主题：Inference on State Variable and Prediction for Linear Gaussian State Space Model with Aggregate and Disaggregate Data
This paper studies several econometrics issues related to data aggregation for linear Gaussian state space model. In particular, it presents conditions under which the disaggregate model dominates its aggregate counterpart for inference on state variables and forecasting aggregate observables in terms of mean squared errors. From a methodological point of view, there are two contributions. First, it generalizes the aggregation results from autoregressive moving average (ARMA) models to linear Gaussian state space models. Second, it provides new theoretical results for inference on state variables, teasing out the channels through which the disaggregate model achieves better inference than the aggregate model. Monte Carlo simulations confirm the theoretical results. An empirical application to aggregate and disaggregate unemployment data reveals the extent of the information loss caused by aggregation.
Dr. Anlong Qin earned his Ph.D. in economics from Boston University in May 2022. His research fields are econometrics, time series analysis, and financial econometrics. Currently, he is interested in Bayesian analysis and machine learning methods for estimation, inference and forecasting with big datasets. He also holds a master’s degree in quantitative economics and a bachelor’s degree in economics from Renmin University of China.
时间：2022.09.22 (周四) 14:30-15:30