Title: Forecasting conditional covariance matrices in high-dimensional data using the general dynamic factor model
Authors: Carlos Trucios - Sao Paulo School of Economics - FGV (Brazil) [presenting]
Abstract: In this paper, we use the general dynamic factor model with infinite-dimensional factor space to develop a new procedure to estimate and forecast the conditional covariance matrix in high-dimensional data. The performance of our procedure is evaluated via Monte Carlo experiments and the results show good finite sample properties. The new procedure is used to construct minimum variance portfolios in a high-dimensional real dataset. The results reveal a better out-of-sample portfolio performance when compared with alternative procedures.