RESEARCH @ HEC LIEGE
Asset and Risk Management (ARM)
What’s new for Asset and Risk Management (ARM) strategic research field over the past academic year?
At ARM, our research projects focus on corporate finance and alternative investments (Prof. Marie Lambert), asset and risk management (Prof. Georges Hübner), and financial risk modelling and econometrics (Prof. Julien Hambuckers). Prof. Julien Hambuckers joined the team in October 2018.
Below are some of the highlights of our domain for this past academic year:
Research collaborations and consolidation of our international networks
2018 – 2019 was a big year for ARM, with the visit of numerous scholars from world-renowned institutions.
Among others, we discussed and shared the latest research on artificial intelligence in portfolio optimization (Prof. Ghysels, University of North Carolina); profitability of private equity funds (Prof. Phalippou, University of Oxford) and the quest for true performance measurement of hedge funds (Prof. Agarwal, University of Atlanta).
On the research side, ARM’s members were quite active as well, with more than 10 papers published over the academic year (see the complete list in ORBI).
In particular, the paper written by Marie Lambert and Georges Hübner, entitled “Performance sharing in risky portfolios: The case of hedge fund returns and fees”, was published in the Journal of Portfolio Management, a leading journal among practitioners and academic in the field of asset management. In this paper, they provided several insights regarding the effects of leverage and short-sales constraints on competition in the asset management industry.
Contributions from our new colleague: Julien Hambuckers
Julien and his colleagues from University of Goettingen (T. Kneib) and Dortmund (A. Groll) studied the factors affecting losses due to frauds, cyber-attacks, legal disputes and human errors in banks, called operational losses.
This topic is particularly important for banks and their regulators, because some of these losses can be so large that they would bankrupt an institution if no capital requirements are put in place. In addition, due to our poor knowledge of operational losses, it is hard to identify a-priori which factors have a role to play in this phenomenon.
To overcome these issues, the authors developed a statistical method focusing on extreme value (called Extreme Value Regression) and combined it with a popular machine learning algorithm (called the LASSO) to select the relevant explanatory variables.
They were then able to easily identify the important determinants of extreme losses (mostly financial uncertainty and economic growth), and to provide a set of recommendations for risk managers.
With their method, a bank will have better capital requirements in times of crisis, providing also recommendations for regulators.
Discover more details in their paper available on ORBI here.
Helping regulators and risk managers anticipate extreme events with the g-and-h distribution
Julien teamed-up with two Italian colleagues from University of Trento (M. Bee) and Sacre Cuoro of Milano (L. Trapin) to further investigate which statistical model provides the best forecasts of operational risk indicators, namely Value-at-Risk (VaR).
VaR is a quantity that measures how bad an adverse event (e.g. a loss due to a fraud) can be, and is routinely used by banks and their regulators to compute capital requirements. Therefore, good forecasts are crucial to be sure that banks own sufficient reserves in case of a crisis.
The authors showed that a model, called g-and-h, is the best to anticipate extreme events, and provided a novel way to calibrate it on operational loss data. These findings offer new methods for regulators and risk managers to improve capital requirements.
The paper is available on ORBI here
Other important research contributions for the finance world include:
What are the effects of environmental risk and moral hazard on staging activity? Discover the trade-offs Venture capitalists face when doing stage financing in Marie Lambert’s paper (co-authored with J. Tennert and H.-P. Burghof) here.
Discover how using LASSO for full probabilistic forecasts could work for you, in this paper by Julien Hambuckers (co-authored with A. Groll, T. Kneib and N. Umlauf). Julien and his colleagues applied their methods using Munich rent data and data on extreme operational losses from the Italian bank UniCredit.