Hoohlo, Mphekeleli2024-09-122024-09-122022Hoohlo, Mphekeleli. (2022). Towards a framework for operational risk management in the banking sector [PhD thesis, University of the Witwatersrand, Johannesburg]. WireDSpace.https://hdl.handle.net/10539/40754https://hdl.handle.net/10539/40754A research report submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy to the Faculty of Commerce, Law and Management, Wits Business School, University of the Witwatersrand, Johannesburg, 2023The objective of this study is three-pronged. One, it investigates the factors that influence capital adequacy as measured by the covariates (exposure, frequency and severity) used in banking operations that accompany firms data-log loss reports. Two, it assesses the differential impact of discretionary (by adding artificial data) and non-discretionary (using real world data) loss disclosure on firms’ value-at-risk. R software is used to determine the value-at-risk. GLM and GAMLSS techniques are employed and subsequent tests of significance derive aforementioned influential factors, accompanied by a data augmentation algorithm in Matlab software to determine the differential impact of artificial and real world operational loss disclosures on firms’ performance in relation to meeting capital requirements. Three, it challenges firms’ risk-neutral assumption inherent in operational risk practice, asserting that; in theory, banking operations are more risk averse. Rattle software is used in a k-means cluster analysis method to determine whether controls compensate for persistent losses due to the firms’ natural risk aversion. The research arrived at estimates on the number of losses and their sizes; whereby exposure positively influences the risk ceded by the bank having “learned” from possible variations in past data, therefore improving operational risk manage- ment frameworks by introducing ex ante forward-looking components, whereas the addition of artificial data points by data augmentation circumvents former dilemmas of large and rare events so one can do more “learning”, notwithstanding the nature of the data’s suspect quality as they are constructs not observations. Nevertheless, the artificial intelligent EBOR framework’s performance improves on (Hoohlo (2014)’s applied data scaling and parametization techniques arrived at a proxy of about ZAR3B) former techniques for capital adequacy calculation of OpRisk opening up exploration modeling beyond historical accounts of significance to incorporate forward-looking aspects. Furthermore, checks and balances set up based on operational negligence slow down operational risk losses over time thereby establishing the move of firm risk tolerance levels away from risk neutrality, suggesting that banks are more risk averseen© 2022 University of the Witwatersrand, Johannesburg. All rights reserved. The copyright in this work vests in the University of the Witwatersrand, Johannesburg. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of University of the Witwatersrand, Johannesburg.Banking operationsRisk managementBanking sectorSDG-8: Decent work and economic growthThesisUniversity of the Witwatersrand, Johannesburg