Alleman__Dominant_Firm_

Dominant carrier market power in US international telephone markets JAMES ALLEMAN, GARY MADDEN{ and SCOTT SAVAGE*} Columbia University, Graduate School of Business, Suite I-A Uris Hall, New York, NY 10027, USA {Curtin University of Technology, Communication Economics and Electronic Markets Research Centre, GPO Box U1987, Perth, Western Australia 6845 and }University of Colorado at Boulder, Interdisciplinary Telecommunications Department, Campus Box 530, Boulder, Colorado 80309-0530, USA An econometric model is used to examine market power in US international telephone markets. Lerner index estimates suggest AT&T’s collection rate-cost margin was between 12% and 24% during 1991 to 1995. Although Lerner estimates imply deadweight welfare losses of up to US $261 million per annum, such losses are small compared to those from the inefficient pricing of international inter- connection. Settlement rate-cost margins on US bilateral markets of approximately 89% translate into a US $4907 million transfer from consumers to carriers in 1995. I . INTRODUCTION Prior to the divestiture of AT&T there were concerns that Regional Bell Operating Companies (RBOCs) would distort competition in long-distance markets through cross-subsidization, and the incumbent would exert market power by providing rivals with inferior access to bottleneck facilities (Brennan, 1987; Hausman, 1995; Ward, 1995). Accordingly, the Modification of Final Judgement pre- vented the newly divested RBOCs from providing international message telephone services (IMTS) and inter-LATA services. The FCC also adopted a system of regulation which classified carriers ‘dominant’ according to their ability to exercise market power. Dominant firms were subject to additional regulation intended to prevent anticompetitive behaviour. For instance, AT&T operated under rate-of-return regulation until 1989, and price-cap regulation for three telecommunications service baskets thereafter.1 Additional regulation and oversight was also imposed on US carriers negotiating operating agreements with foreign monopolies, and US carriers affiliated with foreign carriers (Brands and Leo, 1999). By 1995 new entrants MCI and Sprint had obtained accounting rate agreements for all major markets, and shared ownership of international facilities with AT&T (FCC, 1997a).2 More importantly, several long-distance networks were developed within the US to compete with AT&T for connection with international facilities. In 1993 the FCC concluded there was sufficient competition in ser- vice basket’s (ii) and (iii) to relax rate-regulation, and in 1996, AT&T was declared non-dominant in the provision of IMTS. The decision to reclassify AT&T was primarily based on evidence concerning entry barriers. In particular, the FCC (1996; } 6) argued that ‘domestic competition pre- Applied Economics ISSN 0003–6846 print/ISSN 1466–4283 online # 2003 Taylor & Francis Ltd http://www.tandf.co.uk/journals DOI: 10.1080/0003684022000040957 Applied Economics, 2003, 35, 665–673 665 *Corresponding author. 1 Price-caps cover three service baskets: (i) residential and small business domestic, and international services; (ii) toll free ‘800’ services; and (iii) business services such as WATS, private line and various switched commercial services (Mitchell and Vogelsang, 1991). 2 The accounting rate is the basic ‘unit of account’ from which international settlement payments are made. The settlement rate determines the amount carrier’s pay to access other country networks and can be thought of as an international interconnection cost. A carrier’s net settlement is determined by the difference between incoming and outgoing traffic, multiplied by the settlement rate. The International Settlements Policy (ISP) requires all US carriers to accept the same accounting rate for a particular bilateral market, and that settlement rates are the same for outgoing and incoming calls (50:50 rule). This arrangement removed the incentive for a single competitive US carrier to accept less favourable interconnection conditions as they will apply to all carriers. Proportionate return supports the 50:50 rule by guaranteeing all carriers a share of incoming traffic, according to their outgoing traffic market share. vents AT&T from leveraging control over its domestic net- work to shut out competition on the international segment. In short, it is no longer plausible to view AT&T as con- trolling bottleneck facilities’. Given that AT&T held over 50% of the IMTS market during 1995, the FCC’s (1996) decision to reclassify AT&T non-dominant was somewhat controversial. This study examines competition and the extent of market power in US IMTS markets for a sample of 30 bilateral markets from 1991 to 1995 within the framework of the dominant firm-competitive fringe (DF-CF) pricing model. Econometric methods are used to estimate structural par- ameters which confer market power, namely, the competi- tive fringe’s supply elasticity and the dominant carrier’s residual demand elasticity. Calculation of a Lerner index from these elasticities provides an estimate of the percen- tage retail price (collection rate) markup over marginal cost for AT&T. Lerner index estimates are used to assess the FCC’s decision, calculate potential welfare losses from pri- cing collection rates above cost, and examine the potential for market entry by RBOCs.3 Such findings have policy implications for assessing the efficacy of market deregula- tion, predicting implications of similar deregulation efforts in other countries, and setting the terms of international settlements. The paper is organized as follows. US industry structure and price formation are discussed in Section II. This dis- cussion provides empirical evidence to support the use of the DF-CF model. A model for estimating market power is outlined in Section III. The model is estimated in Section IV, whilst market power, welfare implications and settle- ment rate pricing are discussed in Section V. Section VI contains concluding remarks. II . INDUSTRY STRUCTURE AND PRICE FORMATION The DF-CF pricing model assumes the market for a homo- geneous service is comprised of a dominant price leader (AT&T) and a fringe of quasi-competitive carriers.4 When the nondominant carrier’s market share is small rela- tive to market demand it views own output as having no affect on price. These rival carriers form a competitive fringe and equate marginal cost with the price set by AT&T. The assumption that AT&T held a dominant market share from 1991 to 1995, and faced a group of relatively small carriers, is supported by FCC (1997a) data. In 1991, AT&T’s market share (in minutes) was 72.7%, whilst MCI and Sprint held 16.5 and 6.9% of the market, respectively. Although AT&T’s market share declined from 1991 to 1995, it maintained dominance. At 1995, AT&T had 62% of the market, MCI 26.8%, and Sprint 8.2%. Smaller carriers and over 200 resellers supplied the remaining minutes. Such changes in market share are consistent with the DF-CF model when absolute market growth exceeds that of fringe carriers. Between 1991 and 1995 total US IMTS traffic increased by 6.99 billion minutes, whilst fringe output increased by 4.75 billion minutes. Anecdotal evidence suggests IMTS pricing has charac- teristics consistent with AT&T acting as a price leader to MCI and Sprint. Listed collection rates for AT&T, MCI and Sprint are generally similar across all rate classifications.5 Whilst data limitations prevent the analysis of carrier specific IMTS price formation, supportive anec- dotal evidence can be inferred from domestic long-distance markets. Kahai et al. (1996) provide econometric evidence which supports DF-CF pricing in the interstate market from 1984 to 1993. Hausman (1995) suggests interexchange carrier (IXC) competition, especially for residential and small business customers, has become a situation where AT&T sets prices (subject to price-caps) followed by its main rivals. For instance, following AT&Ts price increases in 1993 and 1994 both MCI and Sprint could have held prices constant and captured market share from AT&T. Instead they chose to increase their prices along with AT&T (Hausman, 1995). The DF-CF model assumes the services offered by the incumbent and new entrants are not differentiated. Service differentiation in long-distance markets arises at the origi- nating point of the circuit when non-AT&T subscribers are required to dial additional digits to place a call (‘non- premium access’). To allow new entrants to provide ser- vices comparable to those of AT&T, RBOCs are obliged to install switching equipment that allows for ‘equal access’ (1þ dialling) by all long-distance carriers. At 1995, almost 99% of telephone lines had been converted to equal access, and non-premium access differentiation between AT&T and fringe carriers has been largely eliminated. Nevertheless, to account for any perceived differentiation by subscribers, an equal access argument is included in the fringe supply equation below. 666 J. Alleman et al. 3 Another approach, typically employed in regulatory proceedings, is to use the flow-through of unit cost to price to assess the competitiveness of long-distance markets (Ford, 1999). 4 Scherer and Ross (1990) and Kahai et al. (1996) suggest a threshold market share threshold of 40% is required for the largest firm in a industry to exhibit dominant firm behaviour. 5 Listed prices, such as those provided by the FCC (1992, pp. 41–46), are not an accurate basis for making reseller price comparisons. Pure resellers route calls through a facilities-based carrier by purchasing bulk capacity at a discount to the listed price. The evidence above suggests that the DF-CF model is a reasonable approximation of US IMTS markets for the period 1991 through 1995.6 The opportunity to apply this model diminishes substantially after 1995 with industry evolution towards a less regulated, competitive structure. Since 1995, service provider market share has grown dra- matically, whilst increased carriage of data traffic has the potential to affect industry structure and competitive be- haviour. Finally, the 1996 Telecommunications Act allows RBOC subsidiaries to enter IMTS markets provided they satisfy competitive requirements under Section 214 and Section 271 of the Communications Act 1934. At 1997 the FCC had authorized the provision of out-of-region (non-local) switched resale and facilities-based services for five RBOCs. 3. EMPIRICAL MODEL The negative inverse of AT&T’s residual demand elasticity ð"dÞ provides an estimate of the degree of market power held by the dominant carrier: L ¼ ðPOMCdOÞ=PO ¼ �1="d ð1Þ where L is the Lerner index of market power, PO is the collection rate for an outgoing call from the USA to a foreign country, and MCdO is AT&T’s marginal cost of providing outgoing IMTS.7 Following Saving (1970) Equation 1 is rewritten as: L ¼ 1=½ð"þ ð1MSdÞ�Þ=MSd � ð2Þ where " is the market demand elasticity with respect to the collection rate, MSd is AT&T’s market share and � is the fringe’s own-price supply elasticity.8 To calculate L, MSd is obtained from FCC data, whilst " and � need to be empirically estimated. Consider a bilateral market for international outgoing calls from the USA to a foreign country. Given the inter- national settlement policy (ISP) that all carriers must accept the same settlement rate, the inverse supply equa- tion for the competitive fringe in bilateral market i ¼ 1; . . . ; n, at time t ¼ 1; . . . ;T is:9 POit ¼ �0i þ �1QfOit þ �2PSit þ �3PAt þ �4dft þ �5QSOt þ �6QSIit þ �7Tt þ eit ð4Þ where Q f O is outgoing IMTS calls from US fringe carriers to the foreign country, PS is the settlement rate for a call between the USA and f (the fringe carrier’s payment to access foreign country networks), PA is the local-exchange access charge, df is the extent of product differentiation between AT&T and fringe carriers, QS are network quality of service indicators at the US and foreign ends of the market, respectively, T is technological advance and e is an additive disturbance term.10 The point-to-point demand functions of Larson et al. (1990) allow calls in one direction to affect return calls through reversion and reciprocity. US outgoing market demand is: QOit ¼ �0i þ �1POit þ �2YOt þ �3QIit þ �4Comit þ �5QSOt þ uit ð5Þ Market power in US IMTS markets 667 6 Use of the DF-CF model is primarily justified by observed market shares during the sample period. It is also possible that a dynamic oligopoly game could generate similar market share and pricing patterns. With the available data it is not possible to estimate a market model suggested by such a game. 7 Equation 1 assumes AT&T set marginal revenue equal to marginal cost. When rate regulation prevents this condition from being met, the observed price may be below the profit-maximizing price, implying a lower value for L. 8 Assuming fringe resale carriers purchase capacity at a discounted wholesale price and resell those minutes at the prevailing retail price set by AT&T, Equation 1 can also be written as: L ¼ 1=½ð"þ ð1�MSd �MSrÞ� f þ ð1�MSd �MSf Þ�rÞ=MSd � ð3Þ whereMSf is the market share of facilities-based fringe carriers,MSr is the market share of resale fringe carriers, � f is the facilities-based fringe carriers own-price supply elasticity, and �r is the resale fringe carriers supply elasticity with respect to the wholesale price. Whilst a lack of resale data prevents the calculation of Equation 3, it can be shown that Equation 2¼Equation 3 when � f ¼ �r ¼ � or MSr ¼ 0. Since resellers purchase their primary input (network access) from IXCs it is reasonable to assume the reseller’s supply elasticity is similar to that of facilities-based carriers. When this is not the case, the relatively low value for MSr over the sample period (annual average is 8%), ensures any bias from calculating L from Equation 2 will be small. 9 The ‘equal treatment for all carriers’ and ‘50:50 accounting rate share’ rules require all US carriers accept the same accounting rate for a particular bilateral market, and that settlement rates are the same for originating and terminating calls. This arrangement removes the incentive for a single US carrier to accept less favourable interconnection arrangements as they apply to all carriers. ISPs are accom- panied by rules directing the monopoly carrier to return traffic to competing carriers in proportion to incoming traffic received. Proportionate return regulation has two main aims: to support the 50:50 rule by reducing the incentive for competitors to secure additional return traffic through a unilateral reduction in accounting rates; and to provide an incentive for unilateral collection rate reductions as an increase in market share will generate a corresponding higher share of incoming calls. 10 The settlement rate is an appropriate measure of international access in an inverse-supply equation. In flow-through models, (mar- ginal) ‘settlement cost’ may be a more appropriate measure. Settlement cost accounts for proportionate return by incorporating settle- ment rates, the ratio of incoming to outgoing traffic, and carrier-specific market share in the cost calculation (Ford, 1999). where QO is total outgoing IMTS calls from the US to the foreign country, YO is US income, QI is total incoming IMTS calls from the foreign country, Com is the calling community of interest and u is an additive disturbance term. To enable consistent estimation of Equations 4 and 5 the endogeneity of price and quantity must be recognized. Accordingly, a market demand equation for incoming calls from the foreign country to the US Equation 6 is specified, and Equations 4 through 6 are estimated as a system: QIit ¼ �0i þ �1PIit þ �2YIit þ �3QOit þ �4Comit þ �5Acompit þ vit ð6Þ where PI is the collection rate for an incoming call from the foreign country to the USA, YI is foreign country income, Acomp is asymmetric competition and v is an additive disturbance term.11 Econometric estimation is on annual data for 30 bilateral markets from 1991 to 1995. The sample period covers the five-year period preceding the FCC’s 1996 decision to reclassify AT&T non-dominant, and the Tele- communications Act 1996. Sampled countries are Argentina, Australia, Austria, Belgium, Canada, Chile, Colombia, Costa Rica, Denmark, Ecuador, Egypt, France, Greece, Honduras, India, Ireland, Israel, Italy, Japan, Korea (South), Mexico, Netherlands, Philippines, Portugal, Singapore, Spain, Sweden, Switzerland, Turkey, and the UK.12 Twelve countries are ‘developing nations’ according to ITU (1996a) classification, whilst eight of the 30 countries allowed facilities-based carrier competition during 1991 to 1995. The bilateral markets represent approximately 54% of total 1995 outgoing traffic, and 65% of incoming traffic. Definitions, sample means and standard deviations for all variables are provided in Table 1. The a priori signs of the arguments included in Equations 4 through 6 require some discussion. The DF-CF model suggests @PO=@Q f O5 0 or fringe carrier willingness to increase supply in response to AT&T collection rate increases. As @PO=@Q f O tends to zero, fringe supply becomes more elastic, reflecting greater potential competi- tion. Carrier access to international and domestic networks is a primary production input, and lower settlements and access charges should shift fringe supply outward implying @PO=@PS5 0 and @PO=@PA5 0. Kahai et al. (1996) find growth in equal access reduces differentiation between 668 J. Alleman et al. 11 PO, Q f O, QO and QI are endogenous. Given stable demand and costs in foreign countries, PO can be treated as endogenous without much bias. 12 Whilst US collection rate and traffic data are publicly available, foreign country collection rate data is scarce and less reliable. Accordingly, the 30 foreign countries are included in the sample as they have reliable and consistent collection rate data for the period 1991 to 1995. Table 1. Variable description and sample statistics 1991–1995 Variable Definition Mean Std dev PO real per-minute peak (standard) outgoing collection rate (US$) 1.12 0.26 PI real per-minute peak incoming collection rate (US$) 0.92 0.46 QO total minutes of outgoing US traffic (millions) 272 513 QI total minutes of incoming traffic (millions) 149 327 Q f O QO less ATT’s minutes of outgoing traffic (millions) 104 209 PS real per-minute settlement rate for outgoing traffic (US$) 0.52 0.19 PA real per-minute price US long-distance carriers pay local-exchange carriers for access to local networks (US$) 0.06 0.003 df the share of telephone lines converted to equal access 0.97 0.02 T time trend – – QSO US network quality index, one plus the percentage of mainlines connected to digital switches 167 12.7 QSI foreign network quality index, one plus the percentage of mainlines connected to digital switches 160 21.1 YO US real GDP (US$billion) 19,880 513 YI foreign country real GDP (US$billion) 9,814 7,864 Com market size, the product of US and foreign country mainlines (billion) 13,619 18,749 Acomp equals one when there is a monopoly facilities-based provider at the foreign end of the market; zero otherwise. – – MSd AT&T’s share of bilateral market traffic (in minutes) 0.64 0.08 Note. Base year is 1987. Source. FCC (1991–1995, 1997c), ITU (1996a, 1996b), World Bank (1997). AT&T and fringe carriers. This enhances the fringe car- rier’s ability to expand output and places pricing pressure on the dominant firm. As such, increased equal access is expected to shift fringe supply outward suggesting @PO=@df 4 0. QS and T account for exogenous shifts in supply from network enhancement and technology change, respectively. Improved network quality through greater US digitization suggests higher collection rates, however, digi- tization may lead to lower collection rates through increased network efficiency.