Long-term causal inference with many noisy proxies. Lal, A., Imbens, G., & Hull, P. January, 2026. arXiv:2601.06359 [econ]![Long-term causal inference with many noisy proxies [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex We propose a method for estimating long-term treatment effects with many short-term proxy outcomes: a central challenge when experimenting on digital platforms. We formalize this challenge as a latent variable problem where observed proxies are noisy measures of a low-dimensional set of unobserved surrogates that mediate treatment effects. Through theoretical analysis and simulations, we demonstrate that regularized regression methods substantially outperform naive proxy selection. We show in particular that the bias of Ridge regression decreases as more proxies are added, with closed-form expressions for the bias-variance tradeoff. We illustrate our method with an empirical application to the California GAIN experiment.
@misc{LalImbensHull2026,
title = {Long-term causal inference with many noisy proxies},
url = {http://arxiv.org/abs/2601.06359},
doi = {10.48550/arXiv.2601.06359},
abstract = {We propose a method for estimating long-term treatment effects with many short-term proxy outcomes: a central challenge when experimenting on digital platforms. We formalize this challenge as a latent variable problem where observed proxies are noisy measures of a low-dimensional set of unobserved surrogates that mediate treatment effects. Through theoretical analysis and simulations, we demonstrate that regularized regression methods substantially outperform naive proxy selection. We show in particular that the bias of Ridge regression decreases as more proxies are added, with closed-form expressions for the bias-variance tradeoff. We illustrate our method with an empirical application to the California GAIN experiment.},
urldate = {2026-01-21},
publisher = {arXiv},
author = {Lal, Apoorva and Imbens, Guido and Hull, Peter},
month = jan,
year = {2026},
note = {arXiv:2601.06359 [econ]},
keywords = {Economics - Econometrics},
}
Downloads: 0
{"_id":"sFbm9v9Gfy5yQW3gB","bibbaseid":"lal-imbens-hull-longtermcausalinferencewithmanynoisyproxies-2026","author_short":["Lal, A.","Imbens, G.","Hull, P."],"bibdata":{"bibtype":"misc","type":"misc","title":"Long-term causal inference with many noisy proxies","url":"http://arxiv.org/abs/2601.06359","doi":"10.48550/arXiv.2601.06359","abstract":"We propose a method for estimating long-term treatment effects with many short-term proxy outcomes: a central challenge when experimenting on digital platforms. We formalize this challenge as a latent variable problem where observed proxies are noisy measures of a low-dimensional set of unobserved surrogates that mediate treatment effects. Through theoretical analysis and simulations, we demonstrate that regularized regression methods substantially outperform naive proxy selection. We show in particular that the bias of Ridge regression decreases as more proxies are added, with closed-form expressions for the bias-variance tradeoff. We illustrate our method with an empirical application to the California GAIN experiment.","urldate":"2026-01-21","publisher":"arXiv","author":[{"propositions":[],"lastnames":["Lal"],"firstnames":["Apoorva"],"suffixes":[]},{"propositions":[],"lastnames":["Imbens"],"firstnames":["Guido"],"suffixes":[]},{"propositions":[],"lastnames":["Hull"],"firstnames":["Peter"],"suffixes":[]}],"month":"January","year":"2026","note":"arXiv:2601.06359 [econ]","keywords":"Economics - Econometrics","bibtex":"@misc{LalImbensHull2026,\n\ttitle = {Long-term causal inference with many noisy proxies},\n\turl = {http://arxiv.org/abs/2601.06359},\n\tdoi = {10.48550/arXiv.2601.06359},\n\tabstract = {We propose a method for estimating long-term treatment effects with many short-term proxy outcomes: a central challenge when experimenting on digital platforms. We formalize this challenge as a latent variable problem where observed proxies are noisy measures of a low-dimensional set of unobserved surrogates that mediate treatment effects. Through theoretical analysis and simulations, we demonstrate that regularized regression methods substantially outperform naive proxy selection. We show in particular that the bias of Ridge regression decreases as more proxies are added, with closed-form expressions for the bias-variance tradeoff. We illustrate our method with an empirical application to the California GAIN experiment.},\n\turldate = {2026-01-21},\n\tpublisher = {arXiv},\n\tauthor = {Lal, Apoorva and Imbens, Guido and Hull, Peter},\n\tmonth = jan,\n\tyear = {2026},\n\tnote = {arXiv:2601.06359 [econ]},\n\tkeywords = {Economics - Econometrics},\n}\n\n\n\n","author_short":["Lal, A.","Imbens, G.","Hull, P."],"key":"LalImbensHull2026","id":"LalImbensHull2026","bibbaseid":"lal-imbens-hull-longtermcausalinferencewithmanynoisyproxies-2026","role":"author","urls":{"Paper":"http://arxiv.org/abs/2601.06359"},"keyword":["Economics - Econometrics"],"metadata":{"authorlinks":{}},"downloads":0,"html":""},"bibtype":"misc","biburl":"https://bibbase.org/zotero-group/jwshan/6108874","dataSources":["v9RrW54b78PPwoLeo"],"keywords":["economics - econometrics"],"search_terms":["long","term","causal","inference","many","noisy","proxies","lal","imbens","hull"],"title":"Long-term causal inference with many noisy proxies","year":2026}