*
* Application: estimation d'une equation d'investissement (voir Greene) / TD1 Econométrie / M1 EFI 2011-2012

cal 1968
*
open data bdd1.txt
data(format=prn,org=columns,verbose) 1968:1 1982:1 year gnp invest cpi interest

*
* gnp is in billions and the cpi is based on 100. To get real GNP in trillions we
* need to divide by cpi, then divide by 1000 (to convert billions to trillions)
* and multiply by 100, to correct the cpi base. Combining those means we multiply
* by .1.
*
set rgnp = gnp/cpi*(.1)
set rinv = invest/cpi*(.1)
set trend = t
set constant = 1

graph 2
# rinv
# rgnp

stat(print) rinv
stat(print) rgnp

print(picture="*.###") / rinv constant trend rgnp interest

*
* The cross products of the variables can be computed using the instruction
* CMOMENT. Here, we list the three explanatory variables first, followed by the
* dependent variable. As you read across the first three rows in the printed
* matrix, you'll get the sums in the 3x3 system of equations on page 21
*
cmom(print)
# constant rgnp rinv


*
* You can get the cross product matrix of deviations from the mean, by using the
* option "CENTER" on CMOM. The CONSTANT can then be omitted. Note the advice on
* the bottom of Table 3.1 - these aren't going to quite match the numbers that go
* into the calculation of b2 and b3 because those are based upon rounded numbers.
*
cmom(center,print)
# rgnp rinv

*
* The elements of this matrix can be obtained by %cmom(i,j) where i is the row
* and j the column. For instance, the byg at the bottom of page 22 can be computed
* by
*
disp "Byx =" %cmom(1,2)/%cmom(1,1)