marginal revolution read out: Crime in Covid Times
Follow into
Save into
Follow into
Today's post is Crime in Covid Times.
It takes up a question that has been difficult for standard economic models of crime: why did American homicides, and gun homicides in particular, become so unusually volatile after 2019?
The first link, Crime in Covid Times, is the post's own framing of that puzzle. The post's author adds no lengthy commentary beyond pointing to the paper's author, Jens Ludwig. The post presents the central argument that the widely held view of crime as a rational weighing of benefits and costs, the framework associated with Gary Becker's 1968 work, does a generally poor job of explaining recent homicide trends. Behavioral economics, the post argues, helps explain what the Becker model cannot. The starting point is that most shootings do not further some larger goal such as robbery or gang wars over drug turf. Those would be instrumental violence. Instead, they are expressive violence: arguments settled with guns. The question then becomes why arguments start or escalate. The behavioral model points to automatic cognition, the fast and effortless mental processing that is sometimes prone to error. The pandemic increased automaticity by increasing distress, a cognitive bandwidth tax. That helps explain why recent homicide trends, driven by changes in expressive violence, are mirrored by similar trends for deaths from drug overdose, car crashes, and suicide, especially for Black Americans, the group most affected by violence. The post calls these deaths of decision-making. It also notes a large rise and fall in gun carrying, which might partly stem from more traditional Becker-like mechanisms. Large changes in handgun sales and police stops, both of which may also have been affected by the killing of George Floyd, changed the odds that a given argument had a gun present.
The second link, By Jens Ludwig, is the underlying National Bureau of Economic Research working paper. It is Working Paper 35827, issued in September 2026 and prepared for the Brookings Papers on Economic Activity conference in Washington. The paper develops the same argument in more detail. It rejects the standard rational weighing view as inadequate for recent American gun violence. Most shootings, it shows, are not instrumental violence in service of robbery or gang wars. They are expressive violence, arguments in which a gun happens to be present. The behavioral model explains why arguments start or escalate through automatic cognition, which is fast but error-prone. The pandemic increased distress and automaticity, imposing a bandwidth tax on judgment. That accounts for the co-movement of homicides with overdoses, car crashes, and suicides, grouped as deaths of decision-making. The paper also notes a large rise and fall in gun carrying. Here the explanation may partly return to Becker-like mechanisms, because changes in handgun sales and police stops, both possibly affected by the killing of George Floyd, changed the odds that an argument had a gun present. The paper's acknowledgements thank James Alan Fox for help with the Supplemental Homicide Report data. The conclusion is that American gun violence is mostly arguments with guns, so homicide trends are due mostly to changes in the number of arguments and the odds that a given argument had a gun present.
Taken together, the links point to a broader theme. Recent swings in American violence may have less to do with criminals rationally calculating incentives and more to do with the cognitive and emotional conditions that make ordinary disputes turn lethal. The pandemic, by taxing bandwidth and raising distress, appears to have shifted the background rate of bad decisions, and that shift shows up not only in homicide but also in overdose deaths, car crashes, and suicide.