David Weiman Richard Levin
Yale University

I. Introduction

Since the publication of John McGee's (1958) provocative paper on the Standard Oil Trust, scholars of antitrust law and industrial organization economists have debated with religious fervor the theoretical rationality and historical prevalence of predatory behavior as a means of securing market power. Using the record of the 1911 Standard Oil case to support his argument, McGee claimed that it is cheaper to acquire a competitor directly than to induce exit or merger by pricing below cost. He rejected the argument that pricing below cost might force a competitor to sell out cheaply, arguing that a price war would partially dissipate the value of the merger. Yamey (1972) responded, without fully rigorous proof, that pricing below cost would nonetheless be a profitable strategy for the monopolist if (1) it lowered sufficiently the price at which the competitor could be acquired or (2) it signalled to competitors in other markets or potential competitors in the primary market the monopolist's determination to eliminate competition, despite the costs of engaging in predatory activity.

Despite recent proofs1 that Yamey's conjectures hold under conditions of asymmetric information about costs, skepticism remains about whether predatory strategies have actually been employed. Most empirical work on this subject has relied on the interpretation of contemporaneous documents and the descriptions of market participants rather than detailed analysis of quantitative evidence. An interesting exception is the work of Burns (1986), who found that in markets where American Tobacco was alleged to have engaged in price wars, its costs of acquiring competitors were reduced. These costs also varied inversely with a (not entirely convincing) proxy for American Tobacco's reputation.

In this paper we analyze quantitative data and interpret descriptive accounts concerning the response of the Southern Bell Telephone Company (SBT) to competitive entry during the decade following the expiration of the basic Bell telephone patents in 1894. Competition arose rapidly in the south, as it did elsewhere, but it proved less durable than in most other regions. We find that in the period from 1894 to 1900 Southern Bell responded to competitive entry, first, by lowering price or expanding investment in response to the threat of entry, and, second, by pricing below cost when entry actually occurred.

This strategy induced exit only occasionally, apparently because Bell had no real cost advantage over independent telephone companies in small markets. After 1900, Bell focused its strategy on investment in its network of toll lines, which had the effect of isolating the independents against whom Bell continued to fight "wars of attrition." The combination of pricing local service below cost and controlling access to toll lines forced most independents to sell out or take licenses on terms highly favorable to Bell. By 1904, William S. Allen, AT&T's watchdog over its competition proclaimed victory in Southern Bell's territory: "the opposition in the South seems to be disappearing and there is no evidence that it will spring up again."2

II. The Data

Our principal source of data is an SBT report detailing the "general conditions" of its local exchanges in 1900.3 Edward J. Hall, SBT's president, appended the "Exchanges Report" to a lengthy memorandum to the parent company, spelling out the problems plaguing the anemic operating company and his ambitious and costly plans to remedy them. Constrained financially by wary investors, he explained, SBT could only respond to competitive entry by "slaughtering" its prices and profit margins, rather than providing improved service. If these conditions persisted, he reckoned that the company would just break even, covering its debt service with gross profits.

Although Hall only alluded to the "Exchanges Report," it systematically documents the weak links in the company's system and the impact of competition on its profitability. For each of SBT's 78 exchanges, the report specifies the vintage and condition of SBT's equipment, the volume of business measured by the number of sub-stations, linkages to regional and national toll networks, and the overall quality of service and relations with the public. The report also indicates the net earnings of each exchange during 1899, as well as the value of the capital stock at the end of that year.

The final entry for each exchange is the most interesting and informative. It offers a brief, candid portrait of competitive conditions in the local market. In cases where entry had occurred, the report indicates the size of the independent exchange and the number of "points" or sub-stations served by both companies. In many instances the report also provides an assessment of the equipment and service, toll connections, and future prospects of the independent. Where no competition was present, the report documents any threats of entry or measures taken to thwart it. The former is evidenced by efforts of investors to raise subscriptions or to petition the city council for a charter, while the latter typically involved successful lobbying efforts by SBT to block passage of the requisite ordinances. In several instances, to secure political victory, SBT exchanges had agreed to improve the quality of local service, reduce rates, or construct toll lines to nearby cities.

SBT's network of local exchanges included virtually all cities with a population of at least 10,000. Its coverage of smaller cities and towns, however, was sketchy at best. Bell service reached only 20% of urban centers with a population of 5-10,000, except in Alabama and Georgia where the figure was 80%. Local exchanges varied greatly in size, from 2 substations to over 2000, averaging 338 per exchange. The average exchange represented a capital investment of $30,000, and earnings in 1899 averaged $1,350 per exchange. Yet, the earnings of individual exchanges ranged widely from a loss of almost $10,000 in Richmond to a profit of $34,000 in Atlanta.

Only 33 (or 42%) of the exchanges operated in "dual" markets, where they competed directly against an independent. This figure seriously understates the extent of competition over local service. An additional 12 (or 15%) exchanges reported a current threat of entry, and 10 exchanges (or 13%) had successfully blocked entry in 1899. The remainder, only 30% of the exchanges, operated in monopoly markets. Competition from independents was strongest in the largest markets and, in general, in the states of the Upper South.4

III. The Impact of Competition on SBT's Earnings

Cross-sectional evidence corroborates Hall's claim about the impact of entry on the profitability of local exchange service. The graph in Figure 1 clearly illustrates his dilemma. The graph divides SBT exchanges into ranges representing competitive conditions.5 Within each range, exchanges are ranked by the level of earnings. In ninety percent of the markets where local exchanges enjoyed unfettered monopoly, net earnings per station were positive, and they were often very large. Only one-third of the exchanges subject to direct competition, however, reported positive earnings. Where entry had been recently blocked or was currently threatened, exchanges generally earned positive profits, but where entry was currently threatened profits were lower on average than those in unchallenged monopoly markets. An analysis of variance summarizes these results. Exchanges in monopoly markets earned average profits of $8.67 per sub-station. The threat of entry reduced the earnings by 40% to $5.25 per station. Finally, and most strikingly, exchanges which competed directly against independents reported average losses of $3.85 per station.6

We develop this finding further through a regression of net earnings per station upon a vector of dummy variables representing the state of competition in each local market, a measure of the independent exchange's market share where competition was present, and a measure of the size of the Bell system to control for possible economies of scale or density. The results conform to our expectations. The estimated coefficients of the market structure variables correctly rank the average profitability of exchanges by competitive conditions. Potential entry more than halved the premium earned by monopoly exchanges, while actual competition forced SBT exchanges to operate in the red. The impact of an actual competitor on net earnings per station was more than 2.5 times that of a potential competitor. Moreover, the more successful were independents at capturing market share, the larger were SBT's losses.7 Finally, the profitability of SBT exchange service was positively correlated with size. We have not yet identified the source of these apparent economies of scale, but the result may explain why AT&T concentrated their investments in larger cities.

The evidence on earnings per station from Hall's 1900 report is corroborated by data from two earlier sources.8 These documents report the net earnings of each Bell exchange in the years 1896 through 1898, and they report the dates of each competitive entry, and, where relevant, exit. Together, these data strengthen the connection between competition and red ink for individual exchanges and the company as a whole. Between 1896 and 1897 average earnings per exchange fell from $764 to a mere $284, or by almost two-thirds. Company profits rebounded sharply in the following year, a point which Hall noted in his cover letter to corporate headquarters. Still, he expressed concern over "complications," presumably arising from the competition. His anxiety, the evidence suggests, was well placed. In the three years covered in these earlier reports, average earnings per exchange averaged $1500 to $2200 in markets where Bell enjoyed exclusive rights to phone service. In competitive markets, by contrast, exchanges reported average losses ranging from approximately $1000 to $3000.9

Without evidence on rates by locality, we can not directly observe SBT's pricing response to actual or potential entry. The lower profitability of exchanges facing competition in 1899 is consistent with three possibilities. One is that SBT may have held prices and service quality constant, lost business to new entrants, and hence failed to cover fixed costs. We can, however, reject this conclusion for the 31 of the 33 competitive exchanges.10 The remaining alternatives are that SBT slashed prices and margins, as Hall suggested, or incurred large outlays to improve the quality of service. In either case the result would be a reduction in current profits and often large losses.11

Aggregate data on monthly revenues per station, also included in Hall's memorandum, offer some clues about the pricing response to new entry.12 As shown in Figure 2, nominal rates fell by 50 percent from January 1894 to September 1900, from $4.80 per station to $2.40, and real rates declined even more sharply. The decline in telephone rates was neither gradual nor smooth. With the exception of the second quarter of 1895 and the fourth quarter of 1896, the real price declined at an average rate of 0.5% per month during the 1894-1897 period. This trend, however, was punctuated by sharp drops of 19% and 11% in the two exceptional periods mentioned. Subsequently, prices declined at the more rapid average rate of 1% per month. The sharp declines in 1895 and 1896, and the steeper trend thereafter, coincided, not surprisingly, with the first wave of entry into the Southern market. Between 1895 and 1897 SBT faced new competition in 20 local markets, or almost two-thirds of those reporting dual service in 1900. Entry waned in 1899, but gained momentum in the following year, as independents moved into the Lower South.

We explore further the dynamics of Bell's response to competitive entry by combining the data from the 1900 survey with the data on the dates of competitive entry. We expand the regression analysis discussed above by regressing the net earnings per station of each SBT exchange in 1899, as before, on a measure of competitive market share and on a measure of the size of the SBT exchange. Rather than include dummy variables for the state of competition in 1899, however, we use a dummy variable to indicate the year of competitive entry. Although the results are quite robust to variations in the specification, they emerge most sharply when the markets are divided into the following categories: those in which entry occurred prior to 1899, in 1899, and in 1900, and those in which no entry occurs. We also distinguish the two markets in which both entry and exit had occurred by 1899.

The results reveal a dramatic response of net earnings per station to entry. Evaluated at the mean values of SBT system size and independents' market share, the coefficients imply the impact of competition on earnings was most severe in the first year of entry. SBT exchanges facing new entry in 1899 experienced average losses of $22.87 per station. In markets where entry occurred earlier, earnings recovered from the initial impact of competition, and losses averaged only $2.58 per station. Exit of competitors restored profits essentially to the levels earned in unchallenged monopoly markets: just $0.69 short of the $9.27 per station that was earned on average in monopoly markets. Finally, the regression indicates that credible threats of entry had a substantial effect on SBT profitability; exchanges that experienced entry in 1900 suffered losses of $0.56 per station in 1899 in contrast to the $9.27 earned in monopoly markets. An alternative formulation indicates that the adverse impact of threatened entry on 1899 SBT earnings was only half as large in markets where the entry failed to materialize in 1900.

IV. Predatory pricing and acquisition.

Interpreting the cross-sectional evidence dynamically, it suggests a common pattern of response by SBT exchanges to competitive entry into their previously protected markets. Their behavior, we maintain, is consistent with a predatory strategy aimed at restoring monopoly control over local exchange service. In anticipation of entry, SBT exchanges braced themselves by reducing their rates and earning lower profits. In response to actual entry, however, exchanges slashed their prices and earnings, signalling their intention to wage a full-scale price "war." After this initial bout of price cutting, SBT exchanges relented, but still set their prices below cost to weaken their competitors. If successful in driving the independent out of business, SBT regained the market power to restore its prices to levels comparable to those in markets in which its monopoly had never been challenged.

In many cases, however, independents were able to withstand SBT aggressive response, as they enjoyed cost advantages over Bell exchanges, especially in smaller markets. Still, SBT's policy deprived the independents of the cash flow required to finance the expansion and improvement of their equipment. This problem was particularly acute for two reasons. First, lower rates stimulated demand for telephone service, and the unexpected volume of business taxed the capacity of independents. Congestion not only resulted in longer delays in completing calls, thus lowering the quality of service, but eroded the cost differentials which enabled them to meet SBT's prices. Second, the independents relied almost exclusively on retained earnings to finance investment.

Around 1900, Hall conceded that SBT's policy of starving the independents would not work. Price wars drained the company of much needed funds to finance its own investment. Even if the policy was successful in eliminating competition, the costs of fighting could not be readily recovered. Although our regression results do not support his view, Allen claimed that exchanges often found it difficult to raise rates after a price war because of government regulations or potential losses in good will.13

More important, SBT discovered that it could not win a war of attrition against the independents with low prices alone. In smaller markets independents operated at lower costs because they adopted "appropriate" and less costly technologies for the provision of local exchange and toll service. In mid-sized markets, such as in the Raleigh-Durham area of North Carolina, independents offered their customers short-range toll service to smaller cities and towns and thus attracted business customers. In both cases, SBT suffered from its parent's corporate strategy of building a national toll system from the top down.

Nonetheless, SBT did not abandon its predatory tactics. Instead, it expanded its strategic repertoire to include preemptive investments in toll lines and the consolidation of its exchange and toll networks through acquisition and sub-licensing. Where necessary SBT continued to resort to low prices. Qualitative evidence suggests that SBT followed a two-pronged strategy. First, SBT undertook an ambitious program of toll line construction to complete its regional long-distance network and strengthen its hold over large urban markets. This investment strategy appears to have been preemptive, intended to increase SBT's market power rather than respond to demand, because at the time SBT reported among the lowest volume of toll service of any Bell operating company.14

Second, SBT acquired or sub-licensed independent exchanges at nodal locations to block the formation of an alternative local or regional toll network.15 SBT continued to engage in predatory pricing against recalcitrant independents in those large urban markets that might have anchored an alternative regional network. In the case of Richmond, the evidence suggests, SBT sustained large losses over the period to force the independent to sell out at an unremunerative price, thus deterring future entry. Hall threatened a similar response in Mobile, after the Home Company refused SBT's initial offer.

SBT also sought to acquire or sub-license smaller exchanges that formed local networks. Although sub-license agreements conceded the independent a monopoly over local exchange service, covering a radial distance of five miles, they prevented the independent from connecting to non-Bell lines or constructing competing lines. SBT pursued this policy with a vengeance. Between 1900 and 1906, it acquired at least 82 exchanges and sub-licensed an additional 59 (or 77). In particular, it acquired three local telephone networks in Alabama, Georgia, and the Carolinas. After this aggressive foray, Allen proclaimed the defeat of the independent movement in the South.


1 See Milgrom and Roberts (1982) for a proof of the second proposition and Saloner (1987) for the first.

2 AT&T Archives, Box 46 02 01, William S. Allen Papers, Allen-Fish, 1/8/1904. Our interpretation supports the findings of Gabel (1991), who analyzes AT&T's strategy in the Midwest. We present an alternative perspective on SBT's strategy of system-building than the one found in Lipartito (1989a).

3 AT&T Archives, Box 1340, SBT Co., "Report on General Conditions of Exchanges, July 19, 1900;" and Hall-Cochrane, 11/2/1900.

4 The chi-square statistic, testing the difference in size distribution of cities with and without dual service, equals 20.18 and is statistically significant. In states below South Carolina competition was limited to the largest cities, whereas in the rest of the territory, it frequently occurred in cities at the lower end of the size distribution.

5 Following Hall, we measure profitability by the ratio of net earnings to sub-stations, rather than value of capital invested. The capital stock data are not very reliable. Competitive conditions are ordered as follows: (1) unchallenged monopoly, (2) blocked entry, (3) potential entry, and (4) direct competition. In the subsequent analysis we exclude exchanges with fewer than five sub-stations.

6 The t-statistic testing the difference in profitability between monopoly and competitive markets equals 4.60 and is statistically significant at the 1% level. Where entry had been blocked, earnings were not significantly different from those in unchallenged monopoly markets.

7 The F-statistic, testing the difference between the coefficient on the actual and potential competition variables, equals 5.86 and is significant at a 5% confidence level. When evaluated at the mean value for dual markets, the market share variable implies an additional loss of almost $3 per station for SBT exchanges, relative to those incurred at very low levels of competitive market share.

8 AT&T Archives, Box 1263, SBT Co., Financing, 1898-1899, Hall-Howe, 5/16/1899, which includes a financial statement and "Statistics of Opposition."

9 The differences between monopoly and competitive markets in earnings per exchange were statistically significant at the 5% level in each of the three years from 1896 through 1898.

10 AT&T Archives, Box 1263, SBT Co., "Statistics of Opposition." In all but two markets, SBT service base grew between the date of entry and 1899.

11 As contemporary accounting methods did not typically distinguish capital expenditures from current expenses, increased investment would immediately depress current net earnings.

12 AT&T Archives, Box 1340, Hall-Cochrane, 11/2/1900.

13 AT&T Archives, Box 46 02 01, William S. Allen Papers, Allen-Fish, 12/27/1902. Allen made this observation in reference to conditions in Parkersburg, West Virginia.

14 In 1902 the ratio of local to long distance messages in the SBT territory was 1.3%, slightly more than half of the national average; U.S. Census (1902). Of the 33 operating companies SBT ranked 25th in the volume of toll service in 1903; AT&T Archives, Box 1348, Operating Companies, Performance Ratings, 1903.

15 As president of SBT, Hall pioneered this new strategy; see Lipartito (1989a 1989b). The evidence on SBT's acquisition and sub-licensing policy comes from numerous reports and memorandums in the AT&T Archives; see, for example, AT&T Archives, Box 1340, SBT Co., Sub-Licensees, 1900; Acquisition of Independent Companies in Georgia, South Carolina, W. Virginia, North Carolina, and Alabama, 1902-1911; Acquisition of Richmond Telephone Co., 1902; Acquisition and Sale of Exchanges in North Carolina, 1903; Home Telephone Company of Mobile, Al., Negotiations, 1905. Aggregate statistics on the number of sub-license contracts and the official AT&T policy on sub-licensing are contained in AT&T Archives, Box 1364, Sublicense Statistics, 1907-1908.