Risk Technology Awards 2023: Bank runs become sprints

When Stas Melnikov left his job as global head of investment risk at Russell Investments in 2019 to head a team focused on asset liability management at SAS, there was much head-scratching among many of his peers.

“Why are you doing this?” asked some senior colleagues, puzzled about why he’d want to work in what was a pretty mature, slow-moving discipline – after all, banks have been trying to marry up the two sides of their balance sheet for as long as classical banking has existed.

On top of that, for much of the preceding decade, the interest rate risks that could have put bank ALM under pressure were in remission. And while liquidity and funding was part of a multi-pronged regulatory push, it was one part of many – bank leadership was arguably more focused on capital and leverage rules.

For banks, ALM is the topic in focus and I don’t anticipate it going away

Stas Melnikov, SAS

“It was boring. Nothing was happening,” says Melnikov of that period. “But times have changed.”

The abrupt collapse of Silicon Valley Bank in March – swiftly followed by the closure of Signature Bank, the forced marriage of Credit Suisse, and the acquisition of First Republic – has thrust ALM onto the front page, and into the boardroom. Bank management and regulators alike are now asking pointed, urgent questions about the adequacy of existing models and looking for new ways to measure the resilience of financial institutions.

“For banks, it is the topic in focus and I don't anticipate it going away,” says Melnikov, who is now head of risk portfolio at SAS.

The subject’s hot-potato status was reflected in this year’s Risk Technology Awards, where the bank ALM category was particularly hard fought. The full roll of honour, along with the awards methodology and this year’s roster of judges, can be found below.

In this article, shortlisted firms describe the modelling questions unleashed by SVB’s implosion, hurried attempts to overhaul outdated practice at smaller banks, and a near-future in which ALM breaks out of its traditional silo to achieve a new status at bigger banks.

A larger world

Alina Preger is a senior partner at Prometeia, the overall winner of the bank ALM award. She has worked in the sector for more than 20 years. As inflation surged in the US and Europe, she had been expecting banks to refocus on ALM. Higher interest rates are generally seen as beneficial for lenders, pushing up the income from their loan portfolios, but also potentially prompting depositors to shop around and forcing banks to offer better rates to keep them.

Lurking beneath these day-to-day profitability questions are the big bond portfolios that banks hold as part of regulators’ attempts to tackle liquidity risk. The idea was that the bonds can be sold to cover a spike in depositor withdrawals, but this is a double-edged sword – buffers stuffed full of fixed-coupon debt would also lose value as interest rates rose. Most banks would be expected to have hedges of some kind in place, but there was clearly the potential for things to go wrong, given the size of these exposures and the pace of central bank hikes.

Some kind of stress could have been forecast, then. The surprise, says Preger, was that “everything happened so fast”.

After suffering $42 billion of deposit outflows in the space of 24 hours – and receiving requests for a further $100 billion before regulators intervened – SVB became the poster-child for asset-and-liability mismanagement. As is now well-known, the bank ramped up its holdings of long-dated debt and ran down its hedges in an attempt to capture more yield, then announced it was raising capital to cover the losses on those bonds, and spooked its highly concentrated depositor base, turning a bank run into a bank sprint.

“It was rather surprising, and it poses questions for some banks on the way they measure their risk exposure and calibrate the models,” says Preger.

While most banks have not experienced anything like the stress SVB was under, many have still been wrongfooted by customers during the rate hiking cycle, some of whom have not behaved as deposit models predict. For instance, many banks expected retail customers would rapidly shift their savings from low-interest current accounts to higher-rate savings accounts as rates rose, but in many countries that has not happened.

“Clients are now saying their analysis needs to explore a wider set of data to support them in evolving their models in a different direction,” says Preger, adding that Prometeia is doing a lot of work on modelling “non-linear responses” of customers when interest rates change.

That could mean tapping new sources of data that banks already hold, such as transaction data, and feeding this information into machine learning tools to gain new insights on customer behaviour that can be used to refine the deposit models.

“Now we can analyse all the transactions a customer has with a bank and use machine learning and all these advanced analytics, so we can actually explore a larger world in this way,” she says.

Up the food chain

As well as revisiting their models, banks are using ALM systems in new ways – there is wider demand for data and analysis, and the results are being scrutinised at a higher level within each organisation.

Nigel Lee, chief executive of Finland-based MORS, says existing customers have been seeking to increase user numbers, allowing more staff to access ALM systems. He has also had more conversations with chief financial officers, who have traditionally kept clear of discussions around ALM analytics even though treasury departments typically fall within their purview.

“What it tells me is that the information the treasury and ALM teams are providing is going very high in the food chain internally in the bank. It's clearly a boardroom discussion,” says Lee.

As banks have started to prioritise ALM, potential clients have been worrying less about the cost of new ALM software and focusing instead on functionality and how quickly it can be rolled out.

“Budget, of course, is always on the agenda, but it is probably secondary now to concerns about time and trust,” says Lee. “Customers were telling us they needed something in place by the midpoint of next year. Now, they are telling us this needs to be in place by the end of 2023, at the latest.”

Smaller banks in particular have a lot of catching up to do, with many still relying on outdated systems, and running relatively rudimentary calculations. Tech vendors see a gap in the market and are trying to sell their software systems to these smaller institutions, which include digital and community banks as well as credit unions, particularly in the US.

I’ve been joking with people, saying this [stress period] has been great, because people are focused on financial risk again, as opposed to cyber risk and operational risk taking all the headlines

Andrew Aziz, SS&C Algorithmics

Mohamed Mehdi Bouasria, director of product management at software company Finastra, says it is “mind blowing” that some smaller banks are still using Excel to run their ALM calculations on a quarterly basis.

Two years ago, the London-based company decided to create a simplified version of its ALM tool on a software-as-a-service (Saas) basis that would allow smaller institutions to run more sophisticated calculations more frequently than the outmoded quarterly run. ALM IQ launched two weeks before the collapse of SVB. Bouasria claims the product can be implemented by a new customer in less than a month.

“The prospects and number of requests we are getting for demos is just massive,” says Bouasria, who relocated from Europe to New York last year to focus on the product. “We felt that banks, especially on the lower-tier end, are very stressed. They are looking at the regulation like a constraint, and the first thing I’m telling these banks is that ALM is an opportunity.”

Bouasria argues banks that have a good understanding of their assets and liabilities can better decide what products to push and which clients to focus on. He is particularly excited that the Saas version of the product allows smaller banks to run ALM calculations on a much more regular basis.

“It's really on-demand, and we really want to push those banks to adopt the approach of having at least a monthly run with weekly monitoring, which is what experts in the field recommend,” he says.

SS&C Algorithmics is chasing the same opportunity. Andrew Aziz, chief strategy officer and head of product, says he is seeing more interest from smaller institutions that “want to do something more sophisticated” with ALM to better understand their businesses.

“We're seeing more demand for a cloud-enabled solution, a managed service solution, where the overall cost of ownership is lower,” he says.

Hidden links

While smaller banks are increasingly interested in purchasing ALM software for the first time, larger institutions that have had access to analytics for many years are starting to rethink what they do with the output.

“I’ve been joking with people, saying this [stress period] has been great, because people are focused on financial risk again, as opposed to cyber risk and operational risk taking all the headlines,” says SS&C’s Aziz. “It's taken everyone back to the fundamental question of how to manage the entire balance sheet and really assess the interplay between risks.”

He says big banks are now thinking more broadly about how they gather and stitch together various forms of risk data, which might traditionally be analysed in isolation.

“Maybe it’s time to make an investment in something that can actually bring the whole picture together,” he says.

That could mean looking at the linkages between ALM, climate risk and credit risk, in an attempt to understand a bank’s overall balance sheet risk.

At SAS, Melnikov agrees that larger banks are increasingly looking at ways to make ALM data more useful for senior managers by integrating it with other financial information, such as cashflow generation, credit risk and market valuations for derivatives.

SAS acquired data and software company Kamakura in 2022 to bolster its efforts in this area. Melnikov says the acquisition and integration of new types of data is helping the company to “build out all of these very sophisticated simulation models that were not really relevant for the past 10 years”.

The idea is to turn ALM from an issue of basic hygiene or compliance, into crucial data that senior leaders use to steer bank strategy.

“The time is right for this technology,” he says. “ALM as a silo is not going to last.”