SYstemic Risk TOmography: Signals, Measurements, Transmission Channels, and Policy Interventions

Final Report Summary - SYRTO (SYstemic Risk TOmography: Signals, Measurements, Transmission Channels, and Policy Interventions)

Executive Summary:
SYRTO stands for SYstemic Risk TOmography, a name which entails the philosophical approach followed in exploring the economic and financial system after the dramatic events of the global financial crisis arose in 2007 and the European crisis erupted in 2010.
We conceived sovereigns, banks with other financial intermediaries (BFIs) and corporations as a complex system to be inspected by sectioning each main part of it, then understanding how these “parts” of the system are related to one another in terms of systemic risk.
Based on this perspective, we analysed the financial system as a “biological entity”, to be monitored closely to identify the main risk signals and provide the right measures of prevention and intervention. In more depth:
1. We assembled an Early Warnings System (EWS) to be used as risk barometer for each sector and countries alike, identifying potential threats to financial stability;
2. We realized a SYRTO Code in order to detect a series of recommendations, also expressed in terms of EWS prescriptions, on: (a) the appropriate governance structures for EU to prevent and minimise systemic risks; (b) the best mechanisms for ensuring an effective interplay between, and coordination of, macro and micro-prudential responsibilities.
The realization of these targets has been conceived in a step-by-step process in which:
First, we inspected idiosyncratic risks within the financial system thereby making clear the more important risk predictors and how these are related to: (a) sovereign risk, (b) banks and other financial intermediaries risk; (c) non-financial corporates risk;
Second, we inspected both the two-way and multi-way risk connections among macro-sectors (sovereign, bank and other financial intermediaries, corporates), by elucidating the risk linkages and related transmission channels;
Third, we assembled an overall EWS and suggested possible normative superstructure for a better EU economic governance including monetary and policy coordination as well as financial market supervision.

Project Context and Objectives:
During the second half of 2011 sovereign and banking risks increased in the Eurozone in an environment of weakening macroeconomic growth prospects. Contagion effects have become substantial also accelerated by the interplay between vulnerability of public finances and financial sector. Furthermore, bank funding pressures increased markedly thus feeding the fear of the Euro collapse.
While the EU’s response to the economic downturn was swift and decisive (e.g. the European Economic Recovery Plan – EERP – launched in December 2008 and ECB monetary interventions in 2011), the future of macro-economic and monetary conditions in Europe was seen as strictly dependent upon the following building blocks:

• Prediction and prevention of systemic crises, implying an in depth understanding of the principal causes of the crisis and how changes in macroeconomic, monetary, regulatory and supervisory policy frameworks could help prevent their recurrence.
• Risk mitigation, to minimise the impacts of systemic risks thus stabilising the financial system and the real economy in the short run by adopting policy interventions and financial supervision.
• Crisis resolution, by adopting (a) financial policies, in order to contain banking problems; (b) macroeconomic policies, in the form of macroeconomic stimulus both monetary and fiscal having in mind the interplay between macro prudential instruments; (c) structural reforms, by boosting potential growth and productivity then positively reflecting on fiscal burden, on deleveraging and balance sheet restructuring, and on cross-country imbalances.

Within this new and complex economic context, the SYRTO project was launched in March 2013 with the general objective of developing core conceptual frameworks, models and tools to handle the three blocks having a system-wide perspective on financial and economic stability. The lesson learned with the global financial crisis was indeed that the financial and economic community neglected such a perspective.
Since micro-prudential approach to financial stability ignores the multidimensional nature of risks, the challenging task for SYRTO was to develop formal measures of system-wide risk, in order to capture the linkages and vulnerabilities of the financial system, and regulate the overall level of risk of the system and the real economy’s exposure to such a risk.
To put into perspective this new way of thinking the financial system, in a sense assumed as a “biological entity” to be monitored closely to identify the main risk signals and provide the right measures of prevention and intervention, SYRTO proposed a stylized macro-financial architecture in which sovereigns, financial intermediaries and corporations are intricately connected through balance sheets, as well as real and financial links. Together, these links form a complex system of many risk dimensions, translating into systemic risk, namely “the risk that financial instability impairs the functioning of the financial system to the point where economic growth and welfare materially suffer” (Costancio, V. (2010). Macro-prudential supervision in Europe. Speech at the ECB-CFS-CEPR conference on “Macro-prudential regulation as an approach to containing systemic risk: economic foundations, diagnostic tools and policy instruments”. Frankfurt am Main, 27/9/2010, available at https://www.ecb.europa.eu/press/key/date/2010/html/sp100927_3.en.html.).
The SYRTO project inspected the relationships between sovereigns – banks and other financial intermediaries (BFIs) – corporations of the European Union with the following objectives:

• Identify the common (fundamental) and the sector-specific (idiosyncratic) risks, and assemble an Early Warnings System (EWS) to be used:
as risk barometer for each sector and countries alike, in order to identify potential threats to financial stability.
as a system of rule of thumbs by monitoring a series of leading indicators so as to minimise the possible negative impacts from systemic crises, preventing contagion effects also including the appropriate mechanisms to restore systemic crisis problems.

• Explore monetary policy and macro-prudential issues relative to systemic risk developing a “SYRTO Code” in order to detect a series of recommendations, also expressed in terms of EWS prescriptions, on:
the appropriate governance structures for EU to prevent and minimise systemic risks;
the best mechanisms for ensuring an effective interplay between, and coordination of, macro and micro-prudential responsibilities.

The articulation of the project has been though to suggest the best way towards a better macro-economic and monetary integration in Europe. Specifically, SYRTO dealt with:

• The EU economic governance including monetary and policy coordination as well as financial market supervision, taking into account the complex institutional and economic integration in the Eurozone;
• The prediction, prevention and mitigation of systemic risks providing an EWS to be next used as the trait d’union between institutional and quantitative features of systemic risk.

In doing this, we assigned a pivotal role to the systemic risk, being the main focus of our research to be inspected in terms of measurements, transmission channels, and policy intervention.

The research activity was articulated in a step-by-step process conceived with the end to lead our findings towards the two main objectives of SYRTO, namely the Early Warning System and the SYRTO Code:

Step 1. The first objective was to inspect the idiosyncratic risks within the financial system thereby making clear the main risk predictors and how these are related to:
a. Sovereign risk.
b. Banks and other Financial Intermediaries risk.
c. Non-Financial Corporates risk.

Step 2. Our aims were next focused to:
• Inspect both the two-way and multi-way risk connections among macro-sectors (Sovereign, Bank & other Financial intermediaries, Non-Financial Corporates).
• Explore the main risk linkages and related transmission channels.
• Propose novel systemic risk measures.
Step 3. Based on major achievements of the previous steps, we finally focused on our two major objectives:
• Assemble an overall Early Warning System (EWS) to be used as risk barometer for each sector and countries alike, identifying potential threats to financial stability.
• Realize a SYRTO Code, namely a series of recommendations and prescriptions on: (a) how to prevent and minimize systemic risks; (b) the best coordination of macro and micro-prudential responsibilities.

Idiosyncratic Risks.
The objective was to detect the fundamental risk sources of the financial system, focusing on the following risk categories:
1. Sovereign risk, both considering economic fundamentals as well as market prices in order to make clear the relationship between risk perceptions, especially during times of market turmoil, and “fundamental” risk inherent to macroeconomic vulnerabilities.
2. Banks and other financial intermediaries risk, exploring how financial and macroeconomic risks change over time due to the strong interconnectedness between financial intermediation, macroeconomic environment and asset price dynamics.
3. Corporate risk, showing how financial sector risks transmit to real business cycle fluctuations using a balance-sheet risk perspective.

Two-Way and Multi-Way Risk Connections
The aim was to explore the risk connections among the system Sovereigns-Banks and other Financial Intermediaries-Corporates (S-B&FI-C), by focusing on, first, the bi-variate relationships, and second, on the joint determination of risks.
1. Two-way risk connections, monitoring the feedback between sovereign debt and bank risks as well as spillover effects between banks and the corporate sector also considering interconnections between sovereign and corporate risks.
2. Multi-way risk connections, exploring the main financial and economic linkages as a channel for propagation of shocks through recent advanced econometric/statistical techniques such as network analysis, data-mining techniques and latent variable modelling approaches.

Systemic Risk Measures
Our aim was to focus on different systemic risk indicators using different methodologies thereby offering a list of potential battery of measures using financial and economic data. The main purpose was to realize Early Warning and NowCasting indicators, to analyse:
• Linkages and vulnerability of the financial system;
• Bi-variate risk relationship among Sovereign Debts, BFIs and Corporate;
• Empirical analysis of bi-variate risk relationship.

In doing this, the ambition was to realize a risk dashboard in order to realize a risk mapping and identify potential vulnerabilities within the Eurozone.

EWS
The purpose was to realize a monitoring risk system for sovereigns, BFIs and the corporate sector, providing a comprehensive risk analysis covering countries and sectors and aggregating the individual risk dimensions. Such a tool was conceived as an interactive web-based platform with the following objectives:

1. visualize the systemic risk measures realized by SYRTO;
2. monitor risk signals using a Regression Tree-based implemented for sovereign risk, as well as financial and non-financial corporate risks;
3. provide a geographical risk mapping for the Eurozone in each main sector (S-B&FI-C).
4. offer an interactive tool through which calculating the risk profile of single sovereigns, banks, and corporates.

Another tangential objective was to come up with a Data Center to collect relevant information to monitor markets, financial institutions and the economy, thereby realizing a data management infrastructure where data are downloaded for empirical analysis and model calibration.

SYRTO Code
Our final objective was to create a possible framework for systemic risk regulation in terms of policy/monetary recommendations and prescriptions, leading towards a formal SYRTO Code, in order to detect possible rules of thumb to limit the triggers, the shock transmission of systemic risks and ex post policy interventions to stabilize the Euro system. This Code was thought in 3 main chapters:
Chapter 1 – Prevention
Identification of a series of rules of thumb in order to limit the triggers of systemic risk.
Chapter 2 – Mitigation
Limiting systemic shocks transmission and prevent conflicts of interest and ineffective policy interventions.
Chapter 3 – Stabilization
Ex post policy interventions to stabilize the Euro system.

Another related objective was to inspect the common perception citizens have relative to economic environment in Europe, thereby inferring some confidence measures about monetary integration and policy interventions. Such a measure is indeed potentially relevant in order to identify potential areas of improvement of the macro-prudential policy making process.

Project Results:
Main Scientific and Technological Results

1. Idiosyncratic Risks
The meaning we attribute to systemic risk in SYRTO is modular-based, since we assume the financial system to be a complex “ensemble” in which sovereigns, banks, other financial intermediaries, and corporations represent macro-sectors characterized by sector-specific systemic risks that interact one with another through spillover effects. In this perspective, our research activity focused on each main part of the system elucidating their risk dynamics using different approaches.

1.1. Sovereign Risk
Methodological Approaches
We followed two approaches: i) the first addressed the question of market’s reaction following a bad credit event; ii) the second analysed the transmission of shocks within the European CDS market, stock exchange markets, and also between these two markets.
Main Results
Our results reveal significant time-variation in distress dependence and spillover effects for sovereign default risk. Market perceptions of joint and conditional sovereign risk around announcements of Eurosystem asset purchases programs reveal a strong perceived impact on joint risk.
Other interesting findings relate to CDS spreads and equity market dynamics. Indeed, our results indicate that aggregate CDS spreads exhibit different behaviours depending on the market segment under consideration. The equity market price and volatility channels both contribute to increase the dispersion of some CDS spreads (i.e. magnifying impact), and the sensitivity of aggregate CDS spreads to SP500 and VIX indexes is larger at upper quantile levels showing time-varying dependencies, especially during extreme movements.

1.2 Banks and other Financial Intermediaries Risk
Methodological Approaches
This topic was explored relative to: i) the credit risk in a large banking system; ii) the importance of the interbank market; iii) the efficiency of the hedge fund market as well as its ability to generate abnormal relative returns; iv) the Dark Pools trading.
Main Results
Exploring the credit risk in a large banking system, we introduced new measures for financial systemic risk based on the time-varying conditional and unconditional probability of simultaneous failures of several financial institutions. By applying the model to assess the risk of joint financial firm failure in the European Union during the financial crisis, we proved that these measures are particularly efficient in providing robust correlation estimates against possible outliers and influential observations. By using frailty and mixture models, in an empirical study of 233 European banks between 2008Q1-2015Q2, we demonstrate that the global financial crisis and Euro area sovereign debt crisis had a differential impact on banks with different business models. In addition, changes in the interest rate environment predict banks’ business models.
The study of interbank market revealed how it affects the incentives to hold bank capital for liquidity risk-sharing purposes. The model predicts an unexpected negative relationship between bank capital and empirical proxies of the coinsurance opportunities offered by the interbank market.
Relative to hedge fund dynamics during crisis periods, our results suggest that about half of the sample exhibit random absolute and relative returns. Hence, for about half of the sample, efficiency is found. This result has strong implications. For the other funds, most exhibit volatility clustering and structural breaks, indicating that the degree of risk exposure changes over time. Very few funds are able to over-perform the market, i.e. to produce abnormal returns.
Concerning Dark Pools trading, i.e. the ability to trade on OTC markets without publicly announcing orders on the market, we pointed out the urgency for monitoring such activity, because of the effect of dark orders on the price dynamics, and then on the price discovery process. Dark pools are directly related to High Frequency Trading (HFT), and the amount of dark pool orders may reveal the importance of HFT. Using milliseconds data, we found a clear relationship between dark orders/volumes and extreme realizations of stock prices.

1.3 Corporate Risk
Methodological Approaches
We implemented i) data mining tools on corporate balance sheet data with the main aim to detect the best tools to predict corporate defaults; ii) frailty models to capture unobserved components in corporate default intensities.
Main Results
We introduced a novel approach for ensemble modelling based on dominant vs. competitive models, suggesting a pooling approach run using the competitive models only. Numerical experiments based on parametric (logit, Bayesian model averaging) and non- parametric (classification tree, random forest, bagging, boosting) model comparison shows that the proposed ensemble performs better than alternative approaches, in particular when different modelling cultures are mixed together (logit and classification tree).
To capture unobserved components in corporate default intensities, we found that if such components are present, part of the fluctuation in corporate bond prices can be attributed to the variation in beliefs about these frailty factors over time. We show clear evidences of a latent frailty process in the default intensities which is robust to the inclusion of both macro and firm specific variables.

1.4 Agent based Modelling and Socio-Finance
Agent-based and socio-financial modeling have been introduced to describe individual decision making and see how individual choices can lead to speculative bubbles. We first measured nonlinear feedback effects between the market performance and changes in sentiments, describing some stylized facts such as volatility clustering and extreme events which are proven to be perceived as arising due to abrupt sentiment changes via ongoing communication of the market participants.
We also focused on the transition in a financial market, i.e. how certain speculative transitions in financial markets can be ascribed to a symmetry break that happens in the collective decision making. Markets are assumed to behave as complex systems whose payoff reflect an intrinsic financial symmetry that guarantees equilibrium in price dynamics (fundamentalist state) until the symmetry is broken leading to bubble or anti-bubble scenarios (speculative state). Such two-phase transition is modelled in a micro-to-macro scheme through a Ginzburg-Landau-based power expansion leading to a market temperature parameter which modulates the state transitions in the market.
We finally introduced a socio-finance Student-t modelling approach which is particularly useful in cases where we believe that communication of different groups of the population about the movement of the market can affect the returns and/or the volatility of the financial assets as well as in cases where the distribution of returns is characterized by large kurtosis, fat tails, or in general deviates from normality.

2. Two-Way and Multi-Way Risk Connections

We first started investigating the two-way risk connections, moving in many directions with different statistical and econometric procedure, including i) factor models with constant, time-varying and stochastic coefficients; ii) multivariate stochastic volatility, to be used a synthesis of bi-variate risk and quantile regressions to equity premium prediction to inspect time-varying and non-linear relationship with predictors.
We also explored some techniques to detect the main financial and economic multi-way risk connections, and specifically: i) multi-equation system with latent variables; ii) network analysis; iii) data-mining

2.1 Main Results – Two-way risk connections
Using inter-system recurrence networks, we explored the two-way risk connections between the European banking and insurance sectors based on geometrical-based distances. Our results on the dynamical transitions and coupling direction between these sectors show that the banking sector is central in risk transmission compared to the insurance sector.
Investigating the time-varying spatial dependence methodology, we introduced and applied measures of systemic risk levels and the effectiveness of non-standard monetary policy actions to break market perceptions of systemic risk linkages. On the period 2009-2014, we found a high, time-varying degree of spatial spillovers in the sovereign CDS spread data, which remains high (with short spells of reduced linkages) over repeated non-standard interventions by the ECB (in particular the long-term refinancing operations). A downturn in spatial dependence (i.e. systemic risk linkages) arises after the first half of 2012, which is consistent with policy measures taken by the European Central Bank (outright monetary transactions announcement).
We also modelled the two-way causal relationships with a new complicated big-data Bayesian hierarchical model. In this model we considered sectors (Sovereign, Banks and Financial Intermediaries, Corporations) and financial assets (CDS and equity returns), building a model in which one can estimate sector systemic risk factors taking into account macro-systemic risk factors.

2.2 Main Results – Multi-way risk connections
We first developed a 3-equation system methodology to describe how two main risk sources Y (dependent variable) and X (independent variables) move together based on a time-varying risk sensitivity (beta) conditioned on some observables z which enter both into Y and X price processes within a Bayesian setting. This approach is strictly connected with the novel dynamic factor model introduced to model the two-way causal relationships (Bayesian hierarchical model), which provides a way of understanding how financial returns are affected by latent, sector-based factors and macro-systemic risk factors.
On the Network Modelling we explored a battery of econometric measures of connectedness based on Granger-causality networks to the changes of sovereign risk of European countries and credit risk of major European, U.S. and Japanese banks and insurers to investigate the evolution of these connections. Credit risk for banks and insurers is measured using a version of the Merton Model (Contingent Claims Analysis) applied to risk-adjusted balance sheets. We highlight connections among banks, insurers, and sovereigns by quantifying the effects of risk transmission within and across countries and financial institutions proposing financial network measures that allow for early warnings and assessment of the system complexity.
We also explored novel research directions based on agent-based modeling, focusing on “synchronization”, a phenomenon which happens through so-called integrate-and-fire oscillators. These are thought to be relevant in a wide range of phenomena in Nature from the coordination of flashes of swarms of fireflies, .... to coordination in neural activity which can lead to epilepsy. In our study we show how synchronization can take place between stock exchanges where each stock exchange acts as an integrate-and-fire oscillator creating correlations in the price dynamics seen in the worlds network of stock exchanges.
To model dependence in high-dimensional multivariate time series and to address over-parametrization in large vector autoregressive (VAR) models, we introduced graphical model combination based on the notion of causality with a new sparsity prior distribution on the graph space to address model selection problems in multivariate time series of large dimension. The results we obtained offer interesting insights for further research into empirical evaluation of macro-financial linkages which has long been the core of the IMF’s mandate to oversee the stability of the global financial system.
Finally, we introduced a novel approach to estimate systemic risk, conceived as a latent variable conditional on a specified set of common covariates. The meaning we attribute to systemic risk is modular-based, since we assume the financial system to be a complex “ensemble” in which sovereigns, banks, other financial intermediaries, and corporations represent macro-sectors characterized by sector-specific systemic risks that interact one with another through spillover effects.
Focusing on the connection between hedge funds and sovereign systemic risks in the Eurozone, and also shed some lights on the hedge fund dynamics relative to the surge in sovereign spreads, we introduced novel systemic risk measures for (a) Eurozone Core countries (France and Germany), (b) GIIPS (Greece, Ireland, Italy, Portugal and Spain), and (c) the hedge fund industry.

3. Systemic Risk Measures

To accomplish this task we computed well known and novel systemic risk measures also proposing a new taxonomy based on the micro- macro-perspective as well as according to mapping procedure which can take the form from micro to macro and viceversa. Specifically, we articulated the entire computation in 3 building blocks:
First, we analysed the time evolution of systemic risk in Europe using the WorldScope lists of the financial industry for the European countries. The analysis is based on the cross-sectional distribution of systemic risk measures such as Marginal Expected Shortfall, Delta CoVaR, SRISK, SES and network connectedness measures. Moreover, we considered Shannon entropy on the estimated cross-sectional systemic risk measures. These measures are conceived at a single institution level for the financial industry in the Euro area and capture different features of the financial market during the period of stress. To estimate systemic risk measures, we use a rolling window approach with a window size of 252 daily observations, which corresponds approximately to one year of daily observations.
Second, we introduced a multivariate linear Gaussian state space model to extract a business cycle and a financial cycle from a panel of economic and financial time series. The panel consists of quarterly observed variables and include GDP, credit, credit to GDP, credit to disposable personal income, and residential property prices. The cycles vary stochastically over time and are identified by loadings and phase shifts that are estimated by maximum likelihood. We focus on four large economies: U.S. Germany, France, and U.K. The cycle lengths are not fixed ex-ante but are estimated from the data.
Third, we proposed a new simple ranking method CorrRank to determine the systemic risk of financial institutions. The method only uses average pairwise stock correlations and turns out to be nearly identical to the more sophisticated network based systemic risk rankings such as Google PageRank. Using data on the European banking sector, we subsequently investigate whether our network-based measures contain additional information to readily available rankings based on book and/or market-based firm characteristics. We find that network measures indeed complement currently available systemic risk ranking measures such as VaR- and BetaRank.
Taken together, the 3 building blocks lead to the following systemic risk taxonomy:

3.1 Micro Systemic Risk Measures – Single Financial Institution:
• Standard Systemic Risk Measures: (i) Equity side (CoVaR, MES, SES, SRISK, Granger Causality); (ii) Bond side (CoRisk, Distressed Insurance Premium).
• New Systemic Risk Measures: (i) PageRank; (ii) CorrRank; (iii) BetaRank and VaRRank; (iv) Markov switching models; (v) Multivariate stochastic volatility model.
3.2 Macro Systemic Risk Measures – Financial System:
• Standard Macro Systemic Risk Measures: (i) Dynamic Granger Causality Index; (ii) Principal Component Analysis (PCA).
• New Macro Systemic Risk Measures: (i) Normalized Rank Reflection Symmetry; (ii) Sovereign Joint Default Probabilities; (iii) Business and Financial cycles; (iv) Dynamic PCA; (v) Dynamic Factor Models; (vi) Leading indicator (LI) of the Euro area IP growth.
3.3 Aggregation/Disaggregation – Mapping Micro and Macro (from Micro to Macro and viceversa):
• Single Institutions → Financial System (Aggregation): (i) Entropy measures; (ii) Stress indices; (iii) Panel regressions.
• Single Institutions ← Financial System (Disaggregation): (i) Graphical Models; (ii) Network analysis.

4. Data Collection

We collected and organized financial and economic data through a building block structure where systemic risk measures and early warning system are then developed. Specifically, we implemented:
• a Data Center, to collect relevant information to monitor markets, financial institutions and the economy, and to evaluate the severity of the risks impact, both considering individual and systemic risks;
• a Data Management Infrastructure, where all the units can download the data for empirical analysis and model calibration.

4.1 Data Center
The research focused on the definition of the SYRTO Data Framework, which is composed by three different hierarchical levels. This structure has been conceived to bring modularity and flexibility to the SYRTO dataset with low time-consuming tasks, which allow modifying at a single level. In this perspective, we have data integration with new variables in Level 1 (type and market) and implementation of new systemic measures or early warning systems (Level 2 and Level 3). Specificlly:
• Level 1 includes “raw” financial and economic data mainly about the Euro area, and the EU28 countries; some other core countries around the world are included as well such as, e.g. U.S. Japan.
• Level 2 is constituted by systemic risk measures estimated on the basis of the first level data (Systemic risk measures); and
• Level 3 is represented by early warning and now casting indicators.

Level 1 contains data on single financial institutions and relevant macroeconomic variables for contagion and systemic risk analysis, and represents the basis of the database, the input for the systemic risk measures and the early warning system. Clearly, the research for representative data has been performed accordingly the view of different institutions (mainly SYRTO units and European Central Bank). Primarily source for the collection of these data are European and International institutional data provider (Eurostat, the European Central Bank). Data provided by such Institutions enable comparisons between countries and regions and they are, in most cases, disseminating free of charge via on-line data bases. The list of the data source used includes also Datastream , a Thomson Reuters database containing global financial data and equity indexes, and Bloomberg database, that provides real time and historical financial market data and economic data, covering all sectors worldwide.
The data are classified into three different classes:
• Macroeconomic data (economic indicators and sensitive variables);
• Bond (sovereign and corporate bonds); and
• Equity (equity indices and STOXX600 data).

Macroeconomic data are used as input for systemic risk measures and target variables for early warning systems. They are classified into two categories: economic indicators, and sensitive variables for banking system and for the financial sector:
• Economic indicators are monthly and quarterly statistics measuring economic and employment developments in the euro area and in the EU28 countries and other core countries around the world are included as well such as, e.g. U.S. Japan. Data, provided by Eurostat, cover seven different areas: Business and Consumer Surveys, Inflation, House prices, Industry, Labour Fource Survey, and National accounts.
• Sensitive variables represent indicators for the financial and banking system. We consider the following:
Composite indicator of Systemic Stress (CISS) by European Central Bank and its constituents (reported in Hollò et al. (2012) ). This indicator is an important proxy for the current state of the European financial system; the data on banks’ access to the ECB marginal lending facility and data on daily liquidity conditions ; the banking crises dataset developed by the ESRB Expert Group on Countercyclical Capital Buffers (Detken et a., 2014) . The dataset has been provided by ECB.
• Bond data include data about corporate and sovereign bonds. On the corporate side, we consider the credit default swap and corporate bonds on the constituents of STOXX Europe 600 – which represents our reference markets. The sovereign bond (2Y, 3Y, 5Y, 7Y, 10Y, 15Y, 30Y) that we consider in SYRTO are based on the Euro countries and the other relevant countries for systemic risk. Data are downloaded from the ECB and Eurostat online database.
• Equity data include both Stoxx600 data and a selected list of the Datastream Global Equity Indices. SYRTO has chosen as benchmark for the financial market the constituents of the STOXX Europe 600 index. This index represents large, mid and small capitalization companies across 18 European countries . Data includes equity and bond market assets that are available in SYRTO at daily and monthly frequency, while accounting and balance sheet data are available on quarterly and annual basis. Also a list of selected Datastream Global Equity Indices covering European markets broken down into different sectors has been defined. Such indices cover a minimum of 75 - 80% of total market capitalisation of each market. Index constituents for each market are reviewed quarterly.

4.2 Data Management Infrastructure
We fixed the process of extracting, transforming, and loading data into the SYRTO data warehouse.
There are three main data sources: i) international agencies, ii) the Thomson Reuters Datastream database and iii) the Bureau Van-Dijk Amadeus database . While working with these data sources, it is important to remember that procedures for data and metadata extraction are an important phase of the data process that must be defined and implemented in order to ensure high level of quality in terms of accuracy, credibility, timeliness, accessibility and interpretability without forgetting efficacy and economic efficiency. For this reason, data extraction procedures should be automated to avoid manual imputation and intervention. Data extraction procedures must also be error-proofed. Possible sources and types of error are analysed, and provisions are put in place to check and correct for errors.
Computationally, the data extraction and integration process has been developed using the Konstanz Information Miner Platform (KNIME) (Berthold et al., 2008; www.knime.org).

Data Extraction
Looking at the SYRTO project, due to the characteristics or limits of the data sources, the degree of automation of the procedures for data extraction varies across data sources .
Starting from international agencies’ databases the first category of data sources, the extraction and integration procedures ensure high-quality standards in terms of output; this is mainly due to the high quality of the data guaranteed by the data providers and the efficiency and effectiveness of the procedures for data and metadata extraction. That said, extracting the data of interest of the research group was achieved by using SDMX queries through SDMX web services, which grant access to Eurostat’s, the ECB’s and the OECD’s online data warehouses. Specifically, the SDMX queries have been implemented through the use of RSDMX, an R package to parse and read SDMX documents in R that provides a set of classes and methods to read data and metadata documents exchanged through the SDMX framework . Being developed in R, routines for data extraction can be coded directly into the data management process developed in the KNIME platform, ensuring high standards in terms of process efficiency and effectiveness.
Moving to the Thomson Reuters Datastream database, the extraction procedures ensure quite good-quality output standards in terms of the efficiency and effectiveness. Automated procedures to extract data from such a data source have been defined using the Advance for Office (AFO) add-ins , a user interface integrated in Microsoft Excel provided within Datastream; it allows the user to define automated TSI requests that can be repeated over time without the need to manually edit the details of such queries. Compared to the procedures described for the data extracted from international agencies’ databases, procedures for extracting data from Datastream have the weak point of not being able to be directly included in one node of the data process defined in KNIME. This requires two steps: first, extracting the data using AFO add-ins, and, second, importing the Excel series into KNIME.
Finally, the procedures for extracting data from the Bureau Van Dijk-Amadeus database currently provide lower-quality standards in terms of efficiency and effectiveness . These are mainly lacking because of the lower degree of automation of procedures for extracting data from this database. For characteristics of the database, especially related to the limits imposed during the download of the data, it is possible to define automatic procedures for the selection of the variables of interest; however, the user cannot define the procedures for automatic download data.

Data Integration and Storage
The process of data integration and storage is defined and managed in KNIME. The aim is to manage all of the data through time-series objects, i.e. data that have the property of being temporally ordered, each of which contains one of the variables discussed in the previous sections not only in terms of data but also metadata, which may help the user to more easily and more quickly understand the characteristics of the series.
From the practical point of view, thanks to the integration of R in KNIME, lists of xts objects have been defined, each of which contains the data and metadata related to one of the variables described in the previous paragraphs. Among the many classes of time-series objects available in R, the choice to define xts objects lies in its simplicity of use, its overall flexibility and in the possibility to define user-added attributes. An xts object can be considered an extension of a zoo object , differing from the zoo class in three key ways: the use of formal time-based classes for indexing, internal xts properties and user-added attributes. Simplified, an xts object contains an array of values comprising the data (often in matrix form) and an index attribute to provide information about the data’s ordering.
The possibility of defining user-added attributes is a feature the importance of which should not be underestimated. With this feature, it is possible to maintain a single object within the values of the series in its metadata, providing the user with detailed information that can simplify data use. Indeed, for each single xts object, the basic metadata are specified (such as the complete name of the variable, the data source, the frequency, etc.) and a link to the complete metadata, if available, is given.
The process of defining, gathering and storing xts objects is handled automatically and with practically zero manual intervention. Automatic procedures are adopted to constantly monitor how data storage is progressing, how the coverage of the data received compares to expectations and how significant any data revisions are in order to take timely action to solve any emerging problems.
Regarding the data transformation process, we focused on the missing data imputation. With the final aim of obtaining complete data sets to load in the SYRTO data warehouse, two imputation techniques have been developed from a methodological point of view and R routines have been implemented: the “Approximate Bayesian bootstrap with Propensity score and Nearest neighbour” and the Symbolic Missing Data Imputation .

5. EWS

The realization of an Early Warning System (EWS) is the core deliverable of the project being a synthesis of the major achievements realized by our group. Indeed, we tried to connect all the techniques, tools, methodologies we developed, inspected, implemented, and tested using data coming from our Data Center. The EWS platform delivers a set of leading indicators, risk signals, and the systemic risk measures introduced in SYRTO with the end to identify potential abnormalities in the Euro financial system, conceived as a complex network of sovereigns, banks with other financial intermediaries, and corporates.
Within the online platform, end-users can deal with interactive graphs concerning the various facets of systemic risks, which are easy-to-read and easy-to-use and include a brief methodological description.
To access the platform from the SYRTO home page www.syrtoproject.eu just click on “platform”.
The platform is organized in four sections:
• Early Warning System – Sovereigns;
• Early Warning System – Banks;
• Early Warning System – Non-Financial Corporates;
• Risk Assessment – Systemic Risk Measures.
The three Early Warning Systems contained in the first 3 sections were run following a two-step procedure: (i) a Random Forest (RF) was grown on the data and (ii) a single Regression Tree (RT, henceforth) was realized using the RF predictions as dependent variable.

5.1 Early Warning System – Sovereigns
Within the Early Warning System for Sovereigns, we report the RT realized using the sovereign CDS spreads (in basis points) as dependent variable and the list of macroeconomic variables contained in the Abbreviation Codes link as covariates (the same link contains also the list of sovereigns with corresponding abbreviation).
More details about statistics for all indicators (also using heatmaps with standardized values), and corresponding risk distribution of the expected CDS spread for each final node are available when clicking on it.
The Risk Calculator is the online tool one can use to compute the expected risk by specifying the values of the selected covariates.
To see the mean values of the predicted risks for all the Euro countries over time, click on “Risk Prediction by Country and Year”. The same values are also visualized in an animated risk map, which depicts the countries with different colours based on their risk level (green is low risk and red is high risk).

5.2 Early Warning System – Banks
This section reports the analysis run for the 139 systemically important banks selected by the ECB where the expected CDS spreads (in basis points) were predicted on the basis of the balance sheet data (source: Bankscope). Since not all the selected banks have a listed CDS, we first computed the tree using all the banks with the CDS, then using the same model to predict the risk for all other banks in the sample.
The Risk Calculator tool is also available here, through which a generic bank can compute its own level of expected risk based on their data.
The section contains an additional link, “Risk prediction by banks and year”, in which we provide a predicted risk table for all the systemically important banks in the Eurozone.

5.3 Early Warning System – Non Financial Corporates
For non financial corporates, we inspected millions of balance sheet data for firms located in the Eurozone (source: Amadeus) computing a RT for each country. The risk prediction is expressed in terms of a risk score ranging from 0 (low risk) to 1 (high risk). The risk cut-off is 0.5: values greater than 0.5 are for firms near to default.
The online Risk calculator is available and works as for sovereigns and banks (of course based on different indicators). The animated risk map is also available for non financial corporates, in which you can see the “risk movie” for all the countries based on the mean values of the risk scores in each country for all corporates.

5.4 Risk Assessment – Systemic Risk Measures
The last section is organized in 4 subsections: (i) Measuring systemic risk in the Euro area; (ii) A decomposition of a panel of economic and financial time series into business and financial cycles; (iii) Network, Market, and Book-Systemic Risk Rankings; (iv) Measuring the Economic Health State (EHS) in the Euro Area. By clicking one of the links you are moved into specific risk measures.

All the graphs are dynamic and the user can define the time horizon, make a zoom on the sub-periods, use the tooltips for facilitating the reading, download the graphs. As an example for the graphical representation of our indicators.
The first subsection contains a first group of systemic risk measures pertaining to Marginal Expected Shortfall, Delta CoVaR, SRISK, SES and network connectedness measures. Moreover, we considered Shannon entropy on the estimated cross-sectional systemic risk measures.
The second subsection contains the second group of measures pertaining to multivariate linear Gaussian state space model to extract a business cycle and a financial cycle from a panel of economic and financial time series. The measures focus on four large economies (United States, Germany, France, and the United Kingdom) and the platform contains the four related dynamic graphs (which have the same characteristics describe above).
The third subsection shows the ranking method CorrRank to determine the systemic risk of financial institutions. The method only uses average pairwise stock correlations and turns out to be nearly identical to the more sophisticated network based systemic risk rankings such as Google PageRank.
In the fourth subsection, we report measures of the European citizens’ perceptions on the National Economy, the European Economy and the Economic Health State for countries of the Euro Area during the period 2005-2015, using the MIMIC-CUB Model. The analyses are based on Eurobarometer Survey data. This last group of measures relate to the specific deliverable of SYRTO concerning the measures on the monetary and policy intervention perceptions in Europe. Indeed, one aim of SYRTO was also concerning the citizens’ perspectives about policy and monetary interventions in the Eurozone, being convinced that inferring some confidence measures relative to these issues is potentially relevant in order to identify potential areas of improvement of the macro-prudential policy making process.
We then developed a measure of the citizens’ perception about the “Economics Health State” (EHS) for the countries of the Euro Area in the period 2005-2015 using the MIMIC-CUB Model introduced in Carpita et al. (2015) and Carpita and Ciavolino (2015) .
The measure EHS can be used with others macroeconomic indicators for economic analysis: increasing (decreasing) values of this estimate is a statistical evidence of more positive (negative) perception of the citizens about the European EHS (Carpita and Ciavolino, 2016).

6. SYRTO Code

The SYRTO Code contains the policy implications and recommendations for the measurement and management of systemic risk. Specifically, it explores the challenges for governance and coordination of macro-prudential policies aimed at systemic risk. The report draws on a wide body of academic research on systemic risk, such as has been done under the SYRTO grant, but not limited to it.
The report starts with six takeaways, that give the most important policy lessons that we can deduce from systemic risk research:

• models give useful early warning signals;
• low financial stress levels are not synonymous to high financial stability;
• the challenge is to make hard decisions based on soft information;
• manage the complexity of the financial system;
• there is evidence for a country-specific financial cycle;
• systemically important institutions are correctly identified.

The main chapters of the report articulate the lessons learned and are structured as an introduction with policy challenges, followed by chapters on (i) prevention, (ii) mitigation and (iii) stabilization.

Chapter 1 introduces the policy challenges regarding governance and coordination of systemic risk. The challenges are then addressed in subsequent chapters that deal with prevention, mitigation and stabilization. Financial firms and financial markets can be triggers and transmitters of systemic risk.

Chapter 2 deals with the triggers, both outside and inside financial institutions, and how to detect them early on. (The direct effect of low-probability events on financial firms). The chapter ends with recommendations on how to improve the governance and coordination of policies aimed at preventing systemic risk.

Chapter 3 deals with the transmission of systemic shocks (low probability/high impact events) through the financial system. Research finds an important role for the financial cycle, and a low-risk anomaly that can be indicative of an impending crisis. The chapter ends with recommendations on how to improve the governance and coordination of policies aimed at the transmission of systemic risk.

Chapter 4 describes the lessons learned on stabilizing the European banking sector. Assuming we can never completely prevent systemic events, we look at possibilities of limiting the damage and breaking transmission chains. The chapter ends with recommendations on how to improve the governance and coordination of policies aimed at the stabilization of the financial system after a crisis.

The Policy Challenges Addressed by the SYRTO Code

I. The Governance and Coordination of Systemic Risk
The financial crisis changed the consensus on the adequacy of traditional bank regulation, which focused on the solvency of a single institution. The basic insight is that the banking system can ‘run on itself’, because of a lack of trust between financial institutions. The old system assumed that the health of banks was adequately captured with risk-based regulation, which turned out to be false. When the losses mounted, it turned out that potential losses were severely underestimated. Moreover, it became quite hard to assess which bank was solvent and which was not. The financial position of multiple banks was threatened at the same time: a systemic crisis.
Systemic risk is the risk of the breakdown in the financial system, by the default of two or more institutions in the same time period. Systemic risks are characterized by (i) initial shocks of modest magnitude, and (ii) the transmission of those shocks between financial institutions that threatens their existence.
For the financial system we can identify two dependency chains in the “Black box” of contagion channels. First, a common shock can affect all institutions, caused by the collapse of an asset price bubble funded by debt. For example, a real estate boom and subsequent bust affects all the banks who have lent to real estate developers. This is a shock to all banks, caused by the common exposure. As such, it is a dependency chain that might not be observable ex ante, when regulators only focus on the health of individual banks.
The second source of dependency is contagion: a shock to just one or a few institutions spills over to other institutions and markets through the networked structure of the financial system. The contagion had no fundamental reason, but was caused by the distressed selling of sugar by a bank that had speculated with grain. In that way, the two market prices started to move together, and the problem of one bank spilled over to other banks, leading to multiple bank failures. A modern-day example is the credit crisis of 2007/2008 which was initially confined to a problem in real estate and CDOs. The losses that ensued led to the selling of other assets so that comovement arose between assets that were otherwise not related
The failure, or near-failure, of multiple banks in the 2007-2010 period have shown that new measurement techniques, policies and institutional structures are necessary to prevent or mitigate systemic risks in the futures. Below we introduce the systemic risk instruments that have been introduced after 2007, and the issues in terms of governance structures and the coordination between micro- and macro-prudential policies.

II. Systemic Risk Instruments
The regulatory reform initiated by the G20 in the aftermath of the crisis is close to finalization. In the EU, the new rules on capital and liquidity represent the first defense for preventing the accumulation of systemic risk. They have incentivized banks to move towards safer business models and required more robust capital and liquidity buffers to those institutions willing to operate in riskier markets. Better capitalized banks are also better positioned for supporting lending and economic growth. There is indeed increasing evidence suggesting a positive correlation between strong capital ratios and banks capacity to sustainably lend into the real economy.
The repair process of the European banking system since 2011 has led to a major strengthening of banks’ capital base. EU banks increased their common equity tier 1 (CET1) ratio between 2011 and 2014 from 9.2% to 12.1%, see European Banking Authority (2015). While banks have further reduced exposures in certain areas or business lines, for instance, in investment banking, total asset volumes increased by about 6% as of December 2014.
New regulations have focused on the following aspects:

A. Mitigating liquidity risk
Liquidity risk has been, if not the source, the main driver of the financial crisis. The combination of poor liquidity management and a reliance on short-term funding led to multiple failures and near-failures when liquidity disappeared. Earlier regulation operated on the assumption that robust capital cushions would shield banks against major shocks.
Already in 2008, the Basel Committee published the Principles for Sound Liquidity Risk Management and Supervision. These provided guidance on the risk management and supervision of funding liquidity risk in order to foster better risk management practices. In addition, the Committee introduced two minimum standards (the liquidity coverage ratio (LCR) and the net stable funding ratio (NSFR)) for liquidity and funding, which pursue the objectives of promoting short-term resilience of a bank’s liquidity positions as well as longer-term funding stability.
Stricter requirements and supervision have also been introduced on banks’ funding plans. Banks are now explicitly requested to develop a funding strategy that provides effective diversification in the sources of funding. While banks should plan their funding strategy under business-as-usual circumstances, they are also required to consider contingency plans to be activated in case of emergency situations, both idiosyncratic and systemic. This mitigates the transmission of systemic shocks through the banking system that could arise from the forced liquidation of (illiquid) assets to cover a funding shortfall.

B. Higher capital ratios
Higher capital requirements have come into force, which mitigates the transmission of shocks. Contingent capital and bail-in capital serve the same purpose. Counter-cyclical capital buffers (CCB) for systemically important financial institutions (SIFIs) lean against the build-up of debt-driven asset price bubbles that are known for triggering systemic problems. The CCB can vary between 0% and 2.5% of risk-weighted assets (RWA) and is switched on by national authorities when deemed necessary.

C. Reducing asset volatility.
For US-banks, the Dodd Frank act limits proprietary trading, which reduces the vulnerability of individual banks to shocks. Ring fencing ensures that consumer banking activities are shielded from more risky banking activities. For the remaining financial market activities of banks, central clearing (CCP) for swaps and credit value adjustments (CVA) reduce the counterparty risk from derivative transactions, which limits the fall-out of a defaulting counterparty to the financial system.

D. Improving supervision and resolution
In the European context the problem of resolution was made harder by the system of national supervision for cross-border banks, which made it hard to assess solvency, liquidity and to estimate the externalities of bank failures. To improve the supervision of large European banks, the banking union has been formed. The single supervisory mechanism (SSM) and the single resolution mechanism (SRM) are designed to ensure a fair and orderly supervision process and increase the objectivity of the decision to close down a troubled bank. The SRM reduces the uncertainty and disruption in case of a looming default, mitigating the transmission of initial shocks through the financial system.

III. Governance Structures for Systemic Risk
The introduction of systemic risk instruments has gone hand in hand with the development of governance structures, such as changes in the ECB’s responsibility, the European Banking Union, the SSM and the SRM. The SYRTO research on systemic risk has consequences for these institutions and the governance of systemic risk.
From the early warning research comes a clear need for a governance mechanism to set the threshold for false warnings. The early warning models produce forecasts on the probability of an impending crisis, but they come with a band of uncertainty. The uncertainty gives rise to two problems, namely that of false warnings (act, but there is no crisis) and that of missed crisis (not acting, but a crisis still occurs). Policy makers need to understand this choice and decide on thresholds for acting. In this dilemma, a clear governance structure is important.
A problem for researchers is that of the missing counterfactual: successful interventions will appear in the data as “no crisis”. It makes statistical inference on crisis-signals harder, and complicates the communication of an institution that is responsible for systemic risk mitigation. The public might argue that the enacted policies have been unproductive, since a crisis did not materialize and the immediate costs were quite visible. For example, higher lending standards have affected the spending power of consumers. The low volatility paradox gives rise to such problems, where a systemic risk supervisor might want to intervene right at the time when market-implied risks are at their lowest.
Two international institutions that have been involved in systemic risk are the Basel Committee on Banking Supervision, and the Financial Stability Board. They have no official jurisdiction, but they have been instrumental in proposing measures to determine which institutions are systemically important. Research in SYRTO has looked at the same issue, i.e. which institutions contribute most to systemic risk, from a more econometric angle.
Note that the effective governance of systemic risk could involve a large role for international institutions. For example, researchers from both the IMF and the BIS have warned for the problems of excessive credit growth, financial innovation and the potential for systemic risk. These institutions are less susceptible to national interests or industry lobbying and can act more independently. To some, this is their biggest weakness, but for systemic risk it could be exactly the right thing to have.
The effectiveness of the single resolution mechanism will become clear in the coming years. We described how coordination in case of a bank default between countries could work. A voting mechanism based on the asset shares or loan shares of banks in each country leads to outcomes that are close to what a supranational supervisor would achieve. One implication is that decisions made by a supranational supervisor, such as the ECB in the European Banking Union, are quite close to that of an optimal voting scheme.
An important governance issue is the role of the ECB in stabilizing European financial markets. It seems that its interventions were mostly successful, and, in terms of results, the ECB as institution serves the purposes of mitigating systemic risk well.

IV. Coordination of Macro and Micro-Prudential Responsibilities
SYRTO research holds lessons for the best mechanisms to coordinate macro- and micro-prudential supervision.
In the traditional view, prudential supervision at the level of a single institution is enough to mitigate the potential moral hazard problems that arise from deposit insurance. As such, in the EU, national supervisors were responsible for the micro-prudential supervision tasks. The EU-passporting agreements arranged for home-country supervisors to take the lead in supervising cross-border banks.
From the perspective on systemic risk it is clear that the micro-prudential approach is not enough. The reason is that the interventions or regulations that are necessary to make the system more resilient depend on the interplay between institutions. Policymakers have dubbed this the “macro-prudential” approach to supervision, where the word “macro” refers to the perspective of the system-as-a-whole.
Macro and micro-prudential policies overlap in that they are both aimed at the stability of financial institutions. They deviate in the area where micro-prudential interests are the protection of consumers of a single bank and macro-prudential interests are the stability of the system as a whole.
On the supervisory structure, we recommends to assign macro-prudential powers to a single body. This facilitates ownership and designates clear responsibilities. To prevent gridlock when micro and macro-prudential concerns do not coincide, it may be appropriate to define a hierarchy of objectives, where the macro-prudential objective takes precedence. This could be a procedure to agree upon for the ECB and ESRB.
Micro- and macro-prudential perspectives can collide in the case of low market-implied risks. Low volatility can be a warning signal for systemic crises. But low volatility is a good thing from a micro-prudential perspective: buffers appear to be high, and risk appetite appears to be low, because the measured riskiness of the assets is low. For the interplay of the two responsibilities, it is important for micro-prudential supervisors to incorporate the systemic risk assessment in their appraisal of the soundness of institutions.
Central clearing has a positive influence on the stability of a single institution, because the uncertainty and unpredictable contagion effect caused by the default of one counterparty is mitigated. All transactions are cleared centrally and the default of one clearing member is borne by the default fund of the clearing, and, ultimately, the clearing members. However, in the case of a large systemic event, there is a risk that the system-wide impact is larger than in the de-centralized setup. For the effective interplay micro-prudential and macro-prudential responsibilities, it is therefore of key importance that the size of the default fund reflects macro-prudential concerns.
Contingent convertibles contribute to the loss-absorbing capital for a bank and are admitted by micro-prudential supervisors to fulfill capital requirements. However, their widespread use could lead to new channels of contagion. Micro and macro-prudential supervisors will need to coordinate on mitigating potential channels for systemic risks to propagate through the system by subsequent triggers of contingent convertible bonds.
In the mitigation of systemic risk, the interplay between micro- and macro-prudential authorities becomes a concern in terms of the financial cycle. The research in SYRTO finds that a financial cycle can be identified as a separate cyclical component in time series of credit growth and house prices. However, the cycle is different per country and per asset class. This creates the need for information sharing between country-level experts and a macro-prudential supervisor. The guidelines from the ESRB incorporate this intuition, by proposing countercyclical capital buffers in capital regulations for banks, which are switched on and off on a country-by-country basis.
Coordination between micro- and macro-prudential tasks is important for stabilization, in the areas of stress-testing, complexity and the policy maker’s loss function. In stress-testing, micro-prudential supervisors need to coordinate with macro-prudential authorities on the appropriate stress scenarios that not only stress a single institution, but include system-wide shocks and take potential channels of contagion into account. Done properly, stress tests are a good crisis management tool that benefits stabilization.
The complexity of the financial system might not always be clear from the micro-prudential view. The complexity of financial products and business practices of banks deserves specific attention from the micro-prudential supervisor. The complexity of interactions and causality chains should be on the radar of the macro-prudential supervisor. The coordination between the two types of supervisors is necessary to obtain a comprehensive assessment of where the largest downside risks related to complexity are.
The policy maker’s loss function defines the trade-off between missed crises and false warnings. These are the typical type-I and type-II errors in statistical inference and, in the policy space, pose a specific challenge to the communication and interventions of the supervisor and regulator. At the one end is a missed crisis, which is obviously of great concern. But at the other end is the fall-out from too many false warning. Systemic risk instruments will be used frequently, and an actual crisis will seldom materialize. This could hurt the reputation of regulators. This is both a matter for the governance of systemic risk in general, as for the coordination between micro- and macro-prudential responsibilities.

Potential Impact:
Our group of researchers assembled a mix of strong competences to explore EU economic governance including monetary and policy coordination as well as financial market supervision.
The way with which SYRTO handled these issues was primarily conceived with the end to create a community in which academic scholars interacts with policy circles and market practitioners.
The main impacts we realized are firstly on knowledge dissemination and scientific networking especially within the Euro area, but also expanding in U.S. to ensure that results generated from research is available and can be fully used to contribute to the knowledge of systemic risks and its impacts on financial stability in the Union. In doing this, the dissemination activities used by the group focused on:

1. Website and other communication tools;
2. Conferences and meetings;
3. Scientific networking.

1. Website and other communication tools

The SYRTO Project website (http://syrtoproject.eu/) is developed in WordPress and contains sections relative to:
Objectives and structure of the project;
The consortium, the research team, and the advisory board (including links to personal web pages);
Programs and materials of the conferences organized by the consortium;
Publications, including policy-related documents and the SYRTO Working Paper Series;
A blog in which the project-related news are published;
The online EWS platform;
A restricted area, accessible only for the SYRTO researchers.

News is communicated to the public via the project blog and social media accounts. To facilitate their dissemination, all the main results of the project are published on the website in the form of both papers (see the SYRTO Working Paper series and related policy documents) and presentations made at conferences and/or seminars (these are published on the blog and also on the SlideShare account connected to the project).
Additional communication tools we made available are on:
Facebook (www.facebook.com/SYRTOproject)
Twitter (@SYRTOproject),
SlideShare (www.slideshare.net/SYRTOproject).

From May 2014, we also publish a bi-annual SYRTO Newsletter using mailchimp. The newsletter contains links to recent publications from the SYRTO team, conferences, and updates on upcoming research activities.

2. Conferences and Meetings

Scientific interconnections were also achieved through conferences and meetings organizations. We specifically organized the following 7 international conferences/meetings which received a great interest in Europe and US.

- First International Conference
The First International Conference of the SYRTO Project was held on June 25, 2013 at the Department of Economics and Management of the University of Brescia (Italy).
The topics of the conference were oriented to some operational instruments that were proposed to identify the areas of financial instability and to address the measures of monetary policy and macro-prudential supervision that are appropriate for preventing managing, and resolving systemic crises in Europe.
The conference program and materials are available on the SYRTO website (http://syrtoproject.eu/category/conferences/first-international-conference/).

- GRETA Conference. 12th International Conference on Credit Risk Evaluation Designed for Institutional Targeting in Finance
The GRETA Conference was held September 26-27, 2013 at the Scuola Grande di San Giovanni Evangelista in Venice (Italy).
The conference focused on risk, regulation, and opportunities in an increasingly interconnected world and provided an opportunity for participants to discuss both the causes and implications of recent events in the financial markets, which might result in the suggestion of fruitful directions for future research.
The conference program and all materials are available on the SYRTO website (http://syrtoproject.eu/category/conferences/greta-conference/) and on the conference website (www.greta.it/credit/credit2013/programme.html)

- SYRTO Code Workshop
The SYRTO Code Workshop was held on July 2, 2014 at the Head Office of Deutsche Bundesbank in Frankfurt am Main (Germany).
The workshop was organized by the SYRTO team jointly with Bundesbank, the European Central Bank (ECB), and the European Systemic Risk Board (ESRB). The aim of the workshop was to discuss with researchers and experts the main provisional results of the SYRTO project, and to select the most relevant topics for the SYRTO Code.
The workshop’s program and materials are available on the SYRTO website (http://syrtoproject.eu/category/conferences/syrto-code-workshop/).

- CSRA Research Meeting
The Consortium for Systemic Risk Analytics (CSRA), jointly with SYRTO project, organized the 2014 semi-annual research meeting to be held at the Massachusetts Institute of Technology in Cambridge, Massachusetts on December 15th 2014. During the Conference some of the SYRTO researchers discussed the main provisional results of the project.
The conference program and materials are available on the SYRTO website (http://syrtoproject.eu/category/conferences/csra-research-meeting/ ).

- SYRTO Conference on Systemic Risk
The SYRTO Conference on Systemic Risk was organized by the Amsterdam School of Finance and Risk Management and the Econometrics Group of VU University Amsterdam and took place in Amsterdam on June, 4-5 2015. The aim of the conference was to discuss the potential for translating statistical findings into concrete policy triggers, as well as how systemic risk measures can help to support policy making
The conference program and all materials are available on the SYRTO website (http://syrtoproject.eu/category/conferences/conference-on-systemic-risk/).

- EFMA Special Session
The SYRTO team organized a special session dedicated to “Systemic Risk Tomography: Signals, Measurements, and Transmission Channels” at the European Financial Management Association 2015 Annual Meetings. The special session took place in Amsterdam on June 25, 2015.
The conference’s program is available on the SYRTO website (http://syrtoproject.eu/category/conferences/efma/ ).

- Final International Conference
The Université Paris1 Panthéon-Sorbonne and our SYRTO team organized on February 19, 2016 the final conference on the SYRTO Project, presenting the main results, including the SYRTO Code and the Early Warning System Platform.
The conference program and all materials are available on the SYRTO website (http://syrtoproject.eu/category/conferences/final/).

3. Scientific Networking

Our group involved researchers outside the Consortium focusing on systemic risk, macro-prudential regulation as well as monetary and policy coordination. As a result, SYRTO plays now a key role in some of the most important scientific networks dealing with such issues. More specifically:
- SYRTO is included within the Systemic Risk Hub (www.systemic-risk-hub.org/) an independent and collaborative initiative that brings together leading researchers, academics, and professionals with the aim sharing the most outstanding developments in the current literature regarding systemic risk, thereby creating a platform for the transfer of knowledge and for professional dialogue.
- SYRTO is the Academic Partner of the CSRA (www.systemic-risk.org/about-csra/csra-academic-board) which is a not-for-profit industry consortium (a Delaware corporation) of financial institutions and academics founded with the objective of pooling expertise and risk analytics to provide greater transparency of systemic risk exposures in the global financial system.
- SYRTO established a networking relationship with ECB (and ESRB), Bundesbank and the Volatility Institute. Together with ECB and Bundesbank we are now exploring future collaborative activities on the same topics upon which SYRTO was launched.

Academics and policy circles are called to create a networking community to advance the “knowledge base that underpins the formulation and implementation of relevant policies in Europe as regards macro-economic and monetary integration in Europe”.
The design of a powerful multi-comprehensive macro-policy model for EU requires to consider the interrelation between monetary and fiscal policy across regions in order to model a macro-policy framework able to serve as prescriptive for policy makers.
The common monetary and policy measures lack in considering risks jointly, since there is no coordination between fiscal and monetary policy as factors affecting the level of systemic risk. As a result, it lacks also in the macro-prudential regulation.
With SYRTO we provided pragmatical tools for the macro-prudential regulation within the European economies taking into account the complex relation among monetary policy, fiscal policy, real economy and systemic disturbances.
SYRTO was then conceived, articulated and devoted to lay the foundations for a better macro-economic and monetary integration in Europe.

The expected impacts of the project are directed towards the developing of conceptual and analytical underpinnings for efficient macro-prudential policies in order to reshaping the European economy and strengthening the political unity of the EU.
Our two main deliverables, the SYRTO Code and the EWS Platforms, have indeed important impacts and implications we are convinced may help identify the impending risk abnormalities while providing the right measures of prevention and intervention.
Our reflections on the policy implications and recommendations for the measurement and management of systemic risk contained in the SYRTO Code comes in fact from a totally independent scientific and policy perspective.
In the same way, we realized a EWS platform based on the most important policy lessons that we deduced from systemic risk research for (i) visualize our systemic risk measures, (ii) monitor their risk signals, and (iii) provide a geographical risk mapping. No matter about a priory economic, financial, econometric, and policy prior beliefs. In this respect, the forthcoming SYRTO Centre we are planning to launch could play an important role in the academic and policy circles, being an independent eye on systemic risks in the Eurozone providing research outputs to academics, practitioners, regulators and policy makers globally. This Centre will use the methodologies developed by the team and the results of empirical analysis to produce indicators about the state of the markets, corporates, banks and financial intermediaries, and sovereigns both through the publication of periodic reports, and through risk indicators.

List of Websites:
www.syrtoproject.eu

Roberto Savona
Primary and Scientific Coordinator of the SYRTO Project
Associate Professor of Financial Markets and Institutions (University of Brescia, Italy)
roberto.savona@unibs.it

Prof. Monica Billio
Scientific Coordinator of the SYRTO Project
Professor of Econometrics (University Ca’ Foscari Venice, Italy)
billio@unive.it

Final Report Summary - SYRTO (SYstemic Risk TOmography: Signals, Measurements, Transmission Channels, and Policy Interventions)

Diese Seite teilen

Herunterladen