GAUSS State-Space Modeling (SSLIB)#

Description#

This package provides tools for estimating and evaluating time-invariant state-space models. In addition to providing the tools for custom state space models, it provides pre-built functions for state-space modeling of:

ARIMA packages
SARIMA packages

Installation#

The GAUSS State Space library can be installed and updated directly in GAUSS using the GAUSS Package Manager. It requires a working copy of GAUSS 22+.

For more information on installing the GAUSS Package Manager please see our video blog, Installing the GAUSS Package Manager.

Dependencies#

This library requires the installation of the following GAUSS libraries:

Usage#

Estimating custom state-space models using the sslib library requires:

Loading data
Setting up the state-space representation
Calling the estimation function ssFit()

See our complete guide to getting started with state-space estimation in GAUSS for a full example.

Commands#

Estimating custom models#

`ssFit()`	Estimates parameters of a state-space model using Kalman filtering and maximum likelihood estimation.
`ssIRF()`	Computes the impulse response functions based on the final estimated parameters.
`ssPredict()`	Computes in sample predictions or out-of-sample forecasts based on the final estimated parameters.
`ssKalmanSmooth()`	Performs the Rauch-Tung-Striebel backward recursion smoother for state variables and state covariances.

Pre-built models#

`ssARIMA()`	Estimates parameters of a state-space model ARIMA(p, d, q) model using Kalman filtering and maximum likelihood estimation.
`ssSARIMA()`	Estimates parameters of a state-space model SARIMA(p, d, q)(p_s, d_s, q_s) model using Kalman filtering and maximum likelihood estimation.

Model evaluation#

`ssgetAIC()`	Computes Akaike’s information criterion from loglikelihood.
`ssgetAICC()`	Computes the corrected Akaike’s information criterion from loglikelihood.
`ssgetBIC()`	Computes the Schwarz’ Bayesian information criterion from loglikelihood.
`ssgetHQIC()`	Computes the Hannan–Quinn information criterion from loglikelihood.
`ssHeteroskedasticityTest()`	Tests the null hypothesis of no heteroskedasticity by comparing the sum-of-squares of the first third of the sample to the sum-of-squares of last third of the sample. Analogous to a Goldfeld-Quandt test.
`ssJarqueBera()`	Performs the Jarque-Bera goodness-of-fit test on model residuals.
`ssLjungBox()`	Computes the Ljung-Box test for autocorrelation.
`ssSkewness()`	Computes the sample skewness.
`ssKurtosis()`	Compute the sample kurtosis.