This paper develops a model to reflect the Argentine market for end-user Internet access and draws conclusions on certain aspects of economic efficiency. After the Introduction, the second section, entitled A Simple Model, provides a descriptive analysis of the market and its related industry, deriving a stylized technology and setting up a basic model. The third part of the paper, An Application to the Argentine Market, applies the model to Argentina, while the fourth section, Lessons from a Linear Demand Case, examines the results of the model under certain conditions. The model and the linear demand case are fully developed in the Appendix.
Contents
Introduction
A Simple Model
An Application to the Argentine Market
Lessons from a Linear Demand Case
Concluding Remarks and Guidelines for Further Research
Notes
References
Appendix
Introduction
In July 1995 Argentina saw the opening of the end-user Internet access market, dominated by the private, legally granted telecommunications monopoly, and consisting of a few more firms which had to purchase international data transmission services from the monopoly. High initial prices soon dropped as more firms entered the fast growing market, and initial per-minute charges disappeared while Internet service providers (ISPs) started offering unlimited access at flat rates to end users.Despite several legal controversies about the status and the extent of the international access monopoly, the market structure has not changed much since its inauguration, with the exception of a single carrier which has been allowed to operate with several restrictions, among which the prohibition of reselling international access services to third parties. The monopoly charges a flat rate to ISPs for international access with no traffic limit beyond the capacity of their link, and these in turn compete among themselves offering unlimited access at flat rates with absolutely no kind of price discrimination between different types of customers, namely firms or households, heavy or occasional users.
At the same time, even an unskilled user can detect serious congestion problems at the local and international levels, as transmission speeds vary widely according to the time of the day. In the meantime, the presence of flat rates immediately rises suspicions of inefficient pricing schemes being applied.
The purpose of this work is to analyze different kinds of equilibria emerging from alternative pricing schemes applied by the ISPs and by the monopoly, and, in the light of this results, explore the possibility of increasing social surplus by introducing modifications in the actual tariff table.
A Simple Model
Technology
Telecommunication services are typically produced in a network that embodies more than a single technology. In the specific case of Internet access, the consumer must have some specific hardware and software and link telephonically to an ISP. The ISP, in turn, must purchase a link from a backbone access provider, which in turn must be interconnected with the several backbones that form the Internet.
The most familiar telecommunications technology are telephone switching networks. These kind of networks have a fixed and well determined capacity, as a telephone call requires a dedicated link, and no other user can take advantage of the network facilities involved in the link until the call is terminated. In contrast, Internet services are produced with packet-switching networks, which can theoretically handle an unlimited number of connections over the same facility, as they operate by splitting data into small packets that contain all the necessary information to reach their destination, where they are reassembled into the original data. This splitting allows to send packets of different origins in interleaved order through the same link, at the expense of a larger transmission time for the same amount of data when the number of connections grows [ 1 ].
The technical differences between switching and packet-switching networks require a differential treatment for congestion. When in a telephone switching network the last free link has been used to establish a call, all further attempts by any user to call another user will fail until one or more of the ongoing calls are terminated [ 2 ]. In a packet-switching network, however, no user should be denied access no matter how many connections are taking place, but transmission times will increase with every additional connection [ 3 ]. This allows us to treat packet-switching network facilities as standard congestible goods, for which a vast literature has been developed, typically in the public finance field [ 4 ].
The above considerations can be used to sketch a stylized technology, characterized by the three following assumptions:
a) There is an only product (Y: end user access to a network); the only input needed for the production of Y is an intermediate congestible good (X: access to an international network)
b) Good X is available only at a certain level of capacity; if a seller of consumer access wishes to expand his business, he must purchase a second unit of good X. This assumption rules out price discrimination at the international access level.
c) Good X is produced with an input F, which is fixed and congestible.
Industry Framework
An important economic characteristic of telecommunications networks is that their operating costs are nearly independent of the flow of services through the facilities (Mitchell and Vogelsang, 1991). This yields an industry which costs are almost completely fixed, so that marginal costs approach zero. Therefore the equilibrium price can’t emerge from a perfect competition solution, as with marginal cost pricing perfectly competitive firms would experience economic losses [ 5 ].
The monopolistic competition framework has proved useful to handle problems with large fixed costs and negligible marginal ones, and has been largely applied to the telecommunications industry [ 6 ].
However, to use monopolistic competition as the theoretical framework for an industry, one must be prepared to argue that "each firm or product has no direct neighbour in the product space" (Tirole, 1988) in order to comply with the assumption that a price change by one firm has negligible effects on the demand of any other firm.
The Internet access market fulfills this requirement to a reasonable extent, as there are hardly any substitutes to online services and many firms try to differentiate themselves by offering bundles and launching introductory packages with other products and services, such as cable television and computer hardware.
A peculiarity of the Argentine market is the legally granted monopoly for international backbone access, which is due to expire in the year 2000 if no drastic change in telecommunications policy or in the agreements between the government and the concessionaires take place. Due to the insignificant size of the local market relative to the whole of the Internet, virtually all backbone access requirements are for international data transmission, making the purchase of a link from the monopoly an unavoidable necessity to set up a service providing firm.
Taking into account the above discussed considerations, we can introduce two more assumptions:
d) Good Y is produced under a monopolistic competition framework. Each firm faces a decreasing demand, makes zero profits and is unable to affect the demand any other firm faces. These assumptions are the standard ones in the monopolistic competition literature.[ 7 ]
e) Good X is produced by a monopolist, who is the owner of the fixed factor F.
For further clarification, figure 1 provides an idea of the structure of the industry described in this model.
Demand
Consumer demand of network services is actually a demand for information transmission. Consumers will take into account the cost of network services (in our model, price of good Y) and the congestion they will face while trying to retrieve information. However, as no single consumer is able to modify the level of congestion prevailing in the network, each of them will take congestion as given. Therefore, the consumer demand function will show the price of network services as its only explicit argument, though implicitly incorporating in its functional form the influence of congestion.
Mitchell and Vogelsang argue that “although an individual consumer’s telecommunications demand appears to an observer as stochastic, the market demand has strong regular daily and weekly patterns.” (Mitchell and Vogelsang, 1991). Similarly, consumer groups are identified according to their demand elasticities and individual demanded quantity. The typical example here is the difference between firms and households, with the formers being charged two or three times the regular tariff the latters pay for basic telephone services.
The main interest of this work are the effects of the introduction of differential pricing according to types of consumers. In our stylized case, we will consider two types of consumers which will impose different loads on the network. A rough distinction may identify these groups as firms and households, but a closer analysis may enlarge the firms group with some households with heavy use of the network, while some firms may use Internet access only as a secondary communication element. To successfully enact differential pricing policies, ISPs must be able to identify the kind to which every consumer belongs. A more sophisticated solution than the household-firm distinction is a self-assessment method, which leaves the choice of the category to the consumer, but applies a high penalty if a preset maximum of traffic is exceeded.
A second and more sophisticated method of dealing with congestion is the smart market solution by MacKie-Mason and Varian, who propose a system that would allow providers to charge consumers the marginal cost of transmitting data corrected for congestion at each moment of time. “In a smart market users only indicate the maximum willingness to pay for network access. [...] The router (the computer controlling network traffic) notes the bid attached to each packet and admits all packets with bids greater than some cutoff value” (MacKie-Mason and Varian, 1994c) [ 8 ]. While our analysis will not go as far as asmart market problem would require, variable pricing solutions as well as differential pricing ones need to be supported by appropriate technological devices which enable firms at every stage of production to measure the traffic each individual user imposes on the network at each moment of time.
In the light of this last discussion, three further assumptions can be added:
f) There are two types of consumers; those who will be “heavy” users of networks (identified in all formal expressions with a superscript H) and those who will be “light” or occasional users (identified with a superscript L). Consumers of each kind will be homogeneous in preferences, so that it will be possible to derive aggregate demands.
g) It is possible to identify which kind a consumer belongs to at no cost.
h) It is possible to measure a single consumer’s use of a network at any point of time. Similarly, it is possible to measure a service provider’s use of the international network at any point of time.
An Application to theArgentine Market
The Argentine market features flat tariffs at both stages of production: the monopolist charges a flat rate to ISPs for international access service, and these in turn charge flat rates to consumers. No traffic restrictions are enacted at either stage.
Prices of regulated monopolies’ products and services in Argentina are always the result of intrincated negotiations which last for a long time and, after they are concluded, they are hardly revised [ 9 ]. During the price setting process, telecommunications companies may agree to cross-subsidize some of their services in order to make the outcome of the negotiations compatible with some political goals. The result are price schemes far away from the dictates of efficiency, and this problem is only aggravated when a partial equilibrium problem, such as the one this work deals with, is focused. Therefore, we will not look into the reasons of the monopolist’s pricing scheme and its price level, considering them as exogenously set and assuming only that they are adequate to cover the monopolist’s fixed cost [ 10 ].
Regarding the competitor’s pricing schemes we will analyze two different variations. First, we will solve an unrestricted problem that will allow firms to set different flat prices for each kind of consumers and then we will compare its results with a restricted case, where competitors are forced to set a unique flat price for all consumers.
In a further stage we will drop the monopolist’s flat pricing scheme and introduce a variable scheme, with the monopolist charging ISPs for every connection they sell.
Problems
We are now in a position to formalize our analysis in terms of three different problems:
Problem 1: With two different and perfectly distinguishable demands for the final product and a fixed cost corresponding to a monopolically provided input, find the equilibrium prices and the number of firms in a monopolistic competition framework.
Problem 2: With two different and perfectly distinguishable demands for the final product and a fixed cost corresponding to a monopolically provided input, find the equilibrium price and the number of firms in a monopolistic competition framework, given the constraint that each competitor must charge the same price to all consumers, regardless of their type.
Problem 3: With two different and perfectly distinguishable demands for the final product , different variable costs according to the type of customer to which the product is sold and a residual fixed cost [ 11 ], find the equilibrium prices and the number of firms in a monopolistic competition framework.
These three problems are fully developed in the Appendix with special emphasis in the cases featuring linear demands, to which analysis we devote the following section.
Lessons from a Linear Demand Case
Though the three proposed problems can be entirely solved for most standard functional forms, we’ve focused on a linear demand case analysis to simplify calculations and make it possible for the non specialist to find the way to the results we present.
A linear demand is defined by two parameters, namely slope and intercept. As shown in the Appendix, it is possible to find the equilibrium values for number of firms and prices in terms of the demand functions parameters and the exogenously set price for the input X, and hence those values corresponding to demanded quantities and consumer surplus. This allows us to perform an accurate comparison between the results of problem 1, which allows for differential pricing, and problem 2, which forces firms to charge an only price to all consumers. It is also possible to compare relative prices between problem 1, where the monopolist charges a flat price to competitors and problem 3, where variable costs are introduced.
When focusing on problems 1 and 2, the first results are not surprising: problem 1 yields a higher equilibrium value for the number of monopolistic competitors than problem 2. Accordingly, the equilibrium price for problem 2 is found between the two equilibrium prices for problem 1.
The exception is the case when the two demands have the same intercept; if this happens, both equilibria coincide as the price charged in the unrestricted case to “heavy” and “light” users will be the same, and it will also be the same as that charged in the restricted case; the equilibrium number of firms will also be identical in both cases. As the demand intercept represents the price at which the quantity demanded equals zero, the coincidence of intercepts appears a difficult condition to fulfill, and would be at most a product of fortune.
For further clarification, let’s consider a transition from a restricted equilibrium to an unrestricted one. The comparison provided in the Appendix shows how, unless the coincidence of intercepts is satisfied, the number of monopolistic competitive firms will grow, “heavy” users will pay a higher price while “light” users will be charged a lower one respect to the one they were paying before the transition. A relevant implication of the model is that consumer surplus will be reduced as a consequence of the transition, as firms will be able to extract more of it by means of the differential pricing. However, as ISPs are monopolistic competitors, they must comply with the zero-profits condition; all extra benefits are therefore transferred to the monopolist via the increased number of firms which purchase an international access link at a flat price.
The welfare analysis reveals that, in the linear demand case, social surplus remains unaffected by the transition, the whole of the loss in consumer surplus being transferred to the monopolist in the form of increased profits. This result can’t be generalized from a linear case, but it provides a hint in the sense that no efficiency gains should be sought at the competitor’s level, and that relevant policies aimed to increase efficiency should be directed to modify the monopolist’s status.
As for the third problem, the exogenously set monopoly prices pose a severe limitation to a comparison with the precedent cases, as no fixed reference is kept for input prices. However, it is still possible to comment on the ratio between prices charged to “heavy” and “light” users. In problem 1, this ratio was equal to the ratio of the demand intercepts; when the monopolist starts charging for every additional connection sold by the monopolistic competitors, the equilibrium price ratio will be the same as in the flat pricing problem if and only if the ratio of input prices in problem 3 is equal to the equilibrium price ratio in problem 1. With input prices exogenously set, nothing can be said about the requirements for this condition to hold.
The Appendix discusses these three problems in full mathematical detail and provides a numerical example for comparison between problems 1 and 2.
Concluding Remarks and Guidelines for Further Research
In the preceding sections we have had an overview of the Argentine market for Internet access services, sketched out a stylized model to capture the main features of its technology, industry framework and consumer behaviour and examined the results of this model for a linear demand case.
The most interesting result is the null welfare increase from the monopolistic competitors’ transition from a unique pricing scheme to a differential pricing one. This result should provide an indication in the sense that little efficiency gains can be expected from a price scheme modification at the end user level, and that relevant actions to increase efficiency should be taken on the monopolist’s pricing scheme. Furthermore, if the government has a utility function which attaches a higher weight to consumer welfare than to firm profits, it should prevent such a transition.
The main shortcoming of our analysis is its limitation to the linear demand case, as some efficiency costs or gains could be hidden in other functional forms. Therefore the results obtained should be considered in the light of this clarification. Also, including congestion in an implicit form in consumer’s demand functions eliminates a valuable indicator at the moment of measuring the sources of changes in welfare. Future extensions of this work should be aimed at giving congestion a treatment which allows for its explicit inclusion, as well as at generalizing the mathematical development of the model.
The Author
Mauricio Drelichman was born in Buenos Aires, Argentina, on July 2, 1975 and graduated in economics at Universidad de San Andres in 1997. This paper is his graduation thesis. He is currently taking the master program in economics at Universidad de San Andres, where he also works as a teaching assistant. His main fields of interests are economic history, industrial organization (with special emphasis on network economics), and money & banking.Electronic mail: mauricio@udesa.edu.ar
URL: http://www.udesa.edu.ar/grads/mauricio/
Post: Malabia 1261, (1414) Buenos Aires, Argentina
Notes
I am extremely grateful to Rolf Mantel for his patient guidance and continuous advice during the development of this work. I am also indebted to Enrique Kawamura for his helpful bibliographical suggestions, as well as to George McCandless and Federico Guerrero for their contributions in the early stages of this research. All errors remain mine.
1. For a detailed analysis of the technological aspects of networks and their economic structure see Economides (1996) and MacKie-Mason and Varian (1994c).
2. A stylized analysis of switching networks technology is provided in Mitchell and Vogelsang. (1991)
3. The infinity of connections over a packet-switching network is actually only a theoretical construction. The software that maintains packet switching connections and several network facilities as well have a"timeout" setting, which drops a connection if a specified amount of time has passed without a response from the other party.
4. See, for example, Stiglitz (1988) or Rosen (1988).
5. For a detailed discussion of efficient princing schemes departing form the perfect competition solution, see Baumol and Sidak (1994) and Varian(1996).
6. See, for example, Economides and Wildman (1995).
7. E.g. Tirole (1988).
8. For further analysis of smart markets see MacKie-Mason and Varian (1994a), (1994b), and (1994c).
9. The most notorious case of a regulated monopoly's price-setting process in Argentina is that of the rebalancing of local and long distance telephone rates, which has been open for over two years with no firm conclusion, featuring several proposals and retreats from both the government and the telephone companies. For more information on the telephone tariff's rebalancing process in Argentina see Artana, López Murphy, Navajas, and Urbiztondo (1996).
10. Price-setting for monopolically provided inputs is discussed in Baumol and Sidak (1994) and in Economides and White (1995).
11. The residual fixed cost is necessary to reflect increasing returns to scale and the necessity of the monopolistic competition framework. It can be interpreted as hardware cost.
References
Daniel Artana, Ricardo López Murphy, Fernando Navajas and Santiago Urbiztondo, 1996. El Rebalanceo de las Tarifas Telefónicas en la Argentina. Fundación de Investigaciones Económicas Latinoamericanas, Working Paper no. 48.
William J. Baumol and J. Gregory Sidak, 1994. Toward Competition in Local Telephony. Cambridge, Mass.: MIT Press.
Nicholas Economides, 1996. “The Economics of Networks”. International Journal of Industrial Organization, Volume 14 Number 2 (March).
Nicholas Economides, and Lawrence J. White; 1995. “Access and Interconnection Pricing: How Efficient is the ‘Efficient Component Pricing Rule’?”. Antitrust Bulletin, Volume 40, Number 3, pp. 557-579.
Nicholas Economides and Steven S. Wildman, 1995. . Monopolistic Competition with Two Part Tariffs. Mimeo.
Jeffrey K. MacKie-Mason and Hal R. Varian; 1994a. “Pricing Congestible Network Resources,”. IEEE Journal on Selected Areas in Communications, Volume13, Number 7 (September).
Jeffrey K. MacKie-Mason and Hal R. Varian; 1994b. “Pricing the Internet,”, in Brian Kahin and James Keller, (eds.) Public Access to the Internet. Cambridge, Mass.: MIT Press.
Jeffrey K. MacKie-Mason and Hal R. Varian; 1994c. “SomeEconomics of The Interntet,”, in Werner Sichel and Donald L. Alexander (eds.), Networks, Infrastructure and the New Task for Regulation. Ann Arbor, Mich.: University of Michigan Press.
Bridger M. Mitchell and Ingo Vogelsang, 1991. Telecommunications Pricing: Theory and Practice. Cambridge: Cambridge University Press.
Harvey S. Rosen, 1988. Public Finance (2nd edition). Homewood, Ill.: Irwin.
J. Stiglitz, 1988. Economics of the Public Sector. New York: W.W. Norton.
Hal R. Varian, 1996. “Differential Pricing and Efficiency,”. First Monday, Volume 1, Number 2 (August).
Jean Tirole; 1988. The Theory of Industrial Organization. Cambridge, Mass.: MIT Press.
Appendix
Copyright © 1997, ƒ ¡ ® s † - m ¤ ñ d @ ¥
