Decomposing Financial Crises with SOTMs

Abstract

The provided models for macroprudential oversight have thus far concerned assessing the cross-sectional or temporal dimensions close to isolation. In this chapter, we turn the focus to exploring cross-sectional dynamics. The Self-Organizing Time Map (SOTM) provides means for visual dynamic clustering and thus also for illustrating dynamics in cross sections of multivariate macro-financial indicators. This is one of the key tasks in risk identification, when the focus is on build-up phases of imbalances in the entire cross section, such as the global dimension in country-level risks and a system-wide focus on data concerning individual financial intermediaries. With respect to the visual analytics mantra, the SOTM can be positioned similarly as the previously discussed Self-Organizing Financial Stability Map (SOFSM). The first decomposition applies the standard SOTM to describing the global financial crisis that started in 2007 in a manner that would be applicable for real-time surveillance. The second section uses a SOTM on time-to-event data to generalize patterns before, during and after financial crises.

This chapter is partly based upon previous research. Please see the following work for further information: Sarlin (2013a, b)