Instrumental variable with between effects

[mona bakrrr]

Dear All,
I hope you are all fine.
My data set is panel data and I am doing my analysis using between-effects model. I have two endogenous variables and 3 instruments. This in addition to 10 other control variables.

I would like to check the relevance and validity of these 3 instruments but I am not able to get estimates for Cragg-Donald Wald test statistic and Sargan-Hansen test.

It would be much appreciated if you could advise me how to get these estimates.
I can get them with xtivreg2 fixed effect or ivreg2 but cannot get these when using between effects.

Thank you very much
Mon

[Phil Bromiley]

You'll increase your chances of a helpful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

If you look at the documentation for xtreg, Remarks and Examples. it seems like you can manually calculate the panel means and then estimate ignoring the panel effects. Then ivreg2 might work.

More is required to justify the between estimator of (2), but the conditioning on the sample is not
assumed because i +i is treated as an error term. Newly required is that we assume that i and xi
are uncorrelated. This follows from the assumptions of the OLS estimator but is also transparent: were
i and xi correlated, the estimator could not determine how much of the change in yi, associated
with an increase in xi, to assign to versus how much to attribute to the unknown correlation. (This,
of course, suggests the use of an instrumental-variable estimator, zi, which

[Phil Bromiley]

Let me note that my suggestion is just what seems reasonable to me. I could be wrong on the statistics.

[Carlo Lazzaro]

Mona:
I would also take a look at the help file that comes with the community-contributed programme -xtoverid- (just type -search xtoverid- from within Stata to access it).

[mona bakrrr]

You'll increase your chances of a helpful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

If you look at the documentation for xtreg, Remarks and Examples. it seems like you can manually calculate the panel means and then estimate ignoring the panel effects. Then ivreg2 might work.

More is required to justify the between estimator of (2), but the conditioning on the sample is not
assumed because i +i is treated as an error term. Newly required is that we assume that i and xi
are uncorrelated. This follows from the assumptions of the OLS estimator but is also transparent: were
i and xi correlated, the estimator could not determine how much of the change in yi, associated
with an increase in xi, to assign to versus how much to attribute to the unknown correlation. (This,
of course, suggests the use of an instrumental-variable estimator, zi, which

Thank you very much

[mona bakrrr]

Mona:
I would also take a look at the help file that comes with the community-contributed programme -xtoverid- (just type -search xtoverid- from within Stata to access it).

Thank you very much.