Simple data format (n=30,000):
INDIVIDUAL ID | PROVINCE | YEAR |
0001 | British Columbia | 1999 |
0002 | British Columbia | 1999 |
0003 | British Columbia | 2003 |
0004 | British Columbia | 2003 |
0005 | British Columbia | 2003 |
0006 | Alberta | 1999 |
0007 | Alberta | 1999 |
0008 | Alberta | 2003 |
The two options I'm considering are:
1. Entering dummy variables for i.province and i.year into an ivreg2 model. I'm not sure this is a valid approach because I haven't seen it used in 2SLS before and I'm not using panel data. The model would be:
Code:
ivreg2 y (x=z) control_1 control_2 i.year i.province
Code:
xtset province year
Code:
xtset province xtivreg y (x=z) control_1 control_2 i.year, fe
0 Response to Trouble adding province-year fixed effects to a 2SLS model using ivreg2
Post a Comment