statsmodels.stats.oaxaca.OaxacaBlinder#

class statsmodels.stats.oaxaca.OaxacaBlinder(endog, exog, bifurcate, hasconst=True, swap=True, cov_type='nonrobust', cov_kwds=None, rng=None)[source]#

Class to perform Oaxaca-Blinder Decomposition

Parameters:

endogarray_like: The endogenous variable or the dependent variable that you are trying to explain.
exogarray_like: The exogenous variable(s) or the independent variable(s) that you are using to explain the endogenous variable.
bifurcate{int, str}: The column of the exogenous variable(s) on which to split. This would generally be the group that you wish to explain the two means for. Int of the column for a NumPy array or int/string for the name of the column in Pandas.
hasconstbool, optional: Indicates whether the two exogenous variables include a user-supplied constant. If True, a constant is assumed. If False, a constant is added at the start. If nothing is supplied, then True is assumed.
swapbool, optional: Imitates the STATA Oaxaca command by allowing users to choose to swap groups. Unlike STATA, this is assumed to be True instead of False.
cov_typestr, optional: See regression.linear_model.RegressionResults for a description of the available covariance estimators.
cov_kwdsdict, optional: See linear_model.RegressionResults.get_robustcov_results for a description of the required keywords for alternative covariance estimators.
rng{None, int, array_like[int], numpy.random.Generator, numpy.random.RandomState}, optional: If rng is None, a new Generator is created using fresh entropy from the operating system. If rng is an int or array of ints, a new Generator is created, seeded with rng. If rng is already a Generator or RandomState instance, that instance is used.

Methods

`three_fold`([std, n, conf])	Calculates the three-fold Oaxaca Blinder Decompositions
`two_fold`([std, two_fold_type, ...])	Calculates the two-fold or pooled Oaxaca Blinder Decompositions
`variance`(decomp_type[, n, conf])	A helper function to calculate the variance/std

Notes

Please check if your data includes a constant. This will still run, but will return incorrect values if set incorrectly.

You can access the models by using their code as an attribute, e.g., _t_model for the total model, _f_model for the first model, _s_model for the second model.

Examples

>>> import numpy as np
>>> import statsmodels.api as sm
>>> data = sm.datasets.ccards.load()

‘3’ is the column of which we want to explain or which indicates the two groups. In this case, it is if you rent.

>>> model = sm.OaxacaBlinder(df.endog, df.exog, 3, hasconst = False)
>>> model.two_fold().summary()
Oaxaca-Blinder Two-fold Effects
Unexplained Effect: 27.94091
Explained Effect: 130.80954
Gap: 158.75044

>>> model.three_fold().summary()
Oaxaca-Blinder Three-fold Effects
Endowments Effect: 321.74824
Coefficient Effect: 75.45371
Interaction Effect: -238.45151
Gap: 158.75044

Methods

`three_fold`([std, n, conf])	Calculates the three-fold Oaxaca Blinder Decompositions
`two_fold`([std, two_fold_type, ...])	Calculates the two-fold or pooled Oaxaca Blinder Decompositions
`variance`(decomp_type[, n, conf])	A helper function to calculate the variance/std