Understanding how households’ car ownership and use respond to changes in income and fuel price is essential for travel demand forecasts and the design of car use policies. The literature estimating the effect of income and fuel prices on car use have almost exclusively studied the decision of owning a car and how much to drive as two separate choices. This may potentially result in the failure to capture the full response of the household or in model misspecification. In this paper, we estimate the income and fuel price elasticities of the joint decision of car ownership and car travel on household registry micro panel data using a discrete-continuous model. 

We use a correlated random effects approach to control for unobservable household heterogeneity. Our discrete-continuous model allows us to estimate the effect of income and fuel price on car ownership, car travel conditional on car ownership, and the total effect on vehicle kilometres travelled (VKT). 

We use a register database covering the Swedish population from the years 1999 to 2018, containing 96.7 million observations. The database includes socio-economic information on all registered adult Swedes, and information about all privately owned cars.

We find an income elasticity of VKT 0.4, half the elasticity comes from the impact of income on car ownership and the other half from the car travel conditional on car ownership. Hence, estimating the income elasticity of VKT on only car owners as earlier studies on micro panel data studies have done can underestimate the elasticity.

The model yields the fuel price elasticity of VKT -0.72 which is larger than has been found by meta-studies, but only slightly larger than the previous micro panel data studies and the previous discrete-continuous studies. We find that fuel prices primarily impact car travel and has a smaller impact on car ownership.