【主讲】范青亮 (厦门大学)
【主题】Estimating A Large System of Seemingly Unrelated Regressions Using Penalized Quasi-Maximum Likelihood Estimation
【时间】2016年4月15日 (周五) 15：30-17：00
【地点】上海财经大学经济学院楼701室
【语言】英文
【摘要】In this paper, we propose using a shrinkage estimator, penalized quasi-maximum likelihoodestimator (PQMLE), to estimate a large system of equations in seemingly unrelated regressionsmodel where the number of equations is big relative to sample size. We develop the asymptoticproperties of the penalized quasi-maximum likelihood estimator for both error covariance matrix andmodel coefficients. In particular, we derive the asymptotic distribution of the coefficient estimatorand the convergence rate of the estimated covariance matrix in terms of Frobenius norm. The oracleproperty of the covariance matrix estimator is also established. Simulation results show that whenthe number of equations is big relative to sample size and the error covariance matrix is sparse, thepenalized likelihood estimator performs much better than conventional estimators. As an illustration,we apply the PQMLE to the study of state level public capital returns in the United States.
