version 6
set more off
capture log close
clear
set memory 50m
log using c:\documents\ellison\demand.log,replace
use c:\documents\ellison\gasoline.dta
**********************
*Data set at a glance
**********************
gen tmp=year1+year2+year3+year4+year5+year6
assert tmp==1
drop tmp
gen tmp=da1+da2+da3+da4+da5+da6+da7+da8+da9
tab tmp
drop tmp
gen tmp=da1+da2+da3+da4+da5+da6+da7+da8+da9+db1+db2+db3+db4+db5+db6+db7+db8
tab tmp
drop tmp
sum lq lp ly da1-da9 db1-db8 year1-year6
*The data set contains 18,110 observations on the consumption
*of gasoline, income and gasoline price. Data are annual.
*Dummy variables identify six years, and 18 geographic locations
*(9 according to classification "a", 8 to "b", 1 base location).
******************
*Demand estimates
******************
reg lq lp ly year1-year6,robust noconstant
reg lq lp ly year1-year6 da1-da9 db1-db8,robust noconstant
gen q=exp(lq)
gen p=exp(lp)
gen y=exp(ly)
reg q p y year1-year6,robust noconstant
reg q p y year1-year6 da1-da9 db1-db8,robust noconstant
*******************
*Kernel regression
*******************
net from http://fmwww.bc.edu/RePEC/bocode/k
net describe kernreg1
net install kernreg1
net get kernreg1
kernreg1 lq lp,k(6) np(100)
kernreg1 q p,k(6) np(100)
*The option k(6) specifies that the weight function (kernel)
*is gaussian. The option np(100) specifies that the number of
*equally spaced points (grid) is 100. For further information
*about kernreg, see Stata Technical Bulletin N.30.
log close