Frontiers of Research in Systemic Risk Forecasting


November 28th 2013

Time: 9.30am - 5.30pm
Venue: LSE, venue tbc to those who register
Organisers: Jon Danielsson and Kyle Moore
Speakers: Rodrigo Guimarães (Bank of England), Jong Ho Hwang (US Department of the Treasury), Anton Korinek (Johns Hopkins University), Bernd Schwaab (European Central Bank), Martin Summer (Oesterreichische Nationalbank), Wolf Wagner (Tilburg University), Chen Zhou (De Nederlandsche Bank)

The conference brings together academics and policy makers working in the area of systemic risk to discuss the latest developments in research in systemic risk forecasting.

Photographs from this event are available to view here.

The programme can be downloaded here.

Session 1

Session Chair: Jon Danielsson (SRC, LSE)

Session 2

Session Chair: Ron Anderson (SRC/FMG, LSE)

Session 3

Session Chair: Jean-Pierre Zigrand (SRC, LSE)

Session 4

Session Chair: Kyle Moore (SRC, LSE)

This conference is generously supported by the Economic and Social Research Council (ESRC) [grant number ES/K002309/1].

Full Report

On the 28th November the SRC organised the Frontiers of Research in Systemic Risk Forecasting conference.

Rodrigo Guimarães (Bank of England) opened the first session of the conference and presented his recent research (co-authored with Oliver Bush BoE, and Hanno Stremmel WHU) on macroprudential policy.  His analysis focuses on finding indicators that can be successfully employed by regulators in order to adjust capital requirements through the business cycle. Guimarães started by highlighting how, at the moment, there is very little theoretical support and empirical evidence to guide regulators’ choice. For example ‘quantity’ measures of risk (such as credit-gap or price-to-rent ratio) seem to be directly associated with distress in the financial sector but the timing of the observations for these measures may not be suitable for regulation purposes, while ‘price of risk’ indicators (market based measures like credit spreads) could be too volatile or partially unrelated to the policy object of interest. The current Financial Policy Committee (FPC) draft on this issue identifies 38 relevant indicators which are classifiable by three broad categories: banking sector balances, non-banking sector balances and market measures. Arguing that such a wide range of indicators may generate communication and accountability issues for policy makers, Guimarães and his co-authors reduced  this large set of measures to only three meaningful indicators via a two-step procedure : first, Principal Component Analysis (PCA) is applied to a subset of the 38 indicators (chosen on the basis of time series availability) , then the three indicators mostly correlated with the first components are used as explanatory variables against which financial crisis events are regressed. Guimarães’ research shows that two of the selected indicators (credit-gap and a VIX-based measure for the ‘price of risk’) are statistically significant (and robust) predictors of financial crisis, whilst the third measure (financial sector leverage) seems to have a lower and less stable correlation with crisis events.
The next presentation was by Bernd Schwaab (ECB) and was closely related to the precedent discussion about predictability of financial sector crisis via quantity or risk based indicators. Schwaab presented ‘Nowcasting and forecasting global financial sector stress and credit market dislocation’, a paper co-authored with Siem Jan Koopman and André Lucas (both VU University Amsterdam). In this paper the authors investigate the dynamics of credit dislocation (defined as a persistent decoupling of the credit risk cycle from macro fundamentals) and of global financial sector risk. By the means of a high dimensional partly nonlinear and non-Gaussian dynamic factor model, the authors identify a risk-based indicator of credit conditions that has in the past preceded episodes of distress in the financial sector. Schwaab argued that the mechanism underlying this link is that a credit cycle disconnected with the business cycle may signal very loose lending standards, increasing the probability of joint default of a large number of financial intermediaries. The new credit risk measure performs well compared to other suggested indicators, such as credit to GDP gap.
Wolf Wagner (Tilburg University) presented his work entitled ‘The disturbing interaction between countercyclical capital requirements and systemic risk’, co-authored with Bàlint Horvàth (Tilburg University). The paper proposes a new approach to tackling systemic risk which is that systemic risk can be better understood and managed if one studies the interaction between its ‘time’ dimension (ie. procyclicality) and its ‘cross-sectional’ dimension (ie. exposures or interconnectedness).  The paper aims at answering the following specific question: how does policy intervention in one dimension of systemic risk affect systemic risk in the other dimension? The authors propose a three-stage partial equilibrium model under the following setting: Banks are subject to a moral hazard problem that is solved by requiring bankers to put capital into the banks. Banker’s endowment with capital determined by prior returns on projects. This creates scope for capital requirements that depend on state of the economy (project returns). There exists a systemic cost: if both banks fail, the economy has insufficient funds to undertake a worthwhile project. Finally, banks can choose the level of correlation between their projects and this choice interacts with capital requirements. Wagner emphasizes that with these minimum ingredients, counter-cyclical macroprudential regulation can increase systemic risk in the cross-section.  The reason for this result is based on the fact that countercylical policies insulate banks from sector-wide fluctuations but not against bank-specific shocks. Indeed, the relative cost of being exposed to idiosyncratic risk increases and leads to more systemic risk-taking. The consequence is that macroprudential policies which improve systemic risk in one dimension (countercyclicality) worsen systemic risk in the other dimension (cross-sectional risk). Ultimately they can even lead to more procyclicality. However, the reverse problem does not arise: policies that reduce cross-sectional risk at the same time lower countercyclicality.  Therefore, cross-sectional policies are preferred.
Martin Summers (Oesterreichische Nationalbank) started his presentation by pointing out that the recent literature on the financial crisis identifies leverage as one of the key drivers of systemic risk. Hence the relevance, in the context of the conference, of his latest research paper ‘Endogenous leverage and asset pricing in double auctions’ co-authored with Thomas Breuer and Hans-Joachim Vollbrecht (both University of Applied Sciences Vorarlberg). The paper proposes a model of double auction trading for collateralized debt, where agents, with exogenous heterogeneous beliefs on the future payoffs of the traded real asset, submit limit buy and sell orders on both loans and real assets. The authors show that, by simulating this trading protocol, the process converges to the level that would have been predicted by a general equilibrium model. Summers emphasised that these findings shed new light on the features of market structure that lead to equilibrium prices and provide a basis for experimental investigation.
Following Summers’ presentation, Anton Korinek (Johns Hopkins University) presented his work ‘The redistributive effects of financial deregulation’, co-authored with Jonathan Kreamer (University of Maryland). By developing a formal model in which the financial sector benefits from risk-taking through earning greater expected returns, they analysed how the risk-taking of banks affects the distribution of surplus in the economy, and investigate what exactly the distributive effects of different financial policies are. They found that risk-taking by the financial sector leads to externalities on the real economy in bad states. The financial sector, however, does not internalise these externalities, resulting in the distributive conflict. Additionally, they identified a novel channel through which bailouts lead to redistributions between the financial sector and the real economy. Korinek concluded his presentation by discussing policy implications.
Session four began with Chen Zhou’s (De Nederlandsche Bank) paper ‘Looking at the tail: price-based measures of systemic importance’, which was co-authored with Nikola Tarashev (BIS). Zhou‘s presentation showed that information about rare events which is extracted from market prices by employing tools of extreme value theory, contributes substantially to measures of banks’ systemic importance. They examined the usefulness of such measures and found that they exhibit strong and intuitive relationships with simple bank characteristics of balance sheets and income statements. At the end of his presentation Zhou showed the correlation between systemic importance and future CDS spreads which generated a lively discussion amongst the audience.
The second paper of the session, titled ‘A proposal for an open-source financial risk model’ was presented by Jong Ho Hwang (US Department of the Treasury). He began by looking at a series of problems that we currently face on financial risk framework. In response to these current shortcomings, the author proposed a new model for gathering, measuring and disclosing financial risk information in the global economy. The proposed model envisions an open-source risk modelling system that is best-in-class, evolving and built from a methodology that is completely transparent. Moreover, the proposed open-source financial risk model separates the dual function that internal risk models perform within financial institutions. Zhou also discussed several potential issues such as herd mentality, governance issues, and economic incentives for desirable capital formation.