Price – Splitting the Pie – Why Not Just Bill and Keep?

Back in the old days when telephony was the province of national monopolies, domestic prices were usually regulated and regulators are subject to political pressure to keep prices down.  One good way to make lots of money was on international calls.  The alternative to calling was telegraph, expensive itself, or taking a boat.

In the very beginning, carriers split the cost of cables between countries and each carrier kept all the revenue from customers placing international calls from their country.  This simple mechanism for splitting the pie is called “bill and keep”. It is used now between most Internet backbone providers and its simplicity is part of what makes the Internet work so well and be so readily extensible.  But bill and keep makes it difficult to charge monopoly rents, as you’ll see below, so the arrangements between carriers for international calls rapidly got more complex.  The descendants of those same carriers would like to make Internet traffic exchange and the Internet itself more complex.

Let’s suppose calls between Country A and Country B start out priced at $4.00/minute in either direction.  For simplicity, we’ll also assume that there are a million minutes of traffic daily in each direction.  Each carrier is billing its own customers $4 million per day and keeping every penny of that.  What’s not to like?

But now Carrier A gets greedy and REDUCES prices to $3.50.  Customers quickly figure this out and, when they have a choice, originate calls from Country A rather than Country B.  Now there are 1.2 million minutes going from A to B and only .8 million minutes going the other way.  Even at reduced prices, Carrier A is now getting to bill and keep $4.2 million dollars/day while Carrier B is now only collecting $3.2 million.  Oh-oh, got to do something about that or someday prices might get down near cost!

What carriers did to avoid competitive discounting was called the settlement rate regime.  Although nominally a way to share revenue, it was actually a way to fix prices.  At its simplest, it would work like this between Country A and B.

If the carriers decide that the “correct” price is $4.00/minute, something called an “accounting rate” would be set at $4.00.  The carrier with more minutes has to pay the carrier with less minutes a settlement fee of half the accounting rate ($2 in this case) for each “excess” minute.  In the example above, since Carrier A ended up with .4 million minutes per day more traffic than Carrier B, A owes B $800,000/day.

This really changes the game.  Carrier A ended up billing $4.2 million but now nets only $3.4 million; Carrier B receives the settlement fee and ends up with the $4 million it was getting before it lost its traffic.  In other words, all the loss from the price reduction falls on the perfidious Carrier A.  You can’t make money by cutting prices to steal traffic once settlement rates have been set up (this does ignore the fact that lower prices tend to encourage more calling).  Of course, the whole point was to discourage price cutting and, for a long time, the settlement rate structure worked just fine and international calls stayed expensive even as the facilities needed to complete them got cheaper.  The International Telecommunications Union (ITU), which the carriers set up to manage all this price fixing, is still around as part of the UN.  The ITU would really like to help make the Internet complicated too.

Settlement rates got more complicated with decolonialization.  Used to be that the phone companies in the colonies belonged to the national carrier in the mother country so it didn’t matter that most calls flowed from the richer mother country to the colony; no settlement needed when it’s all in the family.  But the local carriers ended up belonging to the ex-colonies.  Traffic imbalances were still huge – often as much as 10 minutes from the old mother country for each minute back.  Certainly British Telecom and France Telecom didn’t want to pay enormous settlement charges to their former subsidiaries.  Don’t worry; they didn’t.  Ratios were introduced so settlement would only be due if the ratio were exceeded in each direction.

Settlement got even more complicated when real domestic competition broke out.  British telecom was still an international monopoly when Sprint, MCI and others joined AT&T as international US carriers.  Now how do you settle and with whom if you’re British Telecom?  “Proportional return” had to be invented so that BT would send Sprint, MCI, and AT&T US-bound minutes in proportion to the amounts of UK-bound traffic each of those carriers sent, then settle the difference with each.  Now add competition in the UK.  This really gets complicated.  But it’s all for the good cause of keeping prices high so carriers tried hard.

Trouble is that telephony really was getting competitive.  Another problem was that complexity invites arbitrage.  Some really smart people figured out ways to game all the complexity.  One relatively simple scheme is to route through third countries where there may be either lower total settlement charges or just a temporary shortage of outbound minutes when all the math was done.  Full employment for rocket scientists!  Pretty unproductive but quite renumerative use of brain power.

The ultimate arbitrage was to make international calls appear as domestic since domestic rates were usually much less than international.  Suppose a company has a private line between its home office PBX and another PBX in an office abroad.  High international rates often justified the cost of private lines but they were expensive.  Pretty soon a guy in a leisure suit shows up and tells the company’s telecom manager how he can make some money to help the company offset the cost of the line. See, the PBXes can be programmed so that calls coming over the private line into the PBX in the foreign office can be redialed out as domestic calls in that country.  Technical term is a “leaky PBX”.  Leisure Suit Larry’s employer’ll pay the company that owns the line $1.00/minute to complete these calls.  The calls end up being $2.00/minute to people who buy prepaid calling cards from LSL’s company or someone it resells minutes to.  Better than $4.00.

Even better, send the calls internationally over the Internet and hop off the Internet at some brand new competitive telco as local calls.  Now there’s not even a requirement to buy the expensive private line or route through a zillion third countries.  VoIP was the bale of straw that broke the camel’s back that was the international settlement rate structure.  You have to work hard to find a $4.00/minute call now unless you’re roaming internationally with a mobile phone.

True confessions:  ITXC, the company Mary and I founded, used both the technical and arbitrage advantages of VoIP  to win billions of minutes of wholesale international phone traffic and become the seventh largest carrier in the world in terms of international traffic.  That’s how come I know all this stuff.

I wish the story ended here:  VoIP Does In Settlement Rate Structure! Everyone Talks Free or Almost Free Forever!  World Trade Is Advanced!  Families Stay in Touch!  Trouble is those same old carriers reconstituted would like to return the favor and impose the equivalent of settlement rates on the Internet.  Might break the Internet with complexity; might only make it more expensive.  Win-win as far as they’re concerned.

The first post in this series on price is about the cost of complexity.

The second post is about FON, the WiFi sharing scheme, as a great experiment in pricing.

The third post is about “freeloaders”.

The fourth post is on unseating an incumbent with disruptive pricing.

The fifth post is on access charges.

Comments

[Ted Weitz]

Of course there is another wrinkle to the entire complex pricing structure that turned out to be very important. When Carrier A and Carrier B trade traffic, internationally or domestically, they are really engaged in a swap in which the accounting rate (or its equivalent) is a highly artificial number, and money may or may not change hands. When Wall Street pressures for revenue mount, the temptations to play with the revenue number increase, and the essence of the enormous accounting scandals at WorldCom and Qwest was that they established artificial prices for various swaps and then reported them as revenue. Going back to Tom's example at the outset, the accounting rate structure produced $4 million a day in revenue for each carrier. But if the accounting rate had been halved, the revenue would have been only $2 million a day. In domestic exchanges, or where there was no regulatory agency involvement, the carriers could adjust the accounting and alter their revenue. Indeed, the accounting for such exchanges was so complex even auditors discussing them in good faith could not fully agree. And everyone suffered in the long run.

[Scott]

An unbiased settlement structure could work to everyone's benefit. In the old days, a telco with a huge inflow of international calls and few outgoing calls could purchase the infrastructure to handle the traffic, knowing they will be reimbursed for their expense. Unfortunately, the telcos couldn't resist biasing the process to their benefit. The core problem, as I see it, is that the process focused on balancing revenues, not costs. That will absolutely create a price floor. One can monkey with cost numbers just as easily, so the appropriateness is in the details of the process. And yes, it should be kept very simple.

The Internet has already seen one wave of game playing that may still happen today, I don't know. In the early years with many small ISPs, ISPs would resist carrying data traffic for non-customers, pushing it onto other ISPs' networks for transmission. To each ISP, they were spending their revenue on their customers. But for the Internet as a whole, it was inefficient. With fewer ISPs and a more balanced variety of traffic, I presume the games have subsided some. But if there was a simple, unbiased settlement structure between ISPs (not at the retail level), I think it would actually accelerate new developments.

I just don't think the telcos should be the ones defining and owning the process.

[Aswath]

Two points:
1. I don't recall all the details (never fully understood it then either), but then MCI found it cheaper to carry interstate calls via Canada, but was penalized because it was government traffic and there was a concern about security.
2. Even now you can find $4/min rate if you are careless. Verizon's "non-plan" basic rate to call India is $5.85. This is more than twice what it was under monopoly Ma Bell in late 70s!