Stats Regression instrumental - CyrilB1531/lodestar GitHub Wiki
Instrumental-variables data — Lodestar.Stats.Regression.Instrumental
The data types InstrumentalVariables takes and returns. They are
compiled into Lodestar.Abstractions, as every public data type that carries no logic is
(decision 0003), in their
own namespace so that none of them borrows a name from ordinary least squares: the covariances here
are linearmodels' four, not statsmodels' seven.
Types
| Type | What it is |
|---|---|
IvDesign |
The response and the three blocks: exogenous, endogenous, instruments. |
IvOptions |
The covariance, its kernel and scaling, Fuller's α, GMM's weight and the intercept. |
IvCovarianceType |
Unadjusted, robust, kernel or clustered. |
IvSummary |
The table, R², the model test, the first stage and the overidentification test. |
IvFirstStage |
One endogenous regressor's first-stage diagnostics. |
The tests they report are WaldTests and the kernels KernelTypes,
both shared with the panel estimators in the common types.
See also
- Instrumental variables — the estimators.
- Python → C# equivalence.