*
* Example 4.3 from page 51-52
*
open data tablef4-1[1].txt
data(format=prn,org=columns,skip=36) 1 753 lfp whrs kl6 k618 wa we ww rpwg hhrs ha he hw faminc mtr wmed wfed un cit
*
* Create the variables needed. Earnings are whrs*ww, and we need the log of that.
* We need the square of wa. And we need a dummy for the observations with a
* non-zero number of kids. We can construct that by taking kl6+k18 (number of
* kids) and using the relational operator >. (kl6+k18)>0 is 1 if the number of
* kids is >0 and 0 if the number of kids is equal to zero
*
set logearn = log(whrs*ww)
set agesq   = wa**2
set kids    = (kl6+k618)>0
*
* Run the regression over the sample which has lfp equal to 1
*
linreg(smpl=lfp,vcv) logearn
# constant wa agesq we kids
*
* This shows the effect of the age variables. We create an "x" variable which is
* just equal to the age for 18-70. (We don't want to use the actual data, because
* they aren't sorted by age). Then the age effect variable is created by taking
* the second coefficient from the regression (%beta(2)) times the test value for
* the age, and the third coefficient times the age squared. The SCATTER
* instruction with STYLE=LINE draws the function.
*
set testage 18 70   = t
set ageeffect 18 70 = %beta(2)*testage+%beta(3)*testage**2
scatter(style=line,footer="Age Effect on log(earnings)")
# testage ageeffect