SUREG and restrictions in a three dimensional panel dataset.

[Maite Bou]

I have a three dimensional panel data (combining countries, sectors and years) and I am using a SUREG command in Stata 13.1, in order to estimate 4 different ecuations (in wchich my dependent variable are employment for every each of the four sectors). However, I need to adjust the model in order to add two particular restrictions:

• Constant Returs of Scale
• Symetric estimated coefficients (for sectoral employment).

I have been trying to figure out a way to test those two hypotesis with a SUREG model but I just couldn't find a way, Does anyone by any chance know which comands should I use?
As I am not an advanced user on stata, I also have a smaller problem, in my regression I add a dummy independent variable i.country but I have some colliniarity issues. In order to fix that I should do k-1, however, when it comes to stata acccoding, I can't figure out a way to express it.
Thank you in advance.
Maite

[Carlo Lazzaro]

Maite:
- as far your first question is concerned, your chance of getting helpful replies is conditional on posting exactly what you typed and what Stata gave you back (as per FAQ #12).
- as far as you second question is concerned, with -i.country- Stata avoids the dummy trap automatically. Hence, the collinearity problem you report is probably due to the fact that you already -xtset- your data including -country-. As above, it's hard (for me, at least) to comment further on these issues without knowing more details about your Stata session.

[cengiz hokka]

Hello dear Carlo Lazzaro ,
I am researching the effect of corporate governance practices on the profitability of banks. I have 12 years of data, 146 banks and 17 countries (7 developed countries and 10 developing countries). Can a multidimensional panel be applied to such a data structure? and multidimensional, will we consider banks and countries intertwined or analyze them as non-nested models? Or should I analyze such a data set as classical two-dimensional?
I would be grateful if you can reply to my post

[cengiz hokka]

Hello dear Carlo Lazzaro ,
I am researching the effect of corporate governance practices on the profitability of banks. I have 12 years of data, 146 banks and 17 countries (7 developed countries and 10 developing countries). Can a multidimensional panel be applied to such a data structure? and multidimensional, will we consider banks and countries intertwined or analyze them as non-nested models? Or should I analyze such a data set as classical two-dimensional?
I would be grateful if you can reply to my post

[Carlo Lazzaro]

Cengiz:
please do not bump. Thanks.
That said, if banks are nested within countries, why not givinng -mixed- a shot?

[cengiz hokka]

Cengiz:
please do not bump. Thanks.
That said, if banks are nested within countries, why not givinng -mixed- a shot?

Sorry, my message went twice.
Mr. Lazzaro, by -mixed- do you mean Random effect model?
In my previous stata training, one of our trainers said: When working with multidimensional panel data, we first check which dimensions have an effect with the RE xtmixed command. for example;

xtmixed roa topvar syo lr1 ykb ts ioyo tecrube ly byo gdp ||_all:R.country ||_all:R.bank ||_all:R.year, mle

then we run the model again according to which dimension has an effect. For example, let's assume that all 3 dimensions have an effect.

xtmixed roa topvar syo lr1 ykb ts ioyo tecrube ly byo gdp ||_all:R.country ||_all:R.bank ||_all:R.year, mle

Then we run this model in 3D with a fixed effects estimator. but we cannot predict this model directly with fixed effects. In order to do that, we need to do some transformations. If we transform for the roa variable;
. egen meanroa = mean (roa)
. egen meanroac=mean (roa), by(country)
. egen meanroab=mean (roa), by(bank)
egen meanroay=mean(roa), by(year)
. gene dfroa=roa-meanroac-meanroab-meanroay+2*meanroa

After making these transformations for each variable, we can predict the model with the -reg- command. for example;

reg roa topvar syo lr1 ykb ts ioyo experience ly byo gdp

Is such an approach correct? what are your suggestions?