Main menu

Project structure

The main activity in this WP is to coordinate the efforts in the Consortium to maintain and to ensure the proper level of communication among the partners. This is particularly important since there is a circular workflow from data collection to modelling and validation and visualization. Thereby it is necessary to provide a communication channel to exchange in real time information, data and feedback on the data already collected.

This work package will address the influence of alternative, non-financial factors to systemic risk. In particular, we will analyse unstructured data (i.e. textual documents such as news and blog posts) as well as country level data with the goal to extract indicators relevant for systemic risk forecasting.

The goal of WP8 is to explore the distribution of correlations in a series of large, complex networks of heterogeneous agent models of increasing complexity, up to the study of the correlation structure of more realistic financial networks constructed in the main project (WP2). Ultimately the objective of the workpackage is to produce convincing evidence for the existence of randomly distributed, strong, long range correlations as a typical feature of complex systems, which make it impossible to divide them into weakly interacting parts. In

Eample Work package 


The main goal of this project is to provide a framework of theoretical and software instruments to understand and deal with systemic risk. Such instruments should be made available to policy makers in order to find suitable countermeasures to possibly avoid and/or mitigate the effect of the crisis. The success of this project will be achieved if we are be able to identify a set of indicators of the systemic financial situation, monitor their real-time evolution, and, finally, explore the possible future scenarios following a given countermeasure. To this purpose several competences, not only scientific and technical, but also in also policy making are needed.


Show/Hide the complete project structure descritption.

Firstly, we have to provide a coarse-grained description of the real situation by filtering and analysing the available information on liquidity and solvency of the main actors in credit markets. In addition, mapping financial relations by means of network analysis is particularly useful, to assess properties like robustness, resilience. The starting point of this activity would be data collection realized in WP1 (Data Collection and Database Consolidation) together with the collection of any other extra information that could allow to draw geographical maps of the economic situations. Starting from this “cartography of risk”, we plan to move to the second point that is the development of a technological support in order to visualize and elaborate this information. This online platform will be in line with similar projects that are presently starting in the world. In particular, we will be in contact with various companies (e.g. palantirtech[1] and anacubis[2]) in order to ensure compatibility between our project and similar state-of-the-art activities. A key ingredient of such platform will be the possibility to interact with the data by computing and visualizing the quantities of interest such as the communities formed by financial institutions, the robustness and stability of the system, volume and size of the exchanges, chains of institutions that could generate bankruptcy avalanches. In addition, the researchers of the Consortium will be able to upload models and run simulations.
This activity constitutes the core of WP3 (Network Visualization and Systemic Risk Forecast). The third and final step will be the possibility to forecast the future evolution of the system, coupled with the possibility to test the effect of policy actions such as varying the interest rate, bailing out particular institutions and not others, facilitating the trade etc. While the final evaluation on these subjects is in the hands of policy makers, these instruments are meant to help their decisions by providing quantitative results in a short time. Such role of support in the decision making relies on the work carried out in WP2 (Agent-based Modelling) and WP4 (Empirical analysis and model validation)where the models to be used in this activity will be produced and tested. While in principle the platform will be the final deliverable of the research and development activity, we believe it is crucial to start testing a prototype already during the early stage of the project. This means that all these activities will proceed through a continuous feedback from data analysis to models, validation and visualization and back to data analysis. The management of this workflow will be ensured by the activity of WP5 and WP6 (Management and dissemination). WP7 (Modelling of non-financial risk factors) will address the influence of alternative, non-financial factors
to systemic risk. In particular, we will model unstructured data (i.e. textual documents such as news and blog posts) as well as country macro-level indicators with the goal to extract indicators relevant for systemic risk forecasting. The textual documents analysis will consist of the following tasks: specification and collection of relevant data sources, construction of sentiment vocabularies and sentiment analysis, visualization of large document collections and trends in sentiment evolution, detection and forecasting of complex sentiment and systemic risk related events, and finally qualitative modelling of systemic risk factors based on extracted features. Assembling the data on diverse country level indicators should serve for the forecasting of main trends and movements in different sectors of main
world economies as well as worldwide. We plan to build different forecasting models in order to have more realistic estimates of defaults distributions, which are important for the network based methods of estimation/assessment of systemic risk. In WP8 (Correlations in complex networks) we propose that the three accepted criteria for a financial institution to qualify as systemically important (size, substitutability and connectedness) should be supplemented by the criterion of „strongly correlated“, and in close collaboration with the original project we wish ultimately to identify the strongly correlated cluster of core institutions.
While it is a commonplace that in a crisis every player in the market becomes strongly correlated with every other one, in a heterogeneous, strongly interacting system, such as global finance, strong correlations propagate along several paths indirectly linking sometimes very distant components, thereby overwriting the underlying network of links and creating strongly correlated clusters even in normal times. Furthermore, the growth of these strong correlations and the clusters connected by them may be the precursors of a crisis.