*
* HANSEN.RPF
* Example of Hansen J-test for specification in IV models
*
* RATS User's Guide, example from Section 3.9.
*
cal(a) 1920
allocate 1941:1
open data klein.prn
data(org=columns,format=prn) / cons $
profit privwage invst klagged1 prod govtwage govtexp taxes
set wagebill = privwage+govtwage
set trend = t-1931:1
set capital = klagged1+invst
smpl 1921:1 1941:1
instruments constant trend govtwage govtexp taxes $
profit{1} capital{1} prod{1}
*
* Simple 2SLS. The test statistic will be included in the output.
*
linreg(inst) cons / resids
# constant profit{0 1} wagebill
*
* Doing the test by regressing residuals on the full instrument set.
*
linreg resids
# constant trend govtwage govtexp taxes $
profit{1} capital{1} prod{1}
cdf chisqr %trsq 4
*
* The two test statistics are slightly different because the J-statistic
* produced by the first LINREG uses a degrees of freedom corrected
* estimate of the residual variance. To get the identical result, you
* need to multiply %trsq by (%nobs-4.0)/%nobs
*
* Test with the weight matrix adjusted for serial correlation and
* heteroscedasticity. The test statistic here can't easily be computed
* except by using some set of options for LINREG
*
linreg(inst,optimal,lags=1,lwindow=newey) cons
# constant profit{0 1} wagebill
*
* This will produce the identical results to the included J-test
*
cdf chisq %uzwzu 4