February 28, 2018
MIDAS-PRO-LASSO: Mixed Frequency Data Regression Models with Parameter Profiling and LASSO
Jonas Striaukas, CORE UCLouvain
In this article we propose an estimation method for high dimensional mixed frequency data sampling (MIDAS) regression model and study its properties. An extensive Monte Carlo study shows that our proposed method outperforms MIDASSO approach which is based on unrestricted MIDAS specification in terms model selection properties. The empirical study concentrates on forecasting the Euro area retail interest rates using wide range of daily financial variables. The results show evidence of predictability of retail interest rates using mixed frequency data set.
Please fill the doodle by Tuesday February 21 if you plan to attend and mention whether you would like a sandwich.