For companies, the receivables side has always been a bit of a problem even as new artificial intelligence products and other technologies are making a difference in payables .Processing receivables “still involves a lot of manual work,” says Victor Penna, Head of Regional Solutions & Advisory Team, Asia Pacific, at J.P.Morgan Treasury Services.
Transaction Banking: Helping to Solve the Receivables Blues
Banks like J.P.Morgan are turning their attention to the problem. “For example, we’ve been building technology that enables us to help automate the reconciliation process,” he explains. “Rather than have 30 people sitting there manually matching outstanding invoices to collections, we can now automate a big part of that process.”
In Part 2 of an interview he did with CFO Innovation’s Cesar Bacani, Penna discussed trends in bank automation products for receivables, payments and other processes. Excerpts:
Can the client company outsource cash management to the bank, so institutions like yours would pay the payables, collect the receivables and so on?
Most of the leading banks like us tend to provide the gateway for these services already. We’ll provide the host-to-host link so that the [customer’s] shared service centre can send one payment file to us covering multiple countries and payment types, and we’ll then break it up and send the payments out in each market.
Things like cheque outsourcing are now quite common. Rather than a shared service centre printing and mailing its own cheques, it would include all those cheque payments in a single file that it sends to J.P. Morgan. We would then route those files to each country, so we print a ringgit check in Malaysia and mail it out for the company; a rupiah cheque in Indonesia; an Australian dollar cheque in Australia and so on. The same with electronic payments.
Where we’re seeing more activity is on the receivables side, which still involves a lot of manual work. For example, we’ve been building technology that enables us to help automate the reconciliation process. Rather than have 30 people sitting there manually matching outstanding invoices to collections, we can now automate a big part of that process.
In a sense, the company is outsourcing part of that activity to the bank, and the bank is giving the client the technology to do that more efficiently. With some of the new technology we’re building, the payer doesn’t even need to provide a reference number anymore.
Really? How does that work?
We’re doing a lot of work around what we call ‘virtual reference’ solutions. Normally, if you were a customer, I’d send you an invoice and I’d ask you to pay to a bank account. I’d send an invoice to another customer, and he’d need to pay to the same bank account, and so on. Then I get all these credits into the bank account, and have to review each one individually to work out which payment belongs to which outstanding invoice.
So what we do now is automatically assign a reference number to each payment received. This reference number corresponds to specific customers within the company’s accounts receivable ledger. Bank account statements containing each payment received together with the reference number can now be uploaded into the company’s ERP system enabling automatic posting to the right customer without the need for manual reconciliation.
You can have as many accounts as you have customers?
We’re working with one client at the moment which has 700,000 reference numbers. Theoretically you could have millions. Unless the customer sends the payment to the wrong bank account, it provides 100% identification in terms of which customer made which payments
This is only for electronic payments? If I write a physical cheque . . .
It’s for electronic, and the reason that it is geared towards electronic is that electronic payments are actually harder to identify. When you receive a cheque, you usually know exactly where it came from. The problem is when you get an electronic credit in your account that contains no reference information.
Depending on the clearing system, and whether identifying information has been input into the reference field, you may not have anything to go by. You might have sent out many invoices for US$1,000 and you’ve got no idea who that particular thousand dollar receipt belongs to.
We have also developed technology that allows you to further automate the reconciliation process. This system can take a list of outstanding invoices from an ERP system like SAP for example, and then identify the most logical combination of outstanding invoices that match the amount received.
For banks, one problem with payables is that a large customer can send tens of thousands of instructions to process payments, but the paperwork often is not complete, has data in the wrong fields and so on.
We actually have one of the most sophisticated artificial intelligence systems in our bank. We auto-repair lots of transactions everyday.
The vast majority of the payments are correctly formatted, but there are instances where people input the transaction incorrectly, wrongly transpose a number in the account field, they mix up the beneficiary and account number or they put the wrong branch address.
Our engine can automatically repair those payments. It’s pretty clever; it can use different algorithms to identify where the problems are, and automatically repair the payments, so that defective instructions need not go back to our customer once verified.
But sometimes you do have to return paperwork to the client?
Of course. Sometimes you may not be able to pick out whether you should repair it using Option A or Option B. So we go back to the client and ask: Which way would you like to do it, this way or that way? Once they say ‘this way,’ that would be loaded into our system. Every time we get that same incorrect instruction, we’ll repair it the same way.
In the old days, those transactions would always have to go back, and that required a lot of work. Through the use of these systems, we are increasing the straight through processing rate. For some clients, this means a straight through processing rate of 98-99% versus 70% or less in the old days.
So it doesn’t matter whether the formatting one client or bank uses in one market is different from that used by another in a different market?
You can’t fix everything. It really comes down to what the specific problem is. If it’s a repetitive problem, then often you can fix that problem.
Click here to read Part 1 of this interview