The primary domain of the toolbox is building and implementing structural macroeconomic models (such as DSGEs or other types of rational-expectations models) and model-aided analytical systems. To this end, the toolbox integrates the core modelling functions with support for time series analysis, data management, and reporting. IRIS is an open-source m-file-only toolbox; it's also easily extensible since its design is based on the object-oriented programming Matlab concepts.

The core modelling features include

• flexible modelling language;
• full support for both stationary and non-stationary (balanced-growth path) models;
• easy object-oriented access to models, their properties, and diagnostics;
• first-order accurate solutions combined with exact non-linear (perfect-foresight-like) simulation mode;
• combined anticipated and unanticipated information;
• classical and bayesian estimation methods, including an adaptive posterior simulator;
• optimal time-consistent policy;
• Kalman filter with an exact non-linear prediction step;
• conditional and judgmentally adjusted filtering and forecasting;
• full support for multiple parameterisations.

The IRIS infrastructure furthermore provides tools for

• multivariate time series analysis and forecasting (VARs, SVARs, BVARs, and FAVARs);
• time series and database management and processing;
• command-oriented PDF-based reporting.
• backward-looking econometrics-style models;

Check out the overview of IRIS features.

• Why is IRIS updated so frequently? IRIS is being developed continually, and it has deliberately very short release and development cycles with rather incremental changes made between releases. Short release cycles allow more flexibility in responding to users' requests, fixing bugs, and making improvements in the under-the-hood plumbing and wiring.

• Why is IRIS more than 11 MB large (after unzipping)? About half of it is because of two X12 executables (one for Windows, one for Unix/Linux) distributed with the toolbox. The other half is the toolbox's almost 2,000 m-files. IRIS integrates a large number of tasks and features, and has a complex, interconnected structure of backend functions and classes -- i.e. those not directly visible or accessible to the users of the toolbox.

Copyright © 2007--2012 Jaromir Benes.