Monday, October 19, 2009

Whose Ox is Gored?

In July I wrote two posts on the “inconvenient essential” of invoice verification. (Please see the posts dated July 5 and July 8.)

I acknowledged at that time that many elements of a typical invoice verification process were sacrificed by the TRE model but I also pointed out that there were mitigating factors that, in my view, balanced the risks.

TRE has recently changed some of its invoice verification procedures, reducing the extent of the actual verification process and substituting, in some cases, a statistical sampling approach.

To its credit, TRE provided a comment period for all Members to give their opinions of these changes prior to implementing them. My own opinion was, and is, that the changes served only the purposes of the exchange itself and of Sellers in a hurry to get invoices to market. There is no credible argument, in my view, for a benefit being afforded the Buyers.

I quote below some passages from my comment letter:

“I understand that continuing to verify each invoice in cases where there might have been dozens if not hundreds of invoices paid without incident in a given Seller/ Account Debtor relationship might seem to be a waste of resources. I can also see that some Account Debtors might begin to complain when the number of verification requests begins to get very large.

I can also understand that, absent any problem, a substantial history of successful transactions between a Seller and an Account Debtor makes the probability of the nth invoice of the same type being problematic quite small. In such cases a sampling system can certainly make sense.


 I think it’s a little disingenuous to present this change of process as an “enhancement”…

 It just can’t be argued that this is not a further weakening of the verification process.

 What began as a third-party function, providing some level of independence in the process, morphed quickly into an internal TRE function of limited scope. Even though the scope was limited, it still applied to all invoices. Now the scope remains limited and the application will not be to all invoices.

 I can see that it will help Sellers get deals to auction more quickly and I can see that it will reduce the TRE workload and expense. The only group that I can’t find a benefit for is the Buyers!

 I’m certainly mindful of the need to ramp up volume and to bring on new Sellers and...I think there are situations in which some “sampling” activities might be justified. But based only on what I’ve seen, this makes me nervous...because of what it does to reduce the strength of the process...”

My comments also included reference to some details of the TRE process that Members are required to keep confidential but I think my concern about the changes comes through in the sections I have quoted above.

The Exchange, again to its credit, did follow up with a conversation addressing my concerns. I expect that all who submitted comments were given the same courtesy.

There is no indication that anyone’s concerns had any impact on the procedural changes, however.

I think it is important to point out a fundamental disconnect between the apparent analytical approach of the exchange and that of an individual Buyer.

TRE seems to want to analyze risk at the level of the entire portfolio of transactions that occur on the exchange. That, in my view, is not correct. TRE itself takes no transaction risk. There is no exchange-wide exposure from which a risk assessment at that level can validly be made.

Without the ability to buy what would amount to an “index” position in the entire TRE portfolio, a valid risk assessment can only be made from the point of view of the portfolio of an individual Buyer.

And Buyers will now only be sure that their invoices have been verified under certain defined conditions. Under any other conditions they cannot be certain.

To be sure, Buyers should be able to gain a sense of the probability that their invoices have been verified, but as long as the probability is less than 100%, the uncertainty should be reflected in higher required returns.

So the Sellers should ultimately wind up paying for their increased transaction velocity via higher returns to compensate the Buyers' increased risk.

And, in the end, only the exchange itself really appears to benefit.


  1. Invoice accuracy has long been the step-daughter never seen in the light of day. The typical trade creditor has errors on roughly 8% to 23% of the invoices they generate, most of which come back to haunt them as payment deductions. As a result, credit departments also spend more time reconciling deductions and disputes than calling on truely past due accounts. Moreover, 90% of deductions result in a credit memo or other adjustment in favor of the customer. By the same token, invoices that are clean are generally paid within a week of terms, while those with errors are paid on average 30 to 60 days slow. This is because inaccurate invoices must be handled as exception in AP departments. In short, they create more work for both suppliers and their customers, and they clog up the settlement process.

    As regards The Receivables Exchange, most invoice errors will reveal themselves as payment deductions and be covered by the difference between the advance amount and the invoice total, but they will also tend to cause payments to be past due. Therefore, there is a cost to the supplier/seller for invoice errors. Also, it takes less time to correct these errors before the invoice is sent than post-payment, so there should be an incentive to the suppliers to address invoice errors in their billing processes. There is even software that provides automated checks before invoices are generated (much like the AP 3-way match), and anecdotal evidence indicates such systems facilitate significant reductions in DSO. Even so, we see little movement by suppliers/sellers in this regard. It will therefore be interesting to see if participation in The Receivables Exchange provides an added incentive for suppliers/sellers to clean up their act/invoices.

  2. Hi Dave,

    Thanks very much for the thoughtful comment.

    The typical holdback on an invoice purchase does provide some margin for later-than-expected payment. It probably also provides room for modest-sized errors.

    Significant errors in billing or outright refusal to pay, on whatever grounds, would not be covered by the holdback, though.

    It is that risk i.e. that the TRE verification process does not include obtaining from the Account Debtor a confirmation that the goods/services actually meet the requirements of the contract and that payment is actually confirmed as due, that is the source of the primary systemic risk.

    This current change of procedure just amplifies the magnitude of that risk.

  3. This issue brings to light even more the value of invoices that are uploaded to the TRE platform directly from a supplier network such as Ariba. In the case of an Ariba supplier selling an invoice processed via the Ariba Network, both the issue brouhgt up in the blog (verification) and the issue in the comment (quality of invoices/errors) are rendered moot.

    1) Verification: When the next release of the Ariba Network (AN) comes out (4Q09), Ariba suppliers will be able to click a button and directly upload their approved invoices for sale to the TRE platform. Approved invoices are delivered directly to the AN from the Obligor's ERP, and are then transmitted directly to the TRE platform with no opportunity for the supplier to change any of the data contained within. Given the unbroken electronic chain of data, the need for verification is obviated.

    2) Quality of Invoices & subsequent error correction: The nature of the Ariba Network is such that buyers/obligors set parameters up front that suppliers must meet in order for a submitted invoice to be considered in good order. This systematic quality control ensures that the vast majority of quality issues are filtered out before an invoice is even received by the Buyer/Obligor. So the issues brought up by Mr. Schmidt are largely removed for Ariba invoices.

    Combine the above with the fact that eInvoicing via Ariba reduces invoice approval times down to an average of less than 5 days, and invoices sold by Ariba suppliers directly from the Ariba Network will offer TRE Buyers a qualitatively better option to reduce risk and extend returns.

    I would welcome your thoughts on this.