When Is Neyman-Orthogonal Inference Feasible? Existence and Relevance

Facundo Arga\~naraz and Juan Carlos Escanciano

Papers from arXiv.org

Abstract: Neyman-orthogonal moments underpin debiased and double machine learning, yet their existence and informativeness are typically presumed. We ask three questions: when do nontrivial orthogonal moments exist, which target directions contain first-order information, and does a proposed moment detect all such directions? Restricted local overidentification characterizes existence under attainability. Effective Fisher information determines relevance; its rank counts informative target directions. A rank sandwich combines moment Jacobians with target changes offset by nuisance perturbations, distinguishing information missed by a moment from information absent from the experiment. In sample selection without covariate exclusions, this yields identification-robust inference under rank deficiency. In latent-variable models, the informative target space is an RKHS. For Gaussian repeated measurements, additional measurements enlarge the class of smoothed distributional targets that are regularly learnable, while CDFs, quantiles, and threshold-policy functionals remain uninformative at the parametric rate for every finite number of measurements.

Date: 2023-03, Revised 2026-09
New Economics Papers: this item is included in nep-des and nep-ecm
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://arxiv.org/pdf/2303.11418 Latest version (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:arx:papers:2303.11418

Access Statistics for this paper

More papers in Papers from arXiv.org
Bibliographic data for series maintained by arXiv administrators ().