“Capacity Sales, Bandwidth Markets and Prices.”

Howard Holme
President Bandwidth Market, Ltd.

KMI’s 6^th Annual Fiberoptic Submarine Systems Symposium

Vancouver, British Columbia, June 16, 2000

            Providers, buyers and users have great opportunities to participate in submarine fiberoptic telecommunication systems and thus improve their profits and stock price.   We suggest here using bandwidth markets or exchanges to lease or sell submarine telecom assets.   We also discuss pricing issues, and trading bandwidth as a commodity.  

            Companies can participate in bandwidth markets in several ways.   Companies can “sell” the telecom assets they have or build.   We will use the words “sell” and “buy,” though typically companies sign leases that last different time periods, usually from one to twenty five years.   Companies can also buy assets or services using bandwidth markets.   Finally, companies can, more in the future than now, actively trade bandwidth contracts; similar to the way they may presently trade commodity energy contracts.

Fiber and bandwidth opportunities still exist in areas that are not yet “saturated” with fiber and bandwidth.   Further, fiber and bandwidth have become such vital drivers of the new economy that it is hard to determine what “saturation” would be.   Fifteen to twenty years ago in the US, Sprint directly buried 12 fibers (not in conduit) and thought it had more than the world would ever use.   Three years ago, Qwest started burying two conduits, 96 fibers, and thought it had more than the world would ever use.   Two years ago, Level 3 started burying 8 conduits (each one of which could today be equipped with 1000 fibers), along the same cross-country Union Pacific Lines Qwest used.   Denver Colorado has spawned yet a third company (along with Qwest and Level 3) that will build one of the world’s largest networks.   Aerie says it has 15,000 miles of pipeline and other rights of way and will connect 192 US cities, thus overbuilding Sprint, Qwest, Level 3, and Williams Communications.    At the local level, last year, Level 3 and Nextlink started burying 24 conduits in five loops for Denver’s local traffic network.

We have yet to hear a builder of dark fiber networks who says it built too much.   Companies still have a huge opportunity where major networks are not yet installed.

            Companies that have built large telecom systems, especially dark fiber, are among the most successful companies.   Global Crossing and FLAG are examples.   In the US, Williams has succeeded greatly twice, first in building and selling 11 of its 12 pipeline fibers to LDDS (now Worldcom), and now again, rebuilding the gas pipeline right of way fiber optic system throughout the US and acting as a wholesale carrier ’s carrier.   The Montana Power Company sold its electric generating capacity, and now its transmission capacity and built fiber optic systems.   Its stock rose from 17 to 58 within the last two years.   RCN has partnered with companies in building networks, and its stock rose from 10 to 75 in the last two years.     Metromedia Fiber Network, Inc.’s stock rose from 7 to 90 in the last two years, NextLink (stock up from 27 to 270), Qwest (stock up from 10 to 70), Level 3 (stock up from 45 to 130), and Global Crossing-Frontier (stock up from 10 to 60).   Recent stock market declines have decreased these peak prices, but the companies still have bright futures.

Why Companies Participate in Bandwidth Markets

            Many national telecom monopolies had a captive market, so did not specialize in marketing (the specialty of bandwidth markets).   In prosperous times with labor shortages in many locations, companies can avoid building large sales forces and reduce sales costs.   Companies are often geographically isolated, while bandwidth markets may grow to be global before many national companies do.   In the future, markets, perhaps as much or more than individual companies, will have major influence on market prices.  

            Unlike national monopolies that relied on lifetime customers, competitive telecommunications companies added new definitions to the word “churn.”   As Newton’s Telecom Dictionary says, “Long distance users change their preferred carrier as often as some of them change their underwear . . . Some users have found ways to switch their long distance service often enough so that they never pay for a long distance phone call.”   No wonder companies can use help selling telecom services.

            Ecommerce is revolutionizing business-to-business sales in nearly every industrial product.   GM, Ford and Chrysler opened their automobile parts purchasing exchange.   Enron opened its Internet energy trading market in September 1999 and became the biggest e-commerce site in the world in three months, trading over $20 Billion of energy.   Companies would be very ill advised to build sales forces like the US Regional Bell Operating Companies (RBOCs), during the same time the RBOCs will have to fire most of their salespeople.

According to May 2000, estimates of the Gartner Group, in 1999, worldwide business to business (B2B) e-commerce reached US$145 billion (US$1 = RM3.80), and the North America region accounted for 63 per cent of the market with revenue reaching US$91 billion.    In 2004, worldwide B2B e-commerce is projected to surpass US$7.29 trillion, the North America region will account for 39 per cent of the market with revenue of more than US$2.84 trillion.   B2B revenue in Europe in 1999 totaled US$31.8 billion. In 2004, the European B2B market will be more than US$2.34 trillion.   In 1999, Asia/Pacific B2B revenue totaled US$9.2 billion, and by 2004 the region will have B2B e-commerce revenue of US$992 billion.

The New York and Chicago stock and commodities exchanges are the most efficient markets in the world, but online trading is revolutionizing them.   Electronic Computer Network and Internet trading are forcing these exchanges into consolidation, merger, and closing trading pits.   The French MITEF opened a traditional open outcry trading pit and computerized trading system simultaneously on a new product.   Within two months the trading pit closed.   In light of these facts, companies should not plan for the long term to sell the industrial product of bandwidth, which is tailor made for ecommerce, with a large direct sales force.

  Most of the so-called “bandwidth markets,” or “bandwidth exchanges,” including Arbinet, RateXchange, Band-X, Min-X, and AIG list exclusively, or almost exclusively, international switched long distance minutes for voice conversations.   Soon, most submarine cable operators will not depend on switched minutes.   To the extent these other markets list bandwidth circuits, they typically list small capacity such as T1s (which can carry 24 voice circuits) or DS3s.    In contrast, our Bandwidth Market lists circuits of all sizes, from T1 to OC192 speed, and terms up to 25-year Indefeasible Rights of Use.

Concerning United States bandwidth listings on the Internet, at the time of this writing, Bandwidth Market lists over 300,000 bandwidth circuits.   Arbinet lists none, RateXchange lists about 300, Band-X lists about 6, and Min-X lists none.   Enron lists perhaps 4 or 6.  

            Several bandwidth markets have sites on the Internet, planning to use or using a system of “offers” and “bids.”   Sellers “post” or state offers to sell the services or products, while buyers can state a “bid” or agreement to pay a certain amount for stated services.   When other exchanges list bandwidth, they typically state a speed, a city pair and a price.   Bandwidth Market, on the other hand, allows parties to list many more details.   (For example, parties may specify address, city, state, country, zip code, CLLI code (alphabetic abbreviation for telecom location), LATA (Bell company regional location), NPA-NXX (country or area code plus next three digits of phone number), collocation descriptions, names of companies with which the circuits can interconnect or “peer” at each end of the circuit), speeds, terms of contract, quality measures including latency (time of travel on the circuit), restoration time (affected by SONET rings or meshed circuits), jitter, Bit Error Rate, packet loss rate, and large spaces for companies to state additional terms of the contract.)  

            Bandwidth markets typically are compensated, if and only if a deal is done, by a percentage commission.   They make arrangements with sellers to avoid or minimize “channel conflict,” for example, by agreeing to forego commissions if the seller is already negotiating with the potential buyer.  

Bandwidth Market, at www.bandwidthmarket.com, allows simultaneous sorting of over 300,000 circuits by country, state, city, address, speed, and date of posting.   Contracts can be standardized.   Prices, visible throughout the world 24 hours a day can be changed in real time.   Communications are by automatic and manual emails.   Auctions can be used.  

            Telecommunications, and Internet sales, are global.   Telecommunications are not valuable unless they connect far beyond states or provinces.   A traditional sales force will be geographically concentrated in one area, but some buyers are not.   Internet markets can be effective regardless of local, national, or international scale.

From the “last mile” to the home or business, to the core of the network bottlenecks can develop that will quickly change the value of bandwidth.   The circuit between two cities that had a shortage before the installation of a terabit router (1,000,000,000,000 bits per second) will surely not have a shortage the day after.   Europe is installing petabit networks (one quadrillion bits per second, a million times as fast as a gigabit network).   “Bits” of information, unlike commodities such as corn, cannot be stored for a significant time.   Companies that react slowly, or need layers of bureaucratic decision making to change a price, will lose sales.  

Now, the telecommunications market is very inefficient, with a large spread between the asking price and the bidders’ willingness to pay.   Watching bandwidth markets, managers can react much faster.    Bandwidth markets can supply necessary information, in form of sales or no sales, that are necessary for managers to decide where to build, how much to build, how much to sell, and at what price.   The markets will reduce the “spread” and increase economic efficiency.

All of the discussion thus far deals with standard networks, used by end users, and sold in standard or ecommerce ways.   Formal commodities trading involves some other factors that we now discuss.

Bandwidth Commodity Trading

            Enron and Williams companies have both endorsed bandwidth commodities trading.   Numerous conferences have discussed bandwidth trading as a commodity—and bandwidth trading will happen for reasons including the following.   Enron had a $20 billion success with Internet energy trading.   Enron has made a very large investment in buying a bandwidth network so it can be a market maker, buying and selling its own and others’ bandwidth.   Many commodity traders looking for the next new commodity after energy commoditization think bandwidth will be the new commodity.  

            Commodities trading in bandwidth is just beginning in the US, with a small number of such trades so far.   Contracts are not completely standardized.   Trading or pooling points are just now being established.   Switches are being built and installed.   Some are concerned about buying from or selling to Enron or Williams, or through their switches, because of their networks and market intelligence and experience.   A limited number of routes will be traded in this way, while most traffic goes elsewhere.  

            To consider only the last of these issues, most companies’ networks will not be between New York and Los Angeles, MAE-East (Vienna VA) and MAE-West (San Jose CA), New York and London, or the other cities that will first be traded.   If companies want to be most involved, like Enron, they can buy networks from others on those routes, perhaps as dark fiber or 20-year Lambdas (OC48 wavelengths).   That is a major commitment, costing millions to billions of dollars, which few companies will make.   Also, these most competitive routes are the most likely to decline the most in value.

Or companies can buy, for use or trading, the standard contracts offered by owners of those services.   On the other hand, if the company is not a very large owner or user of the bandwidth commodity on this route, the traders may have relatively few advantages over many other commodities traders working in this or another commodity.

Assuming bandwidth trading develops into the future, one can expect the principles established on the major trading routes to diffuse to other network routes throughout the United States and the world.   In twenty years, one might envision the full panoply of options, swaps, and derivatives operating for hundreds or thousands of bandwidth routes throughout the world.   At that time, however, one should not expect fat profit margins on bandwidth.

How to Participate in Bandwidth Markets

            Most of the markets have web sites with registration procedures.   Talk to the managers, figure out what you want to buy or sell, with what terms and at what prices.   On some sites you can post your own circuits, long distance minutes, or other opportunities.   On others, talk to the managers or populate on line databases.   

            Talk to the managers to determine contractual terms and conditions, procedures during a sale, use of your contractual terms, use of any switches or routers associated with the market, and flow of payments.

One cannot predict the needs of the next customer.   To post ten circuits to try out a system misses the customer who needs your eleventh or twelfth circuit.   To post T1s or E1s may miss the customer needing STM1s (OC3) or dark fiber.  

If you have installed dark fiber, is there some price at which you are willing to sell (and perhaps build more)?   Dark fiber, like everything else in telecommunications, becomes obsolete—as multimode fiber evolves to single mode fiber, to single mode non-dispersion, to LEAF, to fiber without the “water window,” and so on.   Almost all major fiber builders of the past may ask themselves if they held on too long, for too high a profit margin and price, waiting to sell lit bandwidth while the next generation came to build more dark fiber.    

Pricing:   Will there will be a glut?

Bandwidth is increasing faster than any industrial product in world history.   Bandwidth’s huge increases during the last decade invite questions of whether it can continue to grow at such a pace, and whether gluts, price collapses and bankruptcies will soon come.   As an aid to answering these questions, let us look at one of the closest analogues—computer hard drives.   Hard drives have grown in number and capacity faster than nearly any other product, they have benefited from rapid technological improvement, their prices have dropped rapidly; they have faced and overcome various technical, market structure, vertical market integration, and international hurdles; and are central to the information economy.  

From 1957 to 1997, the areal densities of computer hard drives increased 1.3-million-fold.   From 1988 to 1998, the average price per megabyte of storage dropped from $11.54 to $0.04, about 96% over a decade.   From 1988 to 1998, worldwide disk drive sales revenue increased from about $20 billion to about $30 billion.

Thus for computer disk drives, there has been a very large (over 400 times) increase in total sold capacity in the decade, from roughly $20 Billion divided by $11.54/Mbyte (1.7 Petabytes) to $30 Billion divided by $.04 (750 Petabytes).   Notice that in this industry, price elasticity with demand has been positive.   Total sales have increased as price has rapidly declined.   However, notice further that the price elasticity with demand has only been barely positive.   If this technology is a strong analog, the optimism of some companies that the price elasticity with demand may be a positive 1.7 to 2.0 is likely not to be justified.

A 400 times increase in capacity with a 0.5 increase in revenues has meant that computer disk drive makers have been scrambling to stay profitable, and many have failed.   Indeed, of the top seventeen firms building about $1 billion of drives in 1976, all except IBM had failed or been acquired by 1995.

In terms of pricing declines, some might say there has already been a transatlantic glut.   From about November 1998 to March 2000, the price of a STM1 New York to London 20-year Indefeasible Right of Use (IRU) dropped in price from about $12 million to $1.25 million at Bandwidth Market.   The air distance is 3463 miles, the number of DS0 circuits is 2016, and the months are 240.   Thus, the price per DS0/mile/mo at $1.25 million is $0.0007, though there is an Operation and Maintenance charge.   The monthly price was recently $46,000. Paying that rate would pay for the 20 year circuit in less than 3 years.

Companies that hold out for the prices that were available 6 months or a year ago, can hold out forever and never gain any revenue from their assets.   Many people, including us, expect further large declines when the terabit services are Ready For Service later in 2000.   Others expect major declines in transpacific prices when the higher speed transpacific cables are Ready For Service.  

We at Bandwidth Market certainly expect large price declines over time.   Technology is advancing in the number of colors per fiber, the speed of lasers, the number of colors fiber will carry, the distance of transmission without regeneration, and the number of fibers per cable and the number of cables.   Competition is also increasing from government deregulation, the number of companies laying cables, the number of independent owners of submarine cable capacity, and their willingness to sell and compete on price.   We also expect that the extremely high rates of return of some of the early cable builders such as Global Crossing, as well as terrestrial cable builders, will lead to a continuation of building, probably until some suppliers suffer losses or go bankrupt.   GST, Iridium, buyers at the C block auctions and a number of other examples show that there is no guarantee of success in this and related high tech competitive industries.  

We think stock markets, and the economy as a whole, are pushing telecom companies to decrease their cost of sales.   Pressure to cut sales forces are coming in boom times, because new sales people are very expensive.   Pressure to cut sales forces and use ecommerce will come with a vengeance when stock prices or the general economy decline and companies must use the most efficient sales methods to try to maintain profitability.  

How to set prices?  

AT&T built a monopoly network, in a politically regulated system favoring many cross subsidies.   National and international regulatory authorities expected long distance voice to pay Universal Service charges or settlement rates that could subsidize national governments and various telecom users.   Prices were based on cost of network, plus a regulated rate of return, adjusted by subsidies.   Costs were divided by numbers of miles and prices were developed.   Systems were expected to last 20 or more years, with depreciation and bond payouts to match.   Much of that pricing system remains, including per mile or other nationally negotiated or agency prescribed pricing, while the underlying assumptions have disappeared.   We concur with the recent comments in a Guam Philippines Cable System Newsletter that many new carriers from the deregulated sector will purchase submarine capacity, that it is necessary to remove restrictions relating to capacity use and to allow free resale of purchased capacity, and that bandwidth sales and dark fiber sales may be necessary for the diverse requirements of customers in a new and liberalized environment.  

Costs of network may be nearly irrelevant—especially if the network has been grossly overbuilt by the next generation, or if a company controls a bottleneck.   Increasingly, the industry and bandwidth markets avoid various regulatory and political constraints.   Various subsidies, including international settlements, are not clearly applicable to new technologies, are avoided by companies, and are not enforced by governmental authorities.   Competition is very rapidly replacing monopoly, with private company cables competing against consortium based cables, and even in the local loop with DSL versus cable modem, versus future MMDS wireless.   The number of competitors (even if using the same network) can be as important as new routes.   Political subsidies are very hard to maintain in a competitive environment.   Systems do not last 20 years.   The turnover is like that for computers, and bandwidth is growing much faster than computer performance is.

What will replace old pricing models?   There can be only one answer.   Market prices determined by supply and demand.   Some companies will thrive on cheap networks built in the right places, and others will fail with expensive networks built in the wrong places.   Pricing by cost, and pricing per mile is obsolete.   Pricing assuming 20-year amortization, depreciation, and payouts to bondholders is obsolete.   Planning and pricing must assume rapid decline in both the cost of providing, and the revenue from selling a specific bandwidth unit.

In the short term, it seems that those who build big systems, and sell them quickly, prevail.   Most of the national companies could have built fiber systems and sold them, rebuilt and become the national fiber and large bandwidth suppliers.   Instead, most have preserved their margins on low bandwidth voice traffic and lost major market share.   Voice traffic in the US is reported to be much less than (e.g. 20% of) data traffic.   Now the national and regional legacy voice systems may watch their traffic go to Internet and their companies genuinely struggle.

As bandwidth, in the long-term future, does become a commodity, bandwidth pricing will be more like corn pricing.   Providers will not control the price at which their bandwidth sells, any more than a farmer controls the cost at which his corn sells.   All the people, taking all their individual actions, will set prices collectively acting through markets, and providers will react to the market prices.

_____________

Howard Holme is President of Bandwidth Market, Ltd., a two-year old, Internet based, privately held, Denver, Colorado company.   He received an AB in History with distinction, and Honors in Humanities from Stanford University in 1967, and a Juris Doctor from Yale Law School in 1972.   He will speak at KMI’s 6^th Annual Fiberoptic Submarine Systems Symposium in Vancouver B.C. June 16 at 10:30 AM. Bandwidth Market is at Suite 2400, One Norwest Center, 1700 Lincoln St., Denver CO 80203-4524, 303.894.4480, style='color:black> sales@bandwidthmarket.com.   The web site is www.bandwidthmarket.com.