How Effective are Non‐Monetary Instruments for Safe Driving? Panel Data Evidence on the Effect of the Demerit Point System in Denmark

Extended Abstract: Driving behavior plays a central role in instigating traffic accidents. Hence, affecting the driving behavior of individuals remains a top priority for policy makers around the globe. Standard economic and deterrence theory predicts that rational drivers will respond to policy instruments (or reforms) that increase the expected cost of committing traffic violations (see Becker 1968; Polinksy and Shavell 1979). Following this justification, both monetary and non-monetary penalties are commonly used as key instruments to ensure public road safety. The most common monetary instruments involve traffic tickets (fines), whereas driving license revocation based on point recording is a recently emerging non-monetary instrument for safe driving. Many developed countries now use a point recording system for drivers violating traffic rules. In some countries, this point recording scheme is integrated with insurance premiums (see Dionne et al. 2013). Thus, the point recording scheme serves as incentive for safe driving by increasing the expected pecuniary cost in terms of forgone income associated with loss of driving privilege or through indirect costs on insurance premiums and license redemption. Although the implementation of the point recording system varies across countries, it is common that accumulating demerit points above some threshold in some specified period leads to revocation of the driving license. Therefore, the point recording scheme serves a dual purpose of incentivizing car users to drive carefully and incapacitating reckless drivers (see Bourgeon and Picard 2007). Denmark introduced the demerit point system (DPS) in September 2005. The point recording scheme in Denmark applies to different types of traffic violations and three demerit points in three years lead to conditional suspension of the driving license. While policy instruments such as fines and the point recording scheme have long been used, little is known about the effect of these instruments on inducing safe driving and improving public road safety. This is mainly because most of these instruments are endogenously introduced, a problem that saddles evaluation techniques with simultaneity and reverse causality problems. That is, areas with higher records of traffic violations (or accidents) are subject to higher levels of traffic enforcement. In the absence of longitudinal data, this problem can lead to misleading inferences on the effect of traffic (police) enforcement on driving behavior and public road safety. A few recent studies, including those of Bourgeon and Picard (2007) and Dionne et al. (2011), provide theoretical foundations on the efficacy of the point recording scheme. Bourgeon and Picard (2007) present a theoretical model that demonstrates the effectiveness of the point recording scheme in different scenarios. Dionne et al. (2011) extend Bourgeon and Picard’s (2007) model by linking the point recording scheme to insurance pricing, and provide some empirical evidence of moral hazard in public road safety. Despite their innovative approach to testing the prevalence of moral hazard in public road safety empirically, Dionne et al. (2011) do not simultaneously capture (or disentangle) unobserved heterogeneity and behavioral response (moral hazard) effects. This paper uses unusually rich longitudinal traffic offense data for Denmark to estimate the effect of non-monetary penalties based on point recording, on driving behavior. The longitudinal traffic offense data enable us to follow the driving behavior of individuals before and after the introduction of the point recording scheme in Denmark. This  study adds at least two novel contributions to the existing literature on punishment and deterrence: (a) we estimate the behavioral response of drivers to a non-monetary penalty based on demerit points using a detailed longitudinal data which enable us to explicitly disentangle behavioral response (moral hazard) and unobserved heterogeneity (selection) effects associated with different groups of drivers; (b) in addition to evaluating the effectiveness of non-monetary incentives for safe driving, this study estimates enormously heterogeneous response of drivers and highlights short-term as well as long-lasting impacts of non-monetary penalties. Our identification strategy exploits the longitudinal feature of the traffic offense data and uses standard difference-in-differences approach. We compare the driving behavior of individuals subjected to non-monetary penalties through demerit points (treatment group) with those drivers who are only subjected to monetary penalties (control group), before and after Denmark introduced the DPS. We exploit a specific rule introduced by the reform, which assigns speed offenders to one of two types of penalty based on a 30 percent speed threshold. Accordingly, drivers exceeding the speed limit by more than 30 percent are subject to a demerit point and a fine, whereas those exceeding the speed limit by less than (or equal to) 30 percent are subject to only the fine. Interestingly, this rule strengthens our identification strategy by providing some local randomization around the threshold. We measure individuals’ driving behavior using the frequency of traffic violations committed and the probability of committing a traffic offense in some specified period. We find that drivers exhibit substantial behavioral response to each demerit point assigned to their driving license by reducing their frequency of traffic offenses and their probability of committing a traffic offense. We also find that drivers’ effort, and hence response, increases with the number of demerit points accumulated. Depending on the number of points accumulated in the treatment period, affected drivers reduced their frequency of traffic offenses by 15–30 percent. Similarly, affected drivers reduced their likelihood of committing traffic violations by 11–20 percent. We also show that those drivers who are potentially more reliant on their car are more responsive to each demerit point assigned to their driving license. We find that the treatment effects are strong for those individuals who are self-employed and those who commute longer distances to their workplaces. Finally, while previous studies on the effect of monetary penalties for crimes and traffic offenses suggest that fines may be imposed differentially based on individuals’ income and wealth, our results show that non-monetary penalties based on demerit points affect even those with high incomes and wealth. Thus, our results hint that socially optimal non-monetary penalties may differ based on the expected cost and consequences of these penalties for different groups of drivers. More broadly, our findings assert that non-monetary instruments – penalties that could affect driving privilege – are effective in inducing safe driving. This has some important implications for designing deterrence, public road safety and insurance policies. We check the robustness of our results considering different specifications and allowing for enormously rich heterogeneity on the effect of the reform. We test the implication of our identifying assumption using a series of empirical exercises. More specifically, our identification strategy assumes that in the absence of the reform, the treatment and control group drivers would have exhibited similar (average) behavioral evolution (changes) in driving behavior. We believe that this assumption holds fairly well for the following reasons: (1) we find that treatment and control group drivers have statistically similar pre-treatment observable characteristics in terms of a large set of relevant variables. (2) Treatment and control group drivers exhibit similar pre-reform driving behavior. (3) Our placebo regressions indicate that treatment and control group drivers share a statistically identical pre-reform time trend in driving behavior. These placebo regression exercises also suggest that the demerit points assigned to driving license are potentially ineffective once they are no longer valid, an evidence that undermines the long-lasting effects of these penalties. These pieces of evidence suggest that our estimates can plausibly be interpreted as causal effects carrying behavioral response of drivers for each demerit point assigned to their driving license. References: Becker, Gary S. (1968). Crime and Punishment: An Economic Approach. Journal of Political Economy 76 (2), 169–217. Bourgeon, Jean-Marc, and  Pierre Picard (2007).Point-Record Driving License and Road Safety: An Economic Approach. Journal of Public Economics 91, 235–258. Dionne, Georges, Jean Pinquet, Mathieu Maurice, and Charles Vanasse (2011). Incentive Mechanisms for Safe Driving: A Comparative Analysis with Dynamic Data. The Review of Economics and Statistics , 93(1), 218-227. Polinsky, A. Mitchell, and Steven Shavell (1979). The Optimal Tradeoff between the Probability and Magnitude of Fines. American Economic Review 69 (5), 880–891.