*
* Sample ACF and Partial ACF from page 112
*
open data haversmp.rat
cal 1947 1 4
all 1988:4
data(format=rats) / gnph
set pgnp = log(gnph/gnph{1})
*
* Using the BJIDENT procedure, which graphs the autocorrelations
* and partial autocorrelations on a single graph. Note that the
* 2 standard error band isn't flat, but increases slightly with
* the lag. This is because BJIDENT computes this (which is for
* the autocorrelations) using the formula on 4.8.8 for the variance
* of rho(j) given rho(1),...,rho(j-1).
*
@bjident(number=20) pgnp
*
* Doing graphs like those in the text using direct instructions.
*
correlate(number=20,partials=part) pgnp / corr
spgraph(header='Figure 4.2 Sample Autocorrelations and Partial Autocorrelations for US GNP',$
   vfields=2)
graph(number=0,max=1.0,min=-1.0,style=bar,hlabel='Sample Autocorrelations')
# corr
graph(number=0,max=1.0,min=-1.0,style=bar,hlabel='Sample Partial Autocorrelations')
# corr
spgraph(done)