By Geertjan on Jan 06, 2014
jcMod, which stands for Java clustered modelling, is a collaborative modelling platform developed in South Africa. It integrates the many modelling tasks found at a typical inflation targeting central bank in a controlled, multi-user environment and offers advanced functionality for data capture, modelling, simulating and reporting. It also maintains a repository of all past data, models and reports for future reference.
The application is fairly advanced but also highly customized to the needs of a central bank, especially if that bank is an inflation targeting central bank. It can also be used at any econometric modelling institution. The problem is that a typical modelling institution will have to use multiple applications in order to be able to construct the diverse suite of models required today. This leads to huge integration issues. The application offers a lot of integrated modelling, data and report writing functionality and thus allows for easy collaboration.
jcMod supports a large number of models, such as exponential smoothing, OLS, NLS, ARIMA, ARCH, GARCH, VAR, SVAR and VECM models. jcMod can simulate and optimise multivariate models, Kalman filters and DSGE models and can sample, using Gibbs or MCMC sampling, parameters from these models. Its modular nature means jcMod can easily be extended to support more models, with neural networks planned for the nearby future for example.
jcMod's reports are used by both the modelling team and the monetary policy committee (MPC) and its report editor, apart from normal editing, also allows for the layout of equations and advanced plotting of data, such as histograms and fan charts. Reports and images can be exported in bitmapped or vector (EMF and EPS) formats.
jcMod makes heavy use of Java and NetBeans Platform 7.4 but takes care not to introduce any other external dependencies and jcMod clients can run on any recent JVM.