Analysing the demand for carsharing in Bergen : an empirical approach using car-specific and demographic data

Abstract

Background: Car-sharing is gaining popularity throughout the world. Currently, there is limited research on car-sharing in Norway. This paper aims to identify empirically, based on past data, what demographic and car-specific variables determine demand for a car-sharing service in Bergen. Additionally, it aims to predict future car performance for Bildeleringen’s cars. Research question: What are the drivers of demand for the utilization of a car-sharing service? Method: Using data from Bildeleringen and Statistics Norway, we use multiple linear regression to determine drivers of demand. Furthermore, linear regression is also used for predicting future car performance of Bildeleringen’s cars. Linear regression is chosen out of five possible models based on cross-validation error. Results: Several variables significantly increase or decrease car performance of Bildeleringen’s cars. The variables “car type”, “higher average amount of cars in the parking spot”, “electric cars”, “cars with automatic gear shift”, “higher median income”, “spring”, and “households without car” increase performance, while “higher population density”, “age 20-24 years old”, “retired”, “male”, and “child cushion” decrease performance. Analysis: Out of the significant demographic variables, only “households without car”, “age 20-24 years old”, “retired” and “male” affect car performance to a noteworthy degree. Out of the car-specific variables “car type”, “electric cars”, “cars with automatic gear shift”, “spring”, and “child cushion” affect car performance to a noteworthy degree. Conclusion: The relationship between car performance and demographic variables is not strong. Car-specific variables seem to show a higher degree of correspondence with car performance. Key words: Car-sharing, car-sharing locations, car performance, linear regression, demographic variables, car-specific variables, Norway.