Should different software packages give different estimates?

[um4rio]

For example, I have tried a few VAR equations in Eviews and Microfit. Despite using the same data, they give different estimates for the coefficents inside the cointegrating vector. Is this normal? Or should the two software packages be giving the same results? I only ask this because in my elasticity study, eviews give elastic demand (beta is more then 1), whereas microfit gives the inelastic demand that was hypothesied (beta is less than 1). Overall, I prefer eviews but am puzzled as to why the software give different results?

thanks

[whoami]

I have encountered this problem. The suggestion was to change the starting values and check for the stability of the estimates. Try that with both software.