R or MatLab or Gauss? I really need some suggestions, many many many thanks

Dear all,

I want to estimate a nonlinear Euler equation model of investment for Panel data by using first-difference GMM with inequality constraints, in details:

    {a}*(I/K)_it + F_i + e_(t+1) = (1+B_(it+1))/(1+B_it)*[MPK_(t+1) ...
                                 + (1-d)*{a}*(I/K)_(it+1)]

B_it is an unobservable variable measuring the value of shadow cost, I want to parametrize it with firm-level observable variables:

    B = c+c1*cash_(it+1) + c2*debt_(it+1) + c3*size_(it+1)

where I/K is investment to capital ratio;

               MPK is marginal product of capital; 
               a is adjustment cost parameter;
               d is depreciation rate;
               B is value of shadow cost;
               e is expectational error;
               F is fixed effects

In this model, there are both fixed effects terms and (IK_it ; Ik_it+1). Therefore, I have to take first-difference to the whole non-linear equation to eliminate the fixed effects F_i and to calculate the (e_it - e_it-1) or (e_it+1 - e_it ). Then, I want to minimize the quadratic moment conditions.

However, I have to set some inequality constraints during the optimization process. For example, the value of shadow cost should be non-negative:

    B_it>=0 or [c+c1*cash_(it+1) + c2*debt_(it+1) + c3*size_(it+1)]>=0

Previously, I was using Stata, but neither Stata nor Mata language can do it for GMM.

Hope that I have presented my model clearly. People told me that either MatLab or Guass or R can do the estimation, but each of them has its own advantages and disadvantages.

Since this is my first time to touch this sort of software, I am not sure which one is more suitable for my situation.

For instance, on the website, people say that R is relatively weak in Matrix; Gauss is good for Econometrics but MatLab is good at optimization. Also, Gauss is good at doing Panel data and Micro-econometrics while MatLab is good at Macro-econometrics and time series. Lastly, MatLab syntax is the most painless one

Actually, I am very confused right now and I believe that I really need some serious as well as professional suggestions. I have years experiences in doing econometric estimation by using Eviews and Stata. However, this is my first time to do the structural estimation by programming. The model and estimation contain: a huge Panel data; Matrix programming; constrained optimization.

Anyone please give me some suggestions?

Many many thanks!

[dpb]

http://www.mathworks.com/help/optim/ug/choosing-a-solver.html#brhkghv-21

[Zhixiao]

ok, thanks, I have decided to use MatLab, simply because Mata and Stata have used a language very similar with that of MatLab, while Gauss and R, for me, are totally different things. Actually, I have done all the tutorials of three software and found that MatLab has the most smooth learning curve...