SsfPack — State Space Models

About SsfPack

SsfPack is a suite of C routines for carrying out computations involving the statistical analysis of time series models in state space form. SsfPack provides functions for likelihood evaluation and signal extraction of arbitrary user-specified linear Gaussian state space models, allowing the estimation of models ranging from simple time-invariant univariate forms to complicated time-varying multivariate specifications.

Basic functions are available for prediction, moment smoothing and simulation smoothing. Additionally, functions are provided which put standard models such as autoregressive moving average (ARMA), non-stationary ARIMA, time-varying regression, unobserved components (UC) and cubic spline models in state space form. These functions are also provided for multivariate extensions of these models including vector autoregressive (VAR) and dynamic factor models.

The functions from SsfPack can be easily used for implementing, fitting and analysing linear Gaussian models relevant to many areas of econometrics, statistics, data science, time series analysis and forecasting. Further, SsfPack provides tools for analysing and estimating non-Gaussian and nonlinear models for binary, count, categorical and heavy-tailed or extreme data, using simulation-based estimation methods such as importance sampling and Markov chain Monte Carlo (MCMC) methods.

SsfPack is primarily developed as a module for the object-oriented matrix programming language Ox. The library is written in C, which greatly improves execution speed compared to implementation on other software platforms. A free version of Ox/SsfPack for academic research and teaching purposes can be downloaded from this website.

In 2026, an initiative is launched to develop the new version 4 of SsfPack that can also be called from other platforms including Python and R. The latest developments are reported on this website.

Authors

Siem Jan Koopman: is Professor of Econometrics at the VU University Amsterdam and Tinbergen Institute Amsterdam. He is author of the textbooks Time Series Analysis by State Space Methods (with J. Durbin) and An Introduction to State Space Time Series Analysis (with J. Commandeur).
Neil Shephard: is the Frank B. Baird Jr. Professor of Science working in the Departments of Economics and Statistics at Harvard University. He is a Fellow of the Econometric Society, the British Academy, the Society for Financial Econometrics and the International Association for Applied Econometrics.
Jurgen A. Doornik: is a Research Fellow of Nuffield College, Oxford. He is the main developer of OxMetrics, the originator of Ox, an object-oriented matrix programming language, and author (with David F. Hendry) of PcGive and PcFiml.