Hello.
I'm currently using ivreghdfe command.
Below is the equation that I set for the regression.
(Here, x1 & x2 or c1 & c2 are variables with similar concepts but applied to different subjects.
For instance, x1 refers to the income equality index among men in a region. x2 refers to the income inequality index among women in a region.)
Since x1, x2 are endogenous, I use z1, z2 as the instrument variable for each x1 and x2.
So I used the below command.
I got the first-stage results as this.
(results for x1)
(results for x2)
F statistic in the first regression (15392.21) is too high and weird.
When I drop the c1#x1 term, the f statistic of excluded instruments in the first regression is 50.67.
(And when I drop the c2#x2 term, the f statistic in the second regression is 8.71.)
So what is the problem in this extremely high f statistic?
If this problem happens due to the c1#z1 interaction term, how can I fix it?
(Both x1 and x1#c1 interaction term are necessary in my equation.. so neither of them can be excluded.)
I'm currently using ivreghdfe command.
Below is the equation that I set for the regression.
Code:
y = a + b1x1 + b2c1 + b3x1*c1 + u
Code:
y = a + b1x2 + b2c2 + b3x2*c2 + u
For instance, x1 refers to the income equality index among men in a region. x2 refers to the income inequality index among women in a region.)
Since x1, x2 are endogenous, I use z1, z2 as the instrument variable for each x1 and x2.
So I used the below command.
Code:
ivreghdfe y c.c1 (c.x1 c.x1#c.c1 = c.z1 c.z1#c.c1), absorb(region year) cluster(region) first savefirst
Code:
ivreghdfe y c.c2 (c.x2 c.x2#c.c2 = c.z2 c.z2#c.c2), absorb(region year) cluster(region) first savefirst
(results for x1)
(results for x2)
F statistic in the first regression (15392.21) is too high and weird.
When I drop the c1#x1 term, the f statistic of excluded instruments in the first regression is 50.67.
(And when I drop the c2#x2 term, the f statistic in the second regression is 8.71.)
So what is the problem in this extremely high f statistic?
If this problem happens due to the c1#z1 interaction term, how can I fix it?
(Both x1 and x1#c1 interaction term are necessary in my equation.. so neither of them can be excluded.)
