Skip to content

brycewang-stanford/Auto-Empirical-Research-Skills / skills/30-zirui-song-claude-skills/robustness

robustness

Verified

Checklist of empirical robustness tests for finance/economics papers

Robustness Test Checklist

Systematic checklist of robustness tests for empirical research. Use this to ensure comprehensive testing before submission.


Core Robustness Categories

1. Alternative Samples

TestDescriptionWhen to Use
Exclude outliersWinsorize/trim at different levels (0.5%, 2%, 5%)Always
Drop financial firmsExclude SIC 6000-6999If not already excluded
Drop regulated industriesExclude utilities, telecomsIndustry-specific effects
Different time periodsSplit sample pre/post crisis, early/lateResults may be period-specific
Geographic subsamplesBy region, state, countryExternal validity
Size subsamplesSmall vs. large firmsHeterogeneous effects
Balanced panelRequire continuous observationsSurvivorship concerns

2. Alternative Specifications

TestDescriptionWhen to Use
Different fixed effectsFirm, industry×year, state×yearControl for unobservables
Additional controlsAdd variables referees might suggestOmitted variable concerns
Drop controlsVerify not over-controllingMediator concerns
Different clusteringFirm, industry, state, two-wayInference robustness
Different standard errorsBootstrap, Newey-West, Driscoll-KraaySerial/cross-sectional correlation
Nonlinear specificationsQuadratic terms, splinesLinearity assumption
Log vs. levelTransform dependent variableSkewed distributions

3. Alternative Measures

TestDescriptionWhen to Use
Alternative dependent variableDifferent proxy for same conceptMeasurement concerns
Alternative treatment measureContinuous vs. binary, different thresholdTreatment definition
Alternative control measuresDifferent proxies for size, leverage, etc.Standard practice
Scaled differentlyBy assets, sales, employeesScaling choice matters

4. Identification Tests

TestDescriptionWhen to Use
Placebo/Falsification
Placebo timingFake treatment 1-3 years before actualDiD parallel trends
Placebo outcomeEffect on outcome that shouldn't be affectedSpecificity of mechanism
Placebo treatmentRandom assignment of treatmentRule out spurious correlation
Pre-trends
Event study plotCoefficient for each pre/post periodVisual parallel trends
Joint F-testTest pre-period coefficients = 0Statistical parallel trends
Endogeneity
Instrumental variablesFind exogenous variationSelection concerns
Heckman selectionModel selection explicitlySample selection
Propensity score matchingMatch treated/controlObservable selection
Entropy balancingReweight to balance covariatesCovariate imbalance
Regression discontinuityIf threshold existsSharp identification

5. Inference Robustness

TestDescriptionWhen to Use
Wild cluster bootstrapSmall number of clusters<50 clusters
Randomization inferencePermutation-based p-valuesFew treated units
Conley standard errorsSpatial correlationGeographic data
Multiple hypothesis correctionBonferroni, FDRMany outcomes tested

DiD-Specific Tests

For difference-in-differences designs:

  • Event study with pre-treatment coefficients
  • Parallel trends test (formal)
  • Bacon decomposition (staggered treatment)
  • Callaway-Sant'Anna or Sun-Abraham estimator (heterogeneous treatment effects)
  • Placebo treatment timing
  • Vary treatment window
  • Triple-difference if possible
  • Exclude always-treated or never-treated

IV-Specific Tests

For instrumental variables:

  • First-stage F-statistic (>10, prefer >100)
  • Weak instrument robust inference (Anderson-Rubin)
  • Overidentification test (if multiple IVs)
  • Exclusion restriction discussion
  • Reduced form results
  • Compare OLS vs. IV magnitudes

Quick Commands

  • "robustness checklist" - Full checklist for current paper
  • "DiD robustness" - DiD-specific tests only
  • "what tests for [method]" - Tests for specific identification strategy
  • "referee-proof" - Most commonly requested tests
  • "prioritize tests" - Rank by importance for your setting

Referee-Proof Minimum

At minimum, most papers should include:

  1. Sample robustness: Exclude outliers, alternative time periods
  2. Specification robustness: Alternative fixed effects, controls
  3. Measurement robustness: Alternative variable definitions
  4. Inference robustness: Alternative clustering
  5. Identification test: At least one placebo or pre-trend test

Reporting Template

For robustness tables:

Table X: Robustness Tests
Panel A: Alternative Samples
  (1) Baseline
  (2) Exclude financial firms
  (3) Exclude 2008-2009
  (4) Winsorize at 5%

Panel B: Alternative Specifications
  (5) Add industry×year FE
  (6) Control for firm age
  (7) Cluster by industry

Panel C: Alternative Measures
  (8) Alternative dependent variable
  (9) Continuous treatment measure
finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the workStars: 322kUpdated Claude CodeCodexVerifiedInstall
systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixesStars: 322kUpdated Claude CodeCodexVerifiedInstall
test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation codeStars: 322kUpdated Claude CodeCodexVerifiedInstall
skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.Stars: 198kUpdated Claude CodeCodexVerifiedInstall