The Skills Cliff, Part 3: Return of the Trade Finance Jedi

Welcome to Part 3 of the Trade Finance Skills Cliff series. If you haven’t read Part 1 and Part 2, here is the summary:

The incredible people who got into trade finance in the 1990s loved it and never moved on to other areas in the bank. Now that they are retiring or close to it, we find ourselves short of skilled operators, and the industry is facing real challenges filling jobs and modernizing LCs.

Part 2 ended as a bit of a Debbie Downer, like The Empire Strikes Back, so as promised, Part 3 is the Return of the Jedi. Here we look at the technologies coming online and at the Build, Borrow, Buy approach banks are implementing to build resilient trade finance operations for the next generation of business.

Doc Examination 2.0

Picture this: the date is July 2027, and you are at your trade ops desk, ready to review an LC presentation. Instead of a FedEx envelope arriving, a file shows up on your screen. On the left are the scanned or digital original documents that make up the presentation. On the right is a doc examination tool that reviews them and recommends how to handle the presentation.

You hit the “go” button, and after a moment, the results are on the screen. The LC calls for 25 documents. The tool reports 24 plus one it cannot identify, and links directly to the page so you can decide whether it is the missing 25th document or something else. You agree it is the 25th, the system logs your answer, and Step 1 is done: the presentation is as required.

Now the real examination begins, comparing goods descriptions, ownership, delivery, and shipping details. In the background, the system is pinging a vessel database for questionable history, checking OFAC, OFSI, and other sanctions lists for the vessel and the parties, checking that goods descriptions are consistent across the documents, allowing for variations in wording, and checking HS codes and goods keywords to see whether a license might be needed, all without an expert eyeballing every line.

Then the result comes back, and the system says that in its review, the details check out. It plays a bit like a video game, or like track changes in Word with comments: the document is on one side, the tool’s finding is on the other, and you confirm it or move on. You go through it one item at a time and decide whether, with your expertise, it is correct, knowing the rules engine has already put in front of you anything even remotely questionable. If the LC calls for a certificate of quality, and the tool shows it at the top, you confirm it and move on. That is what these tools are for. Then you approve everything, hit submit, and it moves to the next phase.

None of this is science fiction. Many of these elements are running in banks today, and the July 2027 date is really just a question of how fast the rest of the industry catches up.

Across many transactions, what used to take three to five hours now takes closer to two. At a financial crime conference in London in early October, someone asked how long it takes to become a pretty good document examiner. My own theory had been two to three years, but several people there were pushing three to four. A lot of that time went to moving things around, pushing paper, or scrolling through PDFs, which is not an effective way to train a team.

These tools can potentially double your doc examination capacity, going from two to four sets per day. That lets your bank scale its LC business, and it changes how long it takes to build an examiner: if getting pretty good used to take three to four years, doubling the reps brings that down to about 18 months to two years. And it is fun! You are using a computer, you are scrolling through, and it is aligned with how the next generation grew up spending their time. Newer staff examine documents over and over until it becomes second nature, and seasoned experts are brought in on the questions that need them, rather than reviewing everything.

A fair question, recently raised, is what happens to the learning that used to come from checking every document by hand. How much that matters depends on how much of the work the tool takes on, and the tools will keep evolving. The plumbing changes, but the judgment stays: a trainee still makes the call, a more senior person reviews it and makes the final call, and the trainee learns from the difference.

Because of the quality of the tools, the expertise that trained them, and the rules they follow (UCP, ISP, and so on), the floor of your processing ability is very high, and the ceiling can be too. You can be pretty good in under two years, not necessarily an expert, but way more than a generation ago. Examination where potential questions are presented for review, and not decided for you, changes the whole game: a small but mighty team can process a lot of documents, with staff time going where the expertise is needed.

Build, Borrow, Buy

As I see it, banks that want to remain players in trade finance have three options: build, borrow, or buy. Banks can and do choose more than one, and sometimes all three. The banks with a bigger book of business are more likely to “build” part of this in-house, but any bank can “borrow” people or ideas from others, such as part-time expertise, and “buy” access to the elements it needs. The right answer is whatever works best for your bank, but banks have a lot of options, and they need to be exploring them right away.

These tools make a build-and-borrow combination much easier. Picture a recent retiree who comes back a day or two a week and sits between ops and the first line. Junior staff start with the tool, make their calls, and bring in that part-time expert only when they are unsure. Because a set no longer takes three to five hours, the juniors get regular, repeated practice, and the expert’s time goes to the questions that matter most.

One way to look at it: build where you have the experience and the ability, buy where the problem is common to everyone, and borrow where the gap is judgment.

Standbys: Lifecycle, Not Paperwork

In the standby world, these tools are really just beginning to take off. Standbys are different from commercial LCs because they are generally not expected to be drawn on, and there are no shipping documents to present. But many of the same problems apply, and the dollar figures tend to have an extra zero at the end, and then some. The most recent DCW statistics show that banks in the US have issued about $345 billion in standbys against about $17 billion in commercial LCs, a 20-to-1 ratio.

Standbys are lifecycle heavy, not document heavy. They need to be written well from the beginning, based on good drafting, and good practice: are they subject to ISP98, are they using language from the ISP model forms, are there stray clauses that read more like contract language? Once the standby is in effect, the annual work begins: notices of non-extension, correct contact information, current payment instructions in case of a drawing, and signatures from all parties when a renewal needs changes.

Those of us old enough to remember tickler systems will recognize that a lot of these tools work like that once implemented. The clauses you need for a mining or data center standby, for example, are very different from those for aerospace or big agriculture. Some of the sponsors of this issue of DCW provide services in this space, so you can read more from them elsewhere in this issue. 

Banks are applying build, borrow, and buy here too, from services that help draft and review standbys and send the notices, to software that is platform agnostic for documents that need to be originals or digital originals.

The Third Track

Our industry has come a long way, and while it does, in some instances, seem like we have a long way to go, I am quite proud of where we are and where we can go.

The technology has been able to step in and improve upon manual, paper-based processes in a way that is secure, consistent, and ultimately scalable. MLETR-aligned laws now allow for electronic bills of lading, and fully digital trade transactions are happening today. Standbys are now drafted with AI tools, then reviewed and signed by humans. Some of them back the construction of AI data centers, which means AI is helping draft the paper that finances its own house. Financial crime checks are increasingly carried out in real time using some of the same technologies.

What makes this a good story is that expertise still has a role. Automation frees experienced practitioners to apply their judgment where it matters most, while a new generation comes in, learns the work and takes ownership of these essential financing tools.

At the same London conference, I was glad to see I was far from the youngest person in the room. I’m also looking forward to the New York Standby LC Forum in two weeks, where some of the people who were new to the field seven or eight years ago will now be speaking on panels as experts.

The trolley problem says pick a track. In trade finance, there is a third one, where nobody gets run over, the experts stay in the game, and the next generation drives. Full speed ahead.