*
* Example 3.2.8 from pages 96-98
*
open data oshorts.dat
data(format=free,org=columns) 1 57 overs
*
graph(footer="Figure 3-5 Time series of the overshorts")
# overs
*
* The two standard error bands in the BJIDENT output are computed using the
* extension of the formula used in Figure 3-6 to orders higher than one; that is,
* the limit shown from lag 1 is just 2/sqrt(n), but for lag 2, it's 2
* sqrt(1+2r(1)**2)/sqrt(n) (which is the limit shown in figure 3-6) and 2
* sqrt(1+2*(r(1)**2+r(2)**2))/sqrt(n) for lag 3, etc.
*
@bjident(number=20) overs
*
* You can get the autocovariances by using the instruction CORRELATE with the
* option COVARIANCE. METHOD=YULE gives the same method of computing these as
* described in the book.
*
corr(covariance,number=4,method=yule) overs