StatsPAI 0.2.0

The Causal Inference & Econometrics Toolkit for Python

StatsPAI: The Causal Inference & Econometrics Toolkit for Python

StatsPAI is a Python package for causal inference and applied econometrics. It provides a unified, Stata-style API covering the complete empirical research workflow — from estimation to publication-ready tables in Word, Excel, and LaTeX.

It brings R's Causal Inference Task View (fixest, did, rdrobust, gsynth, DoubleML, MatchIt, CausalImpact) into a single, consistent Python package.

Built by the team behind CoPaper.AI · Stanford REAP Program

---

Main Features

Regression Models:

• Ordinary Least Squares with robust / clustered / HAC standard errors
• Instrumental Variables / Two-Stage Least Squares (2SLS), with first-stage F, Sargan, and Hausman tests
• Panel data: Fixed Effects, Random Effects, Between, First Differences (via linearmodels)
• High-dimensional Fixed Effects (via pyfixest)

Causal Inference — Difference-in-Differences:

• Classic 2x2 DID estimator
• Staggered DID with heterogeneous treatment effects (Callaway & Sant'Anna 2021)
• Event study plots and pre-trend tests

Causal Inference — Regression Discontinuity:

• Sharp and Fuzzy RD with local polynomial estimation
• MSE-optimal bandwidth selection (CCT 2014)
• Robust bias-corrected confidence intervals
• RD plots with binned scatter and polynomial fit

Causal Inference — Matching:

• Propensity Score Matching (logit-based PSM)
• Mahalanobis distance matching
• Coarsened Exact Matching (CEM)
• Balance diagnostics with standardized mean differences

Causal Inference — Synthetic Control:

• Abadie-Diamond-Hainmueller SCM
• Penalized / ridge SCM for many donors
• Placebo (permutation) inference with MSPE ratios
• Donor weight tables and gap plots

Causal Inference — Machine Learning Methods:

• Double/Debiased Machine Learning: Partially Linear (PLR) and Interactive (IRM) models with cross-fitting (Chernozhukov et al. 2018)
• Causal Forest for heterogeneous treatment effects (HTE)
• Compatible with any scikit-learn estimator as first-stage ML model

Causal Inference — Other Methods:

• Causal Impact: Bayesian structural time-series intervention analysis (Brodersen et al. 2015)
• Causal Mediation Analysis: ACME / ADE decomposition with bootstrap inference (Imai et al. 2010)
• Shift-Share / Bartik IV with Rotemberg weight diagnostics (GPSS 2020)

Post-Estimation:

• Marginal effects (AME / MEM) with delta-method standard errors, equivalent to Stata's margins, dydx(*)
• Wald test for linear restrictions, equivalent to Stata's test
• Linear combinations of coefficients with inference, equivalent to Stata's lincom

Diagnostics:

• Oster (2019) coefficient stability / selection-on-unobservables bounds
• McCrary (2008) density manipulation test for RD validity

Publication-Quality Output:

• Multi-model comparison tables (equivalent to R's modelsummary / Stata's esttab)
• Coefficient forest plots across models
• Summary statistics tables (equivalent to Stata's tabstat)
• Balance tables for matching / DID / RCT papers
• Cross-tabulation with chi-squared / Fisher's exact test (equivalent to Stata's tab, chi2)
• Export to Word (.docx), Excel (.xlsx), LaTeX (.tex), HTML — all tables, all formats
• Every result object has .summary(), .plot(), .to_latex(), .to_docx(), .cite()

---

Installation

pip install statspai

With optional dependencies:

pip install statspai[plotting]    # matplotlib, seaborn
pip install statspai[fixest]      # pyfixest for high-dimensional FE

Requirements: Python >= 3.9

Core dependencies: NumPy, SciPy, Pandas, statsmodels, scikit-learn, linearmodels, patsy, openpyxl, python-docx

---

Quick Example

import statspai as sp

# --- Estimation ---
r1 = sp.regress("wage ~ education + experience", data=df, robust='hc1')
r2 = sp.ivreg("wage ~ (education ~ parent_edu) + experience", data=df)
r3 = sp.did(df, y='wage', treat='policy', time='year', id='worker')
r4 = sp.rdrobust(df, y='score', x='running_var', c=0)
r5 = sp.match(df, y='outcome', treat='treated', covariates=['age', 'edu'])
r6 = sp.dml(df, y='wage', treat='training', covariates=['age', 'edu', 'exp'])

# --- Post-estimation ---
me = sp.margins(r1, data=df)            # Marginal effects
sp.test(r1, "education = experience")   # Wald test: beta_edu = beta_exp?
sp.lincom(r1, "education + experience") # Linear combination

# --- Tables (to Word / Excel / LaTeX) ---
sp.modelsummary(r1, r2, output='table2.docx')
sp.outreg2(r1, r2, r3, filename='results.xlsx')
sp.sumstats(df, vars=['wage', 'education', 'age'], output='table1.docx')
sp.balance_table(df, treat='treated', covariates=['age', 'edu'], output='balance.docx')
sp.tab(df, 'treatment', 'outcome', output='crosstab.docx')

---

API Summary

Category	Functions	Description
Regression	regress, ivreg, panel, fixest.feols	OLS, IV/2SLS, Panel (FE/RE/FD/BE), High-dimensional FE
DID	did, did_2x2, callaway_santanna	Classic 2x2, Staggered (C&S 2021), Event study
RD	rdrobust, rdplot	Sharp/Fuzzy RD, CCT robust inference, RD plots
Matching	match	PSM, CEM, Mahalanobis, Balance diagnostics
Synth	synth	Abadie SCM, Penalized SCM, Placebo inference
ML Causal	dml, causal_forest	Double ML (PLR/IRM), Causal Forest (HTE)
Other Causal	causal_impact, mediate, bartik	Intervention analysis, Mediation, Shift-share IV
Post-estimation	margins, marginsplot, test, lincom	Marginal effects, Wald tests, Linear combinations
Diagnostics	oster_bounds, mccrary_test	Coefficient stability, Density manipulation
Tables	modelsummary, outreg2, sumstats, balance_table, tab	Multi-model tables, Summary stats, Balance, Cross-tabs
Plots	coefplot, marginsplot, rdplot, result.plot()	Coefficient, Margins, RD, Event study plots
Export	.to_docx(), .to_latex(), output='*.xlsx'	Word, Excel, LaTeX, HTML — all tables, all formats

All causal methods return a unified CausalResult object:

result.estimate       # Point estimate
result.se             # Standard error
result.pvalue         # P-value
result.ci             # Confidence interval
result.summary()      # Formatted text summary
result.plot()         # Appropriate visualization
result.to_latex()     # LaTeX table
result.to_docx()      # Word document
result.cite()         # BibTeX citation for the method

---

Comparison with Stata and R

Task	Stata	R	StatsPAI
OLS with robust SE	reg y x, r	feols(y ~ x, vcov="HC1")	sp.regress("y ~ x", robust='hc1')
IV regression	ivregress 2sls y (x = z)	feols(y ~ 1 | x ~ z)	sp.ivreg("y ~ (x ~ z)")
Staggered DID	csdid y, ivar(id) time(t) gvar(g)	att_gt(y ~ 1, ...)	sp.did(df, y, treat, time, id)
RD design	rdrobust y x, c(0)	rdrobust(Y, X, c=0)	sp.rdrobust(df, y, x, c=0)
PSM matching	psmatch2 treat x1 x2	matchit(treat ~ x1+x2)	sp.match(df, y, treat, covs)
Double ML	—	DoubleML$new(...)	sp.dml(df, y, treat, covs)
Marginal effects	margins, dydx(*)	margins(model)	sp.margins(result, data=df)
Wald test	test x1 = x2	linearHypothesis(...)	sp.test(result, "x1 = x2")
Export to Word	outreg2 using r.doc, word	modelsummary(output="t.docx")	sp.outreg2(r, filename="r.docx")
Summary stats	tabstat y x, s(mean sd)	datasummary(...)	sp.sumstats(df, vars=[...])

---

About

StatsPAI Inc. is the research infrastructure company behind CoPaper.AI — the AI co-authoring platform for empirical research, born out of Stanford's REAP program.

CoPaper.AI — Upload your data, set your research question, and produce a fully reproducible academic paper with code, tables, and formatted output. Powered by StatsPAI under the hood. copaper.ai

Team:

• Bryce Wang — Founder. Economics, Finance, CS & AI. Stanford REAP.
• Dr. Scott Rozelle — Co-founder & Strategic Advisor. Stanford Senior Fellow, author of Invisible China.

---

Contributing

git clone https://github.com/brycewang-stanford/statspai.git
cd statspai
pip install -e ".[dev,plotting,fixest]"
pytest

---

Citation

@software{wang2025statspai,
  title={StatsPAI: The Causal Inference & Econometrics Toolkit for Python},
  author={Wang, Bryce},
  year={2025},
  url={https://github.com/brycewang-stanford/statspai},
  version={0.1.0}
}

License

MIT License. See LICENSE.

---

GitHub · PyPI · Documentation · CoPaper.AI