Instrumental Variable Estimation of Average Partial Causal Effects

Yuta Kawakami, Manabu Kuroki, Jin Tian
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:16097-16130, 2023.

Abstract

Instrumental variable (IV) analysis is a powerful tool widely used to elucidate causal relationships. We study the problem of estimating the average partial causal effect (APCE) of a continuous treatment in an IV setting. Specifically, we develop new methods for estimating APCE based on a recent identification condition via an integral equation. We develop two families of methods, nonparametric and parametric - the former uses the Picard iteration to solve the integral equation; the latter parameterizes APCE using a linear basis function model. We analyze the statistical and computational properties of the proposed methods and illustrate them on synthetic and real data.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-kawakami23a, title = {Instrumental Variable Estimation of Average Partial Causal Effects}, author = {Kawakami, Yuta and Kuroki, Manabu and Tian, Jin}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {16097--16130}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/kawakami23a/kawakami23a.pdf}, url = {https://proceedings.mlr.press/v202/kawakami23a.html}, abstract = {Instrumental variable (IV) analysis is a powerful tool widely used to elucidate causal relationships. We study the problem of estimating the average partial causal effect (APCE) of a continuous treatment in an IV setting. Specifically, we develop new methods for estimating APCE based on a recent identification condition via an integral equation. We develop two families of methods, nonparametric and parametric - the former uses the Picard iteration to solve the integral equation; the latter parameterizes APCE using a linear basis function model. We analyze the statistical and computational properties of the proposed methods and illustrate them on synthetic and real data.} }
Endnote
%0 Conference Paper %T Instrumental Variable Estimation of Average Partial Causal Effects %A Yuta Kawakami %A Manabu Kuroki %A Jin Tian %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-kawakami23a %I PMLR %P 16097--16130 %U https://proceedings.mlr.press/v202/kawakami23a.html %V 202 %X Instrumental variable (IV) analysis is a powerful tool widely used to elucidate causal relationships. We study the problem of estimating the average partial causal effect (APCE) of a continuous treatment in an IV setting. Specifically, we develop new methods for estimating APCE based on a recent identification condition via an integral equation. We develop two families of methods, nonparametric and parametric - the former uses the Picard iteration to solve the integral equation; the latter parameterizes APCE using a linear basis function model. We analyze the statistical and computational properties of the proposed methods and illustrate them on synthetic and real data.
APA
Kawakami, Y., Kuroki, M. & Tian, J.. (2023). Instrumental Variable Estimation of Average Partial Causal Effects. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:16097-16130 Available from https://proceedings.mlr.press/v202/kawakami23a.html.

Related Material