*
* Example from page 180, section 6.6
*
cal(q) 1993
open data table6-3.prn
data(format=prn,org=columns) 1993:1 1998:3
*
* y1 = expenditures on services,
* y2 = expenditures on durables
* y3 = expenditures on non-durables
*
set logy1 = log(y1)
set trend = t
linreg logy1
# constant trend
*
* Graph in log scale
*
prj fitted
scatter
# trend fitted
*
* Graph in levels. Note that, over a five year period with just about a 3% growth
* rate, the exponential curve isn't very apparent. 1.03**5 is only slightly
* different from 1+.03*5.
*
set expfit = exp(fitted)
scatter
# trend expfit
*
* If you want the slope coefficient to be more easily interpreted as annual
* growth, you can change the trend variable so it increases by 1 every year,
* rather than every quarter, that is:
*
set trend = .25*t
linreg logy1
# constant trend
*
* And if you want it to be a percentage, rather than a fraction, you need the
* trend series to be .01 x what we just did, which will cause the coefficient to
* go UP by a factor of 100.
*
set trend = .0025*t
linreg logy1
# constant trend