This paper studies the impact of income-based criminal punishments on crime. In Finland, speeding tickets become income-dependent if the driver’s speed exceeds the speeding limit by more than 20 km/h, leading to a substantial jump in the size of the speeding ticket. Contrary to predictions of a traditional Becker model, individuals do not bunch below the fine hike. Instead, the speeding distributions are smooth at the cutoff. However, I demonstrate that the size of the realized speeding ticket has sizable, but short-lived, impacts on reoffending. I use a regression discontinuity design to show that fines that are, on average, 200 euros larger decrease reoffending by 15 percent in the following six months. The drop in reoffending is driven by high-income individuals facing the highest fine at the cutoff. I estimate that a fixed fine hike that matches the current deterrence effect would raise the fine increase faced by the bottom income quartile by about 300% at the cutoff, relative to the increase they face under the current income-based schedule. My empirical results are consistent with an explanation that people operate under information frictions. To illustrate this, I construct a Becker model with misperception and learning that can explain all the empirical findings.