Robust ARIMA Model
Robust Autoregressive Integrated Moving Average Model · Also known as: robust ARIMA, outlier-resistant ARIMA, robust time series estimation, ARIMA with outlier detection
Robust ARIMA extends the classical ARIMA framework to detect and correct the influence of outliers and structural breaks during estimation. By jointly identifying anomalous observations and re-estimating model parameters, it produces coefficient estimates and forecasts that are far less distorted by isolated shocks or data errors than standard ARIMA.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use Robust ARIMA when you have a univariate time series that you suspect contains outliers, recording errors, structural breaks, or sharp one-off events (e.g. financial crises, policy changes, pandemics). It is particularly valuable when forecasting accuracy matters and standard ARIMA residuals show heavy tails or a few very large errors. It is less necessary when the series is clean and outlier-free, or when the series is short (fewer than 50 observations), which limits the reliability of outlier detection. Do not use it as a substitute for modelling permanent structural change — persistent regime shifts may need intervention analysis or a regime-switching model instead.
Strengths & limitations
- Produces parameter estimates and forecasts that are resistant to the distorting influence of outliers and level shifts.
- Automatically distinguishes between different outlier types (AO, IO, LS, TC), enabling richer post-hoc interpretation of anomalies.
- Iterative joint estimation ensures that outlier effects and ARIMA coefficients are mutually consistent rather than estimated in isolation.
- Retains the interpretability and forecasting power of the ARIMA framework while adding robustness.
- Widely supported in statistical software (R rob-arima, SAS, Eviews), lowering the implementation barrier.
- Outlier detection depends on a critical threshold; too low a threshold flags genuine variation as outliers, too high misses real anomalies.
- Iterative detection-estimation cycles increase computational cost and can converge to local optima.
- Short series (n < 50) provide insufficient information for reliable outlier identification alongside ARIMA parameter estimation.
- Does not handle multiple structural breaks or regime shifts as effectively as dedicated models such as Markov-switching ARIMA.
Frequently asked
How is Robust ARIMA different from standard ARIMA?
Standard ARIMA estimates parameters using all observations equally, so large outliers pull coefficients and forecasts off course. Robust ARIMA detects anomalous observations, models their effects as intervention terms, and re-estimates ARIMA parameters from the cleaned signal, yielding more reliable estimates when outliers are present.
What are the four outlier types and why do they matter?
Additive outliers (AO) affect a single time point only; innovational outliers (IO) propagate through the MA structure; level shifts (LS) cause a permanent step change; temporary changes (TC) decay back to baseline. Correctly classifying the type guides both the statistical correction and the substantive interpretation of the event.
How do I choose the critical threshold for outlier detection?
Common choices are 3.0 to 4.0 standard deviations. For large samples (n > 200) a threshold of 3.0–3.5 balances sensitivity and specificity; for small samples (50–100 observations) prefer 3.5–4.0 to limit false discoveries. Sensitivity analysis across thresholds is recommended.
Can Robust ARIMA handle seasonal data?
Yes. The approach extends naturally to Robust SARIMA by applying the same outlier detection and joint estimation logic to the seasonal ARIMA specification. Most software implementations support both non-seasonal and seasonal variants.
When should I use an intervention model instead?
If you know the exact timing and expected pattern of disruptions in advance (e.g. a known policy date), a pre-specified intervention model is more efficient and interpretable. Robust ARIMA is most useful when the locations and types of anomalies are unknown and need to be discovered from the data.
Sources
- Tsay, R. S. (1986). Time series model specification in the presence of outliers. Journal of the American Statistical Association, 81(393), 132–141. DOI: 10.1080/01621459.1986.10478250 ↗
- Chen, C., & Liu, L.-M. (1993). Joint estimation of model parameters and outlier effects in time series. Journal of the American Statistical Association, 88(421), 284–297. DOI: 10.2307/2290724 ↗
How to cite this page
ScholarGate. (2026, June 3). Robust Autoregressive Integrated Moving Average Model. ScholarGate. https://scholargate.app/en/econometrics/robust-arima-model
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- ARIMA modelEconometrics↔ compare
- Quantile RegressionEconometrics↔ compare
- SARIMA modelEconometrics↔ compare
- State Space ModelEconometrics↔ compare