It is possible to increase traffic and experience shorter travel times under automation.
Our analysis showed that subsidization (Policy 1) always leads to higher social welfare than ‘tax-and-subsidize’ (Policy 2). Nevertheless, a higher level of automation would be achieved by the latter, and this could lead to important secondary benefits — such as a reduction in accidents and environmental pollution.
We also found that the optimal policy will depend on the market price of AVS and the ability of the industry’s infrastructure to service them. Hence, the idea of an ‘optimal’ policy will change as AVS become more affordable and the infrastructure is upgraded to better serve them over time. For instance, altering sharing arrangements (Policy 3) would not be an ideal approach when AVS first enter the market because the market-share would not be large enough to exploit the benefits of sharing; but, as the market share grows and the infrastructure evolves, this policy becomes increasingly powerful.
Some of our insights might appear to be counter-intuitive. First, it is commonly accepted that ‘any increase in traffic will lead to longer travel times’. However, we found that this may not be the case when there is a mix of regular vehicles and AVS on the road. Put simply, it is possible to increase traffic and experience shorter travel times under automation. Second, since AV users will experience more free time and comfort when travelling, the induced traffic in the network may increase. Therefore, as the level of comfort increases (and the value of time decreases), automation becomes more harmful to social welfare. This counter-intuitive assertion results from the predicted tendency of AV owners to travel more because of the extra comfort afforded by these vehicles.
While AVS are expected to benefit society in many ways, their high initial cost may hinder their widespread adoption. Therefore we believe government intervention is required to ensure that AVS are affordable for the public. Policies that endorse automation may use subsidization, taxation or the promotion of vehicle sharing between multiple users.
The optimal policy for AVS depends on the price gap between autonomous and regular vehicles, and the ability of an area’s infrastructure to service AVS. Therefore, the optimal policy will change with time, as infrastructure is improved and AVS become more cost-effective due to mass production.
In closing
Many forthcoming policies in the realm of vehicle automation remain to be investigated, including changes to traffic rules, taxi regulation and land use changes. Continued research in these areas will allow us to better comprehend the future impact of vehicle automation and provide further tools for policymakers to make effective decisions.
An important next step is to investigate the pricing structure of households in each group of vehicle sharers or ‘AV coalitions’. For example, for joint owners of AVS, how should the ownership costs be equitably divided among the members of each coalition to reach an equilibrium where no household benefits from switching to another coalition?
In the meantime, we hope that our analysis and the insights it generates may support government agencies and other regulatory bodies in their study and implementation of policies to optimally control the adoption of AVS.