* PS10.3, using DATA10-2, for the Breusch-Godfrey test for Example 10.4
*
SAMPLE 1 744
*
READ(data10-2) DAYHOUR LOAD TEMP
*
* The LAG function on the GENR command is used to generate a new variable
* called LOAD1 which is the variable LOAD lagged back one time period.
*
GENR LOAD1=LAG(LOAD)
*
* The sample period now begins at observation 2 rather than 1 since an 
* observation is lost when the variable LOAD is lagged back one time
* period.
*
SAMPLE 2 744
*
* The RESID= option on the OLS command saves the regression residuals in 
* the vector called UHAT.
*
OLS LOAD LOAD1 TEMP / RESID=UHAT
*
* The LAG function on the GENR command is first used to generate the lagged
* regression residual back one time period {UHAT1=LAG(UHAT)}, then back two 
* time periods {UHAT2=LAG(UHAT,2)}, then back three time periods {UHAT3=
* LAG(UHAT,3)}, then back four time periods {UHAT4=LAG(UHAT,4)}, then back
* five time periods {UHAT5=LAG(UHAT,5)}, and finally back six time periods
* {UHAT6=LAG(UHAT,6)}.
*
GENR UHAT1=LAG(UHAT)
GENR UHAT2=LAG(UHAT,2)
GENR UHAT3=LAG(UHAT,3)
GENR UHAT4=LAG(UHAT,4)
GENR UHAT5=LAG(UHAT,5)
GENR UHAT6=LAG(UHAT,6)
* 
* The sample period now begins at observation 9 rather than 2 since 6
* observations are lost when the regression residuals, UHAT, were lagged
* back 6 time periods.
*
SAMPLE 8 744
OLS UHAT LOAD1 TEMP UHAT1 UHAT2 UHAT3 UHAT4 UHAT5 UHAT6
*
* The temporary variables for the regression R-Square, $R, and the number
* of observations in the regression, $N, are automatically stored after
* an OLS regression.  The GEN1 command is used with the temporary variables
* $N and $R2 to calculate the LM test statistic.  The PRINT command prints
* this statistic.
*
GEN1 LM=$N*$R2
DISTRIB LM / TYPE=CHI DF=6
*
DELETE/ALL
STOP