Univariate Bayesian Estimation

[EF00]

Dear all,
is it possible to implement a bayesian approach for a univariate estimation model? How?

[EViews Gareth]

There is nothing built in to do it. Yet...

[EF00]

That's okay. Alternatively, how can I manage different time windows? How can I identify them? For example, how can I handle COVID-19 shocks that might affect the estimation of coefficients in a model and the forecasting results?

[startz]

Can you be more specific? You can control the estimation window or forecast window with the

Code: Select all

smpl

command.

[EF00]

For example, although EViews is not specifically designed for Bayesian models, is it possible to implement a "simplified version" of dynamic linear models using state-space structures and the Kalman filter? How?

[EF00]

What I want is to obtain accurate estimates of the coefficients of my equation, which might be influenced by external phenomena.

[startz]

If you want to exclude the pandemic you could set the estimation to be something like

Code: Select all

smpl 1899 2019 2023 2034

Alternatively, you might include a dummy variable that is set to 1.0 during the pandemic.