time but bereft of spending money, rediscover the allure of the five-finger discount. That is my interpretation of these results and nothing about these results seems strange to me or at odds with the earlier findings in our paper on the minimum wage and recidivism.
I don’t think the authors have really done anything wrong here. I could manufacture some of the usual gripes if I really wanted too, but the identification strategy seems at least broadly sound and the data is widely used. The estimated magnitudes seem plausible. If I was going to complain about anything, it would probably be the imputed $766 million dollar price tag placed on the externality, but I’m also not well-versed in the costs of shoplifting (and in case you’re reading something into my tone, I do not think shoplifting can be dismissed as unimportant). If I had to hang my hat on something, though, I’d say that’s probably on the hefty side. In footnote 48 they consider a a more conservative estimate of a $128 million dollar externality. That seems more plausible to me.
The minimum wage literature is one we all, every single one of us, bring our own political and economic baggage to. When our paper found that the minimum wage reduced criminal recidivism, a lot of people latched on to it because what they heard was “minimum wages stop crime”. I’m sure a lot of people will latch on to FSC’s new paper because they want to hear “minimum wages cause crime”. The reality, of course, is vastly more nuanced. We should expect these laws to have heterogeneous effects born of complex interactions, particularly when we stratify populations into those interacting with an institution as rife with peculiarities and pathologies as the US criminal justice system.