Variance decomposition of forecast errors

From Wikipedia, the free encyclopedia
  (Redirected from Variance decomposition)
Jump to: navigation, search

In econometrics and other applications of multivariate time series analysis, a variance decomposition or forecast error variance decomposition (FEVD) is used to aid in the interpretation of a vector autoregression (VAR) model once it has been fitted.[1] The variance decomposition indicates the amount of information each variable contributes to the other variables in the autoregression. It determines how much of the forecast error variance of each of the variables can be explained by exogenous shocks to the other variables.

Calculating the forecast error variance[edit]

For the VAR (p) of form

y_t=\nu +A_1y_{t-1}+\dots+A_p y_{t-p}+u_t

This can be changed to a VAR(1) structure by writing it in companion form (see general matrix notation of a VAR(p))

Y_t=V+A Y_{t-1}+U_t

A_1 & A_2 & \dots & A_{p-1} & A_p \\
\mathbf{I}_k & 0 & \dots & 0 & 0 \\
0 & \mathbf{I}_k &  & 0 & 0 \\
\vdots & & \ddots & \vdots & \vdots \\
0 & 0 & \dots & \mathbf{I}_k & 0 \\
y_1 \\ \vdots \\ y_p \end{bmatrix}
, V=\begin{bmatrix}
\nu \\ 0 \\ \vdots \\ 0 \end{bmatrix}
u_t \\ 0 \\ \vdots \\ 0 \end{bmatrix}

where y_t, \nu and u are k dimensional column vectors, A is kp by kp dimensional matrix and Y, V and U are kp dimensional column vectors.

The mean squared error of the h-step forecast of variable j is


and where

  •  e_j is the jth column of  I_K and the subscript jj refers to that element of the matrix
  •  \Theta_i=\Phi_i P , where P is a lower triangular matrix obtained by a Cholesky decomposition of  \Sigma_u such that  \Sigma_u = PP', where  \Sigma_u is the covariance matrix of the errors u_t
  • \Phi_i=J A^i J', where 
\mathbf{I}_k &0  & \dots & 0\end{bmatrix} ,
so that J is a k by kp dimensional matrix.

The amount of forecast error variance of variable j accounted for by exogenous shocks to variable k is given by \omega_{jk,h} ,

\omega_{jk,h}=\sum_{i=0}^{h-1}(e_j'\Theta_ie_k)^2/MSE[y_{j,t}(h)] .


  1. ^ Lütkepohl, H. (2007) New Introduction to Multiple Time Series Analysis, Springer. p. 63.