PanelVARResult.normality_test#

PanelVARResult.normality_test(variable)[source]#

Jarque-Bera test on one equation’s residuals.

Parameters:

variable (str) – Name of the equation to test.

Return type:

_WaldTest

Returns:

The _WaldTest with two degrees of freedom.

Raises:

SpecificationError – If variable is not in names.

Note

Non-normality does not invalidate the coefficient estimates, which are consistent regardless. It bites on anything that reads the Gaussian likelihood at face value – the information criteria, and any bootstrap that resamples as though the innovations were symmetric.