Checking accuracy of solution

[mch]

Hi all,

I am solving a model with a second-order approximation and want to check the accuracy of the solution since the model has large shock variances. I have read posts saying that the Euler errors or Den Haan-Marcet stat could be computed using simulated data, but I am computing theoretical moments so there is no simulated data. Is it necessary, or even possible, to check the accuracy of the numerical solution in this case?

Thanks,
Michael

[jpfeifer]

You can still easily simulate the required data (use the periods statement). There is no theoretical measure that I am aware of that can be used to generally compute the accuracy of the solution

[mch]

OK jpfeifer - thanks for your reply