*
* Example 10.1.1 from pp 335-338
*
open data ls2.tsm
data(format=free,org=columns) 1 150 lead sales
*
* Difference and pull the means out of the two variables
*
diff(diffs=1,center) lead / x1
diff(diffs=1,center) sales / x2
*
* Fix an MA(1) model to the transformed "lead" series
*
boxjenk(ma=1,maxl) x1 / zhat
*
* This use of BOXJENK will filter the series x2 by the same MA(1) filter
* estimated for x1. METHOD=EVAL just does a single evaluation at the initial
* values provided in the %beta (estimated coefficients) vector. The residuals
* are the result of the filter.
*
boxjenk(ma=1,maxl,initial=%beta,method=eval) x2 / yhat
*
@crosscorr zhat yhat
*
* This does the direct estimate of the coefficients in the transfer function.
*
boxjenk(inputs=1,maxl) x2
# x1 0 1 3
@regcorrs
*
boxjenk(inputs=1,ma=1,maxl) x2
# x1 0 1 3