Central banks and supervisory authorities will need to evolve from having dedicated innovation ‘labs’ or ‘hubs’ to embracing innovative technology and thinking across all their departments, the chief data officer of Australia’s central bank has urged at a conference.
Roman Kovalenko of the Reserve Bank of Australia issued the call during a panel discussion at the ‘Supervision Innovators Conference 2023’, where he was speaking alongside senior representatives from the Banque de France, European Banking Authority and UK’s Financial Conduct Authority (FCA).
The quartet were tackling the topic of ‘How to keep SupTech future-proof’ (where ‘SupTech’ is short for supervisory technology) at the European Central Bank (ECB)-hosted event in Frankfurt, with the discussion spanning topics including organisational structures and skills, as well as the possibilities and challenges created by hot-topic areas such as artificial intelligence (AI).
“A lot of organisations have stood up innovation ‘labs’ and ‘hubs’ and ‘centres of excellence’ – that’s a great first step,” said Kovalenko in his opening remarks (the RBA is among the central banks worldwide with an innovation lab). “But what ‘good’ looks like in the long term is [that] the entire organisation needs to be an innovation lab or hub – that’s the day and age that we live in right now, that’s the only way to move quickly and scale.”
“To get there requires basically an enterprise transformation,” he told the online and in-person audience in the German city on 20 September. “You need to change technology, process and people capabilities.”
RELATED ARTICLE SupTech on the rise: financial supervisors explore innovative technology – write-up of SupTech session at the Global Government Fintech Lab 2023 (the session featured representatives from Norway, Croatia and Bermuda describing how their authorities are using supervisory technology and barriers to greater uptake)
Supervisors and ‘strategic risk-taking’
“The organisation needs to get really good at getting in the door new tools and harnessing them really quickly – something not easy for central banks and supervisory organisations,” Kovalenko said. “Organisations need to get really good at processes like ‘agile’ and ‘design thinking’ and strategically taking risks and failing sometimes in a safe way.”
“‘People capability’ transformation is never easy although we have good people in our organisations,” he continued. “Hopefully, it’s relatively easier than some other organisations,” he continued, saying that “lift[ing] these really big rocks around technology processes and people… takes years, and a lot of investment and a lot of repetition.”
He acknowledged that people can often be sceptical or resistant about adopting and adapting to things with which they are unfamiliar. “Over the course of years, it’s very easy to lose people, and people are going to lose patience, [asking]: ‘what are you doing, you’re building these systems, and you’re talking this nonsense stuff, where are the outcomes?’”.
“It’s a combination of ‘top down’ – driving large transformation, big strategic investments that take a real long time to pay off – but also a bit of ‘bottom-up’ where you try to tie to immediate business outcomes as soon as possible and work with staff at the grassroots level,” he said.
“I think those are those are the challenges – ‘lifting the big rocks’ and winning the hearts and minds in the short term, so you actually get a chance to drive the long-term journey,” he said.
RELATED ARTICLE Sandboxes could ‘amplify problems’: IMF analysis questions many test-spaces’ impact – a news story (11 September 2023) on a 58-page paper, ‘Institutional Arrangements for Fintech Regulation: Supervisory Monitoring’, assessing how supervisory authorities have set up sandboxes and innovation hubs, and their impact
Banque de France first deputy governor Denis Beau opened his remarks by stating that the “digitalisation of finance makes it very urgent and possible to develop an enhanced approach to financial supervision”.
He said that the central bank and Autorité de contrôle prudentiel et de resolution (ACPR), the European Union (EU) member state’s banking and insurance supervisor, takes a “proactive, demand-driven approach to innovation”, which involved the launch of a SupTech programme in 2019.
“We [set] the tone from the top,” he said, saying that this was part of a strategy to overcome internal resistance – which he referred to as “inevitable and understandable” – to the “transformation of established, well-functioning business processes.” To ensure sustained focus on the topic the SupTech programme is one of 30 priorities in the Banque de France’s strategic plan (which was published in 2021 and will run to 2024).
Beau, who is also a member of the ECB’s Supervisory Board, said ‘demand-driven’ innovation focused on addressing “day-to-day operational issues faced by ACPR staff in carrying out their duties”.
This approach, he said, aimed to ensure what he described as the “classic” (general) problem of striving for ‘solutions in search of a problem’ and ending up with those solutions “ending on the shelves” (gathering dust).
RELATED ARTICLE Sink or swim: financial authorities dive into data –write-up of Global Government Fintech Lab 2023’s ‘How can financial authorities capitalise on data to improve what they do?’ session, which featured: Gavin Curran, head of the Central Bank of Ireland’s markets supervision division; Mari-Liis Kukk, head of the innovation department in Estonia’s Financial Supervision and Resolution Authority; and Fintech Ireland founder Peter Oakes
‘Contant need to revise approach’
Marilin Pikaro, director of innovation at the European Banking Authority (EBA), kicked off by describing the Paris-headquartered authority – “historically a policy development organisation” – as “entering a new phase” as it takes on new oversight responsibilities through the EU’s high-profile Digital Operational Resilience Act (DORA) and Markets in Crypto-Assets (MiCA) pieces of legislation.
Pikaro’s overarching point was that supervisory authorities need to become “significantly faster” in the way they adapt to the challenges and opportunities brought about by technological innovation.
“Do we have years to develop this new supervisory approach?,” she asked rhetorically. “It seems to me that the world is changing if not within days then within months. And we constantly need to revise the approach.”
“Not only is it in industry where those who are quicker win, but also supervisors: those who are faster to adopt their approach are tackling risks better,” Pikaro said, emphasising that a ‘risk-based approach’ by definition needs to be focused on risks that reflect current technologies, as opposed to risks when supervisory strategies were being drawn up.
What this means in practice, she said, was for supervisory approaches to be potentially revised at least annually and “maybe even twice per year”.
RELATED ARTICLE ECB looks to ‘Project Olympus’ to transform banking supervision using SupTech – a news story (20 September 2023) on a speech titled ‘The Impact of SupTech on European Banking Supervision’ delivered by ECB Supervisory Board member Elizabeth McCaul at last year’s ECB-hosted ‘Technology and Banking Supervision Connected’ conference
Short- and long-term approaches
On the topic of internal change, Kovalenko (pictured) suggested that, over the long term, the route to try to ensure everyone in an organisation is engaged is “to change mindsets through training and awareness and examples.”
But in the short term, he suggested that one approach could be for leaders to “identify and empower… individual champions in the organisation who are a bit ‘ahead of their time’… so they’re not just some maverick, unorthodox person… but you try to rally them and you give them a bit of ‘air cover’.” After empowering ‘early adopters’, organisations can then, he said, promote successes and ‘build communities’ on the way to ‘formalising structures’.
“It’s also very important that the experts who are driving the innovation capability… speak the ‘business’ language,” he continued. In this way, he said, organisations should be able to avoid “situations where you’ve got an innovation expert or a data expert coming over and saying, ‘hey, do you want to do some innovation?’ and the other person saying ‘well no, not really’.”
As a further proposal, he suggested rotating staff between positions as a way to “cross-pollinate” innovating thinking. “We’ve had some success there – we really want to pursue it more,” he said.
Picking up on these observations about driving internal change, Beau said that “the easiest thing is to find early adopters in an organisation” and that “the most difficult thing is to spread [innovative ways] across the organisation.”
AI in the spotlight
AI’s growing prominence was among the main non-structural topics discussed during the session.
Within the EU the European Commission proposed the 27-member bloc’s first regulatory framework for AI in April and, in June, the European Parliament adopted its negotiating position on the AI Act. Talks are now underway with nation states in the Council on the law’s final form, with the aim being to reach agreement by the end of this year.
“That would mean a timeline [of] an additional two years until it’s going to be applied,” said Pikaro (pictured), expressing concern that technology is developing so fast that it risks being “outdated by the time it’s applied.”
She highlighted an EBA report, published last month, on the use of machine learning for internal ratings-based models. This called for some clarifications in order to ‘reduce legal uncertainty and avoid unintended consequences’ of the EU AI Act.
Beau was confident that AI is ultimately an “opportunity, not a threat” but that “like many other new technologies” a regulatory framework was necessary to “address the risk without stifling innovation.”
“Regulation is good – important, needed – but it’s nothing without effective supervision,” Beau continued. “And when you when you speak about AI, in our corner, as supervisors, there are really major challenges to be addressed in terms of understanding the technology to be in a position to implement our duties correctly.”
RELATED ARTICLE Financial authorities’ SupTech take-up revealed in global survey – a news story (13 January 2023) on the ‘State of Suptech Report 2022’, compiled by the Cambridge Centre for Alternative Finance’s SupTech Lab (based on a survey of 134 financial authorities from 108 jurisdictions)
LLMs on the rise
Large-language models (LLMs) – a type of AI algorithm that uses deep-learning techniques is able to process and understand very large datasets to generate ‘human-like’ text – were also briefly discussed.
“We’ve got to think about the human and machine coming together,” said Helen Packard, data strategy officer at the UK’s Financial Conduct Authority (FCA). “It’s not about ‘cyborg supervision’, this is about coming together and being more insightful, more proactive, creating stronger financial stability – because together the machine is empowered by us.”
“The thing that I spend a lot of my time worrying about is how to empower all the colleagues with the right skills; how to take away the apprehensiveness of the new technology enhancements; how to think around the ethical use of the data,” Packard continued.
“I think where a lot of organisations are right now is there’s a lot of ‘grassroots’ adoption – you might have some IT people wanting to use language models to optimise some code and you might have some data scientists want to do something else,” said Kovalenko. “At a central level, we’re all trying to just take stock of that and make sure it’s safe, and nothing bad happens.”
“It’s a very short-term approach,” he continued. “But I think in the medium to long term, we need to strategically harness this as an organisation, so there needs to be vision and strategy: ‘how are we going to harness these tools across the organisation?’.”
RELATED ARTICLE ‘ECB completes €200m-plus procurement as it looks to “pioneer” SupTech innovation’ – a news story (29 April 2022) on the ECB completing its first SupTech-specific procurement
‘Huge amount we haven’t tapped yet’
Kovalenko also mentioned the ethics of AI, observing that the “great powers” enabled by the “very powerful” technology also imbue those using it with “great responsibilities”.
“I think this is an opportunity for supervisors to lead in the field of ethics around the application of AI,” he said. “So, not just collecting more data, faster [and] richer, but be at the front of ‘just because they can know something, should I know it?’.”
“AI, generally speaking, we should use it. Generative AI? The jury is out,” said Beau, going on to flag that, in May, the BdF had launched a ‘call for contributions’ on the uses and impacts of generative AI on its activities (available in EN).
“There is an abundance of rules and they are interpreted in many different ways by our firms,” said Packard. “If we can think about how we move to machine readable and executable readable, I think there are many insights that we can get from that from a supervisory [perspective] in terms of outliers, trends, proactiveness. There is a huge amount that we haven’t tapped yet. Machine readable is definitely a direction that we need, collectively, to go in.”
The session was moderated by Daniela Schackis, deputy director-general in the ECB’s Directorate-General SSM (Single Supervisory Mechanism) Governance & Operations.