Skip to main content

European early warning system for systemic risk.

Periodic Reporting for period 1 - (European early warning system for systemic risk.)

Reporting period: 2016-06-01 to 2018-05-31

The European financial crisis posed a serious challenge to the stability and prosperity of the euro area and represented a real threat for growth prospects in Europe and for its integrity. The sovereign European crisis started in the second half of 2011, after the global financial crisis (GFC) highlighted the need of an appropriate warning mechanism to prevent financial stability. The main challenge lies in the realization of a sophisticated system to detect and measure marginal changes in each source of systemic risk. Systemic events have a direct impact on the real economy causing potentially unsustainable costs for governments and the international community in terms of GDP (cost of bank bailouts) and in social terms (negative externalities in the economy). Literature proposes several indicators to measure systemic risk; these measures revealed useful predictive information about the probability of future macroeconomic downturns. aims to identify and signal European economic vulnerabilities to allow policy makers to intervene with proper actions. Different approaches have been adopted in measuring the source of systemic risk resulting in new measures and early/nowcasting warning signals. Moreover, the project contributed to a systemic risk platform for the analysis and visualization of the European and Global financial system which is available to scholars, practitioners and the public. This platform allows to analyze through visualization, explanation and download access to the indicators the systemic risk in the Euro and the Global area. The project has been involved in several dissemination activities such as conferences and workshops related to finance, financial econometrics and statistical methods with the presence of researchers from universities and central banks.
The work performed during the project focussed on new scientific contributions and the realization, jointly with the Systemic Risk Lab (SAFE), of a platform that makes available on a regular basis systemic risk measures proposed by literature and developed during the project. First, the project proposed a construction of an overall meta-index for the measurement of systemic risk based on a Sparse Principal Component Analysis of main systemic risk measures, which ultimately aims to provide an index with a more stable dynamic and with an explicit link to severe economic recessions. Second, Dynamic Quantile Factor models were effectively employed. The proposed model is based on the use of the Dynamic Factor Quantiles model (DFQ) which considers dynamics in the quantiles as a function of latent variables. DQF models are then employed to extract the latent signal of systemic risk from a panel of financial institutions belonging to a given financial system. Third, the project introduces a novel Bayesian model to estimate multi-quantiles in a dynamic framework. The main innovation lies in the assumption that the quantile level of a vector of response variables depends on macroeconomic variables as well as on latent factors. The analysis focuses on equity market returns and macroeconomic variables to analyse the dynamic evolution of spillovers in individual Value-at-Risks. Fourth, the project analyses the contagion channels of the European financial system through the stochastic block models, which group homogeneous connectivity patterns among the financial institutions, thus allowing to describe in a compact way the shock transmission mechanisms of the network. Fifth, a Bayesian approach is proposed to the problem of variable selection and shrinkage in high dimensional sparse regression models where the regularisation method is an extension of a previous LASSO model, which allows to include a large number of institutions improving the identification of the relationship. Findings show that changes in the shape of the out-degree distribution of the network over time represent a responsive indicator of the global financial system and a significant predictor of market returns. Sixth, new measures of network connectivity are proposed, that are Von Neumann entropies and disagreement persistence indexes, using the spectrum of normalized Laplacian and Diplacian. Seventh, an extension of the study of the rate of convergence to a consensus of autonomous agents on an interaction network is proposed by introducing antagonistic interactions and thus a signed network. Finally, the project contributed directly to the realization of the Systemic Risk Dashboard (SRDB) platform, providing the code for the systemic risk measures and the structure code (reading, cleaning and computation of the data) using parallelization in MATLAB to automatize routines. The SRDB is regularly updated and upgrades with new measures are already scheduled. The web-application has been realized with the support of the SAFE Data Center. Dissemination activities of the project involved conferences and workshops related to finance, financial econometrics and statistical. makes use of state of the art econometric techniques and other new techniques that may enhance the frontiers of research in related fields. On this ground, the project investigated the topology of the financial networks focusing on the detection of financial communities and community bridges to overcome the weakness of classical connectedness measures. This measurement can lead to a better resilience of the financial system, which implies reduced disruptive effects of recessions on social welfare and therefore stable economic and wealth growth for new generations. The project proposes an approach to improve the identification of the relationships (linkages) among the financial firms by including a very large number of institutions and thus shows that changes in the connectedness of financial networks are significant predictors of the financial market returns. It also refines the dynamic quantiles model which considers dynamics in the quantiles on the VaR and CoVaR to extract the latent signal of systemic risk from a panel of institutions. The factor-augmented vector autoregressive extension of traditional univariate quantile models makes use of equity market returns and macroeconomic variables to analyse the dynamic evolution of spillovers in individual Value-at-Risks. Recent works highlight that the definition of a “good” measure of systemic risk remains unresolved. A solution that recently appeared in the literature on systemic risk to mitigate the model risk is to construct aggregated indexes from different existing metrics. In a scientific contribution, a criterion for comparing competing systemic risk indexes is proposed with an approach that explicitly deals with the known and evidence common behaviour among specific systemic risk indicators. Findings show that the aggregated index we identify has predictive power with respect to economic downturns. thus contributes to monitoring the financial system by providing new tools to policy authorities to make the financial and economic system less vulnerable and more resilient to shocks.
Earliness Logo