Essays in Empirical Macroeconomics

My dissertation is mostly focused on the effects and quantitative importance of news shocks in an open economy DSGE model. The novelty of these models pertain to their ability to analyze business cycle fluctuations in a structural setting with rigorous micro foundations. The first two chapters of my dissertation estimates open economy DSGE models and investigates the role of news shocks in explaining international business cycles. My third chapter focuses on identification and estimation of a partially observable bivariate probit model.;The first paper, titled "Propagation of News shocks in an Open Economy DSGE model" estimates a large open economy DSGE model of US and the Euro area with frictions and news shocks along with other unanticipated structural shocks. The role of news shocks in generating business cycles is an area of ongoing research and has garnered attention as being a major contributor of output fluctuations. In this paper we find that news shocks that originate domestically have an important quantitative role in explaining domestic output, inflation and interest rates. More specifically, news shocks from the US explain about 30% of US output and those from the Euro explain about 35% of Euro output. The international transmission of news shocks however are not important in affecting business cycles across countries.;The second paper, titled "News shocks and Business Cycles in a Small Open Economy of Canada" investigates the role of news shocks in a small open economy of Canada. Moreover we are also interested in the international transmission of such shocks from a large foreign economy such as the US. In this paper, we estimated a small open economy DSGE model with rigidities using Bayesian methods. We find that news shocks from the US have negligible role in explaining aggregate fluctuations in Canada. Nonetheless, we also find that news shocks originating in Canada play an important role domestically.;The third paper is titled " Identification and MCMC Estimation of bivariate probit models with partial observability". Partial observability in a bivariate probit model arises when one can only observe the binary outcome of a paired decision. Following Poirier (1980) we find a host of research applying a version of this model. However, most applications heavily rely on the assumption of independence across equations and forego estimation of the correlation parameter while some report misleading estimates. In this paper we perform Monte Carlo simulations to show that estimating the correlation parameter in a partially observable case is nontrivial as compared to a fully observable case. We also estimate the model using maximum likelihood as well as bayesian MCMC methods and apply to a dataset of Prezeworski and Vreeland (2002) studying the role of IMF.