I recently returned from attending a conference on the indices of riskiness, which was jointly organized by CEAR from GSU, RiskLab at ETH and the Department of Banking and Finance at the University of Zurich.
Clearly, a shift away from the well-trod path of presentations and discussions about VaR, TVaR and its allocation can be observed. In particular in the US, scenario analysis is gaining importance. Also the Swiss Solvency Test has a set of prescribed scenarios that insurance companies need to evaluate.
Scenarios always come with a significant level of subjectivity because, by definition, the frequency and the severity are in the tail of the respective distributions where little or no historic data may be available to calibrate them. But the major issue is the result of the scenario analysis.
A question of ‘what if?’
We have grown accustomed to regulatory systems which use VaR or TVaR to determine the required economic capital of a firm. Scenarios do not lend themselves to be easily integrated into the risk model results from which capital requirements are derived. Should the regulatory system be broadened to include limits on scenarios or should new ways of integrating scenarios into the risk models be developed?
Since the Swiss Solvency Test had already in its original form used a formula – albeit a strange one – to integrate scenarios into the capital requirement calculations, it was no surprise that Damir Filipovic from EPFL Lausanne presented a much improved way of integrating scenarios.
The discussion uncovered much of the regulatory complexity introduced by integrating scenario calculations: Even with a much improved integration, it is unclear whether this system achieves its purpose to be more resistant to firms intentionally or unintentionally exploiting loopholes that allow them to achieve lower capital requirements.
Joining the dots
Regulators regulate a market by imposing solvency rules on the market participants. Historically a lot of research has focused on how to build economic capital models and which risk measure to apply to derive the capital requirements. This leaves out that many companies in a market are interlinked.
More recent research which was presented at the conference discusses how the stability of a network of companies can be assessed. This is directly linked to analytics which can be used to determine systemically important financial institutions (SIFIs), which is quickly turning into a hot topic.
The discussions focused on whether networks of financial institutions should be characterized by the solvency requirements of their intertwined balance sheets or whether cashflow/liquidity constraints should be modeled. The majority considered a liquidity approach to be more meaningful. The lack of available data forces people to work with the balance sheet / solvency characterization of the financial network.