Time-Varying Parameter Fixed Effects Model
Also known as: TVP-FE model, time-varying coefficients fixed effects, TVP panel model, locally time-varying fixed effects
The time-varying parameter fixed effects (TVP-FE) model extends the classical two-way fixed effects panel regression by allowing one or more slope coefficients to change over time while still controlling for unobserved individual heterogeneity. It is used when the effect of a predictor on an outcome is not constant across the time dimension of a panel dataset.
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 the TVP-FE model when theory or preliminary evidence suggests that the partial effect of a regressor on the outcome changes over the observation window — for example, because of structural breaks, regime changes, or gradual diffusion of a policy. It is also appropriate when a standard fixed effects Chow test or parameter stability test rejects constancy of slopes. Avoid it when the time series dimension T is short (T < 10), because the Kalman filter or kernel smoother cannot reliably track parameter trajectories with too few time points. Also avoid it when the panel is strongly unbalanced or when the assumption of a smooth parameter path is implausible.
Strengths & limitations
- Captures structural change and parameter instability without specifying a fixed break date.
- Retains the fixed effects within transformation, so results are robust to time-invariant omitted variables.
- Nests the standard fixed effects model as a testable special case, allowing a formal stability test.
- Compatible with large-N, moderate-T panels common in macroeconomics and international economics.
- Mean-group aggregation (Pesaran-Smith) provides a consistent pooled estimate even under cross-unit slope heterogeneity.
- Requires a sufficiently long time dimension (T ≥ 10 as a rough guideline) to identify parameter trajectories.
- Estimation via Kalman filter involves tuning or estimating the innovation variance Σ_η, which can be sensitive to starting values.
- Computationally more demanding than standard fixed effects, especially for large panels with many regressors.
- Interpretation is more complex: results are period-specific coefficient paths rather than a single summary slope.
Frequently asked
How is this different from a standard fixed effects model?
In a standard fixed effects model the slope coefficients are constrained to be the same in every time period; only the intercepts vary across units. The TVP-FE model keeps unit-specific intercepts but also lets the slopes change over time, capturing parameter instability without dropping the protection against time-invariant confounders.
How do I test whether I actually need time-varying slopes?
The most common approach is a parameter stability test such as a Chow test (for a known break date) or a Nyblom-type CUSUM test (for unknown or gradual change). Rejecting stability in these tests is the standard justification for moving from fixed effects to TVP-FE.
Can I use the TVP-FE model with a short panel (small T)?
Short panels are the main practical limitation. With T < 10 the Kalman filter or kernel smoother has very little information per unit to trace a time path, and the estimates become unreliable. For short panels, a structural break specification or a random-coefficient model may be more appropriate.
What software can estimate this model?
In Stata the tvpfe or xtmg packages support related estimators. In R the plm package covers standard panel models, while the dLagM and KFAS packages implement state-space and Kalman filter approaches. Python users typically combine the linearmodels and pykalman libraries.
Does the TVP-FE model handle cross-sectional dependence?
Not automatically. Cross-sectional dependence (e.g., common global shocks) can bias the time-varying parameter estimates. Augmenting the model with cross-sectional averages (Common Correlated Effects approach) or applying a factor-augmented specification before fitting the TVP-FE is the recommended remedy.
Sources
- Hsiao, C. (2014). Analysis of Panel Data (3rd ed.). Cambridge University Press. ISBN: 9781107038875
- Pesaran, M. H., & Smith, R. (1995). Estimating long-run relationships from dynamic heterogeneous panels. Journal of Econometrics, 68(1), 79-113. DOI: 10.1016/0304-4076(94)01644-F ↗
How to cite this page
ScholarGate. (2026, June 3). Time-Varying Parameter Fixed Effects Model. ScholarGate. https://scholargate.app/en/econometrics/time-varying-parameter-fixed-effects-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.
- Panel Fixed EffectsEconometrics↔ compare
- State Space ModelEconometrics↔ compare