Not logged in.

Quick Search - Contribution

Contribution Details

Type Master's Thesis
Scope Discipline-based scholarship
Title Incorporating Expert Judgement to Model Non-Maturing Deposits
Other Titles Predictive analytics applied to SNB deposit volume and rate data
Organization Unit
Authors
  • Cédric-James Piaget
Supervisors
  • Erich Walter Farkas
  • Dominik Lambrigger
Language
  • English
Institution University of Zurich
Faculty Faculty of Business, Economics and Informatics
Number of Pages 102
Date 2018
Abstract Text Non-maturing deposits (NMD) such as sight and saving accounts make up a significant percentage of a bank's balance sheet. The behavioural optionalities embedded into the NMD drive the uncertainty in the future cash ow arising from client deposit interest income and expenses. This highlights the importance to model the client deposit rate and client volume. The thesis shows that in order to model the client deposit rate and client volume in a regression framework the least absolute shrinkage and selection operator (LASSO) method which is frequently used in machine learning techniques should be considered. The LASSO method leads to a parsimonious model with a high degree of explanatory power and stability with little risk of overfitting. In the context of NMD various regulatory requirements demand from banks to incorporate expert judgement in their risk management processes. Therefore the banks are encouraged to combine expert judgement and quantitative models. Instead of simply combining forecasts this thesis will in addition discuss how quantitative models can be amended to include expert opinion. We conclude that simple regression methods are not outperformed when using methods based on standard artificial intelligence methods such as the Kalman filter. Without a history of expert judgement the Kalman filter contains as much ambiguity as the regression based methods and therefore does not provide more significance. For practitioners this means that regression based techniques are a safe choice when it comes to the incorporation of expert judgement into a quantitative model.
Export BibTeX