pith. sign in

arxiv: 2606.29833 · v1 · pith:QMFP54QQnew · submitted 2026-06-29 · 💰 econ.EM

Sensitivity, Informativeness, and Misspecification in GMM Estimation

Fangzhou Yu , Seojeong Lee This is my paper

Pith reviewed 2026-06-30 04:03 UTC · model grok-4.3

classification 💰 econ.EM
keywords GMM estimationmisspecificationinformativenesssensitivityHansen J-testpseudo-true valuesweight matrixasymptotic variance
0
0 comments X

The pith

Misspecification can lower the share of GMM estimator variance explained by the moments even when the J-test does not reject.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops sensitivity and informativeness diagnostics for GMM that are evaluated at pseudo-true values so they remain valid under misspecification. The central object is an informativeness measure Delta that equals the fraction of an estimator's asymptotic variance attributable to sampling variation in the moments. Delta reaches one under correct specification but can fall below one when the model is misspecified, and this drop can occur even if the Hansen J-test fails to reject. The paper also shows that the choice of weight matrix creates a trade-off: the optimal matrix improves classical efficiency but reduces Delta by introducing variance the moments do not explain, while simpler matrices largely avoid that loss. Applications to three standard models demonstrate that misspecification changes sensitivity orderings and that Delta can flag structural-efficiency problems the J-test misses.

Core claim

The informativeness Delta measures the share of an estimator's asymptotic variance explained by sampling variation in the moments. It equals one under correct specification and can fall below one under misspecification even when the Hansen J-test does not reject. The associated sensitivity matrix nests the Andrews-Gentzkow-Shapiro matrix under correct specification. Influence-function representations are derived for one-step, two-step, iterated, and continuously updating GMM. In minimum-distance problems the optimal weight matrix adds estimator variance that the moments do not explain, lowering Delta, whereas simpler weights largely preserve it.

What carries the argument

The informativeness measure Delta, defined as the share of asymptotic variance attributable to moment sampling variation and evaluated at pseudo-true values.

If this is right

  • In minimum-distance estimation the optimal weight matrix reduces informativeness by adding variance the moments do not explain.
  • Simpler weight matrices largely avoid that variance addition and therefore maintain higher Delta.
  • Misspecification can reorder the sensitivity rankings of different parameters.
  • Delta can detect structural-efficiency losses that the Hansen J-test does not flag.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Reporting Delta alongside the J-test would give practitioners a direct gauge of how much of the reported precision is actually coming from the data moments.
  • The same informativeness logic could be applied to other moment-based estimators such as IV or minimum-distance problems outside GMM.
  • When Delta is low, model refinement or additional moments may be warranted even if the J-test passes.
  • The efficiency-informativeness trade-off suggests that the conventional preference for the optimal weight matrix should be weighed against the loss in Delta in each application.

Load-bearing premise

The derivations assume well-defined pseudo-true values exist and that standard regularity conditions hold for the asymptotic expansions of the GMM estimators.

What would settle it

A simulation or empirical example in which the model is misspecified, the J-test does not reject, yet the computed Delta remains exactly equal to one.

Figures

Figures reproduced from arXiv: 2606.29833 by Fangzhou Yu, Seojeong Lee.

Figure 1
Figure 1. Figure 1: Informativeness of the influence function for one-step ( [PITH_FULL_IMAGE:figures/full_fig_p021_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: AGS and MRS sensitivities of the average markup with respect to the excluded instru [PITH_FULL_IMAGE:figures/full_fig_p024_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Misspecification-robust informativeness ∆bk of every BPP parameter under OMD (hollow circles) and DWMD (filled circles), sorted by the OMD value. The thin line marks the fixed-weight bound ∆k = 1, attained exactly by EWMD. 4.3 Income and Democracy Our third application revisits the income-and-democracy regressions of Acemoglu, Johnson, Robin￾son, and Yared (2008) (AJRY), a leading example of dynamic panel … view at source ↗
Figure 4
Figure 4. Figure 4: Hansen J statistics and informativeness ∆bγ along the iteration paths in the Acemoglu, Johnson, Robinson, and Yared (2008) specification (five-year panel, full sample). Circles: the uncentered efficient-weight iteration; squares: the fully centered iteration, with the weight equal to the inverse cluster covariance updated at each step. Both start from the Arellano-Bond one-step estimate (s = 1) and converg… view at source ↗
Figure 5
Figure 5. Figure 5: Iteration paths of difference GMM in AJRY specification (five-year panel, full sam [PITH_FULL_IMAGE:figures/full_fig_p030_5.png] view at source ↗
read the original abstract

This paper develops misspecification-robust sensitivity and informativeness diagnostics for GMM estimators, evaluated at pseudo-true values. The sensitivity matrix nests that of Andrews, Gentzkow, and Shapiro (2017) under correct specification. The informativeness $\Delta$ measures the share of an estimator's asymptotic variance explained by sampling variation in the moments, a notion of structural efficiency that equals one under correct specification and can fall below one under misspecification, even when the Hansen $J$-test does not reject. We derive influence-function representations for one-step, two-step, iterated, and continuously updating GMM. We show that in minimum-distance estimation, estimating the optimal weight matrix adds estimator variance that the moments do not explain, lowering informativeness, while simpler weight matrices largely avoid it. The choice of weight matrix therefore involves a trade-off between classical efficiency and informativeness. In applications to the automobile demand model of Berry, Levinsohn, and Pakes (1995), the consumption insurance model of Blundell, Pistaferri, and Preston (2008), and the income-and-democracy regressions of Acemoglu, Johnson, Robinson, and Yared (2008), misspecification reorders sensitivity rankings, simpler weights preserve the informativeness that the optimal weight loses, and $\Delta$ detects structural-efficiency losses that the $J$-test does not.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 3 minor

Summary. The paper develops misspecification-robust sensitivity and informativeness diagnostics for GMM estimators evaluated at pseudo-true values. The sensitivity matrix nests that of Andrews, Gentzkow, and Shapiro (2017) under correct specification. The informativeness measure Δ quantifies the share of an estimator's asymptotic variance explained by sampling variation in the moments; it equals one under correct specification and can fall below one under misspecification even when the Hansen J-test does not reject. Influence-function representations are derived for one-step, two-step, iterated, and continuously updating GMM. Applications to the BLP automobile demand model, Blundell-Pistaferri-Preston consumption insurance model, and Acemoglu-Johnson-Robinson-Yared income-democracy regressions illustrate that misspecification reorders sensitivity rankings, simpler weights preserve informativeness lost with optimal weights, and Δ detects structural-efficiency losses missed by the J-test.

Significance. If the derivations hold, the paper supplies useful misspecification-robust tools that extend sensitivity analysis and introduce a direct decomposition-based measure of structural efficiency. The explicit influence-function derivations for multiple GMM variants under standard regularity conditions (pseudo-true values, differentiability, positive-definiteness) are a clear strength and support implementation. The applications demonstrate concrete consequences for sensitivity rankings and the efficiency-informativeness trade-off in weight-matrix choice. These contributions are proportionate to the claims and add diagnostic value beyond the J-test when the central expressions are free of post-hoc adjustments.

minor comments (3)
  1. [§2.2] §2.2, definition of Δ: the decomposition is presented as direct from the asymptotic variance; a short remark confirming that no additional normalization is imposed would aid readers.
  2. [Table 2] Table 2 (BLP application): the reported Δ values for optimal vs. identity weighting would benefit from a column showing the associated standard errors or bootstrap intervals to assess precision of the informativeness gap.
  3. [§6.3] §6.3 (income-democracy application): the text notes reordering of sensitivity rankings but does not report the numerical sensitivity matrix entries; adding these (or a supplementary table) would make the reordering claim easier to verify.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the careful and positive assessment of the manuscript. The summary accurately reflects the paper's contributions on misspecification-robust sensitivity measures and the informativeness statistic Δ for GMM. We appreciate the recommendation for minor revision and the recognition that the influence-function derivations and applications add diagnostic value beyond the J-test.

Circularity Check

0 steps flagged

No significant circularity; Δ defined from standard asymptotic decomposition

full rationale

The paper defines the informativeness Δ explicitly as the share of asymptotic variance attributable to moment sampling variation, derived via standard influence-function expansions under regularity conditions (pseudo-true values, differentiability, positive-definiteness of Jacobian and weight matrix). These conditions are the usual ones for GMM asymptotics under misspecification and are not derived from or equivalent to the paper's target results. The nesting of the Andrews-Gentzkow-Shapiro sensitivity matrix is a direct algebraic consequence under correct specification, not a self-referential fit. No self-citations are load-bearing; no parameter is fitted to data and then relabeled a prediction; no uniqueness theorem or ansatz is smuggled via prior work by the same authors. The derivation chain is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The paper rests on standard GMM asymptotic theory evaluated at pseudo-true values and introduces Δ as a new diagnostic quantity.

axioms (1)
  • domain assumption Standard regularity conditions for GMM asymptotics hold at the pseudo-true values.
    Required for the influence-function representations of one-step, two-step, iterated, and continuously updating GMM.
invented entities (1)
  • Informativeness measure Δ no independent evidence
    purpose: Quantifies the share of asymptotic variance explained by moment sampling variation under misspecification.
    New diagnostic introduced by the paper; no independent evidence outside the derivation is provided.

pith-pipeline@v0.9.1-grok · 5772 in / 1262 out tokens · 56414 ms · 2026-06-30T04:03:34.883195+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

49 extracted references

  1. [1]

    and Segal, Lewis M

    Altonji, Joseph G. and Segal, Lewis M. , journal =. Small-sample bias in. 1996 , publisher =

    1996

  2. [2]

    Journal of Econometrics , volume =

    The large sample behaviour of the generalized method of moments estimator in misspecified models , author =. Journal of Econometrics , volume =. 2003 , publisher =

    2003

  3. [3]

    arXiv preprint , year =

    Sensitivity of regular estimators , author =. arXiv preprint , year =

  4. [4]

    The Annals of Statistics , volume =

    Point estimation with exponentially tilted empirical likelihood , author =. The Annals of Statistics , volume =

  5. [5]

    The Quarterly Journal of Economics , volume =

    Measuring the sensitivity of parameter estimates to estimation moments , author =. The Quarterly Journal of Economics , volume =. 2017 , publisher =

    2017

  6. [6]

    Journal of Econometrics , volume =

    A doubly corrected robust variance estimator for linear GMM , author =. Journal of Econometrics , volume =. 2022 , publisher =

    2022

  7. [7]

    Econometrica: Journal of the Econometric Society , pages =

    Automobile prices in market equilibrium , author =. Econometrica: Journal of the Econometric Society , pages =. 1995 , publisher =

    1995

  8. [8]

    Quantitative Economics , volume =

    The influence function of semiparametric estimators , author =. Quantitative Economics , volume =. 2022 , publisher =

    2022

  9. [9]

    The review of economic studies , volume =

    Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations , author =. The review of economic studies , volume =. 1991 , publisher =

    1991

  10. [10]

    Journal of the Royal Statistical Society: Series B (Methodological) , volume =

    Assessing sensitivity to an unobserved binary covariate in an observational study with binary outcome , author =. Journal of the Royal Statistical Society: Series B (Methodological) , volume =. 1983 , publisher =

    1983

  11. [11]

    Quantitative Economics , volume =

    Minimizing sensitivity to model misspecification , author =. Quantitative Economics , volume =. 2022 , publisher =

    2022

  12. [12]

    Econometrica , volume =

    Counterfactual sensitivity and robustness , author =. Econometrica , volume =. 2023 , publisher =

    2023

  13. [13]

    Review of Economics and Statistics , volume =

    Sensitivity to calibrated parameters , author =. Review of Economics and Statistics , volume =. 2023 , publisher =

    2023

  14. [14]

    Quantitative Economics , volume =

    Sensitivity analysis using approximate moment condition models , author =. Quantitative Economics , volume =. 2021 , publisher =

    2021

  15. [15]

    American Economic Review , volume =

    Sensitivity to exogeneity assumptions in program evaluation , author =. American Economic Review , volume =. 2003 , publisher =

    2003

  16. [16]

    Journal of Business & Economic Statistics , volume =

    Unobservable selection and coefficient stability: Theory and evidence , author =. Journal of Business & Economic Statistics , volume =. 2019 , publisher =

    2019

  17. [17]

    Journal of the American Statistical Association , year =

    Flexible sensitivity analysis for observational studies without observable implications , author =. Journal of the American Statistical Association , year =

  18. [18]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume =

    Making sense of sensitivity: Extending omitted variable bias , author =. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume =. 2020 , publisher =

    2020

  19. [19]

    Review of Economics and Statistics , volume =

    Plausibly exogenous , author =. Review of Economics and Statistics , volume =. 2012 , publisher =

    2012

  20. [20]

    Journal of Business & Economic Statistics , volume =

    A consistent variance estimator for 2sls when instruments identify different lates , author =. Journal of Business & Economic Statistics , volume =. 2018 , publisher =

    2018

  21. [21]

    Econometrica , volume =

    Inference for iterated GMM under misspecification , author =. Econometrica , volume =. 2021 , publisher =

    2021

  22. [22]

    Journal of Econometrics , volume =

    Asymptotic theory for clustered samples , author =. Journal of Econometrics , volume =. 2019 , publisher =

    2019

  23. [23]

    The Review of Economic Studies , volume =

    One-step estimators for over-identified generalized method of moments models , author =. The Review of Economic Studies , volume =. 1997 , publisher =

    1997

  24. [24]

    Journal of econometrics , volume =

    A finite sample correction for the variance of linear efficient two-step GMM estimators , author =. Journal of econometrics , volume =. 2005 , publisher =

    2005

  25. [25]

    The Annals of Statistics , pages =

    On differentiable functionals , author =. The Annals of Statistics , pages =. 1991 , publisher =

    1991

  26. [26]

    2022 , publisher =

    Econometrics , author =. 2022 , publisher =

    2022

  27. [27]

    Journal of econometrics , volume =

    On the structure of IV estimands , author =. Journal of econometrics , volume =. 2019 , publisher =

    2019

  28. [28]

    Econometric Society 2025 World Congress Volume , year =

    The purpose of an estimator is what it does: Misspecification, estimands, and over-identification , author =. Econometric Society 2025 World Congress Volume , year =

    2025

  29. [29]

    American Economic Review , volume =

    Consumption inequality and partial insurance , author =. American Economic Review , volume =. 2008 , publisher =

    2008

  30. [30]

    Journal of the american statistical association , volume =

    The influence curve and its role in robust estimation , author =. Journal of the american statistical association , volume =. 1974 , publisher =

    1974

  31. [31]

    Quantitative Economics , volume =

    Uncertainty quantification and global sensitivity analysis for economic models , author =. Quantitative Economics , volume =. 2019 , publisher =

    2019

  32. [32]

    The annals of mathematical statistics , volume =

    On the asymptotic distribution of differentiable statistical functions , author =. The annals of mathematical statistics , volume =. 1947 , publisher =

    1947

  33. [33]

    Econometrica: Journal of the Econometric Society , pages =

    The asymptotic variance of semiparametric estimators , author =. Econometrica: Journal of the Econometric Society , pages =. 1994 , publisher =

    1994

  34. [34]

    Journal of Applied Econometrics , volume =

    Estimating household consumption insurance , author =. Journal of Applied Econometrics , volume =. 2021 , publisher =

    2021

  35. [35]

    American Economic Journal: Macroeconomics , volume =

    How much consumption insurance beyond self-insurance? , author =. American Economic Journal: Macroeconomics , volume =. 2010 , publisher =

    2010

  36. [36]

    Journal of Econometrics , volume =

    Nearly-singular design in GMM and generalized empirical likelihood estimators , author =. Journal of Econometrics , volume =. 2008 , publisher =

    2008

  37. [37]

    2014 , publisher =

    Measuring the sensitivity of parameter estimates to sample statistics , author =. 2014 , publisher =

    2014

  38. [38]

    Econometrica , volume =

    On the informativeness of descriptive statistics for structural estimates , author =. Econometrica , volume =. 2020 , publisher =

    2020

  39. [39]

    Journal of Economic Dynamics and Control , volume =

    What to expect when you're calibrating: Measuring the effect of calibration on the estimation of macroeconomic models , author =. Journal of Economic Dynamics and Control , volume =. 2019 , publisher =

    2019

  40. [40]

    Econometrica: Journal of the econometric society , pages =

    Large sample properties of generalized method of moments estimators , author =. Econometrica: Journal of the econometric society , pages =. 1982 , publisher =

    1982

  41. [41]

    Quantitative Economics , volume =

    Double robust inference for continuous updating GMM , author =. Quantitative Economics , volume =. 2025 , publisher =

    2025

  42. [42]

    The Rand journal of economics , pages=

    Estimating discrete-choice models of product differentiation , author=. The Rand journal of economics , pages=. 1994 , publisher=

    1994

  43. [43]

    American economic review , volume=

    Income and democracy , author=. American economic review , volume=. 2008 , publisher=

    2008

  44. [44]

    2009 , publisher=

    Matrix mathematics: theory, facts, and formulas , author=. 2009 , publisher=

    2009

  45. [45]

    Econometric theory , volume=

    Generic uniform convergence , author=. Econometric theory , volume=. 1992 , publisher=

    1992

  46. [46]

    Working paper , year=

    How to weight in moments matching: A new approach and applications to earnings dynamics , author=. Working paper , year=

  47. [47]

    Journal of the American statistical Association , volume=

    Identification of causal effects using instrumental variables , author=. Journal of the American statistical Association , volume=. 1996 , publisher=

    1996

  48. [48]

    Econometrica , volume =

    Covariance Matrix Estimation and the Power of the Overidentifying Restrictions Test , author =. Econometrica , volume =. 2000 , publisher =

    2000

  49. [49]

    Handbook of Econometrics , editor =

    Large Sample Estimation and Hypothesis Testing , author =. Handbook of Econometrics , editor =. 1994 , publisher =

    1994