sic

sic

aic(model::Union{VAR,SVAR}) -> Real
aic(Sigma_u::Matrix{<:Number}, num_coeffs::Int, T::Int) -> Real

bic(model::Union{VAR,SVAR}) -> Real
bic(Sigma_u::Matrix{<:Number}, num_coeffs::Int, T::Int) -> Real

sic(model::Union{VAR,SVAR}) -> Real
sic(Sigma_u::Matrix{<:Number}, num_coeffs::Int, T::Int) -> Real

hqc(model::Union{VAR,SVAR}) -> Real
hqc(Sigma_u::Matrix{<:Number}, num_coeffs::Int, T::Int) -> Real

Computes information criteria for model selection in VARs.

Given a fitted VAR model or directly the covariance matrix Sigma_u, the number of estimated coefficients, and sample size T, each function returns the respective criterion value:

• AIC (Akaike Information Criterion):

\[ \text{AIC} = \log\det(\Sigma_u) + \frac{2k}{T} \]

• SIC/BIC (Schwarz/Bayesian Information Criterion):

\[ \text{SIC} = \log\det(\Sigma_u) + \frac{\log(T) k}{T} \]

• HQC (Hannan-Quinn Criterion):

\[ \text{HQC} = \log\det(\Sigma_u) + \frac{2 \log(\log T) k}{T} \]

Here, \(k\) is the number of estimated parameters and \(T\) is the effective sample size.

Arguments

• model::VAR: A fitted VAR model
• Sigma_u::Matrix{<:Number}: Residual covariance matrix
• num_coeffs::Int: Number of estimated coefficients
• T::Int: Number of observations used in estimation