Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

This paper shows that the bias that imperfect product proxies inject into demand counterfactuals can be removed with a post-estimation correction whose standard errors ignore how the proxies were built.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-03 11:40 UTC pith:HJCRCXZR

load-bearing objection A genuinely useful debiasing toolkit for demand counterfactuals with ML proxies; the theory is coherent, but the key local-misspecification condition needs primitive support before the application claims are fully load-bearing. the 4 major comments →

arxiv 2601.05374 v2 pith:HJCRCXZR submitted 2026-01-08 econ.EM stat.ML

From Unstructured Data to Demand Counterfactuals: Theory and Practice

Timothy Christensen , Giovanni Compiani This is my paper
classification econ.EM stat.ML
keywords demand estimationunstructured dataproduct embeddingsbias correctioncounterfactualsproxy variablesLagrange multiplier diagnosticsdiscrete choice
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Demand models increasingly use embeddings of images, text, and reviews as product attributes, but these are proxies for the true, latent dimensions that drive substitution. This paper argues that treating proxies as truth biases counterfactual predictions, and shows how to correct that bias after estimation with a simple additive adjustment. The corrected estimator is asymptotically centered on the true counterfactual and is efficient, and its variance does not depend on how the proxies were constructed, so fine-tuning or data-dependent embeddings require no standard-error adjustment. Two LM diagnostics help practitioners choose which proxies and how many dimensions to use. In an e-book choice experiment, the correction raises the rate of correctly predicted closest substitutes from 40% to 60–70% for the best review-text specification.

Core claim

The paper's central claim is that mismeasurement of product attributes by proxies is a model misspecification, not a classical measurement-error problem, and can be neutralized by reparameterizing the demand model in terms of a composite parameter γ that bundles structural parameters and latent attributes. Once the naive estimate γ̂ = γ(θ̂, ẽ) is in hand, the counterfactual estimator is adjusted by subtracting weighted averages of the estimation moments, with weights chosen to make the adjusted estimator's first-order dependence on γ̂ vanish. Under the condition that γ̂ is within a neighbourhood of γ0 whose radius is negligible relative to sampling error, the adjusted estimator is asymptoti

What carries the argument

The key machinery is the composite parameter γ(θ,e), the low-dimensional combination of structural parameters and latent product attributes through which all choice probabilities and counterfactuals enter the model. By expressing both the naive estimator and the estimation moments as functions of γ̂, the paper chooses correction weights so that the adjusted counterfactual has zero first-order sensitivity to γ̂; this makes the choice of proxy irrelevant to the estimator's leading bias term. The same object drives the diagnostics: LM1 compares γ̂ to the set of composite parameters spanned by the proxies, and LM2 augments the proxy space with an extra direction to test whether the proxy dimensi

Load-bearing premise

The load-bearing assumption is that the composite parameter derived from the proxies lands close enough to the true value — closer than the fourth root of the sample size; if the proxies are too noisy, nothing in the data forces this, and the correction loses its centering property.

What would settle it

Run a controlled simulation with true latent attributes known and proxy noise at roughly ρ=0.5 or higher, increasing sample size; if the bias-corrected estimator's bias does not shrink at the claimed rate or does not remain far below the naive estimator's bias, the local condition is not doing the work in that regime.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Bias-corrected counterfactuals are centered at the true value and come with closed-form standard errors, so no bootstrap or re-estimation is needed after the correction.
  • Standard errors remain valid when embeddings are fine-tuned on the choice data, because the asymptotic distribution does not depend on the proxy to first order.
  • The LM1 and LM2 diagnostics give a practical answer to which unstructured-data proxies to use and how many principal components to keep.
  • In the e-book experiment, the correction raises closest-substitute hit rates from 40% to 60–70% and improves or ties 11 of 13 specifications ruled in by the dimension diagnostic.
  • Even when mismeasurement is not a concern, the estimator offers efficient counterfactual inference with simple standard errors.

Where Pith is reading between the lines

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

  • A natural testable extension is to calibrate the LM1 threshold against a small validation set with known attribute values, rather than the heuristic χ² log T cutoff, to see whether better proxy-selection decisions result.
  • The paper's logic implies that the cost of searching over many embedding choices is lower than usually assumed: model selection should emphasize the LM diagnostics, since the target counterfactual variance is proxy-independent to first order.
  • The method is best suited to attributes that are fixed product characteristics; applying it to time-varying or context-dependent attributes would require the composite-parameter mapping to hold within each market or individual setting, which the paper does not claim.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper develops post-estimation bias corrections and diagnostics for demand counterfactuals when product attributes are proxied by embeddings or other imperfect measures. The key device is a reparameterization in which attributes and structural parameters enter choice probabilities only through a composite parameter γ; the naive counterfactual estimator is corrected by adding a linear combination of the model's own moment conditions, with weights chosen to remove first-order dependence on γ̂ and to minimize asymptotic variance. The main theoretical results (Propositions 1, 3, 5) state that, provided γ̂ = γ₀ + o_p(T^{-1/4}), the corrected estimator is asymptotically centered at the true counterfactual with a variance independent of γ̂, θ̂, and the proxy. Two LM-type diagnostics are proposed to assess whether γ̂ is sufficiently close to γ₀ and whether the proxy dimension is adequate. Simulations and an application to ebook choice data illustrate the method, with closest-substitute hit rates improving from 40% to 70% in the preferred specification.

Significance. If the maintained rate condition holds, the paper gives a practically useful, computationally light way to debias counterfactual inference in demand models with imperfect proxies. The closed-form standard errors, efficiency property within a natural class, and accommodation of data-dependent proxies are genuine strengths. The empirical demonstration against ground-truth second choices is valuable. However, the central theoretical guarantee is conditional on the proxy-induced estimator γ̂ converging to γ₀ at o_p(T^{-1/4}), and the paper does not provide primitive conditions ensuring this for the leading case of fixed embeddings. The diagnostics are useful heuristics but, as formalized, cannot certify the required rate. These issues are load-bearing for the paper's headline claim of valid inference for unstructured-data proxies, though they are fixable by a more careful statement of scope and by adding rate conditions or a local-misspecification framework.

major comments (4)
  1. [§6.1, Propositions 1 and 3; §6.2, Proposition 5; Remark 8] The condition γ̂ = γ₀ + o_p(T^{-1/4}) is maintained without a primitive justification. For a fixed, off-the-shelf embedding ẽ, the natural limit is γ̂ → γ(θ*(ẽ), ẽ), where θ*(ẽ) is the pseudo-true value; nothing guarantees γ(θ*(ẽ), ẽ) = γ₀. Remark 8 appeals to fine-tuning on the same data, but no rate for the convergence of a data-dependent ẽ to e₀ is established, so the required o_p(T^{-1/4}) shrinkage is not derived. Since the paper's leading case is exactly such proxies, the central asymptotic claim is not supported for the main application without additional assumptions. I recommend either providing concrete primitive conditions on proxy construction (e.g., embeddings refined at a controlled rate) or explicitly recasting the theory as local-misspecification asymptotics and stating that the debiasing guarantee applies only when the proxy error is smaller than the sampling error at the
  2. [§2.3.1 and §3.3.1, Proposition 4 and Proposition 6] The LM1 diagnostic is presented as validating the condition that γ̂ is sufficiently close to γ₀. But Proposition 4 is proved under γ̂ = γ₀ + o_p(1). Under fixed proxy mismeasurement γ̂ is inconsistent, so LM1 diverges at rate T and the proposed threshold C_T² = χ²_{dim γ,0.95} log T will reject with probability approaching one. A finite-sample non-rejection then only indicates that the sample is too small to detect the misspecification; it cannot certify the required o_p(T^{-1/4}) rate. The wording in the practitioner's guides (e.g., "conclude that ẽ is sufficiently close to e₀") overstates what the diagnostic can establish. The diagnostic is still useful as a specification check, but the paper should state this limitation and provide either a formal local-power analysis or a bound that is valid under fixed misspecification.
  3. [§6.1, proof of Proposition 1; §4 simulations] Even when the bias correction removes the first-order term, the remaining bias is of order O(∥γ̂−γ₀∥²) under fixed proxy error. If γ̂ fails to converge to γ₀, this second-order bias can be non-negligible; the simulations in Figure 2 indeed show that for ρ > 0.5 the corrected estimator's bias increases. The paper acknowledges this behavior informally, but the theoretical sections do not make explicit the consequence that the distributional results are only approximate for fixed mismeasurement and that the approximation degrades as ∥γ̂−γ₀∥ grows. Please state this as a formal caveat near Propositions 1 and 5, and quantify the remainder as a function of ∥γ̂−γ₀∥.
  4. [§6.1.2, Proposition 4; §2.3.1, threshold choice] The choice C_T² = χ²_{dim(γ),0.95} log T is heuristic. Proposition 4 only gives statements 'wpa1' conditional on a chi-square random variable being below ϵ²C_T²; it does not provide the distributional approximation needed to calibrate the threshold for controlling a false-acceptance probability. The paper should either derive the large-deviation or local-alternative properties of LM1 that justify this threshold, or present it as an ad hoc rule. This matters because the threshold is used in the application to select among specifications.
minor comments (4)
  1. [Proposition 3 (p. 35)] The statement says 'Let Assumption 3 hold', but the result is for the no-microdata case and should refer to Assumption 2.
  2. [Proposition 7 (p. 39)] There is a typo: 'Let Assumptions 3 and 4 bold hold' should be 'both hold'. Also the display uses 'LM' without a subscript; it should be 'LM1'.
  3. [§2.2, Remark 2 and eq. (15)] The variance estimator (15) is stated without derivation. It would help readers to see that it is the sample analogue of the variance expression in Proposition 1, including the cross-term between k_t and the moments.
  4. [§4, Figure 2] The text says 'when mismeasurement becomes very large (ρ>0.5), the bias correction starts to also perform worse,' but the figure appears to show the onset around ρ=0.4–0.5. Please align the text with the simulation grid.

Circularity Check

0 steps flagged

No significant circularity: the bias-correction construction is self-contained and the application is validated against an external ground truth.

full rationale

The target κ0 is not an input to the construction of κ̂bc. The weights ĉ and d̂t in eqs. (12)-(14) are chosen from sample moments and derivatives of kt, ξt, and mt at γ̂, with probability limits solving the orthogonality constraint (33); Proposition 1 then shows the first-order term in (γ̂−γ0) vanishes by the model's own moment condition E[Ztξt(γ0)]=0. This is a standard influence-function debiasing argument, not a tautology: the centering at κ0 follows from those moment conditions, not from plugging κ0 into the estimator. The application uses second-choice data only to score the closest-substitute prediction, so the 40%-to-70% improvement is an external falsification exercise, not a fitted prediction. The only self-citation, Compiani et al. (2025), supplies the data and naive estimates for comparison; it is not used to derive Propositions 1 or 5 and is validated against an external ground truth, so it is not load-bearing. The rate condition γ̂=γ0+o_p(T^{-1/4}) (Remark 8; Propositions 1,3,5) is a substantive regularity/identification assumption limiting applicability when fixed embeddings are mismeasured, but an unsatisfied assumption is a correctness/robustness concern, not a by-construction circularity; LM1 is a diagnostic for this condition, not a proof of it. No step in the derivation chain reduces to its own input.

Axiom & Free-Parameter Ledger

3 free parameters · 8 axioms · 0 invented entities

No free parameters are fitted to make the theory work; the LM1 threshold is the main hand-chosen tuning device. The composite parameter γ(θ,e) is a reparameterization, not an invented entity — no new economic or physical objects are introduced. The weakest ledger entry is the local-misspecification axiom, which is ad hoc to this paper's debiasing strategy and governs its operating range.

free parameters (3)
  • LM1 threshold C_T² = χ²_{dim(γ),0.95} log T = χ²_{0.95} log T (target rate √(log T)/T)
    The practical gate deciding whether a proxy set is 'close enough' for the bias correction to work. Motivated by a heuristic in Proposition 4's proof; the finite-sample behavior of this gate is not analyzed.
  • Number of principal components (dimension of ẽ) in the application
    Chosen per specification with the aid of the LM2 diagnostic; a model-selection choice that the paper's own theory can only assess via a sup-type test with simulated critical values.
  • Mismeasurement design parameter ρ in simulations = 0, 0.1, 0.2, 0.3, 0.4, >0.5
    Hand-set in the DGP ẽ_j=(1-ρ)e_j+√(1-(1-ρ)²)η_j; the 'local mismeasurement' operating range is defined through this design, and the conclusions about when the correction works are tuned to it.
axioms (8)
  • domain assumption IV exogeneity E[ξ_jt | z_jt] = 0 (Eq. 2)
    The GMM moments that the bias correction subtracts must be mean-zero at the true parameter; otherwise the correction itself introduces bias. Standard in BLP-style demand estimation.
  • domain assumption Composite-parameter structure: (θ, e) enter choice probabilities, moments, and counterfactuals only through γ(θ, e)
    Sections 2.2 and 3.2 restrict attention to models with this property (satisfied by the mixed-logit examples). Models without it — e.g., non-separable attribute interactions — are outside the method.
  • domain assumption Latent attributes e do not vary across markets
    Section 2.1: e is a fixed product characteristic. If e varied across markets, the reparameterization γ(θ,e) and the identification argument would need reworking.
  • domain assumption Same dimension and rank: e₀ and ẽ are J×r with full row rank; Γ convex and open
    Section 6.1. LM2 exists precisely because practitioners face this choice, but the maintained theory assumes the dimension r is matched.
  • ad hoc to paper Local misspecification: γ̂ = γ₀ + o_p(T^{-1/4})
    Remark 8 / Propositions 1, 3, 5. Load-bearing: the first-order debiasing is only valid if the proxy-induced discrepancy is smaller than sampling error to order T^{-1/4}; fixed bad proxies fall outside the guarantee.
  • domain assumption Proxies admit a probability limit e* (fine-tuning case)
    Assumption 4(i): data-dependent proxies from fine-tuning must converge to a fixed limit for the asymptotics in Propositions 5-7; the Remark 4 no-SE-correction claim inherits this.
  • standard math Smoothness, moment, and rank conditions (Assumptions 1(i)-(iii), 2, 3)
    Twice differentiability with bounded derivatives, positive definite covariance matrices, and eigenvalue bounds; standard asymptotic machinery.
  • domain assumption Assumption 4(iii): ∥θ̂−θ*∥ ≤ C∥ẽ−e*∥ wpa1 and first-order condition for θ̂
    Needed for Proposition 7 to translate LM1 into a bound on proxy error; it assumes estimation error is dominated by proxy error, which is strong and not discussed in the main text.

pith-pipeline@v1.3.0-alltime-deepseek · 32380 in / 26796 out tokens · 276332 ms · 2026-08-03T11:40:54.185303+00:00 · methodology

0 comments
read the original abstract

Empirical models of multi-product demand rely on low-dimensional product representations to capture substitution patterns, increasingly using proxies built from unstructured data. When proxies are imperfect, standard workflows yield biased counterfactuals and invalid inference. We develop a practical toolkit to address these issues. Our methods apply to market-level and/or individual data, require minimal additional computation, provide simple standard-error formulas, and accommodate proxies from fine-tuned models. Further, we propose diagnostics to assess proxy quality. Our methods yield meaningful improvements in predicting substitution in empirically calibrated simulations and in an application where we assess counterfactual prediction performance against a ground truth.

Figures

Figures reproduced from arXiv: 2601.05374 by Giovanni Compiani, Timothy Christensen.

Figure 1
Figure 1. Figure 1: Distribution of the naive and bias-corrected estimators [PITH_FULL_IMAGE:figures/full_fig_p029_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Bias and RMSE of the naive and bias-corrected estimators [PITH_FULL_IMAGE:figures/full_fig_p030_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Distance between ˆγ and γ0 versus LM1 diagnostic attributes with mixed logit models that leverage proxies extracted from unstructured data. The key findings are that (i) unstructured data is predictive of substitution patterns, and (ii) book descriptions and reviews, when processed with transformer￾based text models, perform particularly well at predicting substitution. The results in Compiani et al. (2025… view at source ↗
Figure 4
Figure 4. Figure 4: Rates of correct closest substitutes predictions. [PITH_FULL_IMAGE:figures/full_fig_p033_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: LM1 diagnostic in empirical application Note: Each bar shows the value of the LM1 statistic for the corresponding specification [PITH_FULL_IMAGE:figures/full_fig_p055_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: LM2 diagnostic in empirical application Note: Each bar shows the value of the LM2 statistic for the corresponding specification. The horizontal segments show the associated critical values. 54 [PITH_FULL_IMAGE:figures/full_fig_p055_6.png] view at source ↗

discussion (0)

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

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Econometrics with Pre-Trained Embeddings for Unstructured Data

    econ.EM 2026-07 accept novelty 7.0

    Pre-trained embeddings are valid in double machine learning when the target nuisance function lies in the span of the source-task representation; under that condition the downstream estimator can converge faster than ...

Reference graph

Works this paper leans on

45 extracted references · 1 linked inside Pith · cited by 1 Pith paper

  1. [1]

    Ai, C. and X. Chen (2012): The semiparametric efficiency bound for models of sequential moment restrictions containing unknown functions, Journal of Econometrics, 170, 442--457

    2012

  2. [2]

    Allcott, H. and N. Wozny (2014): Gasoline prices, fuel economy, and the energy paradox, Review of Economics and Statistics, 96, 779--795

    2014

  3. [3]

    Allon, G., D. Chen, Z. Jiang, and D. Zhang (2023): Machine learning and prediction errors in causal inference, The Wharton School Research Paper

    2023

  4. [4]

    Andrews, D. W. (1994 a ): Asymptotics for semiparametric econometric models via stochastic equicontinuity, Econometrica: Journal of the Econometric Society, 62, 43--72

    1994

  5. [5]

    --- -.1pt --- -.1pt --- (1994 b ): Empirical process methods in econometrics, Handbook of econometrics, 4, 2247--2294

    1994

  6. [6]

    --- -.1pt --- -.1pt --- (2005): Cross-section regression with common shocks, Econometrica, 73, 1551--1585

    2005

  7. [7]

    Angelopoulos, A. N., S. Bates, C. Fannjiang, M. I. Jordan, and T. Zrnic (2023): Prediction-powered inference, Science, 382, 669--674

    2023

  8. [8]

    Chernozhukov, S

    Bach, P., V. Chernozhukov, S. Klaassen, M. Spindler, J. Teichert-Kluge, and S. Vijaykumar (2024): Adventures in demand analysis using AI, arXiv preprint arXiv:2501.00382

    arXiv 2024

  9. [9]

    Conlon, and M

    Backus, M., C. Conlon, and M. Sinkinson (2021): Common Ownership and Competition in the Ready-To-Eat Cereal Industry, NBER Working Paper 28350

    2021

  10. [10]

    Christensen, S

    Battaglia, L., T. Christensen, S. Hansen, and S. Sacher (2024): Inference for Regression with Variables Generated by AI or Machine Learning, arXiv preprint arXiv:2402.15585

    Pith/arXiv arXiv 2024

  11. [11]

    Ferreira, and R

    Bayer, P., F. Ferreira, and R. McMillan (2007): A unified framework for measuring preferences for schools and neighborhoods, Journal of Political Economy, 115, 588--638

    2007

  12. [12]

    Levinsohn, and A

    Berry, S., J. Levinsohn, and A. Pakes (1995): Automobile Prices in Market Equilibrium, Econometrica, 63, 841--890

    1995

  13. [13]

    --- -.1pt --- -.1pt --- (2004): Differentiated Products Demand System from a Combination of Micro and Macro Data: The New Car Market, Journal of Political Economy, 112, 68--105

    2004

  14. [14]

    Berry, S. T. and P. A. Haile (2014): Identification in differentiated products markets using market level data, Econometrica, 82, 1749--1797

    2014

  15. [15]

    4, 1--62

    --- -.1pt --- -.1pt --- (2021): Foundations of demand estimation, in Handbook of industrial organization, Elsevier, vol. 4, 1--62

    2021

  16. [16]

    --- -.1pt --- -.1pt --- (2024): Nonparametric identification of differentiated products demand using micro data, Econometrica, 92, 1135--1162

    2024

  17. [17]

    Brown, B. W. and W. K. Newey (1998): Efficient semiparametric estimation of expectations, Econometrica, 66, 453--464

    1998

  18. [18]

    Carlson, J. and M. Dell (2025): A Unifying Framework for Robust and Efficient Inference with Unstructured Data, arXiv preprint arXiv:2505.00282

    arXiv 2025

  19. [19]

    Hong, and E

    Chen, X., H. Hong, and E. Tamer (2005): Measurement error models with auxiliary data, The Review of Economic Studies, 72, 343--366

    2005

  20. [20]

    Hong, and A

    Chen, X., H. Hong, and A. Tarozzi (2008): Semiparametric efficiency in GMM models with auxiliary data, The Annals of Statistics, 36, 808--843

    2008

  21. [21]

    Morozov, and S

    Compiani, G., I. Morozov, and S. Seiler (2025): Demand estimation with text and image data, arXiv preprint arXiv:2503.20711

    arXiv 2025

  22. [22]

    Conlon, C. and J. Gortmaker (2025): Incorporating Micro Data into Differentiated Products Demand Estimation with PyBLP, Journal of Econometrics, 105926

    2025

  23. [23]

    Dub \'e , J.-P. and P. E. Rossi (2019): Handbook of the Economics of Marketing, vol. 1, North Holland

    2019

  24. [24]

    Hinck, B

    Egami, N., M. Hinck, B. Stewart, and H. Wei (2023): Using imperfect surrogates for downstream inference: Design-based supervised learning for social science applications of large language models, Advances in Neural Information Processing Systems, 36, 68589--68601

    2023

  25. [25]

    (2013): Ownership consolidation and product characteristics: A study of the US daily newspaper market, American Economic Review, 103, 1598--1628

    Fan, Y. (2013): Ownership consolidation and product characteristics: A study of the US daily newspaper market, American Economic Review, 103, 1598--1628

    2013

  26. [26]

    Fong, C. and M. Tyler (2021): Machine learning predictions as regression covariates, Political Analysis, 29, 467--484

    2021

  27. [27]

    (2015): Asymptotic theory for differentiated products demand models with many markets, Journal of Econometrics, 185, 162--181

    Freyberger, J. (2015): Asymptotic theory for differentiated products demand models with many markets, Journal of Econometrics, 185, 162--181

    2015

  28. [28]

    Goldberg, P. K. (1995): Product differentiation and oligopoly in international markets: The case of the US automobile industry, Econometrica, 891--951

    1995

  29. [29]

    Grieco, P. L., C. Murry, J. Pinkse, and S. Sagl (2025): Optimal Estimation of Discrete Choice Demand Models with Consumer and Product Data, NBER Working Paper 33397

    2025

  30. [30]

    Grieco, P. L., C. Murry, and A. Yurukoglu (2024): The evolution of market power in the us automobile industry, The Quarterly Journal of Economics, 139, 1201--1253

    2024

  31. [31]

    Kuersteiner, and M

    Hahn, J., G. Kuersteiner, and M. Mazzocco (2022): Joint time-series and cross-section limit theory under mixingale assumptions, Econometric Theory, 38, 942--958

    2022

  32. [32]

    Han, S. and K. Lee (2025): Copyright and Competition: Estimating Supply and Demand with Unstructured Data, arXiv preprint arXiv:2501.16120

    arXiv 2025

  33. [33]

    Hansen, B. E. (1996): Inference when a nuisance parameter is not identified under the null hypothesis, Econometrica, 64, 413--430

    1996

  34. [34]

    Hausman, J. A. (1994): Valuation of new goods under perfect and imperfect competition, National Bureau of Economic Research Cambridge, Mass., USA

    1994

  35. [35]

    (2025): Generative brand choice, Working Paper

    Lee, K. (2025): Generative brand choice, Working Paper

    2025

  36. [36]

    Lee, R. S. (2013): Vertical integration and exclusivity in platform and two-sided markets, American Economic Review, 103, 2960--3000

    2013

  37. [37]

    McClure, and A

    Magnolfi, L., J. McClure, and A. Sorensen (2025): Triplet embeddings for demand estimation, American Economic Journal: Microeconomics, 17, 282--307

    2025

  38. [38]

    (2017): Targeted vouchers, competition among schools, and the academic achievement of poor students, Working Paper

    Neilson, C. (2017): Targeted vouchers, competition among schools, and the academic achievement of poor students, Working Paper

    2017

  39. [39]

    Neuman, A. M., Y. Xie, and Q. Sun (2023): Restricted Riemannian geometry for positive semidefinite matrices, Linear Algebra and its Applications, 665, 153--195

    2023

  40. [40]

    (2000): Mergers with differentiated products: The case of the ready-to-eat cereal industry, The RAND Journal of Economics, 395--421

    Nevo, A. (2000): Mergers with differentiated products: The case of the ready-to-eat cereal industry, The RAND Journal of Economics, 395--421

    2000

  41. [41]

    --- -.1pt --- -.1pt --- (2001): Measuring market power in the ready-to-eat cereal industry, Econometrica, 69, 307--342

    2001

  42. [42]

    Newey, W. K. (1994): The asymptotic variance of semiparametric estimators, Econometrica: Journal of the Econometric Society, 62, 1349--1382

    1994

  43. [43]

    Newey, W. K. and D. McFadden (1994): Large sample estimation and hypothesis testing, Handbook of econometrics, 4, 2111--2245

    1994

  44. [44]

    (2002): Quantifying the benefits of new products: The case of the minivan, Journal of political Economy, 110, 705--729

    Petrin, A. (2002): Quantifying the benefits of new products: The case of the minivan, Journal of political Economy, 110, 705--729

    2002

  45. [45]

    Zhang, J., W. Xue, Y. Yu, and Y. Tan (2023): Debiasing ML-or AI-Generated Regressors in Partial Linear Models, SSRN Working Paper 4636026

    2023