Creating a positive and negative assymetric variables in panel data

[Hamid muili]

Dear Scholars, I am trying to estimate a non linear Panel VAR using oil price as the independent variable. I intend to create a positive and negative series of oil price representing an increase and a decrease in oil Price.. is there any direct command or package in stata that does this?

[Nick Cox]

Does this help?

. webuse grunfeld, clear

. xtset company year

Panel variable: company (strongly balanced)
 Time variable: year, 1935 to 1954
         Delta: 1 year

. gen change = D.invest
(10 missing values generated)

. l year invest change in 1/20

     +---------------------------+
     | year   invest      change |
     |---------------------------|
  1. | 1935    317.6           . |
  2. | 1936    391.8    74.19998 |
  3. | 1937    410.6    18.80002 |
  4. | 1938    257.7      -152.9 |
  5. | 1939    330.8    73.09998 |
     |---------------------------|
  6. | 1940    461.2       130.4 |
  7. | 1941      512    50.79999 |
  8. | 1942      448         -64 |
  9. | 1943    499.6    51.60001 |
 10. | 1944    547.5    47.89999 |
     |---------------------------|
 11. | 1945    561.2    13.70001 |
 12. | 1946    688.1       126.9 |
 13. | 1947    568.9      -119.2 |
 14. | 1948    529.2   -39.70001 |
 15. | 1949    555.1    25.89996 |
     |---------------------------|
 16. | 1950    642.9    87.80005 |
 17. | 1951    755.9         113 |
 18. | 1952    891.2       135.3 |
 19. | 1953   1304.4       413.2 |
 20. | 1954   1486.7    182.2999 |
     +---------------------------+

[Hamid muili]

Thanks a lot Prof Nick.. What I intend doing is to create 2variables
For instance, modeling oil price shock on inflation.

Inflation = oilprice_pos + oilprice_neg
Where oilprice_pos is the increase in oil price and oil price neg is the fall in oil price.. The two variables will be generated from a single oil price index..

[Nick Cox]

Perhaps you need something like

gen poschange = max(0, change) 

gen negchange = min(0, change)