Detection of income effect in mode choice: Theory and application

Although income effect is likely to be present in mode choice for most of the population within the developing world, presently used approaches do not take it into account. In fact, the income variable that sometimes enters utility specifications has been justified as a proxy for other variables, which makes its role rather ambiguous. In this paper, the problem is restated from its microeconomic foundations, showing that the usual non-income-sensitive specifications can be interpreted as a particular case within a more general framework which provides the basis for a methodology to test the presence of income effect. The methodology, which does not require additional information besides the usual data collected for disaggregate models, is applied to trips to work originated in a middle-income corridor of Santiago, Chile. The results quantitatively confirm previous qualitative analysis, and show empirically the need to employ income-sensitive mode choice models.