Skip to contentScholarGate
LibraryBookshelfDeskReview StudioAssistant
Sign in
On this page
IntuitionHow it worksWhen to use itStrengths & limitationsCommon pitfallsApplicationsFrequently asked🔒 Read the full methodSourcesRelated methods
Cite this pageSpotted an issue on this page? Report or suggest a fix →
Home›Econometrics›Time-Varying Parameter MA Model
Regression modelEconometrics / time series

Time-Varying Parameter MA Model

Time-Varying Parameter Moving Average Model · Also known as: TVP-MA model, state-space MA, Kalman filter MA, time-varying MA

The time-varying parameter moving average (TVP-MA) model extends the standard MA model by allowing the moving-average coefficients to change over time. Cast as a state-space system, it is estimated via the Kalman filter and smoother, making it well suited for series where the shock-transmission dynamics evolve across the sample.

ScholarGate
  1. Regression model
  2. v1
  3. 2 Sources
  4. PUBLISHED
Cite this page →
Tools & resources
Download slides
Learn & explore

Read the full method

Members only

Sign in with a free account to read this section.

Sign in

Method map

The neighbourhood of related methods — select a node to explore.

Time-varying parameter MA model
ARMA modelKalman FilterMoving Average ModelTime-varying parameter A…Time-varying parameter A…Time-varying parameter A…

When to use it

Use the TVP-MA model when residual analysis or economic reasoning suggests that the lag structure of shocks changes over time — for example, around business cycles, financial crises, or major policy interventions. It is appropriate when standard MA diagnostics reveal regime-dependent dynamics or when rolling-window MA estimates drift noticeably. Do not use it when the sample is short (fewer than roughly 80 observations), as the state noise covariance Q cannot be estimated reliably. Also avoid it if a constant-parameter MA model passes all misspecification tests, since the TVP variant adds substantial complexity without benefit.

Strengths & limitations

Strengths
  • Captures evolving shock-persistence dynamics without imposing discrete break dates.
  • Unified state-space framework allows exact likelihood evaluation and formal hypothesis testing on parameter stability.
  • Kalman smoother provides a full historical trajectory of time-varying MA coefficients, useful for structural interpretation.
  • Naturally extends to MA components embedded in larger TVP-ARIMA or TVP-VARMA models.
  • Robust to gradual structural change, which fixed-coefficient MA models would misspecify.
Limitations
  • Requires a moderately large sample (at least 80 observations) for reliable identification of Q.
  • Computationally more demanding than OLS-estimated MA models, particularly for higher MA orders.
  • The random-walk assumption for coefficients may be too flexible, causing over-fitting in low-signal series.
  • Identification of the state noise covariance Q is weak when the true parameters are nearly constant.
  • Model selection (choosing q and specifying Q structure) adds non-trivial complexity.

Frequently asked

How does a TVP-MA model differ from a regime-switching MA model?

A regime-switching MA model assumes parameters jump between a fixed number of discrete states, whereas the TVP-MA allows coefficients to drift continuously via a random walk. TVP-MA is preferable when structural change is gradual; regime-switching is better when distinct, abrupt regimes are expected.

Can I include both AR and MA components with time-varying parameters?

Yes. The natural extension is the TVP-ARMA model, where both autoregressive and moving-average coefficients evolve as latent states in the Kalman filter framework. Each component adds states, increasing the computational burden.

How do I choose the MA order q in a TVP-MA model?

Start with the same criteria used for a constant-parameter MA: examine the ACF of the series, and compare models by information criteria (AIC, BIC) or out-of-sample forecast accuracy. Given the added complexity of time-varying coefficients, prefer parsimonious orders.

What software can estimate a TVP-MA model?

R packages such as KFAS and dlm support general state-space models including TVP-MA specifications. Python users can use statsmodels or pykalman. MATLAB's Econometrics Toolbox also provides state-space estimation routines.

Is stationarity of the observed series required?

The innovations ε_t should be stationary. If y_t is integrated, first-difference it before fitting the TVP-MA model to ensure the error process is well-behaved for the Kalman filter recursions.

Sources

  1. Harvey, A. C. (1990). Forecasting, Structural Time Series Models and the Kalman Filter. Cambridge University Press. ISBN: 9780521321969
  2. Durbin, J., & Koopman, S. J. (2012). Time Series Analysis by State Space Methods (2nd ed.). Oxford University Press. ISBN: 9780199641178

How to cite this page

ScholarGate. (2026, June 3). Time-Varying Parameter Moving Average Model. ScholarGate. https://scholargate.app/en/econometrics/time-varying-parameter-ma-model

Related methods

ARMA modelKalman FilterMoving Average ModelTime-varying parameter AR modelTime-varying parameter ARIMA modelTime-varying parameter ARMA 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.

  • ARMA modelEconometrics↔ compare
  • Kalman FilterBayesian↔ compare
  • Moving Average ModelEconometrics↔ compare
  • Time-varying parameter AR modelEconometrics↔ compare
  • Time-varying parameter ARIMA modelEconometrics↔ compare
  • Time-varying parameter ARMA modelEconometrics↔ compare
Compare side by side →

Similar methods

Time-varying parameter ARMA modelTime-varying parameter ARIMA modelTime-varying parameter AR modelTime-varying parameter SARIMA modelTime-varying parameter ARCH modelTime-varying parameter OLSTime-varying parameter WLSTime-varying parameter GLS

Related reference concepts

Time-Series Models • Dynamic Quantile Regressions • Dynamic Treatment Effect Models • Diffusion Processes • State Space ModelsGaussian Process ModelsBayesian Model AveragingHidden Markov ModelsTime-Series Models • Dynamic Quantile Regressions • Dynamic Treatment Effect Models • Diffusion ProcessesEconometrics

Spotted an issue on this page? Report or suggest a fix →

ScholarGate — Time-varying parameter MA model (Time-Varying Parameter Moving Average Model). Retrieved 2026-07-21 from https://scholargate.app/en/econometrics/time-varying-parameter-ma-model · Dataset: https://doi.org/10.5281/zenodo.20539026
Quick facts
Originator
Harvey, A. C.; Durbin, J. & Koopman, S. J.
Year
1990s
Type
Time-varying state-space model
DataType
Univariate time series
Subfamily
Econometrics / time series
Related methods
ARMA modelKalman FilterMoving Average ModelTime-varying parameter AR modelTime-varying parameter ARIMA modelTime-varying parameter ARMA model
ScholarGate

A content-first reference library for research methods — what each one is, how it works, and where it comes from.

Open data (CC-BY)

Explore

  • Library
  • Search the library…
  • Browse by field
  • Fields
  • Journey
  • Compare
  • Which method?

Reference

  • Subjects
  • Atlas
  • Glossary
  • Methodology
  • Philosophy

Your tools

  • Bookshelf
  • Desk
  • Chat

Company

  • About
  • Pricing
  • Contact
  • Suggest a method

Entries are compiled from published sources for reference. Verifying the accuracy and suitability of any information for your own use remains your responsibility.

© 2026 ScholarGate · A research-method reference library
  • Privacy
  • Cookies
  • Terms
  • Delete account