Robust Estimators for Finite Mixtures of Count Data Regression Models and their Applications
Year of Dissertation:
2010
Finite mixtures of count data regression models have been successfully used for modeling discrete responses arising from heterogeneous populations. But the maximum likelihood (ML) estimator for such models are sensitive to data contamination and extreme values. This dissertation develops two robust estimators for finite mixtures of count data regression models. One is the minimum Hellinger distance (MHD) estimator and the other is the minimum L2 error (L2E) estimator, a special case of the minimum density power divergence estimator. Two Monte Carlo simulation studies show that the MHD and L2E estimators are more robust than the ML one but come at the cost of efficiency. However, the robustness property of the MHD and L2E estimators is deteriorated as the mixing probability approaches one.
Essays in corporate finance
Year of Dissertation:
2010
ABSTRACT
Credit Risk Pricing in Single Name Corporate CDS
Year of Dissertation:
2010
The paper examines the determinants of the dynamics and term structure of credit default swap (CDS) spreads. I focus on roles of the interest rate risk and default risk. I extract interest rate factors from libor-swap curve, based on affine models. With a large data set, I test a three-factor CDS spread term structure model for 100 companies, using exact matching technique. The factors consist of two interest rate factors and one default risk factor. The estimation shows intricate interaction between interest rate factor and credit risk factor in determining the default process. I find that positive shock of the first interest rate factor will increase subsequent default risk factor, while positive shock on the second interest rate factor will decrease the subsequent default risk factor. In terms of direct effect of interest rate factors on the default process, positive shock on the first interest rate factor will decrease the default arrival rate at the moment, while positive shock on the second interest rate factor will increase the default arrival rate at the moment. Most of the effects of the shocks to interest rate factors will be sent to the default process indirectly through their impact on the subsequent move of the credit risk factor. CDS spreads of high credit rating companies are less responsive to the shocks on default risk than those of low credit rating companies. The work enhances our understanding of the process which underlies the CDS spreads movement and regimes. Such knowledge is essential for CDS pricing, risk measurement and management, hedging the related risk.
Three Essays in Health Economics
Year of Dissertation:
2012
This dissertation consists of three essays. In my first essay, I examine the relationship of the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC program) to breastfeeding. Although WIC promotes breastfeeding among its participants through education, counseling, and the provision of enhanced food packages to breastfeeding women, the program has been criticized for discouraging breastfeeding by providing free infant formula. In order to estimate the extent to which participation in WIC discourages breastfeeding, I employ a methodology that disentangles selection bias associated with WIC participation from the incentives associated with the provision of free infant formula. Findings suggest that postpartum entrants are less likely to breastfed for at least 6 months and have shorter breastfeeding durations than non-participants, and the effects are significantly larger among twin mothers than among singleton mothers.
Essays in Unauthorized Immigration and Migration
Author:
Bilesha Weeraratne
Year of Dissertation:
2012
This dissertation consists of three essays. The first essay develops a new methodology defined as the microdata-based methodology to identify unauthorized immigrants in the US. This identification is based on a discrete choice model on observed authorized and unauthorized immigrant data in 1986 with adjustments for time dynamics between 1986 and 2010 and for calibration in new data. This method produces an algorithm to identify unauthorized immigrants in the American Community Survey (ACS) 2010, and estimates that there were 7,700,869 adult unauthorized immigrants in the US on January 1, 2011. The essay includes detailed descriptive analyses of unauthorized immigrants and their children.
Monetary Policy Shocks and Their Effects on Business Borrowings, Aggregate Output, and Prices
Year of Dissertation:
2012
Developed from previous financial acceleration research, this thesis employs two VAR (vector-auto regression) systems to measure monetary policy shocks and their effects. The first uses the conventional Cholesky method and the second uses the methods developed by Bernanke and Blinder (1992) and Bernanke and Mihov (BM 1998). I use the federal funds rate and as the two monetary policy instruments in the VAR systems.
ESSAYS ON THE IMPACT OF LEGALIZED ABORTION AND THE PILL ON CHILDBEARING BEHAVIOR AND BIRTH OUTCOMES
Year of Dissertation:
2011
This dissertation is composed with two essays on reproductive behavior and related health outcomes. Recent studies on the "power of the Pill" have not adequately accounted for the role of abortion in the years between 1970 and 1973. In the first essay, I use rediscovered data on abortions performed in New York State in 1971 and 1972 by age, race and state of residence to demonstrate the remarkable impact of legal abortion services in New York on the fertility rates of young women as far away as Montana prior to Roe v. Wade. My results strongly suggest that laws enhancing access to legalized abortion more than the Pill caused birth rates of young women to fall in the early 1970s.
Essays on Empirical Asset Pricing
Year of Dissertation:
2012
Advisor:
K. Ozgur Demirtas
ESSAYS ON EMPIRICAL ASSET PRICING