From panel-data models and difference-in-differences to IV, RDD, and time-series forecasting, we deliver defendable econometric analyses with transparent assumptions and reproducible code.
Functional form, interactions, clustered/robust SEs, partial effects.
Unit/time FE, Hausman tests, dynamic panels (intro), clustered SEs.
2x2 & staggered adoption (ATT), parallel trends checks, event studies.
2SLS, validity (relevance/exogeneity), weak IV diagnostics.
Sharp/fuzzy RDD, bandwidth & polynomial choices, McCrary test.
Treated vs synthetic comparison; donor pool selection & placebo tests.
ARIMA/ETS/VAR (intro-mid), stationarity tests, seasonality, impulse responses.
Cross-validation, rolling windows, accuracy metrics with CIs.
| Area | Examples | Notes |
|---|---|---|
| Identification | DiD, IV, RDD, Synthetic | State assumptions; run robustness & placebo tests |
| Panel | FE/RE, clustered SE | Check serial correlation (Wooldridge), heteroskedasticity |
| IV | 2SLS, F-stat, Sargan/Hansen | Weak IV; over-identification diagnostics |
| RDD | Local linear, optimal h | Continuity checks, manipulation tests |
| Time-series | ARIMA/VAR | Stationarity (ADF/KPSS), lag selection, residual checks |
| Reporting | CIs, elasticities | Effect sizes, units, economic significance |
We document every modeling choice so your results are reproducible and defensible.
Critique/strengthen your causal strategy + robustness checklist.
EDA, core models (e.g., FE+DiD/IV) + 4–6 tables/figures + write-up.
Time-series/VAR + scenario forecasts + policy memo + plots.
Pricing varies with data condition, model complexity, and turnaround. We provide a clear plan after discovery.
Question, data, institutional/journal requirements.
Identification strategy, robustness plan, pre-analysis outline.
Merges, transformations, codebook; versioned files.
Model fits + diagnostics; feedback loops on interim results.
Tables (estout/gt/texreg), event studies/coeff plots, write-up.
Scripts/notebooks, figures, and change log; submission tips.
Yes we evaluate your data and question to propose an identification strategy with stated assumptions.
Yes modern ATT estimators and event-study plots with correct weighting.
Cluster-robust SEs, Newey-West (time-series), or multi-way clustering depending on design.
We run weak-IV diagnostics and recommend alternatives (stronger instruments or designs).
Yes data-driven bandwidths, polynomial order checks, and manipulation tests (e.g., McCrary).
Intro-to-intermediate ARIMA/VAR with stationarity checks, lag selection, and forecast evaluation.
Absolutely scripts/notebooks, tables/figures, and a change log for reproducibility.
Yes LaTeX (stargazer/texreg/gt) and Word-ready tables, plus publication-style figures.
Small studies: days; full chapters/manuscripts: weeks with agreed milestones.
By scope, data complexity, and timeline. We send a transparent plan and quote after discovery.