art I of this article included this statement: "It is Visa's and MasterCard's level playing field of dollar-volume cost structure that allows for free market pricing that, in turn, gives the small players the opportunity to compete with the big players." The article also went on to discuss how this level playing field provided by Visa and MasterCard is now being challenged in court by First Data with its First Data Net service.

In Part II of this article, I will discuss the variable costs of transaction processing and also pass along some important news about First Data's countersuit against Visa.

What exactly does it cost to process a credit or debit card transaction? The answer depends largely on who is doing what. If you are outsourcing your transaction processing to First Data or Vital or Global or Concord, then your cost is whatever you negotiate and depends on your volume and the tasks you wish your outsource provider to perform.

For authorization and capture (front-end) and settlement (back-end) services, ISOs using outsourced providers pay in the range of 5 to 12 cents per transaction, depending mainly on the volume of transactions provided. Some folks pay a lot less than 5 cents and some pay a lot more than 12 cents, but the bulk of ISOs pay in the 5- to 12-cent range. This, of course, does not include the 10 cents paid to Visa or MasterCard for interchange.

But the real questions I would like to drill down on are: What are the variable costs associated with processing a credit card (or offline debit card) transaction? If all of the infrastructure is in place, what does it cost to process the next transaction?

There are only two variable costs: telecom costs and association fees. If there are other variable costs to process "the next transaction," I am not aware of them. (Member sponsorship fees often are charged by transaction, which will be discussed in Part III.)

Obviously, there are a lot of costs involved in running a processing company. These include salaries, benefits, occupancy, power, leased lines, depreciation and amortization of hardware and software, and the costs of redundancy and disaster recovery sites, etc. Tremendous overhead must be expended to pay for a 24/7/365 payment-processing center. There are huge fixed costs.

But the question remains: "Assuming you have adequate capacity, what is the cost of the next transaction?"

Let's first address the cost of the telephone call. There are lots of ways a payment processor can accept a telephone call, and they all have different costs. The most expensive method (telecom costs only) of delivering a transaction is via an asynchronous dial-up credit card terminal or PC. The least expensive way (variable costs only) to deliver "the next transaction" is via a fixed-cost leased line.

Most VeriFone, Thales, Lipman and POS computer systems are built with asynchronous, dial-up modems that typically run at 300, 1200 or 2400 baud. Hypercom ICE and T7 terminals and some Ingenico Elite 510 terminals use synchronous dial-up modems. Asynchronous modems are "serial" while synchronous modems are "parallel."

Those of you familiar with PCs know that most of the slower input/output devices (such as keyboard, mouse, credit card reader and RS-232 interface) plug directly into serial ports at the back of the computer while the printer normally plugs into a different kind of port - a parallel port. Look at the number of pins on each of these posts sometime. There are a lot more pins on a parallel port, and this allows for larger amounts of data to be transported in a given time frame than through a serial port.

In addition to a higher transport speed of parallel over serial, asynchronous modems are slow because they require special handshake arrangements once the phone call is received by the processor and continuous chatter (ACKs and NACKs) back and forth to validate the integrity of the data. Synchronous modems are not bothered by this kind of "telecom overhead" if properly installed; hence, a synchronous transaction is quite a bit faster.

This is important because the cost of telecom for a dial-up transaction is directly related to the length of connect time. During the connect time after the handshaking is finished, the transaction data is sent to the issuer for authorization, the authorization code is received, the receipt data and authorization code are sent back to the POS, the transaction is saved and the line is disconnected.

All processors burn a lot of calories negotiating low costs for reliable telecom services. Telecom costs have declined markedly in recent years. Even small processors can receive six-second billing, and the cost of telecom is well below 5 cents a minute for almost everyone. Some have speculated that the largest processing platforms pay 1-1.5 cents per minute with one-second billing!

Our experience is that 1200 baud asynchronous devices average 9.8-second connect times while 1200 baud synchronous devices average 6.0-second connect times. With a 6-second billing contract, the asynch devices cost about twice as much because the connect time consumes two billing units compared to one for synch devices. Connect time for slower, asynch devices often consumes three or four billing units.

Along with dial-up authorization costs, batching and uploading costs and telecom costs of downloading terminal software must be added to the mix at the end of the day/shift. These costs are impacted in a similar manner by the speed and protocol of the POS device.

When you do the math on all of this, it is difficult to believe that any of the larger processors are paying more than a half-penny per transaction when taking into account all of the above items. The very largest processors are possibly paying less than a quarter-cent per transaction for telecom costs.

Of course, the higher the percentage of asynchronous devices, the higher the cost of the transaction. I predict that in the not too distant future you will begin to see higher fees being assessed for slower devices to cover the increased costs of telecom.

So what are the other variable costs of transaction processing? They are the fees charged by Visa and MasterCard to the third-party processors and member banks for using their authorization, capture and settlement systems.

MasterCard has a tiered pricing structure based on weekly transaction volumes for the MasterCard settlement system known as INET. While I am contractually prohibited from disclosing these fees, I can tell you that the cost of these fees at the lowest-tier range is higher than the telecom costs of most processors. These fees are charged directly to the member bank each week and may or may not be included in the billing you receive from your processor.

Visa also has a fee for its Base II settlement system. Since writing Part I of this article, I have been able to obtain a copy of the 40-page countersuit filed by First Data against Visa. Here is the last full sentence on page 15 of the lawsuit: "Visa also demanded that First Data guarantee transactions in the event of the insolvency of a Visa member, even though Visa collects processing guarantee fees on every Visa transaction - even for the transactions that First Data (and not Visa) processes."

The Visa Base II fees are also tiered based upon volume. They are quite a bit lower than MasterCard's settlement fees and could even be a bit lower than the actual telecom costs for the larger processors.

Taking all of this and telecom into account, I believe that the cost of processing "the next transaction" is between a half-cent and 1 cent for the larger processors. Of course, this does not cover the very high fixed costs of operating a processing center, and it does not include things like profit and taxes.

FDC's Ominous Lawsuit

I am going to defer discussion of the fixed monthly costs of merchant processing to Part III of this article so I can talk more about the FDC lawsuit against Visa.

In the last issue I speculated that if FDC is successful with Visa litigation, FDC might indeed challenge the Visa dues and assessments for "on-FDC" traffic. I had seen nothing in the trade press on this subject, but it just seemed logical that a private network would logically want to try the failed Citibank strategy I previously discussed.

Bad news for non-FDC acquirers: First Data is already making the challenge! In paragraph 72 of page 21 of the complaint, First Data states, "... Visa also requires merchant acquirers to pay 'assessment fees' based upon gross sales volume (which includes transactions that are authorized and settled outside the Visa network). In 2001, this fee constituted 0.084% of the gross sales volume generated by the acquirers' merchants. ... By requiring merchant acquirers to pay to support Visa's new product development, Visa is attempting to stifle competition: Merchants are forced to underwrite Visa's costs of developing new products that will compete with their own products as they attempt to enter the Visa credit card network services market." (I added the emphasis.)

The conventional wisdom in our industry is that "scale is the primary driver of profitability." That is a valid argument to a point. But there is an important counterpoint to this argument, which I will make specifically in Part III.

To summarize, let's look at the two topics covered in Part I and Part II: dollar-volume-based costs and variable cost of transaction processing.

Dollar-volume-based costs (interchange, dues and assessments) have been a level playing field for all of the acquirers since the beginning days of BankAmericard and Interbank. Transaction costs certainly are scalable in the overhead of running a data center, but the variable costs of transaction processing boil down to (1) telecom and (2) association fees.

If the largest processors are all paying less than a penny a transaction for telecom and association fees, and the smallest processor is paying even two or three times that amount (not likely), scale gets at most 1 cent or 1.5 cents per transaction advantage on the totality of variable costs!

Even with the advantage of huge scale on fixed costs, there isn't enough opportunity to scale down costs to keep the small players from being competitive unless ...

... The big guy can eliminate 8.4 basis points of costs plus the association fees assessed for settlement, as is being attempted in this lawsuit. An advantage of 1 cent to 1.5 cents hasn't been enough to conquer the market. But 8.4 basis points plus no association fees plus a scale benefit of a half-cent to 1 cent for telecom will clinch the deal.

A First Data victory in this lawsuit would mean that none of us would be able to compete for merchants processing any reasonable volume of business with our hands tied behind our backs.

If this isn't a subject to get the attention of the merchant acquirers, I don't know what is. I have never seen a legal challenge to the competitive landscape of merchant acquiring before. This challenge should rally all of us to fight for the level playing field we require to survive.

It looks as if 2003 is going to be a very interesting year.

Bob Carr is the Founder, CEO and Chairman of Heartland Payment Systems, the nation's largest privately owned merchant acquirer and ninth largest overall, with annual revenues exceeding $300,000,000. Heartland was recognized by INC Magazine as the 57th fastest-growing private company in America and is one of the 10 largest INC 500 companies. Bob was a Founder and Vice President from 1988 to '90 of the Bankcard Services Association, which has since become the ETA.

Before entering the bankcard industry in 1986, he developed computer software systems for unattended fuel pumps and created the first integrated accounting applications for PCs. He also started the computer department at the Bank of Illinois and served as the Director of the Computer Center and as a mathematics instructor for Parkland College. He earned degrees in mathematics and computer science from the University of Illinois in 1966 and 1967.

To learn more about Heartland, visit www.hpsteammates.com or www.heartlandpaymentsystems.com, or e-mail Bob at Bob.Carr@e-hps.com.