Panel Data Wald Test
Posted: Mon Nov 19, 2012 11:53 am
Dear all,
I am currently estimating regressions based on a panel data set.
I do have 7 dependent variables (y1-y7) and 6 independent (x1-x6). I am interested in the outcome of a Wald test based on these regressions.
So, the first step is estimating the regressions, which also leads to the first problem:
For panel regressions, one can choose between fixed and random effects. To determine which effect is appropriate, the Hausman test is applied. In case this is rejected, random effects are appropriate. After having chosen either fixed or random effects, I also want to assign weights (GLS weights : cross-section weights, coef cov. method: white cross section).
Next, I want to apply a Wald test with c(1)=0, c(2)=1.
I tried a bit of programming, but do not know how to include the Hausman test. Furthermore, I also want the outcome of the Waldtest to be displayed in a matrix so that I do not have to look up the values by hand. I tried to do it the following way, still excluding the Hausman test as I do not know how to include it:
matrix(42,2) coefficients=na
!i=0
for !y= 1 to 7
for !x = 1 to 6
!i = !i +1
equation reg!y!x.ls(CX=R,COV=CXWHITE) y!y c x!x
freeze(tab!y!x) reg!y!x.wald c(1)=0
coefficients(!i,1) = tab!y!x.@f
next
next
I would highly appreciate your help on this !
I am currently estimating regressions based on a panel data set.
I do have 7 dependent variables (y1-y7) and 6 independent (x1-x6). I am interested in the outcome of a Wald test based on these regressions.
So, the first step is estimating the regressions, which also leads to the first problem:
For panel regressions, one can choose between fixed and random effects. To determine which effect is appropriate, the Hausman test is applied. In case this is rejected, random effects are appropriate. After having chosen either fixed or random effects, I also want to assign weights (GLS weights : cross-section weights, coef cov. method: white cross section).
Next, I want to apply a Wald test with c(1)=0, c(2)=1.
I tried a bit of programming, but do not know how to include the Hausman test. Furthermore, I also want the outcome of the Waldtest to be displayed in a matrix so that I do not have to look up the values by hand. I tried to do it the following way, still excluding the Hausman test as I do not know how to include it:
matrix(42,2) coefficients=na
!i=0
for !y= 1 to 7
for !x = 1 to 6
!i = !i +1
equation reg!y!x.ls(CX=R,COV=CXWHITE) y!y c x!x
freeze(tab!y!x) reg!y!x.wald c(1)=0
coefficients(!i,1) = tab!y!x.@f
next
next
I would highly appreciate your help on this !