* Distributed Lag Models
* Reference: Chapter 15 of
*    R. Carter Hill, William E. Griffiths and George G. Judge, 
*    Undergraduate Econometrics, Second Edition, Wiley. 

SAMPLE 1 88
READ (TAB15-1.shd) / NAMES

* OLS estimation of an unrestricted lag model - Table 15.2, p. 322
* The ANOVA option reports model selection statistics.
OLS Y X(0.8) / ANOVA
* Save the values of the AIC and the SC
* Note that the SHAZAM calculation is different from the formula
* given on page 326 of the text. 
* (See page 16 of the SHAZAM Version 9 User Manual).
GEN1 AIC1=$LAIC 
GEN1 SC1=$LSC

* Calculate correlations among the current and lagged values of the X.
DO #=1,8
 GENR X#=LAG(X,#)
ENDO
* Adjust the sample period
SAMPLE 9 88
STAT X X1-X8 / PCOR

* Estimate an Almon polynomial distributed lag model.
* Table 15.4, p. 325.
SAMPLE 1 88
OLS Y X(0.8,2) / ANOVA
GEN1 AIC2=$LAIC 
GEN1 SC2=$LSC

* Compare the AIC and the SC measures for the two models.
* Lower values are considered desirable.
PRINT AIC1 AIC2 
PRINT SC1 SC2
* The Almon polynomial distributed lag model has lower values
* for both the AIC and SC compared to the unrestricted model.
STOP