Abstract:
MonaS is a fusion between two models from the central bank of Denmark: The bank’s macroeconomic model (Mona, used for macroeconomic analyses) and the Stress test model (S) used for top down stress testing. MonaS incorporates multiple feedback rounds between the banking sector and the real economy. By including feedback it is possible to model the systemic nature of the interaction not only within the financial sector, but also between this sector and the real economy. The system used to solve the model is sufficient agile to also handle other types feedback and contagion. Introducing feedback increases the complexity of the model: without feedback effects, the core of the model consists of 210 simultaneous equations. Introducing feedback increases the size to 673 equations. The full model consists of 1708 equations. The system has also been used for implementing a prototype of the ECB’s macro prudential stress test model (StampE) which comprises more than 500.000 equations. A Python model management system (PYFS) has been developed for handling such large scale models. Models are specified in an algebraic modelling language which allows rapid model development. This also allows incorporating models from a range of other sources. Optimized solvers can handle such large models in milliseconds on standard hardware, enabling extensive sensitivity analyses which can be visualized using the full arsenal of Python data science tools.