Job market paper
Simulation Based Estimation of Multinomial Discrete Choice Model with Fixed Effects.
PDF Supplemental Material Codes
Multinomial discrete choice model, including binary
choice model, is a class of widely used nonlinear models. Including
unobservable heterogeneity in such models is necessary in many applications due
to the unobserved variations in individual’s preference or attributes.
Unfortunately this will cause problem for their point identification. We study
the estimation of a multinomial discrete choice model in panel data with
potentially multidimensional fixed effects that can be set identified on its
parameters and conditional average partial effects for the outcome and choice probability.
The model we study in this paper is general in the sense that it can contain
components without closed form, and its conditional probabilities for each alternatives
and partial effects can be gotten through Monte Carlo simulation. For this
model we propose a simulation based estimator for all the set identified quantities.
We show that our estimator is consistent under general conditions and a perturbed
bootstrap method can be used to implement its inference. A numeric example with
simulated data is given to show the behavior of our estimator and we find that
the estimated bounds of partial effects contain their true effects.
Working paper
Is there an income discrimination between temporal worker and formal worker in Chinese enterprises?
Working in progress
Behavior of the MLE in non-identified binary choice panel models with unobserved heterogeneity. (with Jesús M. Carro)
Set identified dynamic binary choice model with fixed effects and serially correlated errors.
Set identified dynamic binary choice model with fixed effects and serially correlated errors.