[Python] Fixed Effects Regresion without Exogenous Variables in Python

I am trying to fit a TWFE linear regression, so I have an endogenous (y) variable, entity effects and time effects. I just want to fit this regression:

Y_it = \lambda_t + \gamma_i

But, if I do not use an exogenous variable (X), I get the error of "missing exogenous positional argument".

Here is some code to replicate the error:

import pandas as pd
from linearmodels.panel import PanelOLS

# generate fake Panel data
df = pd.DataFrame({
'entity_id': [1, 1, 2, 2, 3, 3],
'time_id': [1, 2, 1, 2, 1, 2],
'y': [10, 15, 10, 20, 30, 40]
})

# Set index for panel data (entity and time identifiers)
df = df.set_index(['entity_id', 'time_id'])

# Define the dependent variable
y = df['y']

# Create and fit the Two-Way Fixed Effects model (without X)
mod = PanelOLS(y, entity_effects=True, time_effects=True)
res = mod.fit()


On how to solve it:

I know I can do it by creating the dummies with pd.concat, but I have a huge dataset so creating dummies is not a good idea. There has to be a way to do it, since in R and Stata this is possible.

Thanks in advance

Continue reading...