This paper considers estimation and inference in panel vector autoregressions (PVARs) with fixed effects when the time dimension of the panel is finite, and the cross-sectional dimension is large. A Maximum Likelihood (ML) estimator based on a transformed likelihood function is proposed and...