Public transport demand elasticities during the recessionary phases of economic cycles. (August 2015)

• Record Type: Journal Article
• Authors: Cordera, Ruben; Canales, Cesar; dell'Olio, Luigi; Ibeas, Angel
• Is Part Of: Transport policy. Volume 42(2015)
• Page Start: 173
• Page End: 179
• Publication Date: 2015-08
• Subjects: Public transport demand -- Elasticities -- Travel behavior modeling -- Business cycle; Transportation and state -- Periodicals; Transportation -- Rates -- Periodicals; 388
• DOI: 10.1016/j.tranpol.2015.05.022
• Languages: English
• ISSNs: 0967-070X

Abstract

Abstract: The serious recession suffered by the Spanish economy has had an important impact on the working of transport systems. This article provides evidence of how the state of the economy, measured using per capita income levels and unemployment rates, can influence the demand for public transport by bus with data from the city of Santander (Spain) for the period 2001–2012. The methodology used has involved the short and long run equilibrium demand elasticities estimation using a log–log regression model considering the presence of autocorrelation in the residuals and the endogeneity of the transport supply. The results show that the demand for transport is sensitive to changes in unemployment rates with an elasticity estimated at 0.133 in the static equilibrium model and of 0.210 in the long run dynamic model. The resulting elasticity for income levels was not significant in all the models, with a parameter estimated in the static model of −0.505 and of −0.861 in the long run dynamic model. The model estimated by two stage least squares validated using data from 2013 with unemployment rate as an independent variable gave the lowest average square error in validation. Highlights: Equilibrium and dynamic public transport demand models are specified. The objective is studying the influence of economic cycles on transport demand. The elasticity of unemployment was significant in modelling trip demand. The two stage least squares model type had a slightly lower mean squared error.