Hey,
I have a sample of 10,346 observations that are representative for the whole of Germany. I am also given a variable whose label translates to "extrapolation/projection factor" and ranges between 130.8 and 17,196. I have used it in the past to extrapolate quantities consumed in the whole economy simply by multiplying the respective variables with the projection variable and the numbers matched with official statistics.
Now I would like to quality adjust prices by regressing prices on proxies for quality preferences. I then obtain adjusted prices by adding the constant and the residuals, but that is not really important. I am wondering whether in order not to distort results, I need to apply the [pweight=extrapolation/projection factor] - here called EF31. I tried regressions with and without the [pweight=EF31] and results do slightly change.
The example code for the price of beef looks as follows:
Best,
Julian
I have a sample of 10,346 observations that are representative for the whole of Germany. I am also given a variable whose label translates to "extrapolation/projection factor" and ranges between 130.8 and 17,196. I have used it in the past to extrapolate quantities consumed in the whole economy simply by multiplying the respective variables with the projection variable and the numbers matched with official statistics.
Now I would like to quality adjust prices by regressing prices on proxies for quality preferences. I then obtain adjusted prices by adding the constant and the residuals, but that is not really important. I am wondering whether in order not to distort results, I need to apply the [pweight=extrapolation/projection factor] - here called EF31. I tried regressions with and without the [pweight=EF31] and results do slightly change.
The example code for the price of beef looks as follows:
Code:
regress p_beef net_income_month hhmembers West male_single agehoh academic [pweight=EF31] gen constant_beef = _b[_cons] predict res, resid gen residual_beef = res gen p_beef_star = constant_beef + residual_beef
Julian
