User Guide

PanelBox provides 70+ econometric models across 11 families, covering virtually every panel data method used in applied research. Each guide below introduces the model family, lists available estimators, and provides quick-start code examples.

All models follow a consistent API: define a formula, pass your data with entity and time identifiers, and call .fit(). Results objects provide .summary(), coefficient tables, diagnostic tests, and export to HTML/LaTeX.

• Static Models

  ---

  Pooled OLS, Fixed Effects, Random Effects, Between, First Difference

• Dynamic Models (GMM)

  ---

  Arellano-Bond, Blundell-Bond, CUE-GMM, Bias-Corrected GMM

• Spatial Econometrics

  ---

  SAR, SEM, SDM, Dynamic Spatial, General Nesting Spatial

• Stochastic Frontier

  ---

  SFA, Four-Component (unique in Python), TFP Decomposition

• Quantile Regression

  ---

  Pooled, Fixed Effects, Canay, Location-Scale, Dynamic, QTE

• Panel VAR

  ---

  VAR, VECM, IRF, FEVD, Granger Causality, Forecast

• Discrete Choice

  ---

  Logit, Probit, FE Logit, RE Probit, Ordered, Multinomial, Conditional

• Count Data

  ---

  Poisson (Pooled/FE/RE/QML), NegBin, PPML, Zero-Inflated

• Censored & Selection

  ---

  Tobit (Pooled/RE), Honore Trimmed, Panel Heckman

• Instrumental Variables

  ---

  Panel IV / 2SLS with first-stage diagnostics

Choosing a Model Family

Your Data	Recommended Family	Guide
Continuous outcome, no dynamics	Static Models	Pooled OLS, FE, RE
Lagged dependent variable	Dynamic GMM	Arellano-Bond, System GMM
Spatial dependence across units	Spatial	SAR, SEM, SDM
Efficiency / productivity analysis	Stochastic Frontier	SFA, Four-Component
Heterogeneous effects across distribution	Quantile	FE Quantile, Canay
Multiple interdependent outcomes	Panel VAR	VAR, VECM, IRF
Binary / ordered / multinomial outcome	Discrete Choice	Logit, Probit, Ordered
Count outcome (0, 1, 2, ...)	Count Data	Poisson, NegBin, PPML
Censored / truncated / selected sample	Censored & Selection	Tobit, Heckman
Endogenous regressors	Instrumental Variables	Panel 2SLS

Quick Example

from panelbox import FixedEffects
from panelbox.datasets import load_grunfeld

data = load_grunfeld()
model = FixedEffects("invest ~ value + capital", data, "firm", "year")
results = model.fit(cov_type="clustered")
print(results.summary())

See Also

• Getting Started -- Installation and first steps
• Tutorials -- Interactive notebooks with Google Colab
• Inference & Standard Errors -- Choosing the right standard errors
• Diagnostics & Validation -- Testing model assumptions