MinnesotaPrior.cross_equation

MinnesotaPrior.cross_equation: float

Relative tightness of cross-variable coefficients \(\lambda_2\).

A cross lag’s standard deviation is the own lag’s times this, so one shrinks cross and own lags alike and smaller values encode more of the belief that a series is explained by its own past. Any value other than one breaks the Kronecker factorization that the conjugate BVAR requires, which refuses the prior at fit; the point-estimate and Gibbs paths accept it. Must be positive.