CoChat

When Is Parallel Trends Sensitive to Functional Form?

Roth, Jonathan, Sant'Anna, Pedro H. C. · 2023 · Econometrica

The DOI matches its Crossref record; 25 of the 25 references it cites were found in a registry or an index; 2 of the 10 sources its claims cite were checked for support.

Abstract

This paper assesses when the validity of difference‐in‐differences depends on functional form. We provide a novel characterization: the parallel trends assumption holds under all strictly monotonic transformations of the outcome if and only if a stronger “parallel trends”‐type condition holds for the cumulative distribution function of untreated potential outcomes. This condition for parallel trends to be insensitive to functional form is satisfied if and essentially only if the population can be partitioned into a subgroup for which treatment is effectively randomly assigned and a remaining subgroup for which the distribution of untreated potential outcomes is stable over time. These conditions have testable implications, and we introduce falsification tests for the null that parallel trends is insensitive to functional form.

Overview

CoChat Summary

## TL;DR

The paper characterizes exactly when the parallel trends assumption in difference-in-differences is insensitive to outcome transformations: this happens if and only if a corresponding parallel-trends condition holds for the untreated outcome distributions’ CDFs. It also shows this condition is satisfied only in special cases, including random assignment, stationary untreated outcomes, or a mixture of the two.

## Key findings

  • Parallel trends is invariant to strictly monotonic transformations if and only if a parallel-trends-type condition holds for the CDFs of untreated potential outcomes.
  • The paper characterizes this invariance as essentially requiring one of three structures:

1. treatment is as-if randomly assigned,

2. untreated outcome distributions are stable over time within each group, or

3. a hybrid mixture of the first two cases.

  • The authors show the CDF-based condition has testable implications: when the outcome has finite support, the null can be tested by checking whether the implied distribution has non-negative mass at all support points.
  • In the minimum-wage illustration using Cengiz et al. (2019), the implied counterfactual density is negative for wages around roughly $5–7/hour in the 2007–2015 comparison, and the null that all implied densities are positive is rejected with p < 0.001; for the 2010–2015 comparison, the null is not rejected (p = 0.29).

## Methods

  • The paper develops a two-period canonical difference-in-differences model with treated and comparison populations, no anticipatory effects, and untreated potential outcomes indexed by group and time.
  • It proves a characterization result linking invariance of parallel trends under all strictly monotonic transformations to a distributional parallel-trends condition for untreated potential outcomes, plus a decomposition result showing when that condition can hold.
  • For empirical testing, it proposes falsification tests based on whether the implied counterfactual distribution is weakly increasing; in the illustration, the authors use state-level minimum-wage and employment-to-population-ratio data from Cengiz et al. (2019) and bootstrap-based moment-inequality testing.

## Limitations

  • The paper states that failing to reject the null does not imply parallel trends is insensitive to functional form; this is a limitation of the proposed tests.
  • It also warns that relying on such tests may create pre-testing problems, analogous to issues in pre-trends testing.
  • Evident from the text: the main characterization is developed for a two-period model, with extensions only noted as immediate implications for multi-period, staggered-adoption, and conditional parallel-trends settings rather than worked out in detail.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3982/ecta19402

Cites

25 of 25 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

2 Valid · 1 Not valid · 5 Not checked · 17 Own results and interpretations

12 more claims

Create a free account to see all 25 claims and to verify the 1 reference still open.

See all claims — free account
  • Probably not a substantial claim. 2
  • It may state two separate claims as one. 6

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

Show all 25 references

Cited by

Metadata

Authors
Roth, J., & Sant'Anna, P. H. C.
Affiliations
  • Brown University (US)
  • Microsoft Research
Venue
Econometrica
Published
2023
Volume
91
Issue
2
Pages
737-747
ISSN
0012-9682 (ISSN-L)
OpenAlex
W3091865267
Cited by
142
References
20
Citations by year
2026: 35 · 2025: 46 · 2024: 31 · 2023: 23
FWCI
27.39
Type
journal article
Language
en
Open access
closed
Locations
Econometrica (publishedVersion)
Topics
  • Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
  • Social Sciences › Economics, Econometrics and Finance › Economics and Econometrics › Spatial and Panel Data Analysis
  • Social Sciences › Public Administration › Evaluation and Performance Assessment
Keywords
parallel trends assumption, difference-in-differences, monotone transformations, cumulative distribution function, falsification tests, random assignment, treatment effects

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.