EconBase
← All papers

Minimax Instrumental Variable Regression and $L_2$ Convergence Guarantees without Identification or Closedness

Andrew Bennett, Nathan Kallus, Xiaojie Mao, Whitney Newey, Vasilis Syrgkanis, Masatoshi Uehara

arXiv 10 Feb 2023 · Statistics — Machine Learning

arXiv:2302.05404 · PDF · DOI · OpenAlex · Extracted main text

Abstract

In this paper, we study nonparametric estimation of instrumental variable (IV) regressions. Recently, many flexible machine learning methods have been developed for instrumental variable estimation. However, these methods have at least one of the following limitations: (1) restricting the IV regression to be uniquely identified; (2) only obtaining estimation error rates in terms of pseudometrics (e.g., projected norm) rather than valid metrics (e.g., $L_2$ norm); or (3) imposing the so-called closedness condition that requires a certain conditional expectation operator to be sufficiently smooth. In this paper, we present the first method and analysis that can avoid all three limitations, while still permitting general function approximation. Specifically, we propose a new penalized minimax estimator that can converge to a fixed IV solution even when there are multiple solutions, and we derive a strong $L_2$ error rate for our estimator under lax conditions. Notably, this guarantee only needs a widely-used source condition and realizability assumptions, but not the so-called closedness condition. We argue that the source condition and the closedness condition are inherently conflicting, so relaxing the latter significantly improves upon the existing literature that requires both conditions. Our estimator can achieve this improvement because it builds on a novel formulation of the IV estimation problem as a constrained optimization problem.

Citation extraction

55
references
142
in-text mentions
55
distinct cited
7
self-citations
7,911
main-text words

appendix boundary found by appendix_command · 62% of the source is main text. Read the extracted text to check this.

Most heavily cited references

The works this paper leans on most, across its whole bibliography — not restricted to papers in our corpus. Ranked by composite intensity, which combines how often a work is mentioned, how many sections mention it, and how much of that falls in the main text rather than the appendix.

ReferenceIntensityMentionsSectionsMain text
1Dikkala, N., G. Lewis, L. Mackey, and V. Syrgkanis (2020) Minimax estimation of conditional moment models1.000203100%
2Bennett, A., N. Kallus, X. Mao, W. Newey, V. Syrgkanis, and M. Uehara (2022) Inference on strongly identified functionals of weakly identified functions self1.000133100%
3Carrasco, M., J.-P. Florens, and E. Renault (2007) Linear inverse problems in structural econometrics estimation based on spectral decomposition and regularization1.00083100%
4Cavalier, L (2011) Inverse problems in statistics1.00053100%
5Florens, J.-P., J. Johannes, and S. Van Bellegem (2011) Identification and estimation by penalization in nonparametric instrumental regression1.00053100%
6Chen, X. and M. Reiss (2011) On rate optimality for ill-posed inverse problems in econometrics0.92843100%
7Liao, L., Y.-L. Chen, Z. Yang, B. Dai, M. Kolar, and Z. Wang (2020) Provably efficient neural estimation of structural equation models: An adversarial approach0.874122100%
8Kallus, N., X. Mao, and M. Uehara (2021) Causal inference under unmeasured confounding with negative controls: A minimax learning approach self0.8434375%
9Ito, K. and B. Jin (2014) Inverse problems: Tikhonov theory and algorithms, Volume 220.84333100%
10Chen, Q (2021) Robust and optimal estimation for partially linear instrumental variables models with partial identification0.81142100%

Showing the top 10 of 55 scored citations.

Cited by, within the corpus

arXiv econ.EM papers that cite this one, ranked by how heavily they lean on it.

Citing paperIntensityMentionsSections
1Penalized GMM Framework for Inference on Functionals of Nonparametric Instrumental Variable Estimators0.64422
2Long-term Causal Inference Under Persistent Confounding via Data Combination0.58531