Search

We consider a regulator driving individual choices towards increasing social welfare by providing personal incentives. We formalise and solve this problem by maximising social welfare under a budget constraint. The personalised incentives depend on the alternatives available to each individual and on her preferences. A polynomial time approximation algorithm computes a policy within few seconds. We analytically prove that it is boundedly close to the optimum. We efficiently calculate the curve of social welfare achievable for each value of budget within a given range. This curve can be useful for the regulator to decide the appropriate amount of budget to invest. We extend our formulation to enforcement, taxation and non-personalised-incentive policies. We analytically show that our personalised-incentive policy is also optimal within this class of policies and construct close-to-optimal enforcement and proportional tax-subsidy policies. We then compare analytically and numerically our policy with other state-of-the-art policies. Finally, we simulate a large-scale application to mode choice to reduce CO2 emissions.

In carpooling systems, a set of drivers owning a private car can accept a small detour to pick-up and drop-off other riders. However, carpooling is widely used for long-distance trips, where rider-driver matching can be done days ahead. Making carpooling a viable option for daily commute is more challenging, as trips are shorter and, proportionally, the detours tolerated by drivers are more tight. As a consequence, finding riders and drivers sharing close-enough origins, destinations and departure time is less likely, which limits potential matching. In this paper we propose an Integrated System, where carpooling matching is synchronized with Public Transit (PT) schedules, so as to serve as a feeder service to PT in the first mile. Driver detours are proposed towards PT selected stations, which are used as consolidation points, thus increasing matching probability. We present a computationally efficient method to represent PT schedules and drivers trajectory in a single General Transit Feed Specification database, which allows to compute multimodal rider journeys using any off the shelf planners. We showcase our approach in the metropolitan area of Portland, Oregon, considering 8k randomly generated trips. We show the benefits of our Integrated System. We find that 10% more riders find a feasible matching with respect to the status quo, where carpooling and PT are operated separately. We release our code as open source., Comment: 15th ACM SIGSPATIAL International Workshop on Computational Transportation Science (IWCTS 2022)

We consider a regulator willing to drive individual choices towards increasing social welfare by providing incentives to a large population of individuals. For that purpose, we formalize and solve the problem of finding an optimal personalized-incentive policy: optimal in the sense that it maximizes social welfare under an incentive budget constraint, personalized in the sense that the incentives proposed depend on the alternatives available to each individual, as well as her preferences. We propose a polynomial time approximation algorithm that computes a policy within few seconds and we analytically prove that it is boundedly close to the optimum. We then extend the problem to efficiently calculate the Maximum Social Welfare Curve, which gives the maximum social welfare achievable for a range of incentive budgets (not just one value). This curve is a valuable practical tool for the regulator to determine the right incentive budget to invest. Finally, we simulate a large-scale application to mode choice in a French department (about 200 thousands individuals) and illustrate the effectiveness of the proposed personalized-incentive policy in reducing CO2 emissions.

While carpooling is widely adopted for long travels, it is by construction inefficient for daily commuting, where it is difficult to match drivers and riders, sharing similar origin, destination and time. To overcome this limitation, we present an Integrated system, which integrates carpooling into transit, in the line of the philosophy of Mobility as a Service. Carpooling acts as feeder to transit and transit stations act as consolidation points, where trips of riders and drivers meet, increasing potential matching. We present algorithms to construct multimodal rider trips (including transit and carpooling legs) and driver detours. Simulation shows that our Integrated system increases transit ridership and reduces auto-dependency, with respect to current practice, in which carpooling and transit are operated separately. Indeed, the Integrated system decreases the number of riders who are left with no feasible travel option and would thus be forced to use private cars. The simulation code is available as open source.

This paper establishes a general equivalence between discrete choice and rational inattention models. Matejka and McKay (2015, AER) showed that when information costs are modelled using the Shannon entropy function, the resulting choice probabilities in the rational inattention model take the multinomial logit form. By exploiting convex-analytic properties of the discrete choice model, we show that when information costs are modelled using a class of generalized entropy functions, the choice probabilities in any rational inattention model are observationally equivalent to some additive random utility discrete choice model and vice versa. Thus any additive random utility model can be given an interpretation in terms of boundedly rational behavior. This includes empirically relevant specifications such as the probit and nested logit models.