Ridesharing a new traffic risk? The introduction of Moto-Uber in the Dominican Republic. 

Abstract
Background: New technologies that change public and private transportation systems may alter the way populations use roadway networks and thus change exposure to road risk behaviors. Ridesharing, a technology, that connects contracted owner-operator drivers with prospective passengers through a mobile application, may theoretically decrease traffic risk by improving the quality of driving behavior of those individuals who offered this transport service, or increase it by augmenting the level of traffic risk exposure associated with the natural growth of the motor vehicle fleet. Only recently have scholars started to study the effects of ridesharing on traffic outcomes such as traffic fatalities or injuries (1-8). While these results have been mixed, with improvements shown in the United States (1-6) and no associations in South Africa (7) and Chile (8), these studies have only considered effects of ridesharing on motor light vehicle crashes. In countries where motorcycle ridesharing exists and is prominent, it is unclear if it has affected motorcycle traffic crashes. The objective of this work is to assess whether Moto-Uber changed motorcycle traffic fatalities in Dominican Republic cities in which this specific model of ridesharing was introduced.

Methods: Monthly numbers of motorcycle fatalities per 100,000 motorcycle fleet from January 2012 to December 2018 were obtained from the National Institute of Statistics of the Dominican Republic and the Dominican Revenue Agency respectively. A hybrid time-series difference-in-difference (DID) analysis using multivariable regression with Newey-West standard errors was used to compare trends in the log rate of motorcycle fatalities in Santo Domingo and Santiago (cities in which Uber moto was introduced) to La Romana, San Cristóbal and La Vega (cities in which Uber is absent and fulfilled the parallel trends assumption), before and after the introduction of Uber in March 2017. Estimates were adjusted for time trends, seasonality, year and city effects. We also ran an interrupted time series (ITS) analysis using pre-trends and seasonal components as controls to compare to our DID estimates.

Results: Estimates from the DID regression model suggest that the introduction of Uber in Santo Domingo – Santiago relative to La Romana, San Cristóbal and La Vega was associated with a 3% increase (0.03, 95% Confidence Interval [CI]:0.00;0.06) in the log rate of motorcycle fatalities per 100,000 motorcycles. Estimates from the ITS model suggests that the introduction of Uber was associated with a 2% increase (0.02, 95% CI:0.01;0.03) in the monthly log trend of the same outcome.

Conclusion: In the Dominican Republic, concomitant increases in motorcycle fatal crashes were observed after the introduction of Motor-Uber in cities in which this service was offered. One potential explanation is an increase in the number of motorcycle trips after the introduction of this ridesharing model. Nevertheless, in this country negative relationships between ridesharing and motorcycle crashes can be counteracted by making helmet use obligatory for drivers and passengers, building exclusive motorcycle lanes, making anti-brake system and daytime running lights mandatory for new imported motorcycles, restricting hours of use or zone of ridesharing and overall promoting better public transit systems.