A0759
Title: Comparative analysis of multivariate mixed causal and non-causal process representations
Authors: Francesco Giancaterini - Maastricht University (Netherlands) [presenting]
Gianluca Cubadda - University of Rome Tor Vergata (Italy)
Alain Hecq - Maastricht University (Netherlands)
Abstract: Two representations of multivariate mixed causal and non-causal processes are examined and shown that certain data-generating processes necessitate only one particular representation. To identify such cases, new theoretical conditions are introduced. Results from Monte Carlo experiments underscore the importance of selecting the correct representation. Specifically, they illustrate that employing an inappropriate specification can result in inaccurate identification and, consequently, inaccurate estimation of the underlying process. Thus, the paper emphasizes the significance of carefully choosing the appropriate representation. Lastly, both specifications are applied to a bivariate process of cryptocurrency prices, revealing discrepancies in identification and estimation based on the selected representation.
