Title: Augment large covariance matrix estimation with auxiliary network information
Authors: Shaoran Li - Peking University (China) [presenting]
Oliver Linton - University of Cambridge (United Kingdom)
Shuyi Ge - Nankai University (China)
Weiguang Liu - University of Cambridge (United Kingdom)
Abstract: The aim is to incorporate auxiliary information about the location of significant correlations into the estimation of high-dimensional covariance matrices. With the development of machine learning techniques such as textual analysis, granular linkage information among firms that used to be notoriously hard to get are now becoming available to researchers. Our Network Guided Estimator combines the banding and thresholding procedures with the help of augment information from other sources. Simulation results show that the new method has smaller estimation errors comparing with other methods in the literature. We empirically apply the Network Guided Estimator to estimate the covariance of the excess returns of S\&P 500 stocks. The constructed global minimum variance portfolio has the smallest volatility among all competing methods.