Reported by the School of Statistics and Data Science
Recently, Associate Professor LIU Yongxin from the School of Statistics and Data Science of NAU, as the corresponding and co-first author, together with Professor KONG Xinbing and others, completed the paper “Matrix Quantile Factor Model”. This paper was published online in the top international journal of statistics and econometrics, Journal of Business & Economic Statistics (JBES). This journal is sponsored by the American Statistical Association and is recognized by the British Business School Association as a four-star (ABS 4) journal. It is an internationally renowned important journal in the field of statistics and econometrics.
This study focuses on matrix-valued data with low-rank structure and innovatively proposes the matrix quantile factor model. The paper estimates the row factor space and column factor space by minimizing the empirical check loss function with orthogonal rotation constraints. The paper theoretically provides the convergence rate of the estimator in the average F-norm and establishes the central limit theorem for the factor and loading estimators. The study also proposes three methods for determining the number of row factors and column factors, providing effective tools for model selection in related fields. At the application level, this model has been successfully applied to the data analysis of global import and export trade networks. The results show that compared with the traditional vector-based factor model, the matrix quantile factor model can significantly reduce the prediction error; at the same time, introducing marginal quantile factors can further improve prediction accuracy, fully demonstrating the important value of quantile information in macroeconomic forecasting.
Associate Professor LIU Yongxin's main research areas include high-dimensional data statistical inference, factor models, and econometric methods. This achievement reflects the continuous deep exploration and academic influence of NAU in the forefront of statistics.
