Publication detail

Management Board Composition of Czech Banking Institutions and Bank Risk: The Random Forest Approach

Author(s): Mgr. Diana Žigraiová ,
Type: Submissions
Year: 2017
Number: 0
ISSN / ISBN:
Published in:
Publishing place:
Keywords: Banks, management board composition, feature selection, machine learning, risk-taking
JEL codes: C45, G21, G34
Suggested Citation:
Abstract: The paper investigates how the management board composition of banking institutions affects their risk in the Czech Republic. For this purpose, we build a unique data set comprising selected biographical information on the management board members of Czech financial institutions holding a banking license over the 2001-2012 period and combine it with individual bank financial data. We apply a machine learning technique – the random forest – to identify the best predictors of bank risk and further interpret the model output. We find non-linear relationships between average directors’ age, average director tenure, the proportion of directors holding an MBA and the proportion of non-national directors and the observed bank risk proxies. As for average directors’ age, it appears to impact bank stability very little beyond a certain threshold. Decreases in average director tenure on board are found to reduce bank stability while increases in tenure enhance stability. In terms of directors’ education, large increases in the proportion of directors with an MBA enhance bank profit volatility. Furthermore, if the majority of directors on board are foreigners, bank risk, captured by profit volatility and the NPL ratio, increases substantially.

Partners

Deloitte

Sponsors

CRIF
McKinsey
Patria Finance