Information-Theoretic Cost-Benefit Analysis of Hybrid Decision Workflows in Finance.

Analyzing and leveraging data effectively has been an advantageous strategy in the management workflows of many contemporary organizations. In business and finance, data-informed decision workflows are nowadays essential for enabling development and growth. However, there is yet a theoretical or quantitative approach for analyzing the cost-benefit of the processes in such workflows, e.g., in determining the trade-offs between machine- and human-centric processes and quantifying biases. The aim of this work is to translate an information-theoretic concept and measure for cost-benefit analysis to a methodology that is relevant to the analysis of hybrid decision workflows in business and finance. We propose to combine an information-theoretic approach (i.e., information-theoretic cost-benefit analysis) and an engineering approach (e.g., workflow decomposition), which enables us to utilize information-theoretic measures to estimate the cost-benefit of individual processes quantitatively. We provide three case studies to demonstrate the feasibility of the proposed methodology, including (i) the use of a statistical and computational algorithm, (ii) incomplete information and humans' soft knowledge, and (iii) cognitive biases in a committee meeting. While this is an early application of information-theoretic cost-benefit analysis to business and financial workflows, it is a significant step towards the development of a systematic, quantitative, and computer-assisted approach for optimizing data-informed decision workflows.