Estimation of VARMA models

Anindya Roy, Mathematics and Statistics
Peter Linton, Mathematics and Statistics
Tucker McElroy, US Census Bureau

Maximum likelihood estimation of the VARMA model under causality constraints. The standard likelihood evaluation via innovations algorithm is fast, but for not fast enough for iterative calulations. Thus, for estimation it is imperative that likelihood evaluations are done fast enough. The objective will be to figure out a way to compute the likelihood within any iterative scheme that would permit estimation in a reasonable time frame.