Wiktor Budziński - Endogeneity of indicator variables in hybrid choice models: Monte Carlo investigation vs. stated preference study.

Endogeneity of indicator variables in hybrid choice models: Monte Carlo investigation vs. stated preference study.

Presented by Wiktor Budziński, University of Warsaw

Friday, 24^th of January, 11:00-12:00.
Fine Art Building SR (2.09), University of Leeds.

All welcome. No booking required.

Abstract

We investigate the problem of endogeneity arising from incorporating indicator variables (e.g., measures of attitudes) in discrete choice models. We demonstrate that although hybrid choice framework can resolve this issue, it requires explicit accounting for in the model, which has not typically been done in applied studies to date. By conducting a Monte Carlo experiment, we demonstrate the extent of the bias resulting from measurement and endogeneity biases. We propose two novel solutions to address the endogeneity problem: explicitly accounting for correlation between structural and discrete choice component error terms (or with random parameters in the utility function), or introducing additional latent variable. Using simulated data, we demonstrate that these approaches work as expected, that is, they result in consistent estimates of all model parameters. We then apply these solutions to a stated preference study regarding extension of the public theater offer in Poland in order to identify an effect of perceived consequentiality on individuals' choices. We find that directly accounting for endogeneity significantly improves model's fit to the data and that it may even change the sign of some coefficients. Nevertheless, the two proposed solutions lead to different conclusions and may not always be easy to interpret. Specifically, it is challenging to assess whether additional latent variables in the second solution captures an effect of the missing variable causing the endogeneity or rather another (latent) dimension of the consequentiality perception.