Introduction

When I was a kid growing up in New Jersey, we would during the summer sometimes visit the boardwalk at Seaside Heights. As some of you may know, the Seaside Heights boardwalk hosts a huge amusement park with all kinds of rides and games: the roller coaster, the ferris wheel, the tilt-o-whirl, skee-ball, all kinds of stuff to do. I remember that there was a fair amount of time that we spent waiting, standing on line buying tickets, or queuing to get on the rides, but when you did get on the rides, it was just great, a wonderful experience.

Well, this is my last week at the FCC, before I return to my regular life as an economics professor at Boston University. Somehow, as the summer approaches, and quite beyond my control, Seaside Heights comes to mind as metaphor for my time at the FCC. It's corny, I know, and sounds a bit flip, although I don't mean it that way, but somehow the metaphor resonates for me. One reason is that there is certainly a lot of waiting at the FCC. Just ask the communications bar. Everything in government takes at least three months, and somethings never happen at all. For example, I still have not been able to replace the blotter on my desk on which I spilled a cup of coffee last December. But the real reason for the metaphor is that the policy issues are just great! The opportunity to participate in policymaking in a constructive way on such important issues has been richly rewarding both professionally and personally.

I have had the good fortune of a two-for-one deal, the honor and pleasure to work for two remarkable leaders, Chairman Bill Kennard and his predecessor Reed Hundt. But the honor and pleasure doesn't stop there. The opportunity to work with the marvelous FCC staff, throughout the agency, has made my tenure a tremendously rewarding experience in a very personal way. I am very appreciative.

This is also the most fascinating of times to watch, close up and personal, the germination of a modern telecommunications policy for this country. There is much I could say about it. But as an academic, what has fascinated me the most is how economics interacts with law and politics to shape policy.

We are a gathering of professional economists. It is clear to us that economics has much to say about good policies. But, we also are gathering in Washington, and we know that law and politics have much to say about what is possible. And we know from mere common sense that the possible trumps the good.

So why bother? You hear it over and over again. It's all politics! But this cynicism is too simple and too myopic, and a bit misguided. Good economics often can find solutions within legal and political constraints, and sometimes can provide an intellectual basis for relaxing those constraints. Politics shift, and the law changes. Such events create opportunities for good economics.

John Maynard Keynes is probably the most famous economist of the twentieth century. He spent much of his career shuttling back and forth between academia and government, between Cambridge and London. I believe he said something like the following: "If economists could be thought of as ordinary competent people, like dentists, that would be splendid!"

Or maybe he said that it would be splendid if economists would think of themselves that way. Certainly, this is not a very comfortable self image for an academic economist. Any academic economist worth his or her salt is something of an iconoclast, and I doubt Keynes thought of himself like a dentist.

But the Keynes' recommended image is not a bad one for a government economist. Its an image I have mostly tried to project at the Commission, no doubt with varying success, and, since today I am still a government economist, its the image I will seek to convey in this speech.

Before continuing, I do need say, one more time, that the opinions I express today are my own and do not reflect those of the Commission or individual commissioners. The same will be true next week, but I won't need to say it.

What follows is something of a wandering discourse. If there is an underlying theme of my remarks, it is simply that competition generally is a good thing, that competition can contribute to egalitarian goals while not always achieving them, and that attempts to achieve egalitarian goals by other means can be counterproductive and impose unintended costs on consumers.

The Telecom Act

The Telecom Act of 1996 is the backdrop for telecommunications policy today. There seem to me to be three important themes running through this legislation. First, an important goal of the Act is to promote widespread participation in telecommunications markets by consumers and firms. Second, competition is a preferred method for achieving this goal. Third, unnecessary and counterproductive regulations ought to be eliminated or modified. That all sounds pretty good, doesn't it?

A clear goal of the Act is to bring the benefits of competition in telecommunications markets to "all Americans." That phrase "all Americans" is a bit mysterious. Competition certainly does expand markets by offering consumers' better value: lower prices, improved service, new products, greater variety. But Congress' seems to have the idea of delivering these benefits to all communities of consumers: to both business and residential markets, to rich and poor communities, and in cities, suburbs, and rural areas.

The principal modus operandi of the Act is to enable new competition by lowering barriers to entry of all sorts: eliminating laws and regulations that prevent firms from participating in telecommunications markets and restrict consumers' choices; implementing new operational procedures, like number portability, that make it easier for consumers to switch between telephone companies; and obliging some incumbent telephone companies to share their networks with other firms.

If the Act works well, then consumers should enjoy more and better choices. This is not an unreasonable expectation. The combination of lower entry barriers and blossoming new technology creates profitable new marketing opportunities for all kinds of firms to serve all kinds of consumers in all kinds of ways.

However, it is unreasonable to expect that free entry and unfettered competition will present all communities of consumers with the same choices. America is not that homogeneous, and free markets are not that egalitarian. Competitive profit-oriented firms will maximize value by trying to match the products they offer to the demands and costs of serving different types of consumers.

But the Act is also rather specific about certain "universal service" goals. Rates for standard telecommunications services should be "just, reasonable, and affordable," advanced services should be available everywhere, and telephone rates should be about the same in rural, high cost areas as in urban areas. Thus, there does appear to be an egalitarian streak running through the Act.

This creates a conundrum for policy. Free entry into markets does not necessarily respect Congress' universal service goals. Competition follows the money. If the cost of serving a community is higher, or consumers' willingness to pay is lower than elsewhere, then the community very well may attract less competition to provide a service, or perhaps no competition at all.

What then? Up till now, a system of regulated monopolies has supported universal service by keeping the price of basic residential telephone service low, and compensating local telephone companies by letting them charge high prices for long distance access and business services.

But people eventually noticed that the monopolies tended to offer the same old products. So, in passing the Telecom Act, Congress rejected this received system of regulated monopolies in favor of competition and deregulation, in order to better promote innovation.

So how can universal service goals be made consistent with competition? An obvious answer is to subsidize competition where universal service is deficient. The positive economics of such subsidies is pretty simple. A firm that receives a subsidy for serving a particular community of consumers would be willing to serve those consumers at a lower price.

The problem with subsidies is that someone has to pay for them. The subsidies could be financed out of general tax revenues, but Congress made a different election. The Act specifies that universal subsidies must be financed by "contributions" from telecommunications companies.

In competitive markets, the companies required to do the contributing inevitably will recover the cost of their universal service contributions from their customers through higher prices. In fact, AT&T announced this to the world this morning. As markets adjust, higher prices may very well be supported by a decreased level of competition, as higher costs thin the number of competitors.

My point is that universal service based on explicit subsidies, will not eliminate cross subsidies in the price structure. When the dust settles, some consumers somewhere must pay above cost, so that other consumers somewhere else can pay below cost. Not surprising, this creates political conflict.

Now, economists have no special expertise for determining which communities of consumers most deserve universal service support. But economists are pretty good at pointing out whose ox gets gored and how much ox blood gets spilled in the process. For this reason, economists tend to focus on the efficiency implications of policy alternatives.

So let me finish this part of my discourse with an economist's point. Raising prices above costs, by any means - whether by regulation, or a tax, or a mandatory universal service contribution, typically creates economic inefficiency. The magnitude of the inefficiency depends positively on how sensitive is demand to the price distortion.

Thus, under the circumstances, the consumers who are implicitly subsidizing universal service would rather just write a check. But, because the demand for local service in price inelastic for most consumers, writing a check is just like paying more for local service. In other words, the most efficient way to subsidize local service is with a fixed end user charge.

Local Competition

The Telecom Act's strategy for reforming local telecommunications markets is very ambitious. The strategy is to move from regulated monopoly markets to deregulated, competitive markets, without breaking up the local monopolies. This is a big problem because the incumbents own most of the facilities that comprise local telephone networks. Thus local markets can become deconcentrated only by the growth of new facilities-based competitors.

The Telecom Act presents new competitors with various possible strategies for entering local markets. An entrant into the local market could serve customers over its own facilities, or over facilities leased from the incumbent, or by reselling the incumbent's services under its own brand name, or some mixture of these.

To understand the nature of these strategies, it is useful to know a little bit about standard telephone service. A telephone is connected to a twisted pair of copper wires called a "loop". Loops are connected to switches that are housed in the central offices of the local telephone company. Different central offices are connected to each other by fiber optic trunks. Trunks also interconnect the networks of different companies. Telephone service uses the switches to open a circuit over the loops and trunks to connect two telephones and enable a two-way voice communication.

Now let's compare alternative entry strategies.

Resale, I am sorry to say, is not a very imaginative strategy. A reseller offers consumers the same telephone service as the incumbent. It markets the service under its own brand name, and pays the incumbent a wholesale price, who continues to provide telephone service on its network.

Resellers act more like distributors than competitors. Resellers compete with the incumbent by providing more efficient customer service. But resellers have no independent ability to lower the cost of providing telephone services, or to introduce new services.

Resale cannot be expected to put much downward pressure on an incumbent's prices. Under the Act, the wholesale prices of resold services are tied to the incumbent's retail price less avoided cost. Thus, if the incumbent's retail price goes up, so does the cost of rivals engaged in resale, creating an obvious incentive for the resellers to follow the price increase.

Resale may provide a stepping stone to facilities-based competition. A firm can resell the incumbent's service to its customers in advance of building out and serving customers on its own network. Resale enables a new entrant to establish a customer base, which makes it easier to attract the capital needed to build its network.

However, resale itself is not a very profitable activity except perhaps for very selective marketing strategies. Resellers most likely have higher marketing costs than the incumbent. This is not due to any inherent inefficiency of the entrant, but rather due to the fact that the incumbent has established relationships with virtually every consumer in its service territory. Even a long distance company, who has established relationships with only a subset of consumers, is at a disadvantage in the local market.

Now contrast a facilities-based competitor.

A facilities-based entrant might either build its own facilities or lease network elements from the incumbent. For example, one promising facilities-based strategy in some markets is for an entrant to collocate in a central office and to connect leased copper loops to its own switch via a trunk joining the collocation cage and the switch.

Facilities-based entry creates much more impressive conditions for retail competition and deregulation than does resale. Cost-based pricing of unbundled network elements breaks the link between the incumbent's price and the competitors' costs. Thus, an increase in retail prices by the incumbent does not raise costs of facilities-based competitors, and, therefore, these competitors can profitably gain market share by not following the incumbent's price increase. In this way, facilities-based competitors do discipline the market power of incumbents.

Facilities-based competitors are also able to innovate. They can lower the cost of providing telephone service by configuring their networks more efficiently, and can introduce new and higher quality services to consumers by increasing the functionality of their switches. Such innovation enables the competitor to gain market share with lower prices and better services.

A facilities-based strategy has other advantages for a long distance company that integrates back into providing local service. The company avoids paying access fees to the local telephone company for originating long distance calls. The facilities based competitor also collects access fees for terminating local calls.

Today, access fees are well above the long run incremental cost of providing access to the local network for long distance calling. Thus, avoiding and collecting access payments is an important profit incentive for facilities-based local entry.

Probably the most controversial entry strategy is for the competitor to provide service over a complete platform of combined network elements leased from the incumbents. Obviously, this looks a lot like resale, because the incumbent continues to provide telephone service on its own network. Nevertheless, the platform does allow possibilities for innovation by utilizing the functionality of the incumbents' switches in new and different ways.

AT&T and MCI and other long distance companies would very much like to provide local service using the platform of network elements and avoid paying access fees to the incumbent local telephone companies.

Not surprising, the local telephone companies are adamantly opposed to the idea. The local companies refer to the platform as resale with a deeper discount. The deeper discount mainly refers to the different treatment of access.

The FCC issued its Local Competition Order in August 1996, six months after the passage of the Act. One of the requirements of the order is that incumbent local exchange carriers must lease to their competitors any combination of network elements, including the complete platform, at prices equal to forward looking economic cost.

The Commission also defined a methodology for calculating these prices based on total element long run incremental cost, or TELRIC. The TELRIC methodology first calculates the present cost of building a modern network taking as given the existing locations of the incumbent's central offices. This is sometimes called a "scorched node" model. The methodology then allocates the cost of the network to the different network elements, and finally converts these network element costs into lease rates by taking appropriate account of the opportunity cost of capital and economic depreciation.

The 8^th Circuit Court in St. Louis vacated the FCC's TELRIC pricing rules on jurisdictional grounds. The Court further determined that, while the Act requires incumbent local telephone companies to provide network elements in a manner that enables competitors to combine them, incumbents are not obliged to provide combinations of network elements. The Court further held that incumbents could disconnect network elements that were already combined and then charge a competitor for recombining them.

The FCC has appealed the 8^th Circuit decisions to the Supreme Court. Pending the Supreme Courts' decision in the coming year, jurisdictional authority for pricing network elements resides with state regulators, and the degree to which incumbents can make it costly for competitors to combine elements, or levy "glue charges" for combinations, is a matter of much dispute.

Well, that's a lot of background. I now want to turn to my interpretation of the economic reasoning behind the vacated FCC rules, and then turn to the implications of the 8^th Circuit rulings for consumer welfare.

The following are three statements about the efficiency properties of TELRIC pricing of combinations of network elements. First, TELRIC pricing creates conditions for efficient retail competition. Second, TELRIC pricing encourages entrants to make efficient facilities investments. Third, TELRIC pricing encourages an efficient utilization of the incumbents network. These three statements have a reassuring rhetorical glow, but what precisely do the statements mean?

TELRIC pricing of unbundled network elements is intended to simulate the competitive conditions of neoclassical investment theory. The logic of the FCC rules is to require an incumbent local telephone company to share the economies of its network by leasing elements to competitors at rates that approximate what economists' call the incumbent's "user cost of capital."

The user cost of capital is simply the opportunity cost of investing in capital today rather than waiting and investing tomorrow. This opportunity cost has three components. First, the investor loses the return that could be earned by investing the capital funds elsewhere. Second, the asset may depreciate with physical use. Third, the investor forgoes a possible lower price of the capital good tomorrow due to improvements in productivity. Therefore, properly designed TELRIC prices reflect these opportunity costs.

If TELRIC prices approximate the incumbent's user cost of capital, then the efficiency claims for TELRIC pricing derive from well known propositions about profit-maximizing behavior in competitive markets.

Proposition 1. If firms lack market power, then competition will equate the retail price of a service to the long run incremental cost of providing the service.

A firm lacks market power if it cannot significantly raise price by restricting output. Such a price-taking firm will find it profitable to expand its production if the retail price of a service exceeds the incremental cost of producing the service, and contract its production if the opposite is true.

If competitors lease the platform network elements at the incumbent's user cost, then the competitors will enjoy the same cost economies as the incumbent. Competition will drive the retail price of service to equal the cost of providing the service under the platform. This cost equals the long run incremental cost of service on the incumbent's network. And pricing at long run incremental cost maximizes economic efficiency.

Proposition 2. If firms lack market power, then profit-maximizing competitors will invest in their own facilities if the user cost of those facilities is less than the rental cost of facilities leased from the incumbent.

The drive to minimize costs means that a competitor will lease elements from incumbents as long as this is cheaper than investing in its own facilities. In making this decision, the competitor will compare its own user cost of capital with the price of leasing network elements form the incumbent.

Proposition 3. If firms lack market power, then profit-maximizing competitors who lease capital goods from the incumbent will do so until their marginal revenue products equals the rental rate of capital.

This is a sample of what lawyers at the FCC call this econo-jargon. The marginal revenue product is the additional revenue the competitor can earn with additional capital. Profit maximizing firms will lease an input up to the point where the marginal revenue product equals the rental rate for the input. This rule implies that inputs are used in an efficient, cost minimizing way.

You will have noticed that these propositions all begin with the phrase: "If firms lack market power..." and follow with a statement of some efficiency property. This phrasing indicates the central policy purpose of the vacated FCC local competition rules: to establish competitive conditions in retail markets by reducing barriers to entry and eliminating the retail market power of incumbent local telephone companies.

In my opinion, the FCC policy is based on pretty good economics. The neoclassical theory of investment does not apply exactly to telecommunications markets but it is a good starting point.

For example, it might be efficient to expand telecommunications networks in discrete lumps, not in continuous increments, as the neoclassical theory assumes. Since discrete capacity expansions are likely to result in excess capacity for a period of time, rigid TELRIC prices generally will not result in a perfectly efficient utilization of the incumbent's network. The prices will be too high when there is excess capacity and too low when capacity is congested.

So TELRIC pricing is not perfectly efficient, and there are some issues that state regulatory commissions need to sort out in applying the methodology. But, nevertheless, the TELRIC methodology properly applied is procompetitive because it lowers entry barriers by enabling competitors to lease network elements on substantially the same terms that the incumbent provides these elements to itself.

In contrast, 8^th Circuit decision on combinations allows incumbent local telephone companies to raise rivals' costs and maintain barriers to entry by forcing competitors to incur unnecessary costs for combining network elements or by setting high "glue charges" for providing elements in combination.

For example, many incumbents have taken the position that the only way a competitor is allowed to combine a loop and switch is by collocating in the central office. Thus, the incumbent would disconnect the loop from its switch, require the competitor to rent expensive collocation space, build a cage enclosing the space, cross connect the loop to a frame in the collocation cage, and then cross connect the loop back from the frame to the switch. All of this expense to disconnect and reconnect the same loop and switch is pure waste. This proposed method of combining the loop and the switch can have no other purpose than to raise rivals costs and foreclose entry.

A possible policy rationale behind the 8^th Circuit's interpretation is that increasing the cost of combined network elements would encourage true facilities competition, and advance the goal of deregulation. I think this is only half true.

The transition to competition is likely to occur differently in different local markets and sub-markets. High volume business markets already have seen a significant and quickening development of facilities-based competition. There are also some stirring of facilities-based local competition in residential markets. New entrepreneurial companies are building state-of-the-art networks to serve high density residential areas, and others are providing wireless service to apartment buildings.

However, in most markets, and especially in lower density residential and small business markets, we have seen little competition develop so far, and the transition to competition is likely to occur in two stages if it is to occur at all. The first stage would feature competition for retail customers based on resale or the platform of network elements leased from the incumbent. Once new entrants succeed in gaining retail market share, the second stage would see a deepening of facilities competition as carriers seek the best means to serve their customers.

The 8^th Circuit rulings raise the cost of the first stage of this two stage entry strategy. This could discourage entry into local markets, thus slowing and perhaps frustrating the transition to competition intended by the 1996 Act. Thus, if the Supreme Court upholds the 8^th Circuit, many residential and small business communities are likely to be deprived of the benefits of competition.

Universal Service

I can't help but come back to universal service. The tension between universal service and competition is the great drama in the Telecom Act. These are like two horseshoe magnets, that, when held face to face, repel each other. Yet, there is an abiding belief that, if one could just turn one of the magnets upside down, and look at it differently, everything would be alright.

In a recent book, entitled Universal Service, Milton Mueller explains the historical metamorphosis of the meaning of universal service. The phrase was invented by AT&T in 1907 as a justification for an integrated, regulated monopoly. AT&T's slogan was "one system, one policy, universal service." AT&T's version of universal service policy provided a justification for its consolidation of the telephone industry. This consolidation resulted in the system of regulated monopoly I described earlier.

Eventually, universal service came to mean high telephone penetration. Today, this policy goal largely has been achieved. Household telephone penetration in the U.S. is about 94%. Even in states like Arkansas, Mississippi, and New Mexico, where telephone penetration is the lowest, the penetration rate is in the neighborhood of 87 or 88%.

Universal service as a policy goal is traced to Communications Act of 1934 whose stated purpose is

to make available, as far as possible, to all the people of the United States, a rapid, efficient, Nation-wide, and world-wide wire and radio communication service with adequate facilities at reasonable charges.

Today, universal service has come to mean cheap basic telephone service for all residential consumers. There's that egalitarianism again. Cheap rates in high-cost residential markets are supported by implicit cross subsidies in the structure of regulated prices. The litany is familiar. Business rates subsidize residential rates. Urban rates subsidize rural rates. Long distance rates subsidize local rates.

The long distance-local subsidy seems self-defeating. After all, the ability to make long distance calls is one of the benefits of basic local service. Thus, raising the price of long distance calls to lower the price of local service could make local service less valuable to consumers. Indeed, Professor Jerry Hausman of MIT argues, based on an empirical estimation of demand elasticities, that eliminating the long distance-local cross subsidy actually would increase telephone penetration.

The urban-rural subsidy seems even more perverse. Low income households living in cities end up subsidizing high income households in rural areas. The Act does say that telephone rates should be "affordable", but, surely what is affordable to a consumer depends on the consumer's income. I doubt anyone seriously disagrees with this principle. It makes no sense to subsidize rich people just because they live on farms.

The Telecom Act also says charges should be "reasonable." Does "reasonable" mean cheap? Not really. Supreme Court, long ago, in interpreting the Sherman Act's prohibition against price fixing, argued that the only reasonable price was a competitive price. In a competitive market, higher costs mean higher prices.

A much more efficient way to maintain high telephone penetration is with targeted subsidies. Indeed, the FCC already administers means-tested subsidy program. These programs are called Life-line and Link-up. The FCC could build on these program to means test high-cost support.

Here's how it might work.

The FCC is currently examining models to estimate the forward looking economic cost of basic telephone service in different areas of the country. Beginning in 1999, the FCC plans to use these estimates to calculate a high cost subsidy on a per line basis.

The plan is to compare the cost per line in a geographic area to a national "revenue benchmark." The revenue benchmark is supposed to represent something like the amount of revenue a local telephone company can expect to earn from an average customer. If the cost per line exceeds the revenue benchmark, then the a carrier serving those customers is to receive a share of the difference as a subsidy from the interstate jurisdiction for which the FCC is responsible.

But instead of making the same high-cost subsidy available to everyone in a geographic area, the FCC could target the subsidy to consumers who actually need it to afford telephone service. For example, it could provide a greater subsidy to households that qualify for Life-line or Link-up. A simple way to do this would be to set a lower revenue benchmark for Life-line and Link-up customers and a higher benchmark for everyone else. Makes sense?

Let's look at the problem from another angle. I mentioned that Arkansas, Mississippi, and New Mexico have among the lowest telephone penetration rates in the nation, about 87 or 88%. And there surely are places in these states and elsewhere where telephone penetration is substantially lower. For example, the penetration rate for Native Americans on many reservation lands is embarrassingly low.

Something should be done about this. But lowering the price of everyone's telephone service would be a very expensive way to increase telephone penetration.

According to Hausman's elasticity estimate, if the price of basic service were increased by 40%, 85% or so of households in Arkansas would still keep their telephone service. Subsidizing this 85% does nothing to promote higher telephone penetration in the state. The subsidies would be more effective if targeted to the other 15% who do not have telephone service or whose subscription decision is sensitive to the price increase.

This could be done by targeting high-cost subsidies to the geographic areas that actually have low penetration rates. This could be implemented by calculating per line subsidies using a lower revenue benchmark for communities with low penetration rates, and a higher revenue benchmark for elsewhere.

I am not necessarily suggesting that the states receive less universal service support from the federal jurisdiction. I am only suggesting that universal service support would be better spent if it were targeted to achieve the universal service goals of the 1934 Act.

The Internet

Let me change the subject one more time and talk about the internet. Its not really a change of subject, because I want to talk about how the internet fits into the universal service conundrum.

In 1998, The FCC began, in effect, taxing the internet to subsidize the high cost of traditional telephone service. The tax is small, but the economics of the policy are questionable.

The FCC currently has mechanisms in place that subsidize basic telephone service in high cost areas to the tune of about $1.7 billion. These subsidies are financed by "mandatory contributions" that are a percentage of interstate telecommunications revenues.

The FCC has declared that revenues from some internet transmission services are counted as interstate telecommunications revenues, and thus are subject to these mandatory contributions.

In the past 100 years, national telephone penetration of households has gone from next to nothing to about 94%. But in the past 20 years telephone penetration has been almost flat. In contrast, in the past three years or so, the number of households with internet connections has grown from next to nothing to a bit over twenty percent. In effect, the FCC is taxing a growing technology to subsidize an established one.

In 1909, telephone penetration was about the same as today's internet penetration. At that time it made sense to subsidize telephone penetration because of the "network externality principle." The network externality principle states that a service become more valuable to any consumer, the more other consumers subscribe to that service. This is easy to understand for telephone service. The more consumers connected to the network, the more people you can call. According to the network externality principle, it made sense in 1908 to subsidize telephone subscription in order to stimulate penetration and increase the value of the service.

AT&T clearly recognized the network externality principle as an intellectual justification for its version of universal service. In its 1908 annual report, AT&T stated the following:

A telephone without a connection at the other end of the line is ... one of the most useless things in the world. Its value depends on the connections with other telephones - and increases with the number of connections.

Cheap telephone prices probably did make sense in 1934 when the Communications Act was passed because of the network externality principle. Telephone penetration appeared to be stuck a low level. Penetration had risen to above 40% a few years earlier, but had plummeted to just above 30% in the wake of the Great Depression. In this situation, cheap telephone service might stimulate penetration and realize network externalities. But it is a dubious proposition that subsidies of today's magnitude are necessary to sustain high telephone penetration.

The network externality principle also works in reverse. Since almost everyone is connected to the telephone network, a telephone is something too valuable not to have. The empirical evidence bears this out. To repeat, the estimated price elasticity of demand for basic telephone service is exceedingly low.

The Internet is the new technology on the block. The internet provides a platform for myriad new information services. It is potentially a powerful growth engine for the economy. Holding back the internet is probably not a good idea.

The network externality principle applies once again, this time to the internet. The more consumers connected to the internet, the more valuable will be the platform to information content providers. And the more information content to be found on the internet, the more valuable is an internet connection to all consumers.

Thus, the network externality principle provides a good policy rationale for subsidizing the internet, to keep down the price of internet access, increase participation, and realize network externalities. From an economic perspective, taxing the internet to pay for a mature traditional service, even by a small amount, is exactly backwards.

Conclusion

With the Telecom Act of 1996, Congress ushered in a new era of competition in telecommunications markets. The politicians, the public, and, yes, even we practitioners of the dismal science, are hoping for a golden age in which all consumers have the luxury of choosing from an array of cheaper, better, and more various services, and an array of independent providers competing their brains out to satisfy consumers' wants. That's the vision. The problem is how to get there.

Beginning Monday, my very able successor, Bill Rogerson, from Northwestern University, will be advising the FCC on the economics of a wide range of difficult policy issues: local competition, access reform, universal service, the internet, and much more. I wish him patience, and hope he enjoys the rides every bit as much as I have.

Thank you.