ARBIMETRA LAB
Augmented Dickey-Fuller Test
Tests whether a series has a unit root, with a choice of deterministic terms and lag selection.
What it does
Tests whether a series has a unit root, with a choice of deterministic terms and lag selection.
Use it before modelling levels of economic or financial series.
Data you need
- One series
Optional: Lag selection
Supported data: Time series. Level: Intermediate.
Formula and method
Δyₜ = α + βt + γyₜ₋₁ + Σδᵢ Δyₜ₋ᵢ + εₜ, H₀: γ = 0
The change in the series is regressed on its lagged level, optional deterministic terms and lagged changes that absorb autocorrelation. The t-statistic on the level is compared with MacKinnon's non-standard critical values.
Assumptions
- Rows are in time order
- No structural breaks in the sample
Input checks
- At least 20 observations
- Unbroken series
How to read the result
Rejecting H₀ supports stationarity.
Worked example
Illustrative run on US quarterly macro 1959-2009. These figures are the actual output of this model on that example, and they say nothing about your own data.
- Statistic = -3.0545, p = 0.0301.
- The evidence points to stationarity at the 5% level, under a constant.
Limitations
- Low power in short samples
Numerical validation
Validated 18 of 18 automated checks passed, including numerical results, invalid input, missing values, small samples, constant variables, charts and the interpretation text.
| Check | Reference | Result | Abs. error | Tolerance | Source |
|---|---|---|---|---|---|
| ADF t-statistic (c) | -3.0545145 | -3.0545145 | 6.2e-15 | 1e-9 | Independent NumPy calculation |
| ADF t-statistic (ct) | -3.14857705 | -3.14857705 | 2.1e-14 | 1e-9 | Independent NumPy calculation |
| ADF t-statistic (n) | -1.69941661 | -1.69941661 | 1.3e-15 | 1e-9 | Independent NumPy calculation |
Computed with python 3.13.2, numpy 2.2.5, scipy 1.14.1, pandas 2.3.3, statsmodels 0.14.4, scikit-learn 1.7.0, in your browser. A passing check confirms the listed comparison only. It does not make any result suitable for your data, and it is not a claim of equivalence with other statistical software.
Questions
What if ADF and KPSS disagree?
They have opposite null hypotheses, so disagreement often reflects limited power. Examine the series and consider both results.
ARBIMETRA LAB is an educational and research tool. Results should be checked by the researcher and are not professional advice.
