FANPAC has been completely rewritten to utilize the structures and n-dimensional array features found in new GAUSS 5.0. Contact Aptech or your Dealer for Pricing and Information

Supports structures and n-dimensional arrays

  • Familiar keyword interface
  • New thread-safe, easier-to-use procedures

New GARCH models

  • ARMA-GARCH models
    The GARCH specification can now be applied to time series with auto-regression and moving average errors.
  • Normal and t-distribution E-GARCH models
    In addition to the log-conditional-variance model with leverage parameters and generalized exponential distribution, there are now such models with normal and t-distribution.
  • AGARCH models
    GARCH models with assymetry parameters for the arch parameters (Glosten, Jangannathan, and Runkle, 1993)
  • Multivariate VAR-diagonal Vec GARCH models
    The diagonal Vec model can now be applied to the multivariate time series with VAR errors.

New simulation bounds method for statistical inference

FANPAC now contains a simulation bounds method for constructing confidence intervals for models with restricted parameter spaces (Andrews, D.W.K., 1999, "Estimation when a parameter is on a boundary," Econometric, 67, 1341-1383)

A special feature of FANPAC is the ability to place constraints on the parameters to enforce stationarity and invertability and positive definiteness of the conditional variances and covariances. Andrews Method is correct for these kinds of models.

Requires GAUSS Mathematical & Statistical System 5.0 or the GAUSS Engine/Engine Pro/Engine for Workgroups/Enterprise Engine 5.0.

Platform: Windows, LINUX and UNIX.
Requirements: GAUSS/GAUSS Light version 3.2 or higher.