Use this URL to cite or link to this record in EThOS:
Title: Essays on partially identifying models and their applications
Author: Kim, Dongwoo
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
This thesis studies partial identification in discrete outcome models and their empirical applications. Chapter 1 investigates popular count data instrumental variable (IV) models. Many methods in the literature ignore the discreteness of count outcomes and thereby suffering from undesirable misspecification problems. To address this problem, a partially identifying count data IV model is developed. The model requires neither strong separability of unobserved heterogeneity nor a triangular system. Identified sets of structural features are derived. The size of the identified set can be very small when the support of an outcome is rich or instruments are strong. The proposed approach is applied to study effects of supplemental insurance on healthcare utilisation. In Chapter 2, partial identification in competing risks models for discretely measured or interval censored durations are studied. These models are partially identifying because of 1) the unknown dependence structure between latent durations, and 2) the discrete nature of the outcome. I develop a highly tractable bounds approach for underlying distributions of latent durations by exploiting the discreteness and I investigate identifying power of restrictions on the dependence structure with no assumptions on covariate effects. Bounds are obtained from a system of nonlinear conditional moment (in)equalities. I devise a solution method that requires much less computational burden than existing methods. Asymptotic properties of bound estimators and a simple bootstrap procedure are provided. Chapter 3 applies the proposed bounds approach in Chapter 2 to re-evaluate trends in cancer mortality by extending the ``war on cancer'' data studied in Honore and Lleras-Muney (2006). I find substantial reduction in cancer mortality. Estimated patterns differ from the original findings. In another application, I investigate the effects of extended unemployment benefits on unemployment spells using data from Farber et al. (2015). Bound estimates support the original finding that extended benefits did not discourage active job seekers during and after the Great Recession.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available