linearmodels/iv-without-exog
| Rule | iv-without-exog |
|---|---|
| Pack | linearmodels |
| Severity | warning |
| Fix | none, edit by hand |
| Library versions | every version |
With exog=None, the instrument model has no intercept, because linearmodels adds none, so the estimates are biased. Pass a column of ones as exog.
model = IV2SLS(y, None, x, z)model = IV2SLS(y, sm.add_constant(controls), x, z)What the rule reports
Section titled “What the rule reports”The rule reports None written as the exog of an instrument model that has an endogenous
regressor. The model then has no intercept.
How to fix it
Section titled “How to fix it”xpd has no automatic fix, so make the change by hand.
Over 2,000 draws of 500 rows from y = 2 + x + error with one instrument, the two forms gave the
following mean estimates:
exog |
Mean estimate of the coefficient on x, which is 1 |
|---|---|
None |
2.002 |
| a constant column | 0.990 |
On the mroz data the coefficient on educ was 0.0928 without exog and 0.0664 with a constant and
exper. Data with mean zero needs no intercept, so the rule is a warning.
Message
Section titled “Message”xpd prints this message with the issue. A name that starts with $ stands for code from your file.
exog is None here, so the model has no intercept, because linearmodels adds none. In a simulation with a true coefficient of 1.0 and an intercept of 2.0, the mean estimate was 2.002 without exog and 0.990 with a constant column. Pass a column of ones as exog, such as sm.add_constant, unless the data has mean zero.Browse the other rules of the linearmodels pack.