6 Post estimation steps
Post estimation steps are crucial for validating estimated parameters, understanding their robustness, and ensuring that results are reliable and meaningful.
After estimating the model, we should determine if it is appropriate. An appropriate econometric model should meet the following criteria:
- estimates are meaningful (positive/negative) and statistically significant
- model fits the data well (according to GOF measures)
- all assumptions are met (requires diagnostic checking)
- estimates are stable (robust)
- If model is not appropriate we should consider re-estimating the model with a new specification, incorporating new variables (or transforming existing variables), or even using an alternative estimation method (other than OLS).
The key post estimation steps are typically taken:
- significance testing
- determining the goodness-of-fit
- diagnostic checking
- stability checking in terms of robustness