midas estimation

[alexl]

Dear Sirs,
I am starting to approach midas estimates (eviews 9.5). I am not interested in forecasting.

In a traditional PDL least squares estimation I find:
- the PDL coefficients, with the usual set of information (standard error, t-statistics, prob.) for the PDL coefficients (coefficients "gamma" of the transformed variable "z")
- the coefficients of interest (beta) at the bottom of the table, together with: (i) a plot of the estimated polynomial (ii) the usual set of information (standard error, t-statistics,prob.) (iii) the "Sum of Lags"

In the Midas estimation with PDL/Almon. I find
- the PDL coefficients, with the usual set of information (standard error, t-statistics, prob.) for the PDL coefficients (coefficients "gamma" of the transformed varibale "z"), which is great.
- the coefficients of interest (beta) at the bottom of the table, together with: (i) a plot of the estimated polynomial BUT NOT the usual set of statistical information (standard error, t-statistic, prob).

The Wald test is obviously on gammas, not on betas.

I am interested in the statistical significance of the betas...is it possible to find the usual statistical information about these betas coefficients (standard error, t-statistics,prob.) or something that can help me to assess the statistical significance of the betas?
Thank you in advance.

[KrilleJ]

The parameters you wish to test are probably nonlinear functions of the estimated parameters.

Then, perhaps the delta method could be of use for solving the problem:

http://www.stata.com/support/faqs/stati ... ta-method/

[alexl]

Nice point. Thank you very much! I'll check the applicability to my research and I'll try.
Thank you again!