Working Paper: CEPR ID: DP13784
Authors: Antonio Cabrales; Ryan Kendall; Angel Sánchez
Abstract: We consider a situation where driverless cars operate on the same roads as human-driven cars. What policies effectively discourage unsafe (fast) drivers in this mixed-agency environment? We develop a game theoretic model where driverless cars are the slowest and safest choice whereas faster driving speeds lead to higher potential payoffs but higher probabilities of accidents. Faster speeds also have a negative externality on the population. The model is used to create four experimental policy conditions. We fi nd that the most effective policy is a mechanism where the level of punishment (to fast drivers) is determined endogenously within the driving population.
Keywords: driverless cars; policy; social norms; game theory
JEL Codes: C90; D62; D63
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| faster driving speeds (R48) | higher potential payoffs (G19) |
| faster driving speeds (R48) | higher probability of accidents (R41) |
| higher probability of accidents (R41) | negative externality for the population (D62) |
| endogenous punishment mechanism (D91) | reduced risky driving behavior (R48) |
| exogenous punishment condition (K49) | no significant change in behavior (C92) |
| endogenous punishment mechanism (D91) | shift moral responsibility to the driving community (R48) |
| endogenous punishment condition (K42) | incentivizes would-be automated drivers to choose faster manual driving options (J33) |