Time-Varying Parameter Factor-Augmented VAR
Also known as: Dynamic factor model with time-varying parameters
TVP-FAVAR is a hybrid framework combining factor-augmented VARs with time-varying parameter estimation via Kalman filtering. Introduced by Bernanke et al. (2005) and refined by Primiceri (2005), it extracts latent economic factors (e.g., a 'common monetary policy shock') from high-dimensional data while allowing VAR coefficients to evolve stochastically over time. This framework captures both reduced-dimensionality patterns and structural instability, making it ideal for studying evolving policy regimes and shock dynamics.
Key highlights
- Jointly handles dimensionality reduction and parameter instability
- Kalman filter provides real-time inference and one-step-ahead predictions
- Time-varying impulse responses reveal evolving shock transmission
- Can distinguish between stable factors and shifting relationships
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use TVP-FAVAR when you have high-dimensional data and suspect time-varying structural relationships. It is particularly useful for monetary policy analysis (where transmission mechanisms evolve) and macroeconomic forecasting. Requires sufficient sample size (at least 100–150 observations) to reliably estimate time-varying parameters.
Strengths & limitations
- Jointly handles dimensionality reduction and parameter instability
- Kalman filter provides real-time inference and one-step-ahead predictions
- Time-varying impulse responses reveal evolving shock transmission
- Can distinguish between stable factors and shifting relationships
- Computationally intensive; requires specialized software or careful implementation
- High-dimensional parameter space can lead to overparameterization
- Results can be sensitive to priors on time-variation magnitude
- Interpretation of latent factors can be ambiguous or time-varying
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
How many factors should I extract?
Use information criteria (AIC/BIC) on a preliminary static FAVAR, or choose based on the fraction of variance explained (often 80–90%). Start conservatively (2–3 factors); add factors if diagnostics show remaining patterns in residuals.
How do I specify priors on time-variation?
Common choices: normal priors on innovations to log-volatility of coefficients, or inverse-Gamma priors on the variance of innovations. Use empirical priors (estimated from data) or sensitivity-check against diffuse priors.
Can I identify shocks in a TVP-FAVAR?
Yes. Use sign restrictions or zero restrictions (zeros in factor loadings or VAR) to identify shocks. Time-varying identification is more challenging; assume stable identification structure across time unless testing otherwise.
How do I visualize time-varying parameters?
Plot factor loadings, VAR coefficients, or impulse responses over time with confidence bands from the Kalman filter. Highlight periods of significant change; relate them to known economic events for validation.
Sources
- 1.Bernanke, B. S., Boivin, J., & Eliasz, P. S. (2005). Measuring monetary policy. Journal of Political Economy, 113(1), 161-208.
- 2.Primiceri, G. E. (2005). Time-varying structural vector autoregressions and monetary policy. Review of Economic Studies, 72(3), 821-852.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 3). TVP-FAVAR. ScholarGate. https://scholargate.app/econometrics/tvp-favar