Question

he following six questions rely on the information below.

Suppose you have a dataset with the following variables. Each firm (distinguishible by fcode) appears 3 times in the dataset; one for each year:

1. year                1987, 1988, or 1989
2. fcode               firm code number
3. employ              # employees at plant
4. sales               annual sales, $
5. avgsal              average employee salary
6. scrap               scrap rate (per 100 items)
7. union               =1 if unionized
8. grant              =1 if firm received funding for training
9. hrsemp        Average hours spent training each employee in the year.
10. lscrap            log(scrap)
11. lsales              log(sales)
12. lrework            log(rework)
13. lhrsemp            log(hrsemp)
14. lavgsal            log(avgsal)

The Dependent Variable is lharsemp.

You use the command xtset fcode year to tell STATA this is a longitudinal dataset. Then you run the follwoing regression of lhrsemp on several explanatory variables. I HAVE SUPPRESSED CERTAIN VALUES OF THE OUTPUT INTENTIONALLY. The output is:

. regress lhrsemp lavgsal lscrap lsales union grant

------------------------------------------------------------------------------

     lhrsemp |      Coef.      Std. Err.     

-------------+----------------------------------------------------------------

     lavgsal |   .2898908   .3507217    

      lscrap | -.0933434   .0793241  

      lsales |   .1914808    .120671    

       union | -.5693663   .2728087   

       grant |   1.796736   .2646883

       _cons | -4.075021   3.844255

------------------------------------------------------------------------------.

Doubling sales will lead to a change in predicted hrsemp by a factor of approximately _____.

Answer 1

The regression equation holding all other variables constant except lsales is,

log(hrsemp_old) = 0.1914808 log(Sales) + constant

Doubling sales,

log(hrsemp_new) = 0.1914808 log(2 * Sales) + constant

log(hrsemp_new) =  0.1914808 log(2) +  log(Sales) + constant

log(hrsemp_new) = 0.1327244 +  log(hrsemp_old)

log(hrsemp_new) = log(exp(0.1327244)) +  log(hrsemp_old)

log(hrsemp_new) = log(1.141935) +  log(hrsemp_old)

log(hrsemp_new) = log(1.141935 * hrsemp_old)

hrsemp_new = 1.141935 * hrsemp_old

Doubling sales will lead to a change in predicted hrsemp by a factor of approximately 1.141935