Research area · programme
Econometrics & Causal Inference
Deconfounding and causal discovery — separating what correlates from what an intervention changes. 26 papers in the vault, 13 drawn here.
13 papers drawn here · 43 curated relations touch the area · 0 authored papers. The graph opens straight on this view; the map and every other area stay one click away.
Loading the full graph…
Papers drawn in Econometrics & Causal Inference (13)
Every paper drawn in this area, sorted by curated degree. Click a title to open its ego graph on the canvas above, or follow the DOI / arXiv link to the publisher record.
| Paper | Authors | Year | Identifier / Venue | Status | Relations |
|---|---|---|---|---|---|
| Difference-in-Differences with Multiple Time Periods DiD estimator for staggered treatment timing with heterogeneous effects. It shows that two-way fixed-effects regressions can be badly biased when already-treated units serve as controls, and it estimates group-time ATTs aggregated into event-study effects. | Callaway & Sant'Anna | 2021 | DOI:10.1016/j.jeconom.2020.12.001 | peer-reviewed (vault claim) | 10 |
| Difference-in-differences with variation in treatment timing TWFE DiD with staggered timing is a weighted mix of all 2x2 comparisons, including already-treated units as controls with possibly negative weights. Warns against pooled before/after-upgrade rate ratios across cohorts. | Goodman-Bacon | 2021 | DOI:10.1016/j.jeconom.2021.03.014 | peer-reviewed (vault claim) | 7 |
| Estimating Dynamic Treatment Effects in Event Studies with Heterogeneous Treatment Effects Proves that two-way fixed effects (TWFE) event-study regressions under staggered adoption contaminate lead and lag coefficients with heterogeneous treatment effects from other cohorts, and introduces an interaction-weighted estimator. | Sun & Abraham | 2021 | DOI:10.1016/j.jeconom.2020.09.006 | peer-reviewed (vault claim) | 6 |
| What’s trending in difference-in-differences? A synthesis of the recent econometrics literature Review of modern DiD: staggered, heterogeneity, parallel-trends violations | Roth et al. | 2023 | DOI:10.1016/j.jeconom.2023.03.008 | peer-reviewed (vault claim) | 6 |
| The State of Applied Econometrics: Causality and Policy Evaluation Survey of applied causal econometrics; argues for supplementary analyses (placebo tests on pseudo-outcomes and pseudo-treatments, sensitivity to unconfoundedness, robustness) beside point estimates. Applied as placebo cluster labels (recurrence), org-level shuffled release labels (relrate) and placebo intervention dates (ITS). | Athey & Imbens | 2017 | DOI:10.1257/jep.31.2.3 | peer-reviewed (vault claim) | 5 |
| The wild bootstrap for few (treated) clusters Cluster-robust t and pairs/wild bootstraps over-reject when few clusters are treated; wild restricted (WCR) and subcluster bootstraps fix much of it. Relevant to relrate (2025.2 rests on 10-17 orgs), fs_infra Azure (6 orgs) and uppath paths (8-9 orgs): percentile pairs-cluster bootstraps there undercover. | MacKinnon & Webb | 2018 | DOI:10.1111/ectj.12107 | peer-reviewed (vault claim) | 5 |
| Sample Selection Bias as a Specification Error Sample selection bias is an omitted-variable problem: estimates from a self-selected sample are biased unless the selection mechanism is modelled. | Heckman | 1979 | DOI:10.2307/1912352 | peer-reviewed (vault claim) | 4 |
| Simple approaches to nonlinear difference-in-differences with panel data Extended TWFE / Poisson for nonlinear DiD with staggered adoption | Wooldridge | 2023 | DOI:10.1093/ectj/utad016 | peer-reviewed (vault claim) | 4 |
| Wild Bootstrap Inference for Wildly Different Cluster Sizes Wild cluster bootstrap with wildly unequal cluster sizes over-rejects when one cluster dominates; relevant to slownode (Rakuten 46% of flagged bundles) and fsync (top 5 orgs 54% of stalls). | MacKinnon & Webb | 2016 | DOI:10.1002/jae.2508 | peer-reviewed (vault claim) | 3 |
| A Practitioner’s Guide to Cluster-Robust Inference Cluster-robust inference guide: with few or unbalanced clusters, standard errors are too small; use cluster bootstrap / CR3 / wild cluster. | Cameron & Miller | 2015 | DOI:10.3368/jhr.50.2.317 | peer-reviewed (vault claim) | 2 |
| A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix HAC covariance used for the monthly ITS gap series (maxlags 2, small-sample t). With about 20 monthly points HAC SEs can be optimistic; show the CI and the minimum detectable effect rather than a significance verdict. | Newey & West | 1987 | DOI:10.2307/1913610 | peer-reviewed (vault claim) | 2 |
| Monotone Instrumental Variables: With an Application to the Returns to Schooling Monotone treatment selection (MTS) and monotone IV assumptions tighten worst-case Manski bounds. Applied to bundle-only rates (Bundle Insights bias/fs_prev): assuming bundled tickets are at least as severe as unbundled (MTS), the bundle-only rate is an upper bound on the queue rate, which the CE validation (26% vs 14%) supports; | Manski & Pepper | 2000 | DOI:10.1111/1468-0262.00144 | peer-reviewed (vault claim) | 1 |
| Some Statistical Models for Limited Dependent Variables with Application to the Demand for Durable Goods | Cragg | 1971 | DOI:10.2307/1909582 | peer-reviewed (vault claim) | 1 |
Curated semantic relation notes touching Causal Inference (15)
complements ↔ The State of Applied Econometrics: Causality and Policy Evaluation ↔ Negative Controls
placebo tests in econometrics are the pseudo-outcome/pseudo-exposure counterpart of epidemiological negative controls
Drafted by: agent draft (gemini)
complements ↔ Interrupted time series regression for the evaluation of public health interventions: a tutorial ↔ Difference-in-Differences with Multiple Time Periods
ITS for single-start interventions; staggered DiD when bot rollout differs by region/product
Drafted by: agent draft (gemini)
uses method → Interrupted time series regression for the evaluation of public health interventions: a tutorial → A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix
HAC SEs are a standard ITS autocorrelation remedy
Drafted by: agent draft (gemini)
extends → The wild bootstrap for few (treated) clusters → Wild Bootstrap Inference for Wildly Different Cluster Sizes
Few treated clusters case of the wild cluster bootstrap.
Drafted by: agent draft (gemini)
extends → Wild Bootstrap Inference for Wildly Different Cluster Sizes → Bootstrap-Based Improvements for Inference with Clustered Errors
Wild cluster bootstrap behaviour with few/unbalanced clusters.
Drafted by: agent draft (gemini)
extends → Training, Wages, and Sample Selection: Estimating Sharp Bounds on Treatment Effects → Monotone Instrumental Variables: With an Application to the Returns to Schooling
Both tighten worst-case selection bounds by monotonicity; MTS/MIV on outcomes vs Lee's monotone selection into observation.
Drafted by: agent draft (gemini)
complements ↔ A More Credible Approach to Parallel Trends ↔ What’s trending in difference-in-differences? A synthesis of the recent econometrics literature
honest DiD sensitivity is one of the tools synthesised in the review
Drafted by: agent draft (gemini)
complements ↔ Estimating Dynamic Treatment Effects in Event Studies with Heterogeneous Treatment Effects ↔ Difference-in-Differences with Multiple Time Periods
Interaction-weighted event study; alternative heterogeneity-robust estimator for staggered adoption.
Drafted by: agent draft (gemini)
complements ↔ Difference-in-Differences with Multiple Time Periods ↔ The use of controls in interrupted time series studies of public health interventions
Controlled ITS and staggered DiD are complementary designs for evaluating a rollout with a comparison group.
Drafted by: agent draft (claude)
extends → Difference-in-Differences with Multiple Time Periods → Difference-in-differences with variation in treatment timing
CS ATT(g,t) estimator is the remedy for the TWFE weighting problem Goodman-Bacon decomposes.
Drafted by: agent draft (gemini)
complements ↔ The Wild Bootstrap with a “Small” Number of “Large” Clusters ↔ The wild bootstrap for few (treated) clusters
Theory for the wild bootstrap with few large clusters.
Drafted by: agent draft (gemini)
extends → A Comparison of Alternative Models for the Demand for Medical Care → Some Statistical Models for Limited Dependent Variables with Application to the Demand for Durable Goods
The RAND two-part model generalises Cragg's two-tier limited-dependent-variable model to medical spending.
Drafted by: agent draft (claude)
extends → Inference for Nonprobability Samples → Sample Selection Bias as a Specification Error
Extends classical two-step selection correction to propensity-weighted inference combining non-probability convenience samples with probability reference surveys.
Drafted by: agent draft (gemini)
contrasts with ↔ Difference-in-differences with variation in treatment timing ↔ Difference-in-Differences with Multiple Time Periods
Bacon diagnoses TWFE bias under staggered timing; CS gives a clean ATT(g,t) estimator
Drafted by: agent draft (gemini)
complements ↔ Sample Selection Bias as a Specification Error ↔ Training, Wages, and Sample Selection: Estimating Sharp Bounds on Treatment Effects
Selection model vs bounds under sample selection
Drafted by: agent draft (gemini)
Full vault context
This page lists the 13 drawn papers and 15 semantic relation notes touching Econometrics & Causal Inference. For the site-wide four evidence chains, 3 explicit refutations, retracted claim, and all 23 authored records, return to The Research Network overview →
Take the data
The graph you are looking at is a static JSON extract of the vault, served from this origin with no runtime requests. Every count on this page is reproducible from these files and scripts/build-research-graph.mjs.