Data-Driven Models
Data-Driven Models
Empirical forecasting models · Sources
Primary papers, model variants, source notes, and review signals behind the Unobserved components model page.
Reference material used for orientation; read primary and academic sources first when claims conflict.
[S1] Reference
Harvey (1985) introduced the structural time-series model and showed how to cast it in state-space form for Kalman filter estimation.
Reference
[S2] Reference
Harvey (1989) provided the definitive textbook treatment: local-level, local-linear-trend, cycle, and trigonometric seasonal components with full MLE methodology.
Reference
[S3] Reference
Clark (1987) decomposed U.S. GDP into trend and cycle using a bivariate UCM, producing one of the first model-based output-gap estimates.
Reference
[S4] Reference
Watson (1986) analyzed the identification and estimation of the UC-trend model and its relationship to the Beveridge-Nelson decomposition.
Reference
[S5] Reference
Durbin and Koopman (2012) gave a modern treatment covering diffuse initialization, simulation smoothing, and diagnostics for state-space models.
Reference
[S6] Reference
Morley, Nelson, and Zivot (2003) demonstrated that allowing correlated trend-cycle disturbances in the UCM reverses the standard finding about the relative importance of permanent versus transitory shocks.
Reference
[S7] Reference
Commandeur and Koopman (2007) provided a practical introduction to structural time-series analysis with worked examples in SsfPack and STAMP.
Reference
Continue reading
Open the concept, data series, policy setting, or neighboring model that anchors this page.