Periodic Reporting for period 2 - FIN-TECH (A FINancial supervision and TECHnology compliance training programme)
Periodo di rendicontazione: 2020-07-01 al 2021-06-30
The Financial Stability Board (2017b) defines FINancial TECHnology as “technologically enabled financial innovation that could result in new business models, applications, processes, or products with an associated material effect on financial markets and institutions and the provision of financial services”.
In the last few years, Fintech solutions that make use of big data analytics, artificial intelligence, and blockchain technologies have been introduced at an unprecedented rate. These new technologies are changing the nature of the financial industry, creating many opportunities for fintechs to offer more inclusive access to financial services (European Commission, 2018).
The advantages notwithstanding, FinTech solutions leave the door open for many challenges such as underestimation of creditworthiness (credit risk), investor protection issues (market risk), and cyber attacks and frauds (operational risk), which represent central points of interest for regulators and supervisory bodies.
The main aim of the FIN-TECH project has been to create a European knowledge exchange research and training programme, able to provide uniform risk management procedures that automatize compliance of Fintech companies (RegTech) and, at the same time, increase the efficiency of supervisory activities (SupTech).
The research and training programme has been built jointly by universities, regulators, supervisors and fintech, covering all 27 European Union countries, plus Switzerland and the United Kingdom.
The FIN-TECH project has successfully achieved its aim and, aa a conclusion of the action, It has provide a collection of 40 research papers, published in internationally refereed journals, and 12 use cases, that show case how to measure fintech risks to make fintech innovations sustainable. The proposed fintech risk measurement models are ready to be adopted as standard risk management practices.
In the last few years, Fintech solutions that make use of big data analytics, artificial intelligence, and blockchain technologies have been introduced at an unprecedented rate. These new technologies are changing the nature of the financial industry, creating many opportunities for fintechs to offer more inclusive access to financial services (European Commission, 2018).
The advantages notwithstanding, FinTech solutions leave the door open for many challenges such as underestimation of creditworthiness (credit risk), investor protection issues (market risk), and cyber attacks and frauds (operational risk), which represent central points of interest for regulators and supervisory bodies.
The main aim of the FIN-TECH project has been to create a European knowledge exchange research and training programme, able to provide uniform risk management procedures that automatize compliance of Fintech companies (RegTech) and, at the same time, increase the efficiency of supervisory activities (SupTech).
The research and training programme has been built jointly by universities, regulators, supervisors and fintech, covering all 27 European Union countries, plus Switzerland and the United Kingdom.
The FIN-TECH project has successfully achieved its aim and, aa a conclusion of the action, It has provide a collection of 40 research papers, published in internationally refereed journals, and 12 use cases, that show case how to measure fintech risks to make fintech innovations sustainable. The proposed fintech risk measurement models are ready to be adopted as standard risk management practices.
D 1.1: Network establishment
A large network of universities, regulators, supervisors and fintechs has been interconnected during the project, and is a strong basis for future developments.
D 2.1: Repository of consortium research papers in Big Data Analytics
11 papers have been selected among those published by the consortium partners in referred scientific journals, which are open access and acknowledge the project.
D 3.1: Repository of consortium research papers in Artificial Intelligence
12 papers have been selected among those published by the consortium partners in referred scientific journals, which are open access and acknowledge the project.
D 4.1: Repository of consortium research papers in Blockchain
17 papers have been selected among those published by the consortium partners in referred scientific journals, which are open access and acknowledge the project.
Other relevant papers will be continuously added to the ìpublication list, after the submission of the D 2.1 D 3.1 D 4.1 deliverables.
D 5.1: Repository of use cases and slides in big data analytics
Three full use cases have been selected for the application of big data analytics to credit risk management, have been presented in the project’s events, have been shared with the project's network and, finally, have been improved based on the received feedback. The use cases form a strong basis to develop risk management standards for peer to peer lending risk management.
D 5.2: Repository of use cases and slides in artificial intelligence
Three full use cases have been selected for the application of artificial intelligence to market (financial) risk management, have been presented in the project’s events, have been shared with the project's network and, finally, have been improved based on the received feedback. The use cases form a strong basis to develop risk management standards for robot advisory risk management.
D 5.3 Repository of use cases and slides in blockchain (D5.3)
Six full use cases have been selected for the application of artificial intelligence to market (financial) risk management, have been presented in the project’s events, have been shared with the project's network and, finally, have been improved based on the received feedback. The use cases form a strong basis to develop risk management standards for blockchain payment risk management.
D 6.1 Research and development environment
After an experimental activity aimed at building an appropriate environment, an open source repository of R codes has been produced for all 12 produced use cases. Each use case is thus fully reproducible, using the provided data, R code and descriptive paper.
D 7.2 Repository of events participation
A total of +6000 professionals, from universities, regulators, supervisors, fintechs and banks, have participated to at least one event of the project.
D 7.5 Repository of events feedback
A total of 1265 professionals, from universities, regulators, supervisors, fintechs and banks, have provided useful feedback to the project and particularly to its use cases, which have been improved.
D 7.8 Advisory board evaluation report
Further constructive feedback on the project and its use case, has been provided by the five members of the advisory board, chosen to represent non-European countries.
D 7.7 External evaluation report
An integrated overall evaluation report has been produced, for a final assessment of the project, and to decide which areas of research and training exploit and continue in the future. In summary, three main areas of development have emerged: a) network modeling to measure contagion risk deriving from fintech platforms; b) explainable AI models to make highly predictive machine learning models acceptable for regulators and supervisors; c) operational risk models to allow blockchain payments to emerge.
A large network of universities, regulators, supervisors and fintechs has been interconnected during the project, and is a strong basis for future developments.
D 2.1: Repository of consortium research papers in Big Data Analytics
11 papers have been selected among those published by the consortium partners in referred scientific journals, which are open access and acknowledge the project.
D 3.1: Repository of consortium research papers in Artificial Intelligence
12 papers have been selected among those published by the consortium partners in referred scientific journals, which are open access and acknowledge the project.
D 4.1: Repository of consortium research papers in Blockchain
17 papers have been selected among those published by the consortium partners in referred scientific journals, which are open access and acknowledge the project.
Other relevant papers will be continuously added to the ìpublication list, after the submission of the D 2.1 D 3.1 D 4.1 deliverables.
D 5.1: Repository of use cases and slides in big data analytics
Three full use cases have been selected for the application of big data analytics to credit risk management, have been presented in the project’s events, have been shared with the project's network and, finally, have been improved based on the received feedback. The use cases form a strong basis to develop risk management standards for peer to peer lending risk management.
D 5.2: Repository of use cases and slides in artificial intelligence
Three full use cases have been selected for the application of artificial intelligence to market (financial) risk management, have been presented in the project’s events, have been shared with the project's network and, finally, have been improved based on the received feedback. The use cases form a strong basis to develop risk management standards for robot advisory risk management.
D 5.3 Repository of use cases and slides in blockchain (D5.3)
Six full use cases have been selected for the application of artificial intelligence to market (financial) risk management, have been presented in the project’s events, have been shared with the project's network and, finally, have been improved based on the received feedback. The use cases form a strong basis to develop risk management standards for blockchain payment risk management.
D 6.1 Research and development environment
After an experimental activity aimed at building an appropriate environment, an open source repository of R codes has been produced for all 12 produced use cases. Each use case is thus fully reproducible, using the provided data, R code and descriptive paper.
D 7.2 Repository of events participation
A total of +6000 professionals, from universities, regulators, supervisors, fintechs and banks, have participated to at least one event of the project.
D 7.5 Repository of events feedback
A total of 1265 professionals, from universities, regulators, supervisors, fintechs and banks, have provided useful feedback to the project and particularly to its use cases, which have been improved.
D 7.8 Advisory board evaluation report
Further constructive feedback on the project and its use case, has been provided by the five members of the advisory board, chosen to represent non-European countries.
D 7.7 External evaluation report
An integrated overall evaluation report has been produced, for a final assessment of the project, and to decide which areas of research and training exploit and continue in the future. In summary, three main areas of development have emerged: a) network modeling to measure contagion risk deriving from fintech platforms; b) explainable AI models to make highly predictive machine learning models acceptable for regulators and supervisors; c) operational risk models to allow blockchain payments to emerge.
Expected results:
- development of a Fin-Tech risk management platform uniform across Europe with a potential improvement in risk mitigation;
- resulting balance between risk and opportunities which makes Fin-Tech innovation socially sustainable.
Both expected results have been achieved, producing 40 research papers on fintech risk measurement, selecting them in 12 use cases that are ready for standardisation, and disseminating the good practices in fintech risk management resulting from the proposed use cases in 87 SupTech, 30 Spinoff workshops, 6 RegTech workshop,s and 6 research conferences.
The main contribution of the project for BDA risk management is the proposition of a methodology based on correlation networks that can measure the contagion risk of a borrower, in terms of its network centrality, and add such risk to a classical default scoring model. The main contribution of the project for AI risk management is the development of a methodology based on clustered Shapley values which, applied to the predictions that derive from a complex machine learning model, make it explainable, in terms of the features that mostly determine each single prediction. The main contribution of the project for blockchain risk management is the proposition of a methodology based on natural language processing, sentiment analysis and regression models, aimed at understanding from web based information which coin offerings are likely to be fraudolent and, in a similar manner, which types of cyber attacks are most likely.
- development of a Fin-Tech risk management platform uniform across Europe with a potential improvement in risk mitigation;
- resulting balance between risk and opportunities which makes Fin-Tech innovation socially sustainable.
Both expected results have been achieved, producing 40 research papers on fintech risk measurement, selecting them in 12 use cases that are ready for standardisation, and disseminating the good practices in fintech risk management resulting from the proposed use cases in 87 SupTech, 30 Spinoff workshops, 6 RegTech workshop,s and 6 research conferences.
The main contribution of the project for BDA risk management is the proposition of a methodology based on correlation networks that can measure the contagion risk of a borrower, in terms of its network centrality, and add such risk to a classical default scoring model. The main contribution of the project for AI risk management is the development of a methodology based on clustered Shapley values which, applied to the predictions that derive from a complex machine learning model, make it explainable, in terms of the features that mostly determine each single prediction. The main contribution of the project for blockchain risk management is the proposition of a methodology based on natural language processing, sentiment analysis and regression models, aimed at understanding from web based information which coin offerings are likely to be fraudolent and, in a similar manner, which types of cyber attacks are most likely.