FinPathlight: Framework for an multiagent recommender system designed to increase consumer financial capability
Introduction
In spite of increased consumer protection regulations in the wake of the Great Recession of 2008, conflicts of interest between human financial advisors and consumers remain pervasive in the industry; with traditional financial advisory firms consistently found to be placing their own interests ahead of their clients [[1], [2], [3], [4], [5]]. With the "trust" bar being set so low by human advisors, FinTech recommendation methods, which may be objectively reviewed and evaluated, may provide a means for a more trusted method of providing financial advisory services than their incumbent human counterparts [6].
Further, within the current landscape of financial technology (FinTech), the usefulness of applications designed to provide consumer financial decision support with financial goal recommendations are limited in scope; mainly to recommendation assistance with basic and limited, generic goals of increasing savings. To our best knowledge, no current FinTech application exists to specifically provide the recommendation of a comprehensive set of useful financial goals designed for wide range of users from a variety of financial backgrounds and situations. The need for this type of financial recommender system is evidenced by a recent ongoing survey by the National Financial Education Council (NFEC), which indicates that 70% of consumers do not understand how to set personal financial goals [7]. Numerous studies indicate that, for a significant number of U.S. households, traditional financial goal recommendations such as “buying a home”, “saving for retirement” or “investing for college education”, are beyond their current level of financial capability. To wit, the results of a 2015 FDIC survey categorized approximately 9.0 million households, or 7% of the U.S. population, as “unbanked”; meaning that they do not even have a savings or checking account [8]. Another approximately 24.5 million U.S. households, or 20% of the population, were deemed ‘underbanked’; meaning that, although they may have a savings or checking account, they still utilize alternative financial services (AFS), such as payday loans, refund anticipation loans, rent-to-own services, pawn shop loans, or auto title loans (often characterized as ‘predatory’ due to their relatively high interest rates and repayment terms) [8]. For these consumers, a more useful recommender system might include recommendations for an array of specific, achievable financial goals such as how to effectively “make ends meet”, or the “appropriate utilization and management of financial products”; financial capability enhancing objectives typically eschewed by professional financial advisory services.
In consideration of the pervasive mistrust of human professional financial advisors and, given the inadequacies in terms of utility of FinTech alternatives for financial goal recommendation, this research addresses a current gap in the landscape of financial recommender systems. To address these gaps in the trustworthiness and utility of current recommendation systems, we specifically aim to address five research questions (RQ) (Table 1).
In answering these research questions, we develop a framework, called FinPathlight, for a FinTech application designed to provide financial goal recommendations for a wide range of consumers with varying levels of financial capability. In order to assess the viability of an application built using the FinPathlight framework in this study, we evaluate the framework in terms of ‘trust’ and ‘perceived usefulness’; constructs adopted from a conceptual model for the evaluation of recommender systems [9]. In doing so, this study provides three primary research contributions (RC) (Table 2).
The remainder of this paper will be organized as follows: Section 2 provides a review of relevant literature and the current FinTech landscape related to this study. Section 3 discusses the design science research methodology applied to this study. In Section 4, we discuss the theoretical foundations of the study. Section 5 elaborates the implemented components of a multiagent architecture for a PFRS designed to increase consumer financial capability. Section 6 provides an experimental evaluation of the FinPathlight framework. Finally, Section 7 presents this study's contributions, limitations and future research.
Section snippets
Recommender systems for financial planning
Within the area of Financial Planning, a subdomain of the domain of Finance, the majority of the extant recommender systems literature has focused on systems for providing suitable financial services or investment portfolio recommendations for professional financial advisors [[10], [11], [12], [13]]. For direct-to-consumer financial planning, the recommender systems literature is relatively sparse. Research on a “personal financial planning tool”, for personal allocation of resources, was
Design Science Research (DSR) methodology
Within the field of IS, Design Science Research (DSR) is a paradigm for the conceptualization, design, evaluation, and communication of the development of technology-based artifacts that solve real-world problems, while providing contributions to the existing body of knowledge. The DSR approach communicated within this research incorporates the work of Hevner [23] and Peffers [24], along with the Information Systems Design Theory (ISDT) suggested by Gregor and Jones [25]. The ISDT identifies
Theoretical framework
In developing the framework, the two major kernel theories which have informed us are: 1) the Financial Capability Model and 2) Multiagent System with Role-Based Agent Design.
FinPathlight: multiagent recommender systems framework
Following the widely adopted role-based agent design method, we have first analyzed the requirements and identified a variety of roles necessary for the framework to perform. During the design phase, we have assigned a number of closely related roles to the same agent in order to reduce communication overhead. Table 5 shows a set of roles identified for the framework and the agents which are assigned to perform those roles.
Our role-based agent analysis and design have developed the FinPathlight
Evaluation of the FinPathlight framework
Xiao and Benbasat [9] have created a conceptual model for the evaluation of recommendation agents. Based on the aim of our research, which is to produce a useful artifact to make an intervening change in the world through the application of constructive knowledge, in this evaluation we focus on a portion of Xiao and Benbasat's model; specifically, the constructs contained within the ‘User Evaluation of Recommendation Systems’ box highlighted in yellow (Fig. 3): 1) ‘Trust’, and 2) ‘Perceived
Conclusion, limitations and future research
Despite concerns with the general lack of trust in human professional financial advisors due to conflicts of interest, and given the inadequacies, in terms of utility, of FinTech alternatives for financial goal recommendations, no recommender systems are currently available to provide consumers with a comprehensive set of trusted and useful financial goals recommendations. As evidenced by our review of the literature, most of the existing recommender system research relating to financial
CRediT authorship contribution statement
Lawrence Bunnell:Conceptualization, Methodology, Data curation, Formal analysis, Writing - original draft, Writing - review & editing, Validation.Kweku-Muata Osei-Bryson:Conceptualization, Methodology, Formal analysis, Writing - review & editing, Validation.Victoria Y. Yoon:Conceptualization, Methodology, Formal analysis, Writing - review & editing, Validation.
Lawrence Bunnell holds a PhD in Information Systems as well as a Master of Science in Information Systems and MBA from Virginia Commonwealth University. He is currently working as a business intelligence and data analytics executive within the financial services industry. As part of his PhD work, he developed a framework as part of a design science research for an ontology-based, multiagent, hybrid-methods recommender systems FinTech application designed to increase consumer financial
References (82)
- et al.
Personalized finance advisory through case-based recommender systems and diversification strategies
Decis. Support. Syst.
(2015) - et al.
A multi-agent system to support evidence based medicine and clinical decision making via data sharing and data privacy
Decis. Support. Syst.
(2016) - et al.
An architecture and category knowledge for intelligent information retrieval agents
Decis. Support. Syst.
(2000) - et al.
Agile Integration Modeling Language (AIML): a conceptual modeling grammar for agile integrative business information systems
Decis. Support. Syst.
(2007) Developing of an ontological interface agent with template-based linguistic processing technique for FAQ services
Expert Syst. Appl.
(2009)- et al.
Design & natural science research on information technology
Decis. Support. Syst.
(1995) - et al.
Counterfeit product detection: bridging the gap between design science and behavioral science in information systems research
Decis. Support. Syst.
(2017) - et al.
The growth of modern finance
J. Econ. Perspect.
(2013) - et al.
Assessing the costs and benefits of brokers in the mutual fund industry
Rev. Financ. Stud.
(2009) - et al.
Is Conflicted Investment Advice Better Than No Advice?, National Bureau of Economic Research Working Paper Series, No. 18158
The supply side of household finance
Rev. Financ. Stud.
“Is your financial advisor working in your best interests?” the New York times, your money
The FinTech Opportunity
Top 3 Most Missed Questions, National Financial Educators Council National Financial Literacy Test Results
FDIC National Survey of Unbanked and Underbanked Households
E-commerce product recommendation agents: use, characteristics, and impact
MIS Q.
FIPS: A Financial Planification System by Case Based Reasoning, Proceedings of the AAAI-99, Workshop on Exploring Synergies of Knowledge Management and Case Based Reasoning
Knowledge-based interactive selling of financial services with FSAdvisor
Generating context-aware recommendations using banking data in a mobile recommender system
Personal choice point: helping users visualize what it means to buy a BMW
Personalized pricing recommender system: multi-stage epsilon-greedy approach
Agent-based computational investing recommender system
Financial planning recommendation system using content-based collaborative and demographic filtering
Designing a robo-advisor for risk-averse, low-budget consumers
Electron. Mark.
The rise of customer-oriented banking - electronic markets are paving the way for change in the financial industry
Electron. Mark.
Mint by the Numbers: Which User Are You? Intuit MintLife Blog
11 Financial Experts Reveal Their Favorite Money Apps
Making innovation more competitive: the case of FinTech
65 UCLA Law Review
A three cycle view of design science research
Scandinavian Journal of Information Systems Research
A design science research methodology for information systems research
J. Manag. Inf. Syst.
The anatomy of a design theory
J. Assoc. Inf. Syst.
Financial capability: What is it, and how can it be created?
Pathways to Financial Well-being: The Role of Financial Capability
An Introduction to Multiagent Systems
The GAIA methodology for agent-oriented analysis and design
Journal of Autonomous Agents and Multi-Agent Systems
User acceptance of computer technology: a comparison of two theoretical models
Manag. Sci.
Understanding the pathways to financial well-being, National Financial Well-being Survey: report 2
Financial Well-being: The Goal of Financial Education
CFPB Financial Well-being Scale: Scale Development Technical Report
Levels of Financial Capability in the UK: Results of a Baseline Survey, FSA Consumer Research Paper 47
A developmental model of financial capability: a framework for promoting financial knowledge and skills during the transition to adulthood
Int. J. Behav. Dev.
Cited by (0)
Lawrence Bunnell holds a PhD in Information Systems as well as a Master of Science in Information Systems and MBA from Virginia Commonwealth University. He is currently working as a business intelligence and data analytics executive within the financial services industry. As part of his PhD work, he developed a framework as part of a design science research for an ontology-based, multiagent, hybrid-methods recommender systems FinTech application designed to increase consumer financial capability. His current research focus is on knowledge classification, data analytics and machine learning. Professional experience includes data analytics high-level code development and data visualizations, as well as having directed digital development and marketing strategy for several fast growth and start-up e-commerce companies. He can be reached at [email protected].
Kweku-Muata Osei-Bryson is Professor of Information Systems at Virginia Commonwealth University in Richmond, VA. He is also currently a Visiting Professor of Computing at the University of the West Indies at Mona, and has also been Visiting Professor of Information Systems at the Ghana Institute of Management & Public Administration. Previously he was Professor of Information Systems & Decision Sciences at Howard University in Washington, DC. He has also worked as an Information Systems practitioner in industry and government in the USA and Jamaica. He holds a Ph.D. in Applied Mathematics (Management Science & Information Systems) from the University of Maryland at College Park; a M.S. in Systems Engineering from Howard University; and a B.Sc. in Natural Sciences from the University of the West Indies at Mona.
His research areas include: Analytics & Data Science, Knowledge Management, Expert & Decision Support Systems, ICT for Development, Cyber-Security, e-Commerce, and Multi-Criteria Decision Making. His research has been published in various leading research journals, and he is author or editor of 5 books. Currently he serves as a Senior Editor of Information Technology for Development, an Associate Editor of the European Journal of Information Systems, a member of the Editorial Board of Computers & Operations Research, and a member of the International Advisory Board of the Journal of the Operational Research Society.
Victoria Y. Yoon is Professor in the Department of Information Systems at Virginia Commonwealth University (VCU). She received her M.S. from the University of Pittsburgh and her Ph.D. from the University of Texas at Arlington. Her primary research interests have been in the application of Artificial Intelligence (AI) to support the complex decision- making process and the managerial issues of such technology. She has published in MIS Quarterly, Decision Support Systems, Communications of the ACM, Journal of Management Information Systems, Journal of Operation Research Society, and other journals. She is the recipient of the VCU School of Business 2016 Faculty Award of Excellence. In 2018 she is named as a Dean's Scholar Professor at the VCU School of Business. She is a senior editor of Decision Support Systems.