The terms to be included in the dynamic shock model relating two
adjacent market levels are determined by examining the Haugh-Pierce
residual cross-correlations for significance. Each term is estimated by
solving for the bivariate regression coefficient (X^) which is a
function of the standard deviation for each residual series and the
corresponding residual cross-correlation estimate at lag k. Once the
dynamic shock model is estimated, the respective ARIMA filter models are
substituted into the expression. The resulting expression is then
solved in terms of the "causal" (lagged) price. The resulting equation
is referred to as the impulse response function. The following
discussion details this procedure for each size class using monthly and
quarterly data.
Monthly Data
31-40 Size Class: Wt f(Pt)
The implied dynamic shock model for the wholesale (Wfc)/ex-vessel
(Pt) price relationship is given as
(1) wt (Aq X1B)e(; + (B) is a polynomial in the lag operator B (which is unique to the white
noise disturbance associated with the dynamic shock model), and at is
178