Resumo

Operational risk management has become a top agenda among Brazilian banks. Extreme operational risk loss events materialized in Bank Barings, Natwest, Allied Irish, Societe Generale, UBS and Well Fargo. The operational risk scenario analysis tool has been specifically required for large financial conglomerates in Brazil since 2017. A deeper study of the scientific qualitative approach for preparing scenarios could complement the prevalent quantitative approach. Qualitative approach presents the challenge of identifying and overcoming cognitive biases when conducting scenario analysis.

During the preparation of operational risk scenarios, objectives of the present research are to identify, prevent and avoid biases. Therefore, there are two research questions in this paper: H1: We will experience 16 cognitive biases in operational risk scenario analysis. H2: Hazard and Operability technique and the literature on biases may list and describe ways to overcome the 16 biases that may emerge in the scenario analysis process. We did not locate academic research on the themes mentioned above treated simultaneously: operational risk scenario analysis, focus groups and biases.

Operational risk (OR) is defined as the risk of loss resulting from inadequate or failed internal processes, people, and systems or from external events. It includes legal risk. OR events also include fraud and business disruption. OR scenarios search for plausible severe losses beyond past experiences. From theories on biases and heuristics, we extracted 16: anchoring, availability, confirmation, peer pressure, framing, group polarization, groupthink, representativeness, satisficing, motivational, hindsight, survivorship, halo effect, overconfidence, base-rate neglect and self-serving.

Instead of a quantitative approach, the choice was a qualitative investigation. Focus group was adopted. As a secondary inspiration, we applied a technique of chemical engineering called Hazop, the prospect of possible deviations in industrial processes in relation to the intended design of the factory. Data were collected in three ways: recorded audio, volunteer profiles and consensus spreadsheets. There were five in-person focus groups. Each group followed a storyline and identified the necessity of raising additional information to streamline estimates in subsequent scenarios.

The research confirmed that both debating participants and scribes manifested biases and heuristics, even with all the efforts of the moderator to avoid these phenomena. Some of the 16 biases emerged: anchoring in group 1; availability in group 3; confirmation in groups 1, 2 and 4; framing in groups 2 and 4; group polarization in group 1; groupthink in group 2; overconfidence in group 5. Consequent propositions were dominated by business improvements and control activities enhancements. Scribes took note of 19 resulting attitudes and new controls that could be used by the analyzed bank.

Research hypothesis H1 was partially confirmed. During the focus groups, out of 16 cognitive biases, seven were noticed: anchoring, availability, confirmation, group polarization, groupthink, framing, and overconfidence. Research hypothesis H2 was confirmed. When knowledge from Hazop is combined with advances about cognitive biases, more tools are available to deal with either the seven encountered biases or the nine others. It is not advisable to generalize the findings. As no participant was linked to Wells Fargo may have allowed more independence to list new attitudes and controls.

Baybutt, P. (2016). Cognitive biases in process hazard analysis. Journal of Loss Prevention in the Process Industries, 43: 372-377. BIS (2021). Revisions to the Principles for the Sound Management of Operational Risk. Basel, mar. COSO (2013). Internal Control-Integrated Framework. Durham, NC. Girling, P. (2013). Operational Risk Management. Hoboken, Wiley, 352 p. Liamputtong, P. (2011). Focus group methodology: principles and practice. Thousand Oaks, Sage, 224 p. Tversky, A., Kahneman, D. (1974). Judgment under Uncertainty: Heuristics and Biases. Science: New Series, 185(4157): 1124-1131.