Simple T-test

[gka]

Hello guys, I want to estimate the Cov(b1,b2) from the models below:

y=a1+b1X1
y=a2+b2X2

Do you know how to do this?
Ps. I want to check if H1:b1=b2 or H2:b1#b2, for t-test=(b1-b2)/sqrt(Var(b1)+Var(b2)-2Cov(b1,b2))

Is there an other way to do this?

[startz]

Put the equations in a System.

[gka]

The problem is that I have beta coefficients from guantile regression... and this is not easy to do something like this.

Thanks...

Is there any other idea?

[trubador]

Startz' suggestion is the way to do what you want. However, based on your last post, I understand that the equations are already estimated (somewhere else) and you happen to have b1 and b2 parameters from a number of equations. If that is the case, then the following resources might be of help:
http://www.eviews.com/help/helpintro.ht ... 10.html%23
http://www.eviews.com/help/helpintro.ht ... 23ww132090

[gka]

Yes the equations already estimated. But the problem is that I have about 100 b (b1 b2 b3 ... b100) from monte carlo simulation I want to check all of this with the first b1. On the other hand, the basic problem is that I have some equations from quantile regression...

[startz]

Can you compute the covariance from the monte carlo simulated coefficients?