Influence Functions: A Unified Differentiation-Based Framework
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Economics. The day in economics: theory, mechanisms and the evidence beneath them.
Ruth: I'm Ruth.
Warren: And I'm Warren, here with Iris.
Ruth: We're looking at the paper: "Influence Functions: A Unified Differentiation-Based Framework".
Title and authors: Ruth: Let's talk about what the paper actually does in summary. Basically, they lay out a direct differentiation-based framework that uses the functional derivative to compute influence functions across all three model types—parametric, nonparametric, and semiparametric >
Warren: And the big payoff here is that for their semiparametric multi-step plug-in estimation, this method automatically gives you locally robust moment functions and an explicit closed-form expression for the adjustment term >
Iris: But they don't rely on classical differentiability here. They use distributional derivatives to handle those non-smooth moment conditions, which is a big deal because that opens up more kinds of models than just the smooth ones >
The paper's summary: Ruth: So, the core takeaway here is that they derive easily verifiable conditions for local robustness in both parametric and semiparametric models using this framework >
Warren: They also revisit Stein's question about when you can estimate a finite dimensional parameter efficiently without knowing an infinite dimensional nuisance parameter, and they show that local robustness answers that question directly >
Iris: That means the framework ensures the asymptotic variance stays the same whether you evaluate your moment function at the true value or at some consistent estimator of that nuisance parameter, which is what local robustness achieves >
The paper's improvements: Ruth: Then they tackle a major debate: joint versus plug-in estimation. They use an orthogonal decomposition of the moment function to directly compare these two approaches >
Warren: They identify a "companion parameter," which is this kind of parameter you need to identify for the joint estimation condition, and they show that an oracle moment function, called morc-LR beta, weakly dominates the locally robust version of the original moment function >
Iris: That dominance is what establishes verifiable sufficient conditions for efficiency equivalence between those two approaches in the semiparametric setting >
Conclusion: Ruth: So to wrap up, this paper with its unified differentiation-based framework provides a way to derive influence functions and establish verifiable conditions for efficiency equivalence between joint and plug-in estimation across parametric, nonparametric, and semiparametric models >
Warren: They construct an oracle-based locally robust moment function that weakly dominates the local robust version of the original moment function, which gives us a principled benchmark for multi-step inference when nuisance parameters are over-identified >
Iris: And they nail it by diagnosing whether plug-in procedures remain adaptive in cases like the average treatment effect, or if they lose efficiency due to a lack of local robustness in situations like the average treatment effect on the treated >
Rutgers University
econ.EM
Submitted: 2025-01-25
Updated: 2026-11-18
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
The gist: This paper introduces a direct differentiation-based framework that unifies the derivation of influence functions across parametric, nonparametric…
Key concepts
- Unified Derivation Framework
- This core method uses direct differentiation to find the influence function for any model type (parametric, nonparametric, or semiparametric). It shows that the Riesz representer of the functional derivative is found by projecting the identification function onto mean-zero functions, leading to a common algebraic form for all influence functions.
- Local Robustness
- Local robustness ensures that the asymptotic variance remains unchanged whether you use the true nuisance parameter or an estimator of it. The framework identifies an 'oracle-based' moment function that weakly dominates the original robust version, providing a reliable benchmark for multi-step inference.
- Joint vs. Plug-in Comparison
- The method uses orthogonal decomposition of moment functions to compare joint estimation with sequential plug-in estimation. It identifies a 'companion parameter' and establishes verifiable conditions under which plug-in estimation achieves the same asymptotic variance as joint estimation, even when nuisance parameters are over-identified.
Terminology
Summary
This paper introduces a direct differentiation-based framework that unifies the derivation of influence functions across parametric, nonparametric, and semiparametric models to establish verifiable sufficient conditions for efficiency equivalence between joint and plug-in estimation <ref:2501. p. 0 Influence Function: Local Robustness and
Efficiency
Xiye Yang and Ruonan Xu
Department of Economics
Rutgers University
May, 2026
Abstract
This paper introduces a direct differentiation-based framework that unifies
the derivation of influence functions across parametric, nonparametric, and
semiparametric models. We show that the Riesz representer of the functional
derivative is obtained by orthogonally projecting the identification function onto
the subspace of mean-zero functions.Where the paper says it
Unified Derivation Framework
The core contribution is a direct differentiation-based framework that unifies the derivation of influence functions across parametric, nonparametric, and semiparametric models <ref:2501.15307#pg3>. This approach shows that the Riesz representer of the functional derivative is obtained by orthogonally projecting the identification function onto the subspace of mean-zero functions <ref:2501. p. 0 Influence Function: Local Robustness and
Efficiency
Xiye Yang and Ruonan Xu
Department of Economics
Rutgers University
May, 2026
Abstract
This paper introduces a direct differentiation-based framework that unifies
the derivation of influence functions across parametric, nonparametric, and
semiparametric models. We show that the Riesz representer of the functional
derivative is obtained by orthogonally projecting the identification function onto
the subspace of mean-zero functions. Consequently, the influence function
emerges as a linear transformation of this centered moment function. The
approach extends seamlessly to infinite-dimensional parameters, revealing
a common algebraic form for influence functions across both finite- and
infinite-dimensional parameters.Where the paper says it
Where the paper says it
Semiparametric Application
Applied to semiparametric multi-step plug-in estimation, the method automatically yields locally robust moment functions and provides an explicit closed-form expression for the adjustment term <ref:2501.15307#pg4>. This procedure does not require classical differentiability, employing distributional derivatives to accommodate non-smooth moment conditions <ref:2501. p. 2 Crucially, this procedure
does not require classical differentiability. By employing distributional derivatives,
the method accommodates non-smooth moment conditions, with the conditional
expectation serves as smoothing operator. By employing distributional derivatives,
the method accommodates non-smooth moment conditions, with the conditional
expectation serves as smoothing operator. Using this framework, we derive easily verifiable conditions for local robustness in
both parametric and semiparametric models and clarify its role in adaptive estimation.Where the paper says it
Where the paper says it
Local Robustness and Efficiency
The paper establishes that local robustness ensures the asymptotic variance remains invariant to whether the moment function is evaluated at the true or a consistent estimator of the nuisance parameter, answering Stein’s question <ref:2501.15307#pg4>. The framework identifies an oracle-based locally robust moment function that weakly dominates the locally robust version of the original moment function in terms of asymptotic variance <ref:2501. p. 2 We employ an
orthogonal decomposition of the moment function to decouple the joint optimization
problem, rendering it structurally comparable to sequential plug-in estimation. In a
general setting where the nuisance parameter may be over-identified, we identify an
oracle-based locally robust moment function for the plug-in estimator that weakly
dominates the locally robust version of the original moment funciton in terms of
asymptotic variance. In a
general setting where the nuisance parameter may be over-identified, we identify an
oracle-based locally robust moment function for the plug-in estimator that weakly
dominates the locally robust version of the original moment funciton in terms of
asymptotic variance. More importantly, we also establish easily verifiable sufficient
conditions for the efficiency equivalence between the joint and the plug-in approaches.Where the paper says it
Where the paper says it
Joint vs. Plug-in Comparison
The orthogonal decomposition of the moment function facilitates a direct comparison between joint and sequential plug-in estimation <ref:2501.15307#pg5>. The condition Ppj:qrmβ,j s“0 identifies a parameter βj, which is termed a companion parameter <ref:2501. p. 4 3) holds for all x in the support of X, we can obtain a conditional moment
condition PrmγpZ, γq | Xs “ 0, without the appearance of generalized functions. Intuitively, this identifies the infinite-dimensional parameter γp¨q, which is a function
of x, as γp¨, PpZ|Xqq. We employ an
orthogonal decomposition of the moment function to decouple the joint optimization
problem, rendering it structurally comparable to sequential plug-in estimation. In a
general setting where the nuisance parameter may be over-identified, we identify an
oracle-based locally robust moment function for the plug-in estimator that weakly
dominates the locally robust version of the original moment funciton in terms of
asymptotic variance.Where the paper says it
Where the paper says it
Treatment Effect Model Diagnostics
The framework precisely diagnoses when plug-in procedures remain adaptive (as in the ATE) and when they incur efficiency losses due to failure of local robustness (as in the ATT) <ref:2501. p. 2 Section 3 develops the differentiation-based derivation across parametric,
nonparametric, and semiparametric models. Section 4 establishes the link between
local robustness, adaptive estimation, and efficiency, and applies the framework to a
range of treatment effect estimators. Section 3 develops the differentiation-based derivation across parametric,
nonparametric, and semiparametric models. Section 4 establishes the link between
local robustness, adaptive estimation, and efficiency, and applies the framework to a
range of treatment effect estimators. We employ an
orthogonal decomposition of the moment function to decouple the joint optimization
problem, rendering it structurally comparable to sequential plug-in estimation. In a
general setting where the nuisance parameter may be over-identified, we identify an
oracle-based locally robust moment function for the plug-in estimator that weakly
dominates the locally robust version of the original moment funciton in terms of
asymptotic variance.Where the paper says it
Where the paper says it
Where the paper says it
Conclusion
In this paper, we introduce a direct differentiation-based framework for deriving influence functions across parametric, nonparametric, and semiparametric models <ref:2501.15307#pg5>. By grounding the derivation in the functional functional derivative and its integral representation, we demonstrate that the influence function is obtained by centering the identification function <ref:2501. p. 0 Influence Function: Local Robustness and
Efficiency
Xiye Yang and Ruonan Xu
Department of Economics
Rutgers University
May, 2026
Abstract
This paper introduces a direct differentiation-based framework that unifies
the derivation of influence functions across parametric, nonparametric, and
semiparametric models. We show that the Riesz representer of the functional
derivative is obtained by orthogonally projecting the identification function onto
the subspace of mean-zero functions. Crucially, this projection reduces to a simple centering operation: subtracting
the unconditional or conditional expectation from the identification function (see
Example 1). The framework extends seamlessly to infinite-dimensional parameters. By employing distributional derivatives,
the method accommodates non-smooth moment conditions, with the conditional
expectation serves as smoothing operator. Using this framework, we derive easily verifiable conditions for local robustness in
both parametric and semiparametric models and clarify its role in adaptive estimation. By employing distributional derivatives,
the method accommodates non-smooth moment conditions, with the conditional
expectation serves as smoothing operator. Using this framework, we derive easily verifiable conditions for local robustness in
both parametric and semiparametric models and clarify its role in adaptive estimation. The influence function ν9pPq admits a multiplicative representation that extends
naturally to nonparametric settings (see Section 3. This structure underpins
our comparison of joint and sequential plug-in estimation in Section 4. The influence function ν9pPq admits a multiplicative representation that extends
naturally to nonparametric settings (see Section 3. This structure underpins
our comparison of joint and sequential plug-in estimation in Section 4. In a
general setting where the nuisance parameter may be over-identified, we identify an
oracle-based locally robust moment function for the plug-in estimator that weakly
dominates the locally robust version of the original moment funciton in terms of
asymptotic variance. More importantly, we also establish easily verifiable sufficient
conditions for the efficiency equivalence between the joint and the plug-in approaches. In a
general setting where the nuisance parameter may be over-identified, we identify an
oracle-based locally robust moment function for the plug-in estimator that weakly
dominates the locally robust version of the original moment funciton in terms of
asymptotic variance. More importantly, we also establish easily verifiable sufficient
conditions for the efficiency equivalence between the joint and the plug-in approaches. We employ an
orthogonal decomposition of the moment function to decouple the joint optimization
problem, rendering it structurally comparable to sequential plug-in estimation. In a
general setting where the nuisance parameter may be over-identified, we identify an
oracle-based locally robust moment function for the plug-in estimator that weakly
dominates the locally robust version of the original moment funciton in terms of
asymptotic variance. We employ an
orthogonal decomposition of the moment function to decouple the joint optimization
problem, rendering it structurally comparable to sequential plug-in estimation. In a
general setting where the nuisance parameter may be over-identified, we identify an
oracle-based locally robust moment function for the plug-in estimator that weakly
dominates the locally robust version of the original moment funciton in terms of
asymptotic variance.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
References
Ackerberg, D., X. Chen, J. Hahn, and Z. p. 1 While tangent-space method (Tsiatis, 2006) is mathematical elegant, it
can be abstract and opaque for complex models. As Hahn (1998) notes, the general
frameworks of Newey (1990, 1994) and Bickel et al. in establishing asymptotic linearity, semiparametric efficiency, and local robustness. Seminal work by Bickel et al.Where the paper says it
Where the paper says it
Ai, C. and X. Chen (2003): “Efficient estimation of models with conditional moment restrictions containing unknown functions,” Econometrica, 71, 1795–1843 <ref:2501.15307#pg43>.
——— (2012): “The semiparametric efficiency bound for models of sequential moment restrictions containing unknown functions,” Journal of Econometrics, 170, 442–457 <ref:2501.15307#pg44>.
Andrews, D. p. 5 1
The influence function is defined similarly across methods, but the challenge lies in
deriving its explicit form. For example, Newey (1994) uses a score-based expression,
which is extended to three-step settings by Hahn and Ridder (2013); Tsiatis (2006)
relies on tangent space methods and Ichimura and Newey (2022) propose a least
1We do not specify the type of functional differentiability mainly because the asymptotic negligibility
of the remainder terms is guaranteed by the stochastic equicontinuity condition (Andrews, 1994;
Newey, 1994; Newey and McFadden, 1994). 1
The influence function is defined similarly across methods, but the challenge lies in
deriving its explicit form. For example, Newey (1994) uses a score-based expression,
which is extended to three-step settings by Hahn and Ridder (2013); Tsiatis (2006)
relies on tangent space methods and Ichimura and Newey (2022) propose a least
1We do not specify the type of functional differentiability mainly because the asymptotic negligibility
of the remainder terms is guaranteed by the stochastic equicontinuity condition (Andrews, 1994;
Newey, 1994; Newey and McFadden, 1994).Where the paper says it
Where the paper says it
Bang, H. and J. M. Robins (2005): “Doubly robust estimation in missing data and causal inference models,” Biometrics, 61, 962–973 <ref:2501.15307#pg44>.
Bickel, P. J.: “On adaptive estimation,” The Annals of Statistics, 647–671 <ref:2501. p. 0 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Bickel, P. J., C. A. Klaassen, Y. Ritov, and J. A. Wellner (1993): Efficient and Adaptive Estimation for Semiparametric Models, Johns Hopkins University Press <ref:2501. p. 2 By employing distributional derivatives,
the method accommodates non-smooth moment conditions, with the conditional
expectation serves as smoothing operator. Using this framework, we derive easily verifiable conditions for local robustness in
both parametric and semiparametric models and clarify its role in adaptive estimation.Where the paper says it
Cao, W., A. A. Tsiatis, and M. p. 1 Stochastic expansions of the estimation error are often algebraically
cumbersome. While tangent-space method (Tsiatis, 2006) is mathematical elegant, it
can be abstract and opaque for complex models. Influence Function: Local Robustness and
Efficiency
Xiye Yang and Ruonan Xu
Department of Economics
Rutgers University
May, 2026
Abstract
This paper introduces a direct differentiation-based framework that unifies
the derivation of influence functions across parametric, nonparametric, and
semiparametric models. We show that the Riesz representer of the functional
derivative is obtained by orthogonally projecting the identification function onto
the subspace of mean-zero functions.Where the paper says it
Where the paper says it
Chen, X. and M. Reiss (2011): “On Rate Optimality for Ill-Posed Inverse Problems in Econometrics,” Econometric Theory, 27, 497–521 <ref:2501. p. 0 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Chernozhukov, V. p. 1 (1993) uses it to characterize semiparametric efficiency
bounds, while recent literature (e. , Chernozhukov et al. (1993) uses it to characterize semiparametric efficiency
bounds, while recent literature (e. , Chernozhukov et al. While tangent-space method (Tsiatis, 2006) is mathematical elegant, it
can be abstract and opaque for complex models. As Hahn (1998) notes, the general
frameworks of Newey (1990, 1994) and Bickel et al. While tangent-space method (Tsiatis, 2006) is mathematical elegant, it
can be abstract and opaque for complex models. As Hahn (1998) notes, the general
frameworks of Newey (1990, 1994) and Bickel et al. 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Chernozhukov, V., J. C. Escanciano, H. Ichimura, W. p. 1 Hahn and Ridder (2013)
extend Newey (1994) to three-step settings but restrict the analysis to differentiable
functions. The method by Ichimura and Newey (2022) ultimately require solving a
least square projection problem, where the nuisance moment function is scalar-valued. The
approach extends seamlessly to infinite-dimensional parameters, revealing
a common algebraic form for influence functions across both finite- and
infinite-dimensional parameters. Applied to semiparametric multi-step plug-in
estimation, our method automatically yields locally robust moment functions
and provides an explicit closed-form expression for the adjustment term.Where the paper says it
Where the paper says it
Darolles, S., Y. Fan, J.-P. Florens, and E. Renault (2011): “Nonparametric instrumental regression,” Econometrica, 79, 1541–1565 <ref:2501. p. 3 For example,
we have Pp2|1q “ PpY |Xq
in linear regression model. In instrumental variables (IV)
estimation, we get Pp2|1q “PpY,X|Wq
.Where the paper says it
Fan, J. and I. Gijbels (1996): Local Polynomial Modelling and Its Applications, vol. p. 3 , βˆ “ βpPnq. The asymptotic properties of βˆ then readily follow from the
well-established empirical process theory (van der Vaart and Wellner, 1996). 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Where the paper says it
Gallier, J.: Geometric Methods and Applications: for Computer Science and Engineering, vol. p. 4 We can write γx “ γpP, δ:
x
q following (2. 2 demonstrates
that different nonparametric methods use different approximations to the generalized
function δ
:
x
. 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Where the paper says it
Hahn, J.: “On the Role of the Propensity Score in Efficient Semiparametric Estimation of Average Treatment Effects,” Econometrica, 66, 315–331 <ref:2501. p. 2 Section 3 develops the differentiation-based derivation across parametric,
nonparametric, and semiparametric models. Section 4 establishes the link between
local robustness, adaptive estimation, and efficiency, and applies the framework to a
range of treatment effect estimators.Where the paper says it
Hahn, J. and G. Ridder (2013): “Asymptotic variance of semiparametric estimators with generated regressors,” Econometrica, 81, 315–340 <ref:2501. p. 5 1
The influence function is defined similarly across methods, but the challenge lies in
deriving its explicit form. For example, Newey (1994) uses a score-based expression,
which is extended to three-step settings by Hahn and Ridder (2013); Tsiatis (2006)
relies on tangent space methods and Ichimura and Newey (2022) propose a least
1We do not specify the type of functional differentiability mainly because the asymptotic negligibility
of the remainder terms is guaranteed by the stochastic equicontinuity condition (Andrews, 1994;
Newey, 1994; Newey and McFadden, 1994).Where the paper says it
Hall, P. and J. L. Horowitz (2005): “Nonparametric Methods for Inference in the Presence of Instrumental Variables,” Annals of Statistics, 33, 2904–2929 <ref:2501. p. 0 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Huber, P. J.: Robust Statistics, John Wiley & Sons, 1st ed <ref:2501. p. 0 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Ichimura, H. p. 1 Hahn and Ridder (2013)
extend Newey (1994) to three-step settings but restrict the analysis to differentiable
functions. The method by Ichimura and Newey (2022) ultimately require solving a
least square projection problem, where the nuisance moment function is scalar-valued. Influence Function: Local Robustness and
Efficiency
Xiye Yang and Ruonan Xu
Department of Economics
Rutgers University
May, 2026
Abstract
This paper introduces a direct differentiation-based framework that unifies
the derivation of influence functions across parametric, nonparametric, and
semiparametric models. We show that the Riesz representer of the functional
derivative is obtained by orthogonally projecting the identification function onto
the subspace of mean-zero functions.Where the paper says it
Where the paper says it
Imbens, G. W., W. K. Newey, and G. p. 1 While tangent-space method (Tsiatis, 2006) is mathematical elegant, it
can be abstract and opaque for complex models. As Hahn (1998) notes, the general
frameworks of Newey (1990, 1994) and Bickel et al.Where the paper says it
Newey, W. K.: “Semiparametric Efficiency Bounds,” Journal of Applied Econometrics, 5, 99–135 <ref:2501. p. 0 Applied to semiparametric multi-step plug-in
estimation, our method automatically yields locally robust moment functions
and provides an explicit closed-form expression for the adjustment term. Finally,
we leverage this framework to revisit the joint versus plug-in estimation debate,
establishing verifiable sufficient conditions for their semiparametric efficiency
equivalence even when nuisance parameters are over-identified.Where the paper says it
——— (1994): “The Asymptotic Variance of Semiparametric Estimators,” Econometrica, 1349–1382 <ref:2501. p. 7 Linearity guarantees the existence of asymptotically linear estimators. Square
integrability ensures finite variance for estimators with the parametric convergence
rate; for nonparametric estimators, which have slower convergence rates, this may be
adjusted (see Section 3.Where the paper says it
Newey, W. K. and D. McFadden (1994): “Large Sample Estimation and Hypothesis Testing,” in Handbook of Econometrics, ed. p. 5 1
The influence function is defined similarly across methods, but the challenge lies in
deriving its explicit form. For example, Newey (1994) uses a score-based expression,
which is extended to three-step settings by Hahn and Ridder (2013); Tsiatis (2006)
relies on tangent space methods and Ichimura and Newey (2022) propose a least
1We do not specify the type of functional differentiability mainly because the asymptotic negligibility
of the remainder terms is guaranteed by the stochastic equicontinuity condition (Andrews, 1994;
Newey, 1994; Newey and McFadden, 1994). 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Where the paper says it
Newey, W. K. and J. L. Powell (2003): “Instrumental variable estimation of nonparametric models,” Econometrica, 71, 1565–1578 <ref:2501. p. 2 Section 3 develops the differentiation-based derivation across parametric,
nonparametric, and semiparametric models. Section 4 establishes the link between
local robustness, adaptive estimation, and efficiency, and applies the framework to a
range of treatment effect estimators.Where the paper says it
Oaxaca, R.: “Male-female wage differentials in urban labor markets,” International economic review, 693–709 <ref:2501. p. 0 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Pavlíková, S. and D. Ševčovič (2023): “On the Moore-Penrose pseudo-inversion of block symmetric matrices and its application in the graph theory,” Linear Algebra and its Applications, 673, 280–303 <ref:2501.15307#pg43>.
Robins, J. M. and A. Rotnitzky (2001): “Comment on Inference for semiparametric models: Some questions and an answer, by P. p. 0 Influence Function: Local Robustness and
Efficiency
Xiye Yang and Ruonan Xu
Department of Economics
Rutgers University
May, 2026
Abstract
This paper introduces a direct differentiation-based framework that unifies
the derivation of influence functions across parametric, nonparametric, and
semiparametric models. We show that the Riesz representer of the functional
derivative is obtained by orthogonally projecting the identification function onto
the subspace of mean-zero functions. in establishing asymptotic linearity, semiparametric efficiency, and local robustness. Seminal work by Bickel et al. 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Where the paper says it
Where the paper says it
Robins, J. M., A. Rotnitzky, and L. P. Zhao (1994): “Estimation of regression coefficients when some regressors are not always observed,” Journal of the American statistical Association, 89, 846–866 <ref:2501.15307#pg43>.
Rudin, W.: Real and Complex Analysis, McGraw-HiII Book Company, 3rd ed <ref:2501.15307#pg43>.
Sobolev, S. L.: Some applications of functional analysis in mathematical physics, vol. p. 1 Stochastic expansions of the estimation error are often algebraically
cumbersome. While tangent-space method (Tsiatis, 2006) is mathematical elegant, it
can be abstract and opaque for complex models. 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Where the paper says it
Stein, C.: “Efficient nonparametric testing and estimation,” in Proceedings of the Third Berkeley Symposium on Mathematical Statistics and Probability, Volume 1: Contributions to the Theory of Statistics, University of California Press, vol. 3, 187–196 <ref:2501. p. 0 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
Tsiatis, A. p. 1 Stochastic expansions of the estimation error are often algebraically
cumbersome. While tangent-space method (Tsiatis, 2006) is mathematical elegant, it
can be abstract and opaque for complex models. We equip L2pPq with the inner product
x¨, ¨y. Following standard semiparametric theory, we define functional derivatives via
pathwise differentiability (Hampel, 1974; Huber, 1984; Bickel et al.Where the paper says it
Where the paper says it
Tukey, J. W.: “A Survey of Sampling From Contaminated Distributions,” Contributions to Probability and Statistics, 448–485 <ref:2501. p. 0 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. It plays a central role
1
arXiv:2501.Where the paper says it
van der Vaart, A. p. 3 , βˆ “ βpPnq. The asymptotic properties of βˆ then readily follow from the
well-established empirical process theory (van der Vaart and Wellner, 1996). , βˆ “ βpPnq. The asymptotic properties of βˆ then readily follow from the
well-established empirical process theory (van der Vaart and Wellner, 1996).Where the paper says it
Where the paper says it
Additional Discussions on ATE
Example 10 (Third-order moment). The third-order moment β:“ PrpZ ´ γq 3 s is not locally robust to the mean γ:“ PrZs <ref:2501. p. 0 Keywords: Semiparametrics, influence function, locally robust, efficiency. 1 Introduction
The influence function is a foundational object in modern statistics and econometrics,
characterizing the first-order effect of an infinitesimal perturbation to the datagenerating distribution on a target parameter or estimator. matrix of full row rank, and assume A PrBνms is invertible, which constitutes a strong
identification condition. Although the implicit function νpPq typically lacks a closedform expression in this case, its influence function ν9pPq admits an explicit analytic
representation.Where the paper says it
Where the paper says it
Example 9 (IPW and Regression)
For the regression-based estimator, if the conditional outcome functions τ1pXq and τ0pXq were known, the asymptotic variance would reduce to Varpτ1pXq´τ0pXq´τATEq <ref:2501.15307#pg47>. This quantity is strictly smaller than VarpY p1q ´ Y p0qq, the variance that would obtain if both potential outcomes were directly observable for each unit <ref:2501.15307#pg47>.
Example 9 (AIPW)
The doubly robust AIPW moment function coincides exactly with the AIPW specification: morc ATE = mAIPW ATE <ref:2501.15307#pg47>. The only non-companion nuisance parameter is πpXq, because mATE,2 = 0 in this decomposition, its projection onto the propensity score moment vanishes <ref:2501.15307#pg47>.
Example 9 (ATT)
The ATT admits distinct locally robust and oracle moment functions: mLR ATT and morc ATT <ref:2501.15307#pg47>. The variance increase occurs precisely because morc ATT is not locally robust <ref:2501. p. 2 We employ an
orthogonal decomposition of the moment function to decouple the joint optimization
problem, rendering it structurally comparable to sequential plug-in estimation. In a
general setting where the nuisance parameter may be over-identified, we identify an
oracle-based locally robust moment function for the plug-in estimator that weakly
dominates the locally robust version of the original moment funciton in terms of
asymptotic variance.Where the paper says it
In this paper, we introduce a direct differentiation-based framework for deriving influence functions across parametric, nonparametric, and semiparametric models to establish verifiable sufficient conditions for efficiency equivalence between joint and plug-in estimation <ref:2501.15307#pg41>. This algebraic characterization replaces the traditional reliance on tangent-space projections, score-based Riesz representation arguments, and ad hoc guess-and-verify constructions <ref:2501. p. 5 1
The influence function is defined similarly across methods, but the challenge lies in
deriving its explicit form. For example, Newey (1994) uses a score-based expression,
which is extended to three-step settings by Hahn and Ridder (2013); Tsiatis (2006)
relies on tangent space methods and Ichimura and Newey (2022) propose a least
1We do not specify the type of functional differentiability mainly because the asymptotic negligibility
of the remainder terms is guaranteed by the stochastic equicontinuity condition (Andrews, 1994;
Newey, 1994; Newey and McFadden, 1994). This work establishes a cohesive analytical framework, grounded in functional
differentiation, for the unified derivation of influence functions across parametric and
nonparametric specifications. We define the target parameter as a statistical functional of the probability distribution P and compute its functional derivative under
local perturbations of the form P
ϵ “ p1 ´ ϵqP ` ϵQ.Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Direct differentiation-based framework for influence functions can be used to derive closed-form expressions across parametric, nonparametric, and semiparametric models without relying on tangent-space methods or ad hoc guess-and-verify constructions.
-
The system can automatically yield locally robust moment functions in semiparametric multi-step plug-in estimation by computing conditional expectations of partial derivatives using distributional derivatives instead of classical differentiability.
-
The framework enables verifiable sufficient conditions for the semiparametric efficiency equivalence between joint and plug-in estimation, establishing a principled benchmark for comparing these procedures.
-
The system can construct an oracle-based locally robust moment function that weakly dominates the local robustness version of the original moment function, providing a principled benchmark for multi-step inference when nuisance parameters are over-identified.
-
The AI system can diagnose whether plug-in procedures remain adaptive (as in the ATE) or incur efficiency losses due to failure of local robustness (as in the ATT) by checking if morc ATT is locally robust.
Abstract
This paper introduces a direct differentiation-based framework that unifies the derivation of influence functions across parametric, nonparametric, and semiparametric models. We show that the Riesz representer of the functional derivative is obtained by orthogonally projecting the identification function onto the subspace of mean-zero functions, so the influence function is a linear transformation of the moment function. This approach extends seamlessly to infinite-dimensional parameters and reveals a common algebraic form across both finite- and infinite-dimensional cases. It demonstrates that the slower nonparametric convergence rate arises from approximating the infeasible Dirac delta function. The same framework further applies to semiparametric multi-step plug-in estimation, where it automatically yields locally robust moment functions. Crucially, by employing distributional derivatives, the method accommodates non-smooth moment conditions without imposing classical differentiability.
Related papers
- Stationarity with Occasionally Binding Constraints
- Backtesting High-Dimensional Expected Shortfall
- Higher-Order Neyman Orthogonality in Moment-Condition Models
- Online Bayesian Inference for a Dirichlet Process Mixture of Binary Logit Models
- Debiased Conditional Independence Tests for Causal Effects
- Sharp Testable Implications of Encouragement Designs