My dissertation is composed of three chapters that contribute to the ﬁelds of Applied Econometrics and Macroeconomics.
The ﬁrst chapter, “A Copula Model for Discrete Duration Data with Sample Selection,”
presents a copula model to account for sample selection in a model of unemployment duration data. I apply two Markov Chain Monte Carlo (MCMC) methods to determine posterior distributions for the model parameters. In particular, a version of the Gibbs sampler is applied to evaluate the integrals that result from the copula representation of the likelihood, and the Random Walk Metropolis Hasting (RW-MH) algorithm for sampling from the posterior distribution. The model is applied to discrete data on unemployment duration from the 2011 Current Population Survey. Joint estimation of the selection and duration equations indicates that selection bias is present, and the data are informative about the model parameters.
The second chapter, “Monetary Policy and Equity Prices in a Multivariate GARCH Model,”develops a model for the high-frequency analysis of a fundamental relationship in macroeconomics, between monetary policy and equity prices. A market-based approach to estimating daily changes in unexpected monetary policy is incorporated into a multivariate copula GARCH model. This allows for eﬃcient estimation of parameters in the mean equations for each of the variables, as well as the conditional heteroskedasticity and spillovers in volatility. I ﬁnd little evidence for a relationship between monetary surprises and equity prices on this scale, with a mean correlation of -0.0503 between the two time series, which does not vary systematically over time.
The third chapter, “Federal Reserve Communication and its Time-Varying Impact on the
Yield Curve: A Dynamic Nelson-Siegel Model for Daily Data,” implements the ﬁrst Dynamic Nelson-Siegel (DNS) model for daily data. This allows for estimation of the eﬀect of daily shocks to monetary policy and macroeconomic factors. Time-varying coeﬃcients account for the changing degree to which the zero lower bound (ZLB) constrains medium-term yields.
My results indicate that yields were most sensitive to shocks to the future path of monetary policy in 2006-2007 with a rapid decline in sensitivity from 2008 to 2013. By the second quarter of 2013 the eﬀect of a one standard deviation path shock on maturities under two years had fallen by more than 50% from the fourth quarter of 2007. As of July 2014 the sensitivity of 1-year yields began to show a modest increase, suggesting the market expectslift-oﬀ from the ZLB by mid-2015.