A method for modelling operational risk with fuzzy cognitive maps and Bayesian belief networks. (January 2019)

• Record Type: Journal Article
• Authors: Azar, Adel; Mostafaee Dolatabad, Khadijeh
• Is Part Of: Expert systems with applications. Volume 115(2019)
• Page Start: 607
• Page End: 617
• Publication Date: 2019-01
• Subjects: Operational risk -- Causal modelling -- Bayesian belief networks -- Fuzzy Cognitive Maps; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2018.08.043
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: Fuzzy cognitive map is used to improve Bayesian network capability in poor data issues. We proposed a new migration method from FCM to BBN. The proposed method extracts BBN parameters from FCM ones. The proposed method is capable to model operational risk as a data-poor issue. Abstract: A main concern of risk management in financial institutions is measurement of operational risk and its value at risk as a requirement of Basel II accord. Besides risk quantification, identifying causal mechanism leading to operational loss is necessary to plan risk mitigation activities. Bayesian belief networks (BBN) is a causal modelling method able to achieve both goals simultaneously. Eliciting BBN causal model and its parameters from expert knowledge is an alternative to data driven models in case of data scarcity. However, there is still a problem with parameter extraction for complex models with a number of multi parent and multi state nodes. In this paper, we proposed a method combining fuzzy cognitive maps (FCM) and BBN in order to improve BBN capability in modelling operational risks. In the first phase, a causal model is constructed by applying FCM and then a new migration method is proposed to translate FCM parameters to BBN ones. A case study of an Iranian private bank is then given to examine and validate the proposed method.