pymte

Instrumental variables and marginal treatment effects in Python.

Documentation: https://pymte.readthedocs.io

pymte estimates treatment parameters such as the ATE, ATT or a policy relevant treatment effect from instrumental variables data, using the marginal treatment effect (MTE) framework of Heckman and Vytlacil (2005) and the moment-based implementation of Mogstad, Santos and Torgovitsky (2018). The user specifies parametric or nonparametric (spline) marginal treatment response functions, optional shape restrictions such as boundedness or monotonicity, and a set of IV-like estimands. The package returns either a point estimate (when the model is point identified) or sharp bounds obtained by linear or quadratically constrained programming, together with bootstrap confidence intervals and specification tests.

This package is a port of the R package ivmte by Joshua Shea and Alexander Torgovitsky, released under the GPL-3. It reproduces the R package's estimands, options and numerical results; the API follows Python conventions.

Installation

pip install pymte

Linear programs are solved with HiGHS (through SciPy) and quadratically constrained programs with Clarabel (through CVXPY); both are installed automatically. MOSEK and Gurobi can be used instead when licensed:

pip install "pymte[mosek]"
pip install "pymte[gurobi]"
pip install "pymte[plots]"   # matplotlib for plot_mtr, plot_mte, plot_weights

Quick start

import pymte

ae = pymte.load_ae()
r = pymte.ivmte(
    ae,
    target="att",
    m0="~ u + yob",
    m1="~ u + yob",
    ivlike="worked ~ morekids + samesex + morekids*samesex",
    propensity="morekids ~ samesex + yob",
)
print(r)

Bounds on the target parameter: [-0.1028836, -0.07818869]
Audit terminated successfully after 1 round(s)
MTR coefficients: 6
Independent/total moments: 4/4
Minimum criterion: 0
Solver: highs

Add bootstraps=50 for confidence regions, point=True for GMM when the model is point identified, or replace ivlike with outcome="worked" for the regression approach.

The full user guide, a theory primer and the API reference are on Read the Docs.

Contributing

pymte is a new project. If you find a bug, want a feature, see a way to make something faster, or have an application that the package does not yet cover, open an issue or a pull request on GitHub. Small fixes, new examples, and documentation improvements are as welcome as new estimators. If you would prefer to talk first, or want to discuss a larger piece of work, email me at alex.ronczewski@gmail.com.

Citing

If you use this package, please cite the papers that developed the methodology:

• Mogstad, M., A. Santos and A. Torgovitsky (2018). Using Instrumental Variables for Inference About Policy Relevant Treatment Parameters. Econometrica 86(5), 1589-1619. doi:10.3982/ECTA15463.
• Shea, J. and A. Torgovitsky (2023). ivmte: An R Package for Extrapolating Instrumental Variable Estimates Away From Compliers. Observational Studies 9(2), 1-42.
• Heckman, J. J. and E. Vytlacil (2005). Structural Equations, Treatment Effects, and Econometric Policy Evaluation. Econometrica 73(3), 669-738.
• Mogstad, M. and A. Torgovitsky (2018). Identification and Extrapolation of Causal Effects with Instrumental Variables. Annual Review of Economics 10, 577-613.

BibTeX entries are in the documentation under "Citing". The AE dataset is derived from Angrist and Evans (1998), American Economic Review 88(3).

Licence

GPL-3.0-only. This package is a derivative work of the R package ivmte, Copyright (C) 2018-2024 Joshua Shea and Alexander Torgovitsky, which its authors distribute under GPL-2 or GPL-3; this port takes it under version 3 and is distributed under the same version. See LICENSE for the licence text and NOTICE for the copyright notices.