No abstract

This paper studies drivers' responses to a 'notched' penalty scheme in which speeding penalties are stepwise and discontinuously increasing in speed. We present survey evidence suggesting that drivers in Germany are well aware of the notched penalty structure. Based on a simple analytical framework we analyze the impact of the notches on drivers' optimal speed choices. The model's predictions are confronted with data on more than 150,000 speeding tickets from the Autobahn and 290,000 speed measures from a traffic monitoring system. The data provide evidence on modest levels of bunching, despite several frictions working against it. We analyze the normative implications and assess the scope for welfare gains from moving from a simple, notched penalty scheme to a more complex but less salient Pigouvian scheme.

No abstract

We develop a theoretical model to identify and compare partial and equilibrium effects of uncertainty and the magnitude of fines on punishment and deterrence. Partial effects are effects on potential violators' and punishers' decisions when the other side's behavior is exogenously given. Equilibrium effects account for the interdependency of these decisions. This interdependency is important since, in the case of legal uncertainty, higher fines may reduce the willingness to punish, which in turn reduces the deterrence effect of high fines. Using a laboratory experiment, we identify these effects empirically by means of a strategy-method design in which potential violators can condition their behavior on the behavior of potential punishers and vice versa. All our experimental findings on both partial and equilibrium effects are in line with the hypotheses derived from the theory.

We use a laboratory experiment to test the impacts of uncertainty, the magnitude of fines and aversion against making type-I and type-II errors on legal decision making. Measuring uncertainty as the noise of a signal on the defendant's guilt observed by legal decision makers, we observe that a supposed wrongdoer is less likely to be punished if fines and uncertainty are high. Furthermore, judges care far more about type-I errors and violators steal far less often than expected payoff maximizers would. While our results support the theoretical predictions on average, a cluster analysis provides evidence for heterogenous behavior of participants, many of whom don't respond to changes in the parameters or are far more driven by uncertainty than the magnitude of fines.