Refine
Document Type
- Working Paper (4)
- Article (2)
- Doctoral Thesis (1)
Language
- English (7)
Keywords
- ADS (1)
- AI (1)
- Al (1)
- Algorithms (1)
- Decision support systems, Costs, Decision making, Multitasking, Usability (1)
- Fintech (1)
We examine government decisions to support troubled banks. Our contribution is the examination of how federalism can affect decisions to classify banks as systemically important. Whether a bank is viewed by politicians as ‘systemically important’ varies based on how its failure would affect supporters of the government. How a federation is designed has a strong influence on which banks are given public assistance. Where the top level of government is solely responsible for banks, there will be fewer systemically important institutions and so more banks will be allowed to fail. Where lower levels are responsible, governments will allow fewer failures. We use this approach to understand government support for failing banks in Germany. Our findings are relevant for the European Banking Union.
Robo-financial advisors are complex algorithmic decision-making systems with a high potential for mass adoption due to their low operating costs and multitasking abilities. The quantitative aspects of our study measure the efficacy and usability of explanations and qualitative aspects determine the effect of explanations on users and system usability.
The COVID-19 pandemic is posing unique challenges to policymakers across the globe, necessitating efficient action in short timeframes. During such crises, having the right data at the right time is crucial to making informed policy decisions. Traditional economic indicators can be inadequate owing to issues of timeliness, granularity, and difficulty in collection. There is a need therefore for higher-frequency and more granular data to track economic activities. These “alternative” or “proxy” high-frequency indicators could help assess the economic impact triggered by COVID-19, shape new economic policies, and understand the road to recovery. This brief argues for the creation of a publicly accessible, multi-sectoral dashboard based on sectoral, high-frequency indicators.
As the Covid-19 pandemic began to unfold in February, India’s dependence on Chinese inputs for the production of pharmaceutical products was debated intensely. This special report argues that the narrow discussion has fallen short in capturing India’s crucial role in global health as a provider of health-related goods to many developing countries. The report analyses trade data on over 200 categories of health-related goods, and provides quantitative evidence for the extent of many countries’ health-related import dependence on India. Given this de facto position of India, this report concludes that the current health crisis calls for a comprehensive Indian global health strategy that will allow the country to expand on the existing multilateral system.
As India completes 50 days of lockdown, this report presents the findings of a data-driven enquiry into the extent to which the lockdown has achieved its health objectives and arrested the spread of COVID-19. The success is measured on four parameters: flattening the curve, reducing the growth rate of new cases, containing the spread, and improving healthcare capacity. The findings show that while the lockdown has flattened the curve to an extent, it has failed to reverse the trend or contain the disease. Significant changes in strategy would have to be adopted to arrest the spread of COVID-19.
Robo Advisors are financial advisory apps that profile users into risk classes before providing financial advice. This risk profiling of users is of functional importance and is legally mandatory. Irregularities at this primary step will lead to incorrect recommenda- tions for the users. Further, lack of transparency and explanations for these automated decisions makes it tougher for users and regulators to understand the rationale behind the advice given by these apps, leading to a trust deficit. Regulators monitor this pro- filing but possess no independent toolkit to “demystify” the black box or adequately explain the decision-making process of the robo financial advisor.
Our paper proposes an approach towards developing a ‘RegTech tool’ that can explain the robo advisors decision making. We use machine learning models to reverse engi- neer the importance of features in the black-box algorithm used by the robo advisor for risk profiling and provide three levels of explanation. First, we find the importance of inputs used in the risk profiling algorithm. Second, we infer relationships between inputs and with the assigned risk classes. Third, we allow regulators to explain decisions for any given user profile, in order to ‘spot check’ a random data point. With these three explanation methods, we provide regulators, who lack the technical knowledge to understand algorithmic decisions, a method to understand it and ensure that the risk-profiling done by robo advisory applications comply with the regulations they are subjected to.
The rise of automated decision-making systems has been extraordinary in the last few years, both in terms of scale and scope of operation. This is especially true in the field of fintech, where robo financial advisors have gained prominence with their claim to “democratise” finance, with their low operating costs, multi-tasking abilities and potential for mass adoption. This series of three research papers is focused on demystifying algorithmic explainability in the field of fintech, by diving deep into both theoretical and practical aspects of the phenomenon, and contextualising the discussion to India.
The first paper explores the trade-off that emerges between performance and explainability for robo financial advisors, through a detailed review of available literature that allowed comparisons between relevant processes adopted around the world, and interviews with various stakeholders within India to understand the evolving domestic situation. The paper finds that it is not quite a question of if ADS will play a significant role in India’s financial services sector but more a question of when that will happen. The second paper operationalises algorithmic explainability in the particular context of risk profiling done by robo financial advisory applications. Here, an approach towards developing a ‘RegTech’ tool is outlined, which can explain the robo advisor’s decision-making, using machine learning models to recognise and reconstruct different levels of explanations. Finally, the third paper evaluates the effectiveness of user-centric explanations in conveying the decision-making logic of complex algorithmic systems in fintech. The paper demonstrates the usefulness of such explanations from the perspectives of both novice and seasoned investors, and goes on to differentiate between white- and black-box explanations.
In sum, this three-paper series, using a range of tools and approaches, examines in detail the under-appreciated regulatory and operational challenges that emerge during the use of algorithms in the field of fintech, and explores ways to resolve them. While the first paper looks at the present and future of AI in Indian fintech, the second paper develops a tool to explain a robo advisor’s decisionmaking, and the third finds factors that determine users’ comprehension and confidence in such systems. The results of these three approaches in the three papers become vital when seen in the context of the rapid rise of artificial intelligence across products, services and industries globally. If humans are to work alongside machines in this changing world, explainability becomes an important aspect to consider for companies, regulators and users alike, in order for humans to trust the algorithmic systems in place and improve outcomes — in this case, financial outcomes — for all.