The Case for Public Subsidy of the Internet

by Sandra Schickele

THE ECONOMIC CASE FOR PUBLIC SUBSIDY OF THE INTERNET

I THE THEORY OF PRICE AND PRIVATIZATION OF THE INTERNET

It has been argued that little or no case for public subsidy of the Internet 1 exists and that the free market can produce an adequate supply of network services at a price to be determined by market forces. This argument depends on a misunderstanding of the way in which markets allocate resources. It is the purpose of this paper to clear up that misunderstanding and present the model underlying the argument in historical context so that it can be explicity referred to in discussion of the economics of the Internet.

The ongoing debate about privatization of the Internet involves implicit allusion to the neo-classical model of price determination in free markets. This same model is called by economists the "theory of price", or "price theory". Its conclusions are referred to in literature, the press, election campaigns, and private conversation without any reference to the model's origins, assumptions, or limitations. Any person or group wishing to influence the direction of economic policy of the United States Government must marshall arguments based on the price theory model, or at least be able to counter arguments advanced by individuals claiming to possess complete command of its meaning.

This paper presents the neo-classical model of price determination in a context which stresses its historical development as a model. The model's assumptions and limitations and their impact on economic analysis of Internet funding or pricing are spelled out. The neo-classical model of price determination asserts that private ownership of industry coupled with the sale of goods and services at prices determined by unregulated market pricing mechanisms will, under certain circumstances and subject to certain limitations, result in an economically efficient and socially optimal allocation of scarce economic resources.

A model is only as useful in practice as its assumptions permit. Meaningful prediction or interpretation of real world outcomes depends on correct understanding of premises, on knowledge of any inherent limitations, and on formation of a judgment as to whether or not the model's stated preconditions are met in a particular circumstance.

The conclusion of this paper is that the assumptions which must hold if the free market is to be efficient are fundamentally violated by the nature of the Internet and any likely successor to it, and that market prices cannot by themselves efficiently allocate resources for the production and use of the Internet.

The paper is divided into five sections. The first describes parts of the theory of price and outlines the rest of the paper.

The second section lists and defines several of the assumptions that must hold in order for the efficient allocation of resources hypothesized by price theory to occur. In this section of the paper the price theory model is taken as a given and the questions addressed are: on what assumptions is the model based; when do these assumptions hold; do they hold for the Internet?

The third section of the paper considers fundamental limitations of the price theory or free market model itself. Basic problems associated with relying on the model as a guide to public decision making in an environment of rapid change are described. Problems inherent in the nature of the price or market system are shown to affect analysis of potential Internet funding mechanisms.

The fourth section turns to an analysis of some special characteristics of the Internet which cause it to violate several of the assumptions necessary to the efficient functioning of free markets as described earlier in part two. This finding leads to the conclusion that in the absence of a subsidy the private market is not an adequate producer of public network services. The fundamental limitations of the entire price theory model discussed in section three also affect analysis of Internet funding opening the possibility that prices determined by the private market may lead to a socially undesirable outcome.

In section five the outcome of the analysis overall -- that it is not possible for the private market alone to create an efficient or socially optimal production and distribution of network services of the type represented by the Internet -- is presented. It is the case that whatever allocation mechanism eventually is used to distribute network services, primary reliance on market prices will not generate an economically desirable outcome. II CONDITIONS NECESSARY FOR EFFICIENT MARKET PRICES

Briefly stated, the theory of price holds in part that:

The privately owned firms which make up an industry will, under competition, sell their goods at whatever price the market will bear. In a competitive market that price will ultimately in equilibrium just be equal to the marginal cost the producer incurred in the process of production and distribution.2 Firms will not have any control over the price they receive. It is assumed that the firm's only real decision will be whether to sell or not to sell at the price established by impersonal market forces. A firm can stay in business or shut down. The market does not care or notice. If individual firms leave, their leaving will not have any effect on the price at which the good is sold, because there are many producers and no one of them has enough power to control market price. It is assumed that prices and wages are perfectly flexible. Market prices will be simultaneously established in all industries; and the factors of production (labor, capital, and raw materials) will flow into those industries whose goods are in greatest demand where the labor, capital, and raw materials can earn the highest wages, interest, and rent. An economically efficient and socially desirable allocation of resources will result.

The chief benefit and major achievement of the price or free market system in an environment of competitive unregulated markets is said to be allocative efficiency. In other words the market allocates the scarce resources available in such a way that an equilibrium is reached in which no one can be made better off without some one else being made worse off (the condition called Pareto optimality is attained).3

A particular good, such as the Internet, can only be efficiently produced through the free market, according to the theory of price, if the good's production and consumption meet the first nine conditions listed below. If the production or consumption of the good does not meet these conditions the good cannot be efficiently produced through a free unregulated market.

Some of the preconditions which must exist in order for allocative efficiency to be said to occur in a single industry and in the economy overall are the following:

1. Perfect competition exists in all product and factor4 markets. No seller or buyer is large enough to have an effect on market price, even if he leaves the industry.5

2. Perfect information (complete, accurate, and freely available) on relevant prices and characteristics of products and factors is obtainable by all buyers and sellers of goods and factors.6

3. Perfect mobility of resources exists. Labor, capital, and raw materials flow unimpeded to their highest paid use.7

4. No externalities (positive or negative) are present in the production and consumption of goods. The production and consumption of a good has no effect on any individual or group except the actual producer and consumer. No effects fall on third parties.8

5. No public goods can exist.9 National defense and clean air are examples of pure public goods. That is, if they exist, they are available to everyone in the society. In order for the free market to function properly the product being produced and sold must have the "exclusion property", meaning that everyone but the buyer of the good must be able to be excluded from its benefits. Some goods have the characteristics partly of private goods (no effects or spillovers fall on third parties in the case of a pure private good) and partly of public goods. They have sometimes been called quasi-public goods. It is possible for the free market to produce quasi-public goods to a limited extent, but not at an appropriately high output level. Education is an example of a quasi-public good, and it is the contention of this paper that the Internet is the same type of good.

6. No direct interdependence exists in the consumption or production of any good. Consumer A's decision to consume a good is not directly interdependent on Consumer B's decision to consume the same type of good or any other good. Producer C's decision to produce a certain amount of a good is not interdependent with producer D's decision to produce a certain amount of the same or any other good.10

7. No significant economies of scale can exist. Costs faced by large producers are not significantly lower per unit of output than costs faced by small producers. Stated somewhat differently, this assumption implies there are increasing costs in all industries at a relevant level of production. If decreasing costs existed, the first producer would find that his costs fell continually as output increased, and a condition of natural monopoly would exist, precluding meaningful competition.11

8. Indivisibilities in production must be small. All factors (such as capital investment) must be able to be varied by small amounts. There must be no factor which has to be used in large indivisible doses.12

9. Transaction costs must not be so great as to cause market failure. Transaction costs are the costs incurred in enforcing property rights, finding customers and suppliers, and actually carrying out a transaction. They are the costs associated with exchanging rather than producing goods and services. If potential transaction costs, notably the costs of enforcing property rights, are high, it may not pay private producers to create a market in the good in question. Private markets can be established only for goods for which property rights can be guaranteed and easily exchanged.If there are goods which the society desires to have but which have large transactions costs, these goods will of necessity have to be produced outside the free market framework.13

10. The prevailing distribution of income must be accepted as a given in the analysis.14 Efficiency criteria in economics are dependent on the assumption that the pre-existing distribution of income will not change. Changes in the distribution of income will affect the efficiency of a given set of prices and a given allocation of resources. In other words, a particular set of market determined prices can only be said to be efficient for one unique distribution of income. The same prices may become economically inefficient if or when the distribution of income changes.

Few economists would assert that these conditions are completely met in the modern industrial market system. If any of the first nine conditions listed above do not hold a condition called market failure exists.15 Market failure is said to occur when the choices made by buyers or sellers in a market result in an inefficient outcome. Neo-classical economic theory postulates that the appropriate time and place for government intervention16 in the market is when the market fails.

In section four below it is argued that the Internet as a good violates conditions 1,4,5 (in part), 6 and 7, and may violate conditions 8 and 9 as well.

III FUNDAMENTAL LIMITATIONS OF THE FREE MARKET MODEL

There are important limitations on the entire neo-classical theory of price which must be understood before attempts are made to use it to analyze or justify any particular economic analysis or policy. The limitations most important for our purposes are:

1. Serious questions have been raised about the effectiveness with which the market mechanism allocates resources between the present and the future. In theory long and short term interest rates are the prices which allocate resources between present and future--between short and long term investment. Yet the market may devote too many resources to current consumption. The future gets short shrift in the market economy. Individual investment projects represent a much greater risk to the individual investor than to the society collectively. Individual investors may choose not to undertake risky but potentially very valuable projects. If citizens want more resources devoted to long term investment and the creation of future wealth for the society than private markets and private individual behavior would provide, they must correct the market's shortcomings through the ballot box.17 Interest rates as allocation mechanisms are also flawed because government policy manipulates interest rates frequently without regard to the effect on the allocation of resources.

2. Some economists contend that the market model has no theory of technological change, no way to account for it or create it.18

3. The socially optimal allocation of scarce resources called Pareto optimality depends on a static economic equilibrium with an economic pie of fixed size. Price theory has historically been based on static analysis, not allowing for economic growth or change in the distribution of income.19

4. Modern economists have worked long and hard to incorporate the effect of uncertainty of outcomes into economic analysis, but it is very difficult to do so. Analysis of uncertainty in modern economic theory tends to depend on the assumption that a producer or consumer can buy insurance against any uncertainty.20In other words, uncertainty is essentially assumed away. The greater the uncertainty associated with future outcomes the more unjustified it is to base economic policy on the belief that the price system will provide an efficient and socially desirable result.

5. As mentioned above, price theory assumes that the initial distribution of income is given. Price or free market theory applies equally well in a society in which the top one percent of income recipients have ninety nine percent of total income, and a society in which income is distributed completely evenly. Resource allocation through the market is equally efficient with any given income distribution, yet market prices congruent with the pre-existing income distribution may at times provide an unacceptable outcome in a democracy. Each unique income distribution generates a different price structure. A given set of prices is only efficient for one given distribution of income. Price theory has nothing to say about the desirability of any particular income distribution. Decisions about redistribution of income are considered by economists to be entirely political in nature. This point should be remembered in any serious analysis of the economic impact of the Internet, because any pricing mechanism which restricts access of the general public to a national network may only exacerbate the widening polarization in the distribution of income.That is, the distribution of goods through the market mechanism of price simply excludes low income individuals from the market.

6. The definition of efficiency in economics is different in kind from the definition of efficiency in engineering. Efficiency in economics means efficiency in the allocation of resources based on the concept of Pareto optimality which says an economic system is efficient when no one can be made better off without making someone else worse off. Some might describe this as a subjective definition of efficiency. Scientists and many others have at times been misled by assuming that some objectively verifiable definition of efficiency is involved in economic analysis. That is not the case.

7. The theory of price only describes one of several methods of rationing goods and services. Price is a system of resource allocation. Price is a rationing device. As a rationing device it has turned out to be much more useful most of the time than direct rationing with ration cards or other methods. It is certainly an easier system to manage. But it is not the only rationing device ever invented. Other things being equal, it is the most efficient. Where externalities, market power, incomplete information, lack of mobility of resources, direct interdependence in consumption or production, public goods, indivisibility in factor use, large economies of scale, or large transaction costs exist, other things are not equal and market prices cannot efficiently ration the good in question.

All the limitations and complications of the free market or price theory model of production and distribution listed above impact on the Internet (or the NREN) and any future national network as well.

IV THE INTERNET AND THE MARKET

In this section preconditions for the efficient functioning of the private market from section two and potential limitations of the market model from section three above are examined with regard to how they affect potential Internet funding mechanisms.

In the case of the Internet the market fails. Market failure is created by the existence of large positive externalities, possible significant economies of scale, the fact that the Internet is a quasi-public good, the presence of some degree of market power among producers and consumers as well, interdependence in production and consumption, possible indivisibilities, and possible high transactions costs.

A. Marginal Cost Pricing and the Internet

Is the marginal cost pricing which is required for an efficient market solution likely to prevail in a private market for Internet access and services?

The reality of business pricing activities is quite different from the model. In practice, whether a company is recovering its marginal cost of production or not will simply determine whether it is making enough to stay in business. Many large-scale U.S. industries are characterized by prices in excess of the marginal cost of production. The very existence of the marketing and advertising industries, of price discrimination (special "sale" prices, software "upgrade" prices, etc.), and of business schools which teach thousands of potential businesspeople how to establish appropriate "mark-ups" on their products indicates that U.S. industry does not always sell its products at the marginal cost of production.

To the extent that competition has in fact prevailed in the electronics industry, the individual firms in that industry experience great insecurity. This very uncertainty and the price flexibility associated with it are indicators that segments of the electronic industry currently exist in a classic competitive situation. There may however be economies of scale in production of Internet related goods or services. If that is the case, then there is likely to be a resulting "shakeout," among the firms in the industry, with the survivors able to price products well above marginal cost. These same survivors will then dominate the market until another new round of technological development threatens their hold or sweeps them away. Many years ago the economic historian Joseph Schumpeter characterized the periodic bouts of fierce development in capitalism as the process of "creative destruction."21

It is also the case that there is a wide disparity in market power among the different producers and consumers in the industry for Internet related services and goods. Ultimately, the disparity in market power among the different sub-groups involved is likely to affect the level of market prices.

The potential exists for the development of a classic oligopoly 22 in unregulated long distance carrier service. Although the price setting behavior of firms in an industry dominated by a few large producers may vary, oligopoly prices are often determined by the behavior of a "price leader." It is well established in price theory that conscious collusion is not necessary to maximize price in an oligopoly industry.23 One company, usually the largest, announces its prices, and the others follow suit. When the members of an oligopoly group disagree among themselves a price war results. This phenomenon can be seen in the behavior of the deregulated airline industry. A more successful oligopoly price strategy, however, is the price leadership method of price determination. Market prices in an oligopoly environment are not usually determined by marginal cost, or even by average cost of production. A typical oligopoly industry is characterized by prices well above marginal cost.

There is a potential mitigating factor which can influence oligopoly behavior. The theory of contestable markets24 hypothesizes that if entry into and exit from the industry are costless and unimpeded, it may be possible for small firms to enter the market and pull away some of the market share of the large firms if prices set by the oligopoly are far above average cost. The theory is that the possible entrance of smaller firms will act to discipline oligopoly behavior.

There are industry sub-groups involved in production for the Internet to whom the theory of contestable markets probably applies. It seems doubtful that long distance carrier service falls into that category, however.

There are many other companies and sub-groups of producers involved in the Internet "industry" who do not possess market power. Their situation is sufficiently different from the largest companies involved that the two groups should always be distinguished from each other for analytical purposes.

There is also a wide disparity of market power among Internet "consumers." The federal government and the various state governments, among others, have market power. They can negotiate for prices.

The reality is that an attempt to produce Internet network services through the free market will result in negotiated prices based on relative market power for the largest and most powerful participants in the market. Other producers and consumers in the market will for the most part merely experience the end result of the negotiated outcomes.This paper will argue that if access to a future Internet is determined by the ability to pay market prices, most users will almost certainly not be paying a price equal to the marginal cost of their use. They will be paying what the market will bear in a situation not characterized by competitive equilibrium. They will pay an amount which may be well above marginal cost. A market price substantially above marginal cost of production is one indicator that market failure has occurred.

B. Social Benefit, Externalities, and the Internet

To give one simple concrete example of positive externalities associated with access to the Internet: if teachers at all levels of education have network access, they have the ability to train students in network use. When the students graduate they become workers. If it is desirable to future employers that these workers possess familiarity with network use, then the employers benefit from the students' education in network skills. The employers are third parties who receive positive externalities (or spillovers) from the students' education.

The benefits of basic research are not appropriable to a single producer and basic research itself is not susceptible to production through the market mechanism. Any social institutions which contribute to basic research activities are generating positive externalities by doing so. Any system, such as the Internet, which helps such institutions pursue basic research development, is generating positive externalities in the process.

It seems likely that the development of "virtual communities" in the form of highly specialized sub-groups which regularly "meet" and "talk" through the Internet contributes to the creativity of the groups in question. Such groups are capable of a division of labor and a specialization of skills beyond any which have ever existed. There is no reason to believe that useful specialization has lost its capacity to increase economic output since it was described by Adam Smith in his report on the contribution of the increased division of labor to the productivity of a pin factory in An Inquiry into the Nature and Causes of the Wealth of Nations.

Alan McAdams of Cornell University believes that all who use the Internet, and those who may use it or its successor in the future, benefit from (receive positive externalities from) the contributions made to its usefulness by other persons on the Internet. In other words, a substantial group of the (usually heavily subsidized) current users are creating positive externalities through their creative contributions to the net.

An initial very rough approximation of the size of the social benefit (and some indication of the positive externalities) being generated by the Internet today may be calculated by setting a dollar value on the time which individuals are devoting to the network. Thousands of individuals who are not paid to maintain the network nonetheless devote tens of thousands of hours to using it. If we assume they are rational consumers, we must assume they set a high value on its usefulness. Since most of the non-student adults using the net probably earn the average wage for professionals, technicians, and managers, their time is worth $20-$50 per hour (or more).Thousands of students and graduate students also spend time voluntarily on the net. If we value their time at only $4-$6 per hour, a dollar estimate of the value they set on the use of the net will still be considerable.

Even those who are using the net for recreational or purely social purposes are indicating by their use of their time that social benefit is being generated by the availability of the net. The dollar value of the social benefits attributable to the development and maintenance of public parks, for example, consists in part of estimating the recreation value of the park to the public. That value is considered to be in excess of any price set on the use of the parks.

In other words, even the recreational use of the net generates social benefit in the economic sense. In that case the benefits accrue to the immediate users and are not in the form of externalities falling on third parties, but since they cannot at the moment be measured in dollars of payment through the market, they must be added to our estimate of social benefit for purposes of cost/benefit analysis. Attempts to charge for purely recreational use of the net, as opposed to non-recreational use, are going to present very difficult problems.

It seems probable that some distinction in types of Internet use will eventually have to be made. Those uses generating primarily private benefits are better able to be produced and sold through a private market than those uses likely to generate positive externalities. It is not by chance that the anticipated successor to the Internet is called the NREN, or National Research and Education Network. Questions of just how much of high-tech industry should be included on such a network at minimal private cost are complex. The author's personal view is that the positive externalities associated with widespread low cost access to the Internet extend far beyond the traditional research and education community but that it will prove impossible in the present national economic circumstance to provide as large a subsidy as a national network really merits.

C. The Internet as a Quasi-Public Good

That the Internet is inherently a quasi-public good is demonstrated by the fact that production of Internet services and consumption of the Internet as a product have been and currently are being financed jointly by the public and the private sector. The Internet is like the education system, the health care system, the transportation system. The Internet has some of the characteristics of both a public and a private good. As a quasi-public good it will never be produced in sufficient quantity or be as generally available as true allocative efficiency would require unless it receives a generous enough subsidy to reach an optimum level of output and use. Since user fees have already been decided on as a financing mechanism, one can only reiterate that reducing or eliminating access for large groups of users in either the public or private sector may amount to killing the goose that lays the golden egg. This is the case if McAdams is correct in his assertion that the current rapid growth and generally conceded usefulness of the Internet are direct results of positive externalities created by the individuals who have used the Internet the most heavily, usually on a subsidized basis.

Today we have a national network infrastructure being jointly developed by private companies (who presumably are selling their goods at a profit), non-profit institutions including private universities, and federal and state government entities. The State of California pays for the portion of the Internet that the author uses -- CSUNET. The United States Government pays for part of the NSFNET backbone CSUNET is connected to, while numerous private companies provide access for other public and private sector participants. Both public and private sector users can readily communicate with other public and private sector users and companies through the network.

The Internet is clearly an example of a mixed-use, mixed-funding good: a good used and funded by the public and private sector jointly. If the current level of funding and subsidy to the U.S portion of the Internet from all parts of the public and non-profit sectors were to be totaled, it is quite possible that such spending makes up the largest share of spending on the net. Although it is the case that private companies now make up the largest number of attached network hosts on the NSFNET 25, the data about different hosts' expenditures and uses of the net are not complete enough to permit much analysis. The lack of availability of economic data about expenditures on the network by U.S governmental units at all levels, and by U.S. non-profit entities and businesses as well, makes generalizations difficult.

The fact that the Internet is by its very nature global in scope, and that its international characteristics are the source of much of its value, adds complexity to the attempt to formulate appropriate public policy in the U.S. concerning the Internet. Nonetheless, it is the responsibility of U.S. policymakers to consider first and foremost the economic and political impact on U.S. citizens of decisions about future funding and regulation of the network. For purposes of federal government decision making in the U.S., the Internet must be viewed as a U.S. based quasi-public good.

D. Other Sources of Possible Market Failure

Indivisibilities in investment can be seen in the fact that it took the U.S. Government to build the initial network backbone. Is it really the case that no further large scale investment is required? Can future investment proceed in small divisible steps? Doesn't eventual conversion to a completely fiber optic system require massive investment? If so, then this infrastructure cannot be built in small increments. It experiences indivisibilities in production. Are potential large transactions costs associated with any national electronic network? What will be the costs of attempting to enforce property rights in a system built on electronic documents? If property rights are going to be a nightmare to enforce on the Internet, then there are high transactions costs associated with the net as a good.

Interdependence in production of Internet goods and services is revealed in the announcement sent to various network mailing lists on January 5, 1993, concerning NSF network information services awards made by the National Science Foundation to AT&T, Network Solutions, and General Atomics. This announcement states that a consortium of private for-profit companies and non-profit organizations overseen by government entities will split the job of creating necessary products to increase the usefulness of the network by cooperating in the Network Information Services Managers Project called INTERNIC. Decisions to produce seem often to be interrelated on the net. Decisions to use the network (consume its services) are interrelated as well. The value of the Internet to most users (including the author) and a user's decision to "consume" it as a good are often directly dependent on who else uses the net and in what way.

Interdependence in production always exists by definition when an industry is an oligopoly.26 To the extent that portions of the Internet "industry" will consist of oligopoly groups, there is no doubt that there will be interdependence in the decision to produce. A case of market failure will arise.

The general problems with reliance on the market or price system listed in section three affect the Internet. The bias against long term investment displayed by markets is liable to make the market seriously deficient in allocating resources between the present and the future. Markets tend to allocate too much to the present, not enough to the future. A study of the history of infrastructure development reveals that governments have had to give large grants and subsidies to secure major economic infrastructure, or just build infrastructure directly. Heavy market discounting of risky long term future cash flows mitigates against long term private sector investment and has historically had to be compensated for by goverment investment. Privately owned firms are subject to fierce short term constraints. They need to show current profits to retain their market value. Important legal and tax factors also discourage private investment in expensive new plant and equipment. For example, the accelerated cost recovery schedule for depreciation built into current tax code does not in actuality do much to encourage long term investment. The term "accelerated" is viewed by some in the finance field as bordering on the humorous when applied to the current code which requires that plant and equipment be depreciated over a period of up to 10 years. At the time of this writing even equipment to be used in research and computer equipment must be depreciated over 3 years.

Markets experience great difficulty in dealing with rapid technological change. U.S. competitors in the EC27 and Japan cooperate with their governments to follow industrial policies developed outside the framework of free markets which may smooth economic and technological change for them.

Considerations pertaining to the distribution of income are very important with regard to pricing Internet services. Provision of network access solely on a private market basis will ultimately exacerbate the social tensions created by U.S. income inequality. Distribution of network services on a market basis may even contribute to an increase in inequalilty in the distribution of income, because market provided services will effectively screen out a huge portion of the population.

While all of the necessary conditions for efficient functioning of the private market are relevant to analysis of appropriate funding for the Internet, the requirement that there be "perfect information" access may perhaps be specially pertinent. The possibility exists that free or very low cost access to the Internet might increase efficiency in all sectors of the economy by ultimately creating the first society actually to approach a condition approximating perfect information availability. Note that without "perfect" or at least very high quality information the allocation of resources will by definition be inefficient and the market (any market without excellent access to relevant information) will fail to function properly. It is certainly the case that fast access to high quality information would increase the efficiency of the public sector as well as the private sector. A relatively new field in economics studies the importance of information availability for the efficiency of economic performance: information economics. Although the economic analysis of information necessary to achieve efficiency is oriented at this time toward the private sector, and treats information merely as an input to be purchased in the process of production of other goods, some of the findings of information economics shed light on the unique character of the Internet.

Information economists have struggled with some of the special characteristics of information in their efforts to incorporate it into economic analysis. They have found that information is not easy to define. The quantity of information obtainable from a particular action is not clearly defined. Information which is obtained is not homogeneous among its users. Forms of economically useful information are too varied to permit the kinds of price-quantity characteristics that economists use for evaluating markets for most goods. The technical properties of information as a good create further difficulties. Most information is durable and retains value after it has been used. Sometimes information itself has the characteristics of a pure public good. Information may be nonrival in that others may use it at zero cost and it may be nonexclusive in that no individual can prevent others from using it. The technical qualities of information imply that market mechanisms often operate imperfectly in allocating resources to information provision and acquisition.28

These findings of information economics are relevant to analysis of funding mechanisms for the Internet because the Internet is a system for distributing information. Yet information as a good has unique characteristics which cause it not to be amenable to simple market based production, distribution, or sale.

V CONCLUSION

The Internet is a system of many products, uses, and suppliers. Yet it is a whole. There may be, and probably will be, products sold through it. It is itself the connection mechanism through which information and, potentially, goods (such as on-line books) will flow.

The Internet has something basic in common with the electric power industry, the highway system, the education system, and the health care system. None of these systems has been created without subsidy or managed without government regulation. They are all quasi-public goods.The Internet has some of the characteristics of both a public and a private good. As a quasi-public good it will never be produced in sufficient quantity or be as generally available as true allocative efficiency would require unless it receives a generous enough subsidy to reach an optimum level of output and use.

Public sector providers of information on the net, including universities, are considering charging for access to such services as on-line catalogs. If this trend spreads, as seems likely because the public sector is in terrible financial straits, widespread use of the national network will slow to a crawl in the absence of complete federal or state subsidy. This stagnation will occur not because the information is not valuable, but because other public sector users simply do not have the funding to permit use.

In order for public investment to be socially desirable it must be the case that the discounted present value of the benefits of the investment, to whomsoever they may occur, should exceed the discounted present value of the costs, to whomsoever they may occur. Costs and benefits may and often do fall heavily on separate groups and be subject to great uncertainty. The opportunity cost of public funds is the (properly calculated) return they could earn in their next most productive use. Estimates of the benefit/cost ratio for public investment may use too high a discount or interest rate, and fail to value or weigh future benefits properly against more easily measured present dollar costs.Such flawed analyses have worked against public investment at times, and need to be guarded against. There have also been cost/benefit studies done by public agencies in the past which have erred in the opposite direction.

A literature and a history of past practice exist which permit a reasonable analysis of benefits and costs. Estimation of a genuinely appropriate discount rate taking into consideration some of the factors mentioned in this paper, as well as inclusion of all relevant externalities, are the keys to producing a useful analysis. In a highly politicized atmosphere in which powerful vested interests are competing for advantage, it is difficult to find support for an objective examination of the alternatives.

Public subsidy of the Internet is being discussed in a context and in a country which is historically suspicious of publicly financed infrastructure if any profit is to be made, and if the private sector is interested in owning the infrastructure.

Several areas need more analysis. 1)If price is going to be used as a rationing mechanism a more carefully thought out pricing policy is needed. 2)Rationing mechanisms other than price deserve careful examination, including the possibility of setting priorities in message transmission if necessary. 3) Problems associated with provision of commercial products on a publically subsidized network are still unresolved. 4) A study of the size and nature of possible government subsidy of network users, either public or private, by a grant or voucher system, needs to be done. 4) A study of the implications and costs of private versus public ownership of a backbone network has never really been done.

At this point the economic data needed to make sensible statements about what is being spent on the existing structure, by whom, and from what source, do not exist. The study of a systematically collected sample of material on expenditures and subsidies, direct and indirect, would provide very useful information. Such a study should include data gathered from local, state, regional, and national public and private users and providers. An appropriate study could be carried out by the General Accounting Office, a special workgroup in the Office of Science and Technology Policy, or some other agency. The design and collection of appropriate economic data is an urgent need.

Those who know the Internet well are in the best position to provide information which will make it possible to judge the Internet's potential as a force in economic development, estimate the size of the externalities it can create, the economies of scale associated with it, the degree of interdependence involved, the size of indivisibilities in production and consumption, and the nature and size of transactions costs. Economists can provide economic data collection and analysis stills and analytic tools to help in the decision making process.

The Internet and any likely successor to it constitute, among other things, the public library system of the future. The Internet will be the repository of all public information: the public information gathered with by the use of taxpayer money. It may not be too extreme to say that there cannot be a democracy in the United States if the public does not have almost unlimited access to the public infomation that is on the Internet now and will be there in the future. Other countries will eventually be in the same situation.

Within the past few months increased pressure has been brought in the U.S. Congress to require expanded public electronic access to government databases currently sold to private information providers, who then resell them to the public at a fee.29 Public pressure and public scrutiny are not going to go away. Decisions about funding and control of quasi-public goods invariably cause profound controversy in our economic system.

Can the Internet as a system be produced and maintained by the private sector? The application of price theory to the Internet leads to one principle analytical conclusion: market prices cannot by themselves efficiently allocate resources for the production and use of the Internet. In order for an appropriate level of production and use to occur, direct public investment and/or user subsidies are necessary.

ENDNOTES1 The term "Internet" in this paper refers to the system of networks and gateways using the TCP/IP protocol suite which function as a single, cooperative virtual network. Although the Internet is by its nature international in scope, the U.S. network system can be spoken of separately for purposes of analysis. The U.S. network system and possible future variants of it are also referred to here as the net and the national network.2 In the case of the Internet the marginal cost would be equal to the incremental or extra cost of carrying one more message.3 MICROECONOMIC THEORY: BASIC PRINCIPLES AND EXTENSIONS, 5th edition. Walter Nicholson, author. Published by Dryden Press, 1992. See page 224.4 The factor markets are the markets for labor, capital, and raw materials including land. 5 Economics: Principles and Policy, 3rd edition. William J. Baumol and Alan S. Blinder, authors. Published by Harcourt Brace Jovanovich, 1985.See pages 468- 469.6 Ibid.., page 468. See also THE ECONOMICS OF THE PUBLIC SECTOR by Robert Henry Haveman, published by John Wiley & Sons, Inc., 1970. Page 26.7 Haveman, op. cit., page 26. There is no better introduction in plain English to the market system than Haveman's book and no clearer explanation of some of the free market's limitations than his second chapter titled: The Conditions For Efficient Market Operation.8 ECONOMICS, 2nd edition. David N. Hyman, author. Published by Richard D. Irwin, Inc., 1992. See pages 394-403. See also Baumol and Blinder, op. cit., pages 539-543.9 Hyman, op. cit., pages 403-407. Also Baumol and Blinder, op. cit., pages 543-544.10 MULTIPLE PURPOSE RIVER DEVELOPMENT: STUDIES IN APPLIED ECONOMIC ANALYSIS, John V. Krutilla and Otto Eckstein, authors. Published by Johns Hopkins Press, 1958. See pages 43-44 and p.60. This text for economic practitioners has an outstanding and clearly written second chapter on the concept of economic efficiency and its practical meaning in evaluating public investment projects.11 Haveman, op. cit., pages 24-25. See also Baumol and Blinder, op. cit., pages 506-507.12 Krutilla and Eckstein, op. cit., pages 44-46 and p.60.13 Hyman, op. cit., pages 134- 135..14 Krutilla and Eckstein, op. cit., pages 50-51 and p. 125.15 Hyman, op. cit., Ch. 15, Market Failure and the Role of Government in Allocating Resources, pages 394-411.16 Economic theory assumes government intervention will take the form either of direct government regulation, or subsidy or tax. It is assumed a subsidy will be provided in the case of a socially desirable good, and that a tax would be appropriate to discourage use in the case of a socially undesirable good. See footnote 15 for reference.17 THE MARKET SYSTEM, 2nd edition. Robert Henry Haveman, author. Published by John Wiley & Sons, Inc., 1970. See pages 258-259. See also Krutilla and Eckstein, op. cit., pages 47-48, and 122- 126.18 AN INTRODUCTION TO MODERN ECONOMICS, McGraw Hill, 1973.Joan Robinson and John Eatwell, authors. See bottom of page 215, top of page 216.19 A HISTORY OF ECONOMIC THEORY AND METHOD by Robert B. Ekelund, Jr. and Robert F. Hebert. McGraw Hill Publishers, 1975. Page 330 states:"Pareto utilized...Edgeworth's consumer 'indifference curves'...to show that in the case of a fixed supply of goods a welfare optimum in exchange would obtain when no individual could benefit from trade without injuring someone else." (Emphasis added.) See also Krutilla and Eckstein, op. cit., pages 50-51 and p.125; and Nicholson, op. cit., page 224.20 Nicholson, op. cit., pages 252-262.21 Capitalism, Socialism, and Democracy, Third Edition, by Joseph Schumpeter. Published by Harper and Brothers Publishers, 1950. See page 83, and Chapter VII "The Process of Creative Destruction."22 An oligopoly is a market dominated by a few sellers, several of which are large enough relative to the total market to be able to influence the market price.23 Hyman, op. cit., pages 356-375. See also Baumol and Blinder, op. cit., p.523-535 24 Baumol and Blinder, op. cit., p 533.25 Internet System Handbook,Addison-Wesley Publishing Company, Inc.,1993. Daniel C. Lynch and Marshall T. Rose, editors. See page 723, figure 18.5.26 Baumol and Blinder, op. cit., pages 524-526. The interdependent nature of oligopolistic decision making is the very factor which makes analysis of the resulting allocation of resources under an oligopoly extremely difficult.27 Consider the case of the European Airbus.28 Nicholson, op. cit., pages 269-270.29 See Internet mailing list "com-priv psi.com" April 26, 1993.