Celsus provides an econometric analysis service for postgraduate researchers, academics and applied economists working on theses, dissertations, journal articles and funded projects. We specify, estimate and diagnose time-series, panel-data and cross-sectional econometric models in Stata or EViews, and deliver results you can defend in front of a supervisor, an examiner or a referee: the right estimator for your data, the pre-tests that justify it, a complete diagnostic battery, and publication-ready tables with interpretation. Our econometricians guide the modelling decisions with you; the research question, the theory and the final judgement remain yours.
What's included
- Time-series econometrics: stationarity and unit-root testing (ADF, PP, KPSS, structural-break tests), lag selection, ARIMA and VAR models, impulse-response and variance-decomposition analysis.
- ARDL and bounds testing: model specification with mixed I(0)/I(1) regressors, the F- and t-bounds tests, long-run coefficients, the error-correction model and CUSUM/CUSUMSQ stability diagnostics.
- Cointegration and causality: Engle–Granger and Johansen procedures, VECM estimation and interpretation of error-correction terms, Granger and Toda–Yamamoto causality tests.
- GARCH and volatility modelling: ARCH-family models including GARCH, EGARCH, GJR-GARCH and TGARCH, distributional assumptions, asymmetry and leverage effects, and volatility forecasting.
- Panel-data econometrics: pooled OLS, fixed and random effects with the Hausman test, cluster-robust and Driscoll–Kraay standard errors, dynamic panel (Arellano–Bond, system GMM) and panel unit-root and cointegration tests.
- Endogeneity and causal designs: instrumental variables and 2SLS with weak-instrument and over-identification tests, difference-in-differences, and regression models with heteroskedasticity- and autocorrelation-consistent inference.
- Diagnostics and robustness: serial correlation, heteroskedasticity, normality, functional-form and multicollinearity tests for every model, plus alternative specifications reported as robustness checks.
- Deliverables: estimation tables in APA 7 or journal house style, figures, a plain-language interpretation, and fully reproducible Stata do-files or EViews programs.
How we work
- Scope: we review your research questions, hypotheses, data frequency and sample structure, agree the model family and estimation strategy, and set a delivery date.
- Preparation: we clean and transform the series or panel, run the pre-tests (unit root, cross-sectional dependence, panel diagnostics) and confirm that the planned estimator is valid for your data — or recommend the correct alternative.
- Estimation: we estimate the model in Stata or EViews, run the full diagnostic battery and robustness checks, and keep every decision traceable in commented, reproducible code.
- Reporting and aftercare: we deliver tables, figures and interpretation in your target format, then support revisions and answer supervisor or reviewer queries on the econometric methodology.
Econometric methods we cover
| Model family | Example methods | Software | Typical deliverable |
|---|---|---|---|
| Unit root and stationarity | ADF, PP, KPSS, Zivot–Andrews, panel unit-root tests | EViews, Stata | Integration-order table with lag choice and deterministic terms |
| ARDL and bounds testing | ARDL(p,q), F-bounds and t-bounds tests, ECM, CUSUM | EViews, Stata | Long-run and short-run coefficient tables, ECT, stability plots |
| Cointegration and causality | Engle–Granger, Johansen, VECM, Granger, Toda–Yamamoto | EViews, Stata | Trace and max-eigenvalue tables, VECM output, causality matrix |
| Volatility models | ARCH, GARCH, EGARCH, GJR-GARCH, TGARCH | EViews, Stata | Variance-equation table, asymmetry tests, volatility forecasts |
| Static panel models | Pooled OLS, fixed effects, random effects, Hausman test | Stata | Estimator comparison table with robust standard errors |
| Dynamic panel models | Arellano–Bond, Blundell–Bond system GMM | Stata | GMM table with AR(2) and Hansen/Sargan diagnostics |
| Endogeneity and causal designs | IV/2SLS, difference-in-differences, event-study designs | Stata | Instrument diagnostics, treatment-effect table, parallel-trend plot |
Distribution of econometric projects
Related guides
If you would like to understand the methods before commissioning the analysis, our open-access econometrics guides walk through each one: stationarity and unit-root tests (ADF, PP, KPSS), the ARDL bounds test: when and how to use it, cointegration and causality: Johansen, VECM and Granger, GARCH volatility modelling and panel data: fixed or random effects and the Hausman test.
Frequently Asked Questions
How long does an econometric analysis take?
It depends on the number of series or panel units, the model family and how much pre-testing the data require. A single ARDL or fixed-effects model is usually completed within a few working days; multi-model studies with VECM, GARCH or system GMM take longer. We confirm a firm delivery date when we agree the scope.
How is the price determined?
We quote on the scope of the project rather than from a price list. After reviewing your data structure, the model family and the reporting format, we send a scoped fixed quote within one working day, with no surprise extra charges.
Is my data secure?
Yes. Your dataset is used solely for the agreed analysis, we sign a non-disclosure agreement (NDA) on request, and the work is carried out in line with UK GDPR and KVKK principles. All data and code are deleted on completion if you wish.
Is this service academically ethical, and how do I start?
Our service supports, never substitutes, the author's own work: we provide econometric method, estimation and reporting support, while you approve the modelling decisions and defend the results. To start, complete the contact form or message us on WhatsApp with a short description of your data and research question.