The last weeks have seen a couple of exciting workshops and conference at the Florence School of Banking and Finance.
a workshop discussed the role of artificial intelligence (AI) and machine learning (ML) in banking and finance and brought together academics, supervisors and consultants in this field. My colleagues Maria del Carmen has written up a summary
of this discussion on the FBF blog. Here are some additional insights that I took away: when it comes to the use of AI and ML in practice, a robust data governance framework is needed. There seems a large gap between technical capacity and ‘explainability’
of the techniques, which has to be bridged and which requires humans in the loop. Put differently, it is not sufficient to simply use these techniques, but a degree of ‘explainability’ is called for (including towards senior management and board
members). There is also a critical need to sensitise senior management to digital approaches (both benefits but also risks). Finally, one important lesson for supervisors is to take a critical view vis-à-vis the use of AI and ML in the institutions
they regulate, while at the same time they increasingly are using such techniques themselves (Suptech and Regtech). Certainly a tension that has to be addressed carefully.
Second, our first Annual Conference in three years brought together practitioners, academics and policymakers to discuss topics in digital and green finance under the heading “The Future of Finance
– Finance for the Future”. In her keynote and to set the stage, Anneli Tuomen (former Director General of the Finnish FSA and now ECB representative on the Supervisory Board of the ECB) pointed to the risks for European banks stemming
from the Russian aggression and rising energy prices, but also the much stronger capital buffers that European banks have available compared to a decade ago. Further, European banks have come stronger out of the pandemic than initially feared.
She also pointed to new sources of risks, including cyber-risk and increasing dependence of banks on third-party service providers (e.g., cloud services). The increasing digitalisation in finance was an important topic at the conference, including the increasing
role of BigTech companies and new legislative and regulatory initiatives on the European level, including the new crowdfunding regulation. What are the implications for competition from an increasing role of BigTech companies? Discussions on digital
currencies showed a clear contrast between regulators and practitioners who focus on potential uses and technical details of such currencies and academics who see little if any value added from digital currencies on the retail level (to be paraphrased as:
‘what problem is the digital euro the answer to?’). The transition to a sustainable financial system was a second major theme. To which extent can we rely on the financial system to undertake the necessary shifts in asset allocation and to
which extent are regulatory tools needed, including stress tests or even adjustments in capital requirements?
Third, we had a fascinating Women in Finance workshop.
Reuter journalists Brenna Hughes Neghaiwi and Tom Sims presented their recent research on the limited ascent of women in Europe’s banking system. One important
reasons for this can be summarised as an implicit bias in recruitment: “Tim goes away, so who looks like Tim?... It's usually not a Sarah.” However, this bias and limited diversity comes at a cost, not only in the financial system, but also,
as EUI’s Costanza Hermanin pointed out, in the public sector and political sphere. One important question is how to bring more women into leading roles; mentors seem to play an important. EUI’s
Johanna Gautier discussed the strong of women in international financial institutions, with Ann Krueger having been chief economists of the World Bank back in the 1980s and later deputy managing director at the IMF. IMF, World Bank, EBRD and OECD all have
had female chief economists. One explanation for the stronger role for women in these institutions might be that they require higher cultural sensitivities and have thus somewhat less space for the biases described above.