After news broke about Citibank’s loan processing error, back in August 2020, it wasn’t long before a WhatsApp group of mine started to ping.

We’d been through months of lockdown and these messages from ex-colleagues from my lending ops days were a welcome excuse to catch up – and share stories of our own heinous balls-ups. 

We experienced the stress of keeping the boss in the dark until cash was recovered. Margin ratchets were corrected, missing trades found, followed by a sheepish but hurried call to Hippopotamus Capital CLO (or something) who would hopefully return the funds before cut offs and overnight recs outed the ‘oopsie’. 

My own oopsie saw me nervously pushing a negative few million of Swedish Kroner around a nostro for three days before ‘Hippo Dave’ agreed my calcs and wired the cash back. A couple of sleepless nights was the price I paid. Minor, in comparison to the $400m fine levied against Citibank by the OCC following their incident.

My first job in loan ops was in 1998. We’d track loan lifecycles on handwritten ledgers, physically collect signed drawdown docs from neighbouring borrowers, and seal everything in waterproof and fireproof cabinets before leaving for the day. With that level of manual processing, operational errors were all but priced into the deals. 

Mistakes happened because we were busy, especially at month or quarter end, and in my case, largely because I was 19 years old!

Roll on almost 25 years to Citibank’s blunder. With an operating model which mirrors that of most Tier 1 banks today, it includes an outsourced, offshore processing team for their ‘maker > checker’ inputs (four eye check) and, for sums of the value in question, an onshore, bank-side team for a last six eye check.

Someone had initiated a repayment instead of a rollover and had sent lenders the full loan principal instead of just the due interest.

Four eye check missed it. Six eye check missed it.

I sympathise and shudder when I put myself in their shoes. The dawning realization that you’ve just pushed a sum greater than a small nation’s national debt through your loan booking system and out the door in error… I’m certain large holes for crawling into were wished for that day.

Do you know what Citibank’s lawyer said as the incident explanation to the appeals court?

“This was a mistake. Humans make mistakes.”

So whether a bank’s underlying system has a process with four, six – or however many – eyes in place, it will do nothing to mitigate the impact of an error of this incident’s scale. As Citibank’s lawyer very plainly explained, if we allow humans to be the final safety net, mistakes will be made.

Commercial banks must ask themselves, why, with the sheer amount of innovation and technology available on the market, do we still allow errors like this to occur?

An agent like Citibank needs to manage hundreds, sometimes thousands of lenders on a single credit line on behalf of their customers and the tools they use are creaking under the pressure. Whether it’s an in-house solution built on an ageing tech stack or a third party monolith that’s been around longer than I have – it’s simply not good enough. 

Look what technology has done for the retail sector in the last decade. OCR for cheque clearing, facial recognition for account opening and verification, AI-powered algorithms to authenticate customers’ identities, standardised API’s for open banking. The list goes on.

Why has commercial banking, especially lending, not taken advantage of new technologies at the scale it needs? Yes, yes I know global corporations are far more complex animals than us simple retail customers, and I know bank systems are fragmented by the very nature of the complex products and data those customers need. But how much longer are we able to justify the service a commercial lender can provide to its customers?

Did you know that corporate customers of the biggest banks are employing teams to second check the data sent to them by their banks? They maintain their own lending data because they can’t be sure their banks are accurate.

Again, I don’t think that’s good enough. 

What do you think? Will we be reading about these kinds of colossal ‘oopsie’s in 2045?

Or can we work to adopt new thinking and next generation technology to spare ourselves another sleepless night?

Written by James Mellor, Oneiro Solutions Sales Engineer