‘This risk management tool does calculate financial contagion’

17 January 2014

The CEO of private bank Van Lanschot, Floris Deckers, already admitted it in 2008; during a parliamentary hearing about the crisis with the captains of the Dutch financial industry, he acknowledged that the risk management systems were not able to capture the interdependences between risks.

This resulted in a dangerous underestimation of these risks. In 2008 they were ‘just’ talking about the banking crisis, that later turned into the Eurozone sovereign debt crisis. Roger Laeven, Professor of Risk and Insurance at the University of Amsterdam School of Economics specialises in financial risk management. Laeven: “We know that a shock in one financial market (referred to as a  jump) often sets off a cluster of shocks over time affecting  other markets as well. Financial markets and economies are so interwoven that financial contagion is inevitable and quickly spreads. The current models used in risk management are not able to catch that excitation phenomena appropriately.” Together with international colleagues Laeven created a new econometrical model that does take into account the risk of financial contagion and systemic risk.

European Systemic Board 

The model that Laeven, Yacine Aït-Sahalia of Princeton University and Loriana Pelizzon of the University of Venice created, is based on credit default swaps. These are popular derivatives that are used as credit insurance for sovereign bonds. The model can be used to calculate credit risk in general. “Right now we are working hard on translating our findings towards policy makers. Recently I presented our model to institutions such as the European Systemic Risk Board in Frankfurt,” says Laeven. This research is a follow up on Laeven’s and Aït-Sahalia’s earlier research in which they modelled excitation in the international stock markets. 

Laeven: “The data shows very clearly the excitation effect and investors know that the phenomenon is there. It’s just very complicated to include in risk management systems. We try to construct models that can do that. Our approach towards excitation of shocks is quite extraordinary. There is a lot of research on the subject itself. Where does it originate? Which influence is dominant? Is it copy-cat behaviour caused by panic or is the influence of trade restrictions after major losses the strongest driver?  A lot of research tries to explain the phenomena. We contributed a robust model based on probability calculus, statistics and econometrics.”

Where to intervene? 

Laeven says the biggest challenge for now is having the model put into operation. The research itself is in an advanced stage of being published in an established journal and the model has already been used in several dozen papers around the world. Laeven: “We receive quite a lot of requests for the computer code, mainly from scientists at this stage. Hopefully the central banks and governments will follow soon. Except for stress tests and risk analyses the model is also suitable for use the other way around: to calculate which financial impulse creates the most beneficial chain effect. This can help determine in which country to intervene. Our model gives insight in the interaction between markets.” Greece proved to be the best country for a financial impulse, creating positive reactions in Portugal, Spain, Italy and Ireland. It’s not the case that the interaction always works both ways. An intervention in Ireland may not be of much help for Greece and Europe is very vulnerable for shocks in the United States (‘A sneeze from the US gives Europe a cold’) while European jumps have less influence on markets in the US.

Necessary buffers

As early as 2007 Laeven mentioned in a speech the fact that the world missed out on important risk information by not being able to measure the excitation risk of shocks. “If our model is used for future scenarios, for example in stress tests of banks, it will give a much better picture of the required buffers. In our estimates, they can easily be about  20% larger. Considering the implications of our model for necessary buffers, I can imagine bankers won’t be jumping with joy about our model.” Having said this Laeven  adds that of course pleasing the banking sector isn’t the purpose of scientific research in the first place. Laeven: “As a scientist, I get the most pleasure out of constructing models and pushing forward these kind of econometrical ideas. But naturally, in the end I want them to be used as well. In that sense, we really hope we came up with the new standard for models to predict the impact of shocks on financial markets. Simply because it works.”

Amsterdam School of Economics