RISK ASSESSMENT IN THE EU: NEW INDICES BASED ON MACHINE LEARNING METHODS
The project takes as its starting point the observation that in the EU countries there is a lack of instruments to monitor the risk of each EU financial market and the risk of the EU financial market as a whole. In fact, only a few countries (mainly from northern and central Europe, the most developed ones) adopt a volatility index traded in the internal stock market, and none of the countries adopt a more advanced index in order to measure additional tail risk (such as the Chicago Board Options Exchange SKEW index). Moreover, an aggregate EU volatility index has not yet been developed. In fact, the way volatility and skewness indices are computed is obsolete and calls for a radical change. The volatility and skewness indices have proven to have a low forecasting power on future returns and to measure a limited portion of the overall risk of a country market index (see Elyasiani et al. (forthcoming)), giving conflicting signals most of the time. On the other hand, investors, firms and regulators need a simple and single measure of risk assessment that gives a clear and unambiguous signal of market risk.
The main obstacle to overcome for the construction of such indices is the limited availability of option-based data for European peripheral countries. In addition, it is important to respond to the need for new techniques accounting for uncertainty in data and data processing and the narrow focus of the above indices, which treat financial markets as compartmentalized and overlook macroeconomic variables such as inequality indexes (see Kumhof et al. 2015) and other important determinants in risk assessment.
The vision behind this project is to transfer into risk assessment recent breakthroughs from machine learning (ML) methods. The project is intended to open an entirely new field of research, capable of offering the academic community and a range of stakeholders new insights not only in the present field of application, but also in other fields of research that are currently characterised by similar constraints.
references updated 01-04-2019-1
PARTICIPANTS: University of Modena and Reggio Emilia, Italy (UNIMORE) + University of Ghent, Belgium (UNIGHENT)
DO YOU WANT TO BE A PARTNER? PLEASE CONTACT: