With the gradual end of state aid in the context of the covid-19 pandemic, inflation-stagflation, and the deterioration of the economy due to Russia's invasion of Ukraine, “the pressure will come from all sides, from shareholders, regulators, and salespeople,” predicts Matt Moran, partner and deputy advisory lead, M&A head of insurance at PwC Luxembourg.
The company invited the London-based fintech Galytix to set foot in Luxembourg and take advantage of an ecosystem with very short decision-making circuits. Last week, the startup published a 48-page white paper entitled “Banks Must Act on their Early Warning Systems or Risk ROE Downturn”.
“The coming crisis will not be a liquidity crisis but a credit crisis,” said Galytix CEO Raj Abrol, speaking at the PwC HQ Crystal Park during his visit to Luxembourg. “In the UK, more than 5,000 business failures have been recorded, the highest level since 1960. The European Central Bank has asked the banks to prepare themselves.”
The Lego method
For Moran and Abrol, banks must move on from their current risk assessment systems, which are too often based on internal bank data, with indicators that are too late, have too many false positives, not enough external data and are too often in silos. The study cites the cases of Thomas Cook, NMC Healthcare, Wirecard or Hertz to illustrate the damage of a too slow and archaic system.
“Risk professionals spend 60% of their time finding relevant data”, because efforts to find the data are conducted in different places and 95% of the data is in documents that have to be manually extracted after they are found.
The report is an opportunity for Galytix to capitalise on the development of its product based on the data of a large enough financial institution to be relevant.
The Lego method stands for “leverage” (globally, the elements of a company’s financial balance sheet), “external indicators” (market prices, forecasts, ratings, financial analysts, projections by sector), “governance” (indicators linked to governance and legal risks) and “ontology” (the more technological part or the integration of 10,000 sources of data specific to the company, to an economy in which it is developing, to its sector of activity, etc).
The "O" translates into Galytix’s approach--a data-driven loop architecture, which goes from information curation to feedback for the risk analyst at the bank (but the model will be just as useful to an insurer) in nine steps that allow to qualify the data and build general IT models with more granularity.
A DataFactory in Luxembourg?
The paper mentions the case of a bank that implemented Lego with 240 sources on 12 signals and gained 40% predictive insights on financial data, real-time analysis of regulatory impacts on the company’s business, supervision of 300 subsidiaries and 100% of management changes. The result is a kind of risk report card in three colours: green when everything is going well, orange when an indicator needs to be monitored and red when a risk element is detected.
“It is impossible for a bank to identify all its risky customers before they default,” the report concludes. “However, it is possible for banks to establish a prudent system and processes to identify and monitor a significantly higher proportion of accounts with default potential. Establishing a state-of-the-art EWS is a journey that requires decision making at both strategic and tactical levels. The next few years are crucial for any bank that aspires to be on the right side of the credit cycle.”
Having recruited Rupak Ghose, ex-Credit Suisse, ICAP and Financial Markets Standards Board as chief operating officer last August, Galytix is set to launch its GX Algorithmic DataFactory, the nuclear heart of its risk analytics products. At least one of these DataFactories is expected to be launched in Luxembourg, Abrol says.
This story was first published in French on . It has been translated and edited for Delano.