*
* Numerical example from section 8.4
* pages 265-269
*
open data auto1.asc
calendar(q) 1959:1
data(format=prn,org=columns) 1959:1 1992:1 gcng gcngq gdc gydq pop16 x2 x3 y
*
graph(vlabel="Log expenditure")
# y
graph(vlabel="Log gasoline price")
# x2
graph(vlabel="Log disposable personal income")
# x3
*
@dfunit y
@dfunit x2
@dfunit x3
*
* Engle-Granger "spurious regression"
*
linreg y * 1990:4 resids
# x2 x3 constant
graph(footer="Figure 8.2 Residuals from Engle-Granger Cointegrating Regression")
# resids
*
* ARDL model
*
linreg y * 1990:4
# x2{0 to 5} x3{0 to 5} y{1 to 5} constant
*
* The sums of the coefficients can be analyzed using SUMMARIZE. The supplementary
* card lists the variables and lags to be summed. The included test statistic is
* for a hypothesized value of 0, which is what we want for the x2 and x3 lags, but
* not for the y lags.
*
summarize
# y{1 to 5}
cdf(title="Test of A(1)=1") ttest %sumlc/sqrt(%varlc) %ndf
summarize
# x2{0 to 5}
summarize
# x3{0 to 5}