linearmodels.panel.results.PanelEffectsResults.estimated_effects¶
- property PanelEffectsResults.estimated_effects : DataFrame¶
Estimated effects.
Notes
The combined effect of all of the effects in the model, in one column. It is the sum of the entity, time and other effects of each observation. The effects are not available separately. See Estimated Effects for how to recover the entity and time effects of a model that has both.
The estimated effects have no standard errors or any other inference. They are not estimated consistently when the number of time periods is fixed. The effect of an entity is estimated from its observations, so the error does not shrink when the number of entities grows. To make inference about the effects of a small number of entities, include them as dummy variables in the regression.
The values are NaN for models that do not have effects, which are PooledOLS, FirstDifferenceOLS and FamaMacBeth. They are 0 for a PanelOLS model without effects.
The effects of RandomEffects are theta times the entity mean of the difference between the dependent variable and the fitted values, where theta is in the results. They are all 0 if the estimated variance of the random effect is negative and so it is set to 0, since theta is then 0 and the model is pooled OLS.
variance_decompositionshows whether this is the case.