pith. sign in

arxiv: 2409.02311 · v3 · pith:7YALJ6L5new · submitted 2024-09-03 · 💰 econ.EM · stat.ME

A simple distributional difference-in-differences estimator for univariate and bivariate outcomes

Iv\'an Fern\'andez-Val , Jonas Meier , Aico van Vuuren , Francis Vella This is my paper

Pith reviewed 2026-05-23 21:09 UTC · model grok-4.3

classification 💰 econ.EM stat.ME
keywords difference-in-differencesdistribution regressiontreatment effectsparallel trendsbivariate outcomesminimum wageheterogeneous effects
0
0 comments X

The pith

A distribution regression estimator for treatment effects in difference-in-differences designs assumes parallel trends on transformed distributions.

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

The paper develops a distribution regression method to estimate treatment effects under a difference-in-differences design. This is useful because treatment effects can vary across different values of the outcome, such as employment levels or wages. The estimator handles covariates and extends to bivariate outcomes when the untreated distribution has no group and time interaction. Sympathetic readers care because it allows seeing how policies affect the full distribution rather than just averages, which can reveal important heterogeneity in economic impacts.

Core claim

We provide a simple distribution regression estimator for treatment effects in the difference-in-differences (DiD) design. Our procedure is particularly useful when the treatment effect differs across the distribution of the outcome variable. Our proposed estimator easily incorporates covariates and can be extended to settings where the treatment potentially affects the joint distribution of multiple outcomes. Our key identifying restriction is that the untreated outcome distribution does not exhibit an interaction effect of group and time, resulting in a parallel trend assumption on a transformation of the distribution. We highlight the relationship to the changes-in-changes approach and re

What carries the argument

Distribution regression estimator based on the assumption of no group-time interaction in the untreated outcome distribution, which implies parallel trends on a transformed distribution.

Load-bearing premise

The untreated outcome distribution does not exhibit an interaction effect of group and time.

What would settle it

Observing different time trends in the distribution of outcomes for different untreated groups would contradict the key identifying restriction.

Figures

Figures reproduced from arXiv: 2409.02311 by Aico van Vuuren, Francis Vella, Iv\'an Fern\'andez-Val, Jonas Meier.

Figure 1
Figure 1. Figure 1: Distribution of Total employment index for these employment levels are -0.0095 and -0.0101 respectively. Following the increase in the minimum wage, this negative relationship becomes stronger, with the cor￾responding estimate values of -0.1709 and -0.2402. Moreover, despite the relatively small number of observations the difference in the Spearman’s correlation is statistically signif￾icant at the 10 perc… view at source ↗
Figure 2
Figure 2. Figure 2: Part-time employment 0 10 20 30 40 50 0 0.2 0.4 0.6 0.8 1 Fulltime employment FY0|G,T (·|1, 1), FY1|G,T (·|1, 1) FY0|G,T (·|1, 1) FY1|G,T (·|1, 1) [PITH_FULL_IMAGE:figures/full_fig_p026_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Full-time employment 26 [PITH_FULL_IMAGE:figures/full_fig_p026_3.png] view at source ↗
read the original abstract

We provide a simple distribution regression estimator for treatment effects in the difference-in-differences (DiD) design. Our procedure is particularly useful when the treatment effect differs across the distribution of the outcome variable. Our proposed estimator easily incorporates covariates and, importantly, can be extended to settings where the treatment potentially affects the joint distribution of multiple outcomes. Our key identifying restriction is that the untreated outcome distribution does not exhibit an interaction effect of group and time. This assumption results in a parallel trend assumption on a transformation of the distribution. We highlight the relationship between our procedure and assumptions with the changes-in-changes approach of Athey and Imbens (2006). We also reexamine the Card and Krueger (1994) study of the impact of minimum wages on employment to illustrate the utility of our approach.

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

2 major / 2 minor

Summary. The paper proposes a simple distribution regression estimator for treatment effects under a difference-in-differences design. The estimator targets heterogeneous effects across the outcome distribution and extends to the joint distribution of bivariate outcomes under the key identifying restriction that the untreated outcome distribution exhibits no group-by-time interaction (inducing parallel trends after a suitable transformation). The approach is explicitly related to the changes-in-changes model of Athey and Imbens (2006) and illustrated via reanalysis of the Card-Krueger (1994) minimum-wage data.

Significance. If the identifying restriction holds, the estimator supplies a computationally straightforward route to distributional DiD effects, including for multiple outcomes, without requiring parametric assumptions on the outcome law. The explicit linkage to the changes-in-changes framework and the empirical illustration with a well-known data set are strengths that aid interpretability and replicability.

major comments (2)
  1. [Abstract and §2] The abstract and introduction state the identifying assumption and the resulting parallel-trends restriction on a transformed distribution at a high level, but the manuscript supplies no explicit derivation showing how the distribution-regression estimator recovers the counterfactual untreated distribution for the treated group. A step-by-step identification argument (e.g., in §2 or §3) would be required to confirm that the estimator is consistent under the stated condition.
  2. [§4] For the bivariate extension, the assumption that the untreated joint distribution has no group-by-time interaction is asserted to suffice, yet the text does not detail how the distribution-regression procedure is adapted to the joint CDF or density and whether additional regularity conditions are needed for identification of the counterfactual joint. This is load-bearing for the claim that the method extends directly to multiple outcomes.
minor comments (2)
  1. [§2] Notation for the transformation of the distribution that induces parallel trends should be introduced earlier and used consistently when relating the estimator to the changes-in-changes model.
  2. [§5] The empirical illustration would benefit from reporting standard errors or confidence bands for the estimated distributional effects to allow assessment of precision.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments. We agree that the identification arguments would benefit from greater explicitness and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract and §2] The abstract and introduction state the identifying assumption and the resulting parallel-trends restriction on a transformed distribution at a high level, but the manuscript supplies no explicit derivation showing how the distribution-regression estimator recovers the counterfactual untreated distribution for the treated group. A step-by-step identification argument (e.g., in §2 or §3) would be required to confirm that the estimator is consistent under the stated condition.

    Authors: We agree that an explicit step-by-step derivation is missing and would improve clarity. The manuscript states the no group-by-time interaction assumption and notes its relation to changes-in-changes, but does not derive how this yields the parallel-trends restriction on the transformed distribution or how the distribution-regression estimator recovers the counterfactual. In the revision we will insert a detailed identification argument in Section 2, showing the mapping from the assumption to the counterfactual untreated distribution for the treated group and establishing consistency of the estimator. revision: yes

  2. Referee: [§4] For the bivariate extension, the assumption that the untreated joint distribution has no group-by-time interaction is asserted to suffice, yet the text does not detail how the distribution-regression procedure is adapted to the joint CDF or density and whether additional regularity conditions are needed for identification of the counterfactual joint. This is load-bearing for the claim that the method extends directly to multiple outcomes.

    Authors: We acknowledge that the bivariate section asserts the extension without sufficient detail on implementation and regularity conditions. The same no-interaction assumption on the untreated joint distribution permits the parallel-trends logic to carry over, but the manuscript does not spell out the adaptation of distribution regression to the joint CDF or any extra conditions. In the revision we will expand Section 4 to describe the procedure for the joint distribution, state the required regularity conditions, and confirm identification of the counterfactual joint. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper states its central identifying restriction directly as an assumption (untreated outcome distribution exhibits no group-by-time interaction, inducing parallel trends on a transformed scale) and relates it to the external Athey-Imbens (2006) changes-in-changes model. The distribution-regression estimator is then constructed from this assumption plus standard DiD logic; no equation reduces by construction to a fitted parameter or self-defined quantity, no uniqueness theorem is imported from the authors' prior work, and no ansatz is smuggled via self-citation. The bivariate extension follows the same non-circular route. The derivation chain is therefore self-contained against the stated assumption and external benchmark.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on one domain assumption about the absence of group-time interaction in untreated distributions; no free parameters or invented entities are described in the abstract.

axioms (1)
  • domain assumption The untreated outcome distribution does not exhibit an interaction effect of group and time.
    Explicitly identified in the abstract as the key identifying restriction that yields parallel trends on a transformed distribution.

pith-pipeline@v0.9.0 · 5675 in / 1168 out tokens · 28002 ms · 2026-05-23T21:09:38.989896+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

25 extracted references · 25 canonical work pages

  1. [1]

    Inside the war on poverty: the impact of food stamps on birth outcomes

    Almond, D., H.W. Hoynes, and D.W. Schanzenbach (2011), “Inside the war on poverty: the impact of food stamps on birth outcomes”, Review of Economics and Statistics 93, 387–403

    work page 2011

  2. [2]

    Causal Models for longitudinal and panel Data: a survey

    Arkhangelsky, D. and G. Imbens (2024), “Causal Models for longitudinal and panel Data: a survey”, The Econometrics Journal 27, C1-61. 29

    work page 2024

  3. [3]

    Arkhangelsky, D. S. Athey, D.A. Hirshberg, G.W. Imbens, and S. W a- ger (2021), “Synthetic difference-in-differences, American Economic Review 111, 4088–18

    work page 2021

  4. [4]

    Identification and inference in nonlinear difference- in-differences models

    Athey, S. and G.D. Imbens (2006), “Identification and inference in nonlinear difference- in-differences models”, Econometrica 74 431–97

    work page 2006

  5. [5]

    R ¨ummele, and B

    Biewen, M., M. R ¨ummele, and B. Fitzenberger (2022), “Using distribution re- gression Difference-in-Differences to evaluate the effects of a minim um wage in- troduction on the distribution of hourly wages and hours worked, w orking paper, N¨ urnberg

    work page 2022

  6. [6]

    Evalu- ating the employment impact of a mandatory job search program

    Blundell, R., C. Meghir, M. Costa Dias and J. V an Reenen (2004), “Evalu- ating the employment impact of a mandatory job search program”, Journal of the European Economic Association 2,569–606

    work page 2004

  7. [7]

    Quantile treatment effects in difference in differences models with panel data

    Callaway, B. and T. Li (2019), “Quantile treatment effects in difference in differences models with panel data” Quantitative Economics 10, 1579–1618

    work page 2019

  8. [8]

    The Impact of the Mariel Boatlift on the Miami Labor Mark et

    Card, D. (1990), “The Impact of the Mariel Boatlift on the Miami Labor Mark et”Industrial and Labor Relations Review, 43, 245—57

    work page 1990

  9. [9]

    Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania

    Card, D. and A.B. Krueger (1994), “Minimum Wages and Employment: A Case Study of the Fast-Food Industry in New Jersey and Pennsylvania” American Eco- nomic Review 84, 772—93

    work page 1994

  10. [10]

    The effect of mini- mum wages on low-wage jobs

    Cengiz, D., A. Dube, A. Lindner, and B. Zipperer (2019), “The effect of mini- mum wages on low-wage jobs ”, Quarterly Journal of Economics 134, 1405–54

    work page 2019

  11. [11]

    Inference on coun- terfactual distributions

    Chernozhukov, V., I. Fern ´andez-V al, and Melly B. (2013), “Inference on coun- terfactual distributions”, Econometrica, 81, 2205–68

    work page 2013

  12. [12]

    Distribution regres- sion with sample selection, with an application to wage decompositions in the UK

    Chernozhukov, V., I. Fern ´andez-V al, and S. Luo (2019), “Distribution regres- sion with sample selection, with an application to wage decompositions in the UK”, working paper, MIT, Cambridge (MA). 30

    work page 2019

  13. [13]

    Generic inference on quantile and quantile effect functions for disc rete outcomes

    Chernozhukov, V. I. Fern ´andez-V al, B. Melly, and K. W ¨uthrich (2020), “Generic inference on quantile and quantile effect functions for disc rete outcomes”, Journal of the American Statistical Association 115, 123–37

    work page 2020

  14. [14]

    Minimum wages and the distribution of family incomes

    Dube, A. (2019), “Minimum wages and the distribution of family incomes”, American Economic Journal: Applied Economics 11, 268—304. Fern´andez-V al I., J. Meier, A. van Vuuren and F. Vella (2024a) “Bivariate distribution regression with an application to intergenerational mob ility”, working paper, Boston University. Fern´andez-V al I., J. Meier, A. van Vu...

    work page 2019

  15. [15]

    The conditional distribution of excess returns: an empirical analysis

    Foresi, S. and F. Peracchi (1995), “The conditional distribution of excess returns: an empirical analysis”, Journal of the American Statistical Association , 90, 451–66

    work page 1995

  16. [16]

    (2021), “The long-run effects of childhood insurance coverage: medicaid implementation, adult health, and labor market outcomes, American Eco- nomic Review 111, 2550-93

    Goodman-Bacon, A. (2021), “The long-run effects of childhood insurance coverage: medicaid implementation, adult health, and labor market outcomes, American Eco- nomic Review 111, 2550-93

    work page 2021

  17. [17]

    Federalizing benefits: The introduc- tion of supplemental security income and the size of the safety net

    Goodman-Bacon, A. and L. Schmidt (2020), “Federalizing benefits: The introduc- tion of supplemental security income and the size of the safety net .”, Journal of Public Economics 185, 104174

    work page 2020

  18. [18]

    Difference-in-differences estimator of quan- tile treatment effect on the treated

    Kim, D. and J.M. Wooldridge (2024), “Difference-in-differences estimator of quan- tile treatment effect on the treated ”, Journal of Business and Economic Statistics , forthcoming

    work page 2024

  19. [19]

    How effective is the minimum wage at supporting the poor?

    MaCurdy, T. (2015), “How effective is the minimum wage at supporting the poor?” , Journal of Political Economy 123, 497–545

    work page 2015

  20. [20]

    The Impact of recentraliza- tion on public services: A difference-in-differences analysis of the ab olition of elected councils in Vietnam

    Malesky, E.J., C.V. Nguyen, and A. Trahn (2014), “The Impact of recentraliza- tion on public services: A difference-in-differences analysis of the ab olition of elected councils in Vietnam”, American Political Science Review 108 144–68. 31

    work page 2014

  21. [21]

    The changes-in-changes model with covariates

    Melly, B. and Santangelo (2015), “The changes-in-changes model with covariates”, working paper, Bern University

    work page 2015

  22. [22]

    When Is parallel trends sensitive to func- tional form?

    Roth, J. and P.H.C. Sant’Anna (2023), “When Is parallel trends sensitive to func- tional form?”, Econometrica 91 737–47

    work page 2023

  23. [23]

    An optimal transport ap- proach to estimating causal effects via nonlinear difference-in-diffe rences, working paper

    Torous, W., F. Gunsilius, and P. Rigollet (2024), “An optimal transport ap- proach to estimating causal effects via nonlinear difference-in-diffe rences, working paper”, University of California, Berkeley

    work page 2024

  24. [24]

    Simple approaches to nonlinear difference-in-differences with panel data

    Wooldridge, J.M. (2023), “Simple approaches to nonlinear difference-in-differences with panel data”, Econometric Journal 26 C31–66

    work page 2023

  25. [25]

    Analysis of contingency tables having ordered response categories

    Williams, O. D. and J.E. Grizzle (1972), “Analysis of contingency tables having ordered response categories”, Journal of the American Statistical Association 67, 55–63. 32

    work page 1972