SURE/Panel/Random effect for a new Stata user, lets try to explain it for everybody

Hi everybody

I'm a new Stata user (and also a non english native, i apologize for my english), and i'm trying to make a SURE regression.

I) What i want to do:
-SURE regression on balanced panel datas. Panel means : G equations, T years and I industries.
- Random effect

- My system is a classical cost/demand function (generalized leontief case but my question also works for a translog case) :
Y = b0+b1p1+b2p2+b3p3+b4(p1p2)^0.5+b5(p1p3)^0.5+.b6(p2 p3^0.5)
a1 = b1 + b4(p2/p1)^0.5+b5(p3/p1)^0.5
...
=> i need to make cross section restrictions on all of my "bi" that appears 2 or 3 times.
II) What i know:
1)xtsur : makes SUR regression with a random effect on unbalanced panel

=> But : it seems that we can't constraint the "bi" to be equal between equations . Am i wrong? are they ways to solve it?

For exemple :
with pp1=p1^0.5 and mp1 = p1^-0.5
xtsur (y p1 p2 p3 pp1pp2 pp1pp3 pp2pp3) (a1 p1 mp1pp2 mp1pp3) (a2 mp2pp1 mp2pp3)
Seemingly unrelated regression (SUR) in panel data set

One-way random effect estimation:
------------------------------------------------------------------------------
Number of Group variable: 1 Number of obs = 234
Panel variable: ncty Number of eqn = 3
Time variable : year Number of panels = 1

Random effects u_i ~ Gaussian
corr(u_i, e_it) = 0 (assumed)
Panel type : strongly balanced

------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
y |
p1 | -467136.8 248468.6 -1.88 0.060 -954126.4 19852.78
p2 | -1153521 200116.1 -5.76 0.000 -1545741 -761300.5
p3 | -1823076 315600.5 -5.78 0.000 -2441642 -1204511
pp1pp2 | 7334.495 349830.5 0.02 0.983 -678320.7 692989.7
pp1pp3 | 841590.7 445020.1 1.89 0.059 -30632.72 1713814
pp2pp3 | 2584924 433375.6 5.96 0.000 1735523 3434324
-------------+----------------------------------------------------------------
a1 |
p1 | .0141656 .0076403 1.85 0.064 -.0008092 .0291404
mp1pp2 | -.0828314 .0100321 -8.26 0.000 -.1024939 -.0631688
mp1pp3 | -.0089125 .0162751 -0.55 0.584 -.0408111 .0229861
-------------+----------------------------------------------------------------
a2 |
mp2pp1 | .0002708 .0019449 0.14 0.889 -.0035411 .0040826
mp2pp3 | .0069883 .0017342 4.03 0.000 .0035894 .0103873
-------------+----------------------------------------------------------------
sigma_u | see e(sigma_u)
sigma_e | see e(sigma_e)
------------------------------------------------------------------------------
Dependent variables: y a1 a2
Independent variables: p1 p2 p3 pp1pp2 pp1pp3 pp2pp3 mp1pp2 mp1pp3 mp2pp1
mp2pp3


Is there a way to impose cross section restrictions on the coef? something like pp1pp2/y=coeff mp1pp2/a1=coef mp2pp1/a2 ?

=> could we use balanced panel data or are their some minors controls to make?

- 2) sureg : makes SUR regression, no random or fixed effect, unless we ask for it. But not designed for panel datas
Moreover we cant neither constraint the "bi"

3) nlsur:
We can impose cross section restriction!!!
Contrary to xtsur and sureg we can use {} to impose cross section restrictions ("{ invalid name" with xtsur and surreg):
ex: nlsur ( y = {b1}p1 +...+{b4}pp1pp2 ...)(a1={b1}+..+{b4}mp1pp2)...

But it seems that this command doesnt work for panel datas, and i'm not sure that we can add random effect(?)


III) To go further
Is there way to do the same thing but with random effect on coefficient (Random coefficient approach, see Biorn (2004))?

If anyone has answers, ideas ...please help me/us

Thanks a lot