statsmodels.tsa.vector_ar.var_model.FEVD

class statsmodels.tsa.vector_ar.var_model.FEVD(model, P=None, periods=None)

Compute and plot Forecast error variance decomposition and asymptotic standard errors

Parameters:

modelVARResults

The VAR results used to compute the FEVD.

Pndarray, optional

Matrix such that Sigma_u = PP’, used to orthogonalize the impulse responses. Defaults to the Cholesky decomposition of Sigma_u.

periodsint, optional

Number of periods to compute the FEVD for.

Attributes:

decompndarray

The forecast error variance decomposition, with shape (neqs, periods, neqs). decomp[j, h - 1, k] is the share of the h-step ahead forecast error variance of variable j that is due to the orthogonalized shock to variable k, so the first step ahead is stored at position 0. The shares sum to one over the last axis.

neqsint

Number of endogenous variables.

nameslist of str

Names of the endogenous variables.

Methods

cov()	Compute asymptotic standard errors
plot([periods, figsize])	Plot graphical display of FEVD
summary()