The increase in concentration within the cable industry has been paralleled by growth in vertical integration. A recent study reports that vertical integration between MSOs and cable program suppliers is extensive and the extent has markedly increased since the mid-1980s (Waterman & Weiss, 1993b). In fact, vertical integration has long been one of the major concerns in the realm of mass media, tracing back to the historic Paramount case (U.S. v. Paramount Pictures, 1948) in the motion picture industry. Hence, neither the phenomenon nor the controversies of vertical integration are unique to the cable industry. But, partly because of the growing importance of its product, partly because of the very numerous nonintegrated competitors, suppliers, and customers interspersed throughout it, the cable industry provides the most controversial, if not the purest, example of the problems that arise when size and market dominance are associated with vertical integration.
For the policy maker, vertical integration raises issues of preserving freedom of access by information providers to the public and of achieving maximum diversity of information products (Brennan, 1990). For the economist, on the other hand, vertical integration raises issues of economic efficiency (Waterman & Weiss, 1993a). Virtually all discussion regarding vertical integration issues, thus far, seems to focus primarily on the former area-access of program suppliers to the public (Klein, 1988; NTIA, 1988; FCC, 1990; Waterman & Weiss, 1993b). As local monopolists, however, integrated MSOs have a variety of other opportunities to behave differently towards affiliated and non-affiliated networks when both are offered on its menu (Waterman & Weiss, 1993a). For example, since the Cable Consumer Protection and Competition Act of 1992 and underlying antitrust acts allow cable programming networks to charge cable operators differential rates for their programming services for "legitimate business reasons," large integrated MSOs can get volume discounts from their network affiliates. In reciprocity, the network affiliates can be placed on better channel positions on the local systems.
To date, these issues have been treated from the point of the bilateral bargaining relationship between the networks and the operators. In other words, the consequences of vertical integration from the consumer's standpoint have been largely ignored or untested. Thus, the present study seeks to be more inclusive by exploring the direct effect of vertical integration on consumer welfare in price and content diversity. Specifically, how do consumers get these benefits from vertical integration? Is massive vertical integration in any sense a necessary or socially desirable consequence of the economic and physical characteristics of cable television production and distribution? The answers to these questions are sharply divided into two different theories.
Procompetitive Consequences of Vertical Integration
Since Ronald H. Coase's pioneering research (1937), many economists have viewed the abandonment of the market implied by vertical integration as a socially desirable adaptation because vertical integration tends to reduce the misallocation of resources (Spengler, 1950; Bork, 1954; Comanor, 1967; McGee & Bassett, 1976; Perry, 1978; Williamson, 1971, 1974, 1979, 1985). For example, in current antitrust enforcement, there is a general presumption that vertical mergers are not anticompetitive. The successive monopoly model, frequently attributed to Spengler (1950) and Machlup and Taber (1960), is a major intellectual impetus for that policy. It says that vertical integration results in a reduction of the final good price by eliminating the double marginalization (Salinger, 1991). In this view, therefore, integration eliminates a dead weight loss resulting from double monopoly markups on product price and increases consumer welfare because the price of the final product is reduced in the direction of its marginal cost (Waterman, 1993).
Reinforcing this rationale is a more efficient contracting between buyer and seller. In his well-known article, Coase (1937) suggested that there were a number of costs of using the price mechanism, among which were the cost of finding out what the relevant prices are and the cost of negotiating and concluding a contract. In cases where the cost reduction is not obvious, Williamson (1971, 1974, 1979, 1985) applies "transactions cost" theories to explain the vertical extent of a firm. According to Williamson (1985), the economic institutions of capitalism have the main goal of economizing on transactions costs. In this view, the routing of transactions inside firms raises profits together with social welfare. Williamson therefore argues that integration serves to reduce contracting costs or to avoid the risk of opportunistic behavior. By providing a setting for joint profit maximization, vertical integration may eliminate or mitigate the exercise of market power and lead to efficient input use. Longterm contracts may produce a similar effect in eliminating the exercise of market power. However, in the presence of uncertainty, it is difficult or impossible to write complete contracts that adequately cover future contingencies and ex post adaptation to realized events may result in inefficient, opportunistic behavior (Lieberman, 1991; Kerkvliet, 1991). Vertical integration through ownership is more assured than through contract. On the basis of these theories, in fact, the National Telecommunications and Information Administration (1988) asserts that "vertical integration allows the cable firm to avoid the transaction costs of obtaining programming" (NTIA, 1988, p. 90).
If there are transactions costs, there may be an incentive to combine a number of events or activities into one bundle by arranging long-term contracts. A number of sellers and buyers can become one market unit by integration, and the firm which arises does not have to hunt for buyers and sellers in its internal transactions as a result (Malmgren, 1961).
The main deficiency of these explanations, however, is that transactions costs are not well enough defined so as to be measured, leading to difficulties in testing them quantitatively (Flaherty, 1981). In addition, even if firms are integrated to internalize both technological and transactional economies, market imperfections caused by externalities, imperfect competition, or imperfect information remain an important determinant of vertical integration. Vertical integration in response to such technological or transactional economies generally increases welfare. On the other hand, in response to market imperfections, it may increase or decrease welfare, depending on its use. Thus, public policy inquiries sorting out these effects become the primary interest (Perry, 1989).
The Source of Program Diversity
These efficiency theories have been championed by both the NTIA (1988) and the FCC (1990) who have contended that vertical integration has expanded the supply of cable programming in an elastic fashion and thus significantly improved diversity of viewing choices for cable subscribers. The FCC also states that vertical integration has increased not only the quantity but also the "quality" of program services available to the viewing public. In the views of the NTIA and the FCC, horizontal concentration and vertical integration produce significant benefits and strong purchasing value for cable subscribers. The correspondingly higher concentration levels in the cable industry have enabled cable companies to take advantage of valuable economies of scale and foster investment in more and better program sources, more original programming and a wealth of new viewing options for cable subscribers (FCC, 1990). Cable concentration thus is considered an economic "good" rather than a "bad," yielding superior performance for the industry. In more specific terms, MSOs are said to provide critical financial support to sustain such companies as Turner Broadcasting, The Discovery Channel, BET, and C-SPAN in their expanded programming efforts.
The proponents of vertical integration also note certain "synergies." In addition to providing needed capital and a ready subscriber base for new programming services, cable operators can more easily share information with producers about viewer taste, reaction to programs and desire for new programs (Klein, 1989). More generally, executives and managers who are skilled at cable system operation may also be skilled at cable network operation (Waterman & Weiss, 1993b).
In sum, with regard to service prices, it is argued that vertical integration is likely to result in lower consumer prices. On the issue of diversity, it is suggested that all of these advantages of vertical integration ultimately benefit the viewing public because they are likely to result in an expansion of the amount and quality of the programming that is available to the public and increased consumer satisfaction (Klein, 1989).
Anticompetitive Consequences of Vertical Integration
Entry Barrier and Foreclosure Effect
Countering these arguments that vertical integration increases efficiency and welfare, a number of economists (Stigler, 1951; Edwards, 1953; Adams & Dirlam, 1964; De Chazeau & Kahn, 1959; Kahn, 1971; Litman, 1979a, 1992a; Eccles & White, 1988; Rotemberg, 1991) pay more attention on inefficiencies and possible anticompetitive consequences associated with vertical integration. It is first argued that not only are internal transactions not free, but also integration may bring problems of coordination and other managerial diseconomies. Thus, vertical integration is considered more costly and riskier than operating at a single stage; there may be considerable inefficiencies which result from trying to operate additional stages at optimal levels. For example, often the economies of scale do not coincide between adjacent stages (Litman, 1992a).
Furthermore, there exist situations where, given market imperfections, the firm's private profitability creates undesirable social by-products. When the market is distorted, firms generally can earn rents. Thus, firms would willingly tolerate organizational inefficiencies if such inefficiencies help them capture these rents (Rotemberg, 1991). Consumers can be made better off when there is vertical integration in the market system, only if competition insures and compels that such cost savings are indeed passed on to them (Carlton, 1979). Unfortunately, lacking competition in the local exhibition stage, structural imperfections cannot compel good performance for the cable operating systems.
Moreover, it is possible that vertical integration increases the difficulty of entry by new firms, by increasing the capital and knowledge necessary to conduct several types of operation rather than depending on rivals for supplies or markets (Stigler, 1951). Established firms may use vertical integration strategically to increase financial requirements and thereby to discourage entry if potential entrants feel compelled, as a condition of successful entry, to adopt the prevailing structure to enter at multiple stages.
Furthermore, integration confers advantages of a purely strategic character, by giving a company more certain access than nonintegrated competitors enjoy to materials and markets, and by protecting it against the squeezes on prices and margins that may afflict firms confined to one particular stage of production. Therefore, to the extent that integration narrows the alternatives intermediate markets offer to nonintegrated competitors, it becomes necessary for them to integrate as well (De Chazeau & Kahn, 1959; Comanor, 1967). Hence vertical integration may breed vertical integration apart from any gains in efficiency (Litman, 1979a). It is particularly likely to do so if it carries no offsetting disadvantages. If, for example, integrated firms are very large and at the same time operate with decentralization of management, they may enjoy the advantages of market specialization in their own department while retaining the strategic benefits integration confers: protection, assurance, and balance Pe Chazeau & Kahn, 1959).
Recently the rising trend in vertical integration can be explained in part by the increasing difficulty encountered by new programming networks in gaining access to the steadily expanding universe of competing cable services. Even though some independent networks such as ESPN and USA are among those with the highest subscribership, these popular independent networks were established in the early years of cable before extensive vertical integration and a variety of cable niche programming had been developed in the cable industry. They thus had the built-in advantage of "being there first."
One example of an integrated MSO's abuse of power to curtail programming services competing with their network siblings involves the attempt of NBC and Cablevision to launch CNBC in 1989. In response to the new competition in news programming service, a number of large MSOs insisted, as a condition of carriage, that CNBC not become a general news service in direct competition with CNN, which is owned partly by TCI, Time Warner, Viacom, and other MSOs (FCC, 1990).
By contrast, CNBC's contracts did not preclude it from providing programming that might duplicate or compete with FNN, another consumer news service that was independently owned. The outright intent of the contract was to protect CNN from direct competition to its general news service and to allow CNBC to provide programming that could compete with other unaffiliated program services. This illustrates how vertical integration and associated self-preference can distort cable programming decisions at the margin when other conditions and factors are roughly equal.
The Source of Discriminatory Conduct
The apparent influence of vertical integration on information product retailers is frequently observed. For example, ownership of cable programming could give a cable operator the incentive to carry affiliated services to the exclusion of unaffiliated services. Certain channels such as ESPN, USA, and HBO are, for all practical purposes, economic "must carries" for all cable systems. Other network siblings constitute strategic "must carries" for vertically integrated parent cable operations.
In the Waterman and Weiss studies (1993a; 1993b), MSOs that vertically integrated with cable networks were found to carry their affiliated networks more frequently than rival networks. More importantly, Waterman and Weiss reveal that when both affiliated and rival networks are carried on a system, subscribership is often skewed toward the former. Due partly to favored promotion and channel position, cable networks that are vertically integrated enjoy significantly higher subscribership and correspondingly higher ratings than new nonintegrated networks. Only three post-1984 Cable Act channels appear among the 15 channels with the highest subscribership and ratings; all three, TNT, The Discovery Channel, and Nickelodeon, are vertically owned (FCC, 1990). This implies that there are many subtle advantages afforded to the integrated networks.
Moreover, almost all of the contracts between unaffiliated program services and cable operators give the operators "deletion rights"--the right to discontinue carriage of a program service. Such deletion rights provisions, however, appear to be absent from affiliation agreements between integrated MSOs and their own programming services. The Nostalgia Network lost a chunk of its subscriber base because Time Warner Cable decided to drop the network from its giant New York City operation in January 1994. The same cable system dropped The Travel Channel a year before (Higgins, 1994). Needless to say, these two programming networks are not financially affiliated with Time Warner. The biggest cable players thus have virtual veto power over access and can also utilize reciprocity where necessary to insure successful launches of new affiliated parent networks. As horizontal power continues to consolidate among MSOs through large scale mergers, this strategic advantage and leverage will assuredly become enhanced.
In sum, the network branch of the integrated firm might use its cable system ties strategically to disadvantage existing or potential network rivals. Not only can the MSO branch of the firm use its network ties to create barriers to entry but it may also disadvantage competing programming services and alternative multichannel video distributors (Waterman & Weiss, 1993b), regardless of conduct sanctions warned by the FCC. Thus, vertically integrated services, on average, tend to achieve greater market penetration in their first two years than non-integrated services. Among fourteen new services since 1990, integrated services increased their average market penetration from 7.2% at launch to 11.1% after a year and 20.8% after 2 years, while services with no MSO affiliation started with a lower rate at 5.4%, but only reached to 8.7% after a year and 7.8% after two years (FCC, 1994). The inevitable result is that vertical integration may reduce the supply of cable programming, disarm and disadvantage new program sources, or limit the diversity of program services available to cable subscribers (NTIA, 1988). Further, parent MSOs' propensity for favoring affiliated networks may lead to a distorted programming package rather than optimal mix of programming available to consumers.
However, there also are counter arguments to the foreclosure effect and discriminatory conduct by integrated MSOs. The cable industry generally relies on the Klein study (1989) to rebut these types of accusations of program access abuse stemming from vertical integration. Klein argued that MSOs did not discriminate against unaffiliated programming networks. Even though Klein concedes that virtually all networks are carried more frequently on integrated MSOs than on nonintegrated MSOs, he insisted that the differences were "quite small" (Klein, 1989, p. 47).
A more thorough and extensive study on this subject was conducted by Waterman and Weiss (1993b). They found that integrated cable systems do tend to favor programming with which they have ownership ties, either by more frequent carriage of those networks or by lowering pricing or more vigorous marketing of their affiliated programming. Nevertheless, their research findings did not support the view that vertical integration is instrumental in preventing or retarding entry of alternative multichannel video delivery systems into competition with established cable systems. Waterman and Weiss suggest that integrated systems have little incentive, pro- or anticompetitive, to disadvantage unaffiliated networks. They concluded that although vertical integration could play an important role in the exercise of anticompetitive behavior in the cable industry, "its role is fundamentally ancillary to that of buying, that is, monopsony, power of MSOs in the programming market" (Waterman & Weiss, 1993b, p. 2). It is this latter contention which requires empirical testing and is justification for the present study.
Hypotheses and Methods
To provide exploratory data and hopefully to resolve the apparent disputes on the vertical integration in the cable industry, this study sought to find out the consequences of vertical integration by looking from the standpoint of the consumer, not the cable operator or the network. Consumer welfare rather than business welfare is the cornerstone for public policy and provides a more objective criterion for judgment. Second, it is more productive to focus on the consumer welfare because economic efficiency can be better measured in the final stage of production by observation of the consumer rather than the business entity. Cable television has provided a striking example of a consumer welfare insensitive industry for years (Jacobs, 1995). In fact, continued disputes about vertical integration stem from the fact that the consumer welfare has been largely ignored. Clearly, an objective measure has been missing to assess the competing arguments.
On the basis of aforementioned theoretical discussions, two hypotheses are formulated:
[H.sub.1]: Cable subscribers of integrated MSOs pay lower subscription fees per channel, compared to those of nonintegrated MSOs.
In contrast to the successive monopoly model of Spengler (1950), when a multiproduct monopolist vertically integrates into the upstream stage, the final price may increase or decrease. Conceptualizing three effects of vertical integration in multiproduct industries such as the cable television industry, Salinger (1988, 1991) found that the monopolist might either lower both prices of the affiliated and the unaffiliated firms' good, lower the price of the affiliated firm's good and raise the price of the rival's good, or raise both prices. In short, the effects of vertical integration are an empirical question (Litman, 1979a; Salinger, 1988, 1991). The second hypothesis states that:
[H.sub.2]: Cable subscribers of integrated MSOs enjoy a greater degree of channel diversity than those of nonintegrated MSOs. The more heavily integrated the system is, the smaller the expected diversity.
If the cable consumers benefit from the vertical integration of the cable industry, the subscribers of integrated MSOs should have greater diversity than those of nonintegrated MSOs. it should be recalled that the proponents of vertical integration argue that MSOs' investments in the upstream distribution stage have facilitated more cable programming networks. Therefore, the subscribers of integrated MSOs deserve to enjoy greater diversity.
Unfortunately, the unique structure of the cable industry appears not to allow the subscribers to enjoy greater diversity. If there were transaction efficiencies from the integration, the profit-maximizing incentives would command the monopolist to carry the affiliated networks, thereby narrowing its freedom of program choice. Only by sheer accident would diversity be unaffected. Hence, vertical integration is not co-extensive with program diversity. Considering the differences in the number of integrated networks among the MSOs, more heavily integrated MSOs should correspondingly show less diversity.
This pattern can be strengthened by the de facto monopoly situation in most local cable markets. Thus, the backward integration of local monopoly is likely to provide cable subscribers with some grotesque, "ungrateful" services for a period of time. In this sense, integration and long-term contracts are oftentimes associated with low quality (Rotemberg, 1991). Eccles and White (1988) report that low quality and internal transactions go hand in hand. They also suggest that transaction costs actually increase when firms engage in internal transactions.
Several studies have attempted to measure the effect of market structure on program diversity. As these have been extensively reviewed elsewhere (Litman, 1992b) and are not germane to the issue at hand, their results are omitted here. Nonetheless, two important findings are worth mentioning. First, Owen, Beebe and Manning (1974) and Owen and Wildman (1992) demonstrated that diversity was functionally related to market structure and technological factors affecting the number of available channels and the financing method of generating revenues. More important is Litman's (1992b) insightful finding that diversity is a commodity. He aptly explicates the abstract concept of program `diversity' by assuming that diversity is more like an economic product than a policy or performance goal. As an economic product, it operates in a marketplace and generates positive utility to those who consume it. Most importantly, "diversity obeys the laws of supply and demand" (Litman, 1992b, p. 147). For the basic cable and other multichannel television industries on the horizon, according to Litman, "diversity is the product itself" (Litman, 1992b, p. 148) and there is an economic tradeoff between the extant level of programming diversity and its quality.
For this study, the term vertical integration is operationalized as the cable system operator's ownership of cable programming networks. More specifically, vertical integration is defined as a cable system operator's possession of at least 5 percent share of the programming network's equity.
The 25 largest MSOs were divided into two groups. The first group consists of TCI, Time Warner, Comcast, Cox, Newhouse Broadcasting, and Viacomaethe so-called "integrated MSOs" that own at least four networks. Based on the list of cable systems in Television and Cable Factbook 1993 Edition, 250 local cable systems from these MSOs were selected based on a weighed random sampling method.
The second group consists of nine MSOs with no financial interest in cable programming networks. This "nonintegrated group" is composed of Cablevision Ind., Adelphia Communications, Falcon Cable, Century Cable, Sammons, Colony, Crown Media, Intermedia, and Prime Cable. Again, 250 local cable systems were selected for inclusion in the sample. Due to incomplete or out-of-date information on systems initially selected in the sample, 59 cases were eliminated, resulting in the analysis of a total of 441 cases.(1)
There still remain 10 MSOs that have ownership interests in more than one cable programming network, but less than four. For comparison reasons, these MSOs are excluded from the sample. This threshold of four networks is not arbitrary. No MSO in this group is affiliated with the top 25 cable networks in terms of subscribership reach. Rather, they have ownership ties with very minor networks such as Mind Extension University, Prevue Guide, or Viewer's Choice Networks.
Consumer welfare has two dimensions in this study: the service price and the channel diversity, the dependent variables of this study. The price was measured in two dimensions: the monthly subscription fee for the basic programming and the price per channel. The monthly subscription fee cannot by itself be a comparable measure for the effect of vertical integration because it is dependent on the number of channels, the number of subscribers, and other media environments. A more precise measure for the comparison should be the price per channel, which was measured as the "basic" subscription fee divided by the number of "basic" channels. Here the term "basic" channel includes broadcast channels and advertiser-supported cable channels but excludes the public-educational-governmental access channels.
Diversity is divided into two categories: absolute and relative. The absolute measure indicates how many different channel types are found in a cable system. To promote comparisons, however, Levin (1971, 1980) defined absolute diversity as the number of different channel types carried by a cable system divided by the total number of channel types for the entire cable industry. This measure indicates how close to the total potential diversity any system gets. Following Levin (1971), a second measure, "relative diversity," was defined as the number of different channel types divided by the channel capacity of the individual cable system. This relative diversity indicates how many diverse programming services a system carries, given its available channel capacity. For the comparison of channel diversity across systems, a single measure is more helpful. Average diversity developed by De Jong and Bates (1991), is defined as the simple average of the absolute and relative diversity measures.(2)
Results and Analysis
Table 1 represents general system differences between integrated MSOs and nonintegrated MSOs. The two groups show statistically significant differences in many system characteristics. The most remarkable differences are the number of total basic subscribers and the plant size, integrated MSO systems serve more than two times larger geographical areas than do the nonintegrated MSO systems, with the former having two and half times more subscribers than the latter. The integrated systems' larger plant size and enhanced subscriber base are associated with 4.74 more channels than their counterparts.
Table 1 Comparison of Systems, Prices, and Diversity
Variables Integrated MSOs Nonintegrated MSOs SYSTEM CHARACTERISTICS Number of Basic Subscribers 25,717.7 10,055.4 Total Channel Capacity 40.2 40.2 Number of Channels Operated 35.1 30.4 Year of Operation 20.5 18.2 Miles of Plant 444.2 208.1 Number of Broadcast Channel 8.5 7.9 Number of Basic Cable Channel 21.8 18.0 Total Number of Basic Channel 30.3 26.0 Number of Premium Channel 4.8 4.4 PPV Capability (%) 39.0 26.3 PRICES: Basic Service Fee 19.06 19.24 Price per Channel .68 .83 Price per Diversity 1.14 1.44 DIVERSITY: Basic Diversity 17.77 15.34 Total Diversity 20.24 17.66 Absolute Diversity .63 .55 Relative Diversity .58 .58 Average Diversity .60 .57 Variables Differences SYSTEM CHARACTERISTICS Number of Basic Subscribers 15,662.3(**) Total Channel Capacity - Number of Channels Operated 4.7(*) Year of Operation 2.3 Miles of Plant 236.1 Number of Broadcast Channel .6(*) Number of Basic Cable Channel 3.8(*) Total Number of Basic Channel 4.3(**) Number of Premium Channel .4 PPV Capability (%) 12.7(**) PRICES: Basic Service Fee .17(**) Price per Channel .15(**) Price per Diversity .30(**) DIVERSITY: Basic Diversity 2.43(*) Total Diversity 2.57(**) Absolute Diversity .08(**) Relative Diversity .00 Average Diversity .03
(*) p < .05; (**)p < .01
Interestingly, the differentials in the number of channels operated come from "cable" programming channels, not from the differences in the number of "broadcast" signals. Since the total system channel capacity is virtually the same at 40 channels, the integrated systems have higher capacity utilization rates of over 12 percent. On the other hand, there were no significant differences in the number of premium channels or age of the systems.
PPV capability represents the percentages of cable systems that offer the subscriber pay-per-view options. To the extent that PPV capability represents technological progressiveness and undoubtedly greater efficiency as a byproduct, integrated MSOs are more technologically progressive and efficient (39%) than are nonintegrated MSOs (26%).
An analysis of the correlation coefficients among the variables included in the analysis demonstrated that the number of subscribers is highly positively related with the plant size (r = .84). It is moderately related with number of basic channels (r = .44), and hence the absolute diversity (r = .46) and the average diversity (r = .38). The price factors, such as the average basic subscription fee and the average price per channel, showed lower relationships with the number of subscribers.
As commonly expected, channel capacity of the cable system was a major factor explaining the number of basic channels (r = .58), and hence the absolute diversity measure (r = .55). The number of basic channels is highly positively related with the absolute diversity (r = .92) and highly negatively related with the average price per channel (r = -.79).
Performance and Economies of Scale
To test the research hypothesis, the monthly retail price of integrated MCOs was compared to that of nonintegrated MSOs. A lack of substantial discounts in the price per channel for the subscribers of the integrated MCOs means either the transaction cost reduction does not exist or cost efficiencies gained from vertical integration are kept by the system operator in its revenue pocket, resulting in no consumer welfare.
As shown in Table 1, nonintegrated systems charge significantly higher price ($0.17, p [is less than] .01) for their basic cable service and when combined with the fact that they have fewer channels, this price effect is intensified when measured relatively on a price per channel basis. Consequently, nonintegrated systems charge $0.15 more per channel relative to the integrated systems. This is a natural result because nonintegrated systems provide fewer channels at nearly the same price with the integrated systems. Considering the differences in system plant sizes and the number of subscribers, this result can be attributable to the economies of scale in the cable services.
The diversity measures show similar results with the price measures. Table 1 shows that integrated MSOs provide greater diversity than nonintegrated MSOs. Integrated MCOs provide their subscribers with approximately 18 different basic program types, while nonintegrated MCOs offer slightly more than 15 program types (p[is less than] .05). Integrated MSOs showed better performance in total diversity that includes basic channels, premium channels, and PPV options. But the difference in total diversity comes primarily from the basic channels (2.43) since premium channels widened the gap only by .13 different program type.
In terms of the absolute diversity, integrated MSOs show better performance than their counterparts do. Though the difference is very small, it is statistically significant (.08, p [is less than] .01). This is also a result of the differences in the number of channels operated. No difference in relative diversity supports this assertion and there is no significant difference in average diversity.
However, these performance measures interact with the economies of scale because the price per channel is closely related to the number of channels operated in each system (r = - .79). Furthermore, it is shown in Table 2 that nonintegrated systems are more likely to be smaller than integrated systems. About half (50.3%) of the nonintegrated systems are serving markets of less than 3,500 subscribers, while only one-third (34.6%) of the integrated systems are serving in the same market size. Moreover, the number of integrated systems serving more than 50,000 subscribers are five times that of the nonintegrated systems. Thus, these overall differences in the number of channels provided and price per channel come most largely from these size differences. Interestingly, when nonintegrated systems reach economies of scale with more than 50,000 subscribers, however, they provide 1.5 more channels than integrated systems and charge almost the same price per channel.
Table 2 also shows interesting relationships among total basic subscribers, the number of channels, basic service fee, and price per channel. Both integrated MSOs and nonintegrated MSOs charge about $19 for their basic services regardless of the differences in the number of channels and subscribers. Though the number of channels provided is increased according to the number of subscribers, the costs remain at approximately $19 all across the systems. The most outright interpretation of this finding is that cable operators prefer to charge similar or fixed prices all across the systems and differentiate their product or provide value with the provision of different numbers of channels, that is, diversity.
Table 2 Economies of Scale in Price per Channel
Number of basic channels(*) Basic fees Total Systems 28.1882 19.1512 Integrated MSOs 30.2632 19.0665 -1000 20.5122 17.5998 3500 26.9474 19.0021 10000 31.7209 19.0453 50000 33.4225 19.0213 +50001 37.0857 20.9723 Nonintegrated MSOs 25.9671 19.2418 -1000 18.6379 18.4981 3500 23.4490 18.9714 10000 29.4643 19.5902 50000 31.9524 19.6362 +50001 38.6250 21.7813 Price/Channel(*) Cases (%) Total Systems .7516 441 Integrated MSOs .6750 228(100.0) -1000 .9277 41 (17.9) 3500 .7329 38 (16.7) 10000 .6229 43 (18.9) 50000 .5811 71 (31.1) +50001 .5707 35 (15.4) Nonintegrated MSOs .8335 213 (100.1)(**) -1000 1.1015 58 (27.3) 3500 .8913 49 (23.0) 10000 .6949 56 (26.3) 50000 .6304 42 (19.7) +50001 .5733 8 (3.8)
Total Cases 441 (*) p < .001 (**) exceeds 100.0 due to rounding
It seems that the principle of scale economies does not hold in this pricing strategy. Rather, cable operators tend to reduce or increase the number of channels provided according to their specific market situations. For example, very large systems serving more than 50,000 subscribers charge similar prices with smaller systems, but provide more channels and greater diversity. Subscribers served by smaller systems, that is, rural areas, pay the same amount of money for significantly fewer channels and little diversity. In other words, the cable operators are justifying unachieved economies of scale by reducing their programming costs rather than by charging higher prices--lowering the quality of the product rather than the price. This strategy clearly reflects the market power cable systems enjoy in rural areas, and the competition and greater public scrutiny cable operators face in urban areas. In short, the cable operators use the adjustment of diversity, while fixing the service fee at similar prices, in response to each market condition. Litman's (1992b) contention that diversity is commodity and product itself can be better understood in this context.
Then, why do the cable operators prefer to adjust diversity rather than price? First, it seems that differentials in prices among markets are more sensitive and easily detectable than are those in diversity. A slight difference in subscription fees for the "same basic service" in the consumers' eyes may lead to complaints from the subscribers and serious marketing problems for smaller systems. But differentials in diversity may go undetected, thereby allowing the cable operator to avoid complaints from the consumer and strict scrutiny from franchising authority, despite substantial product differences among markets. The result is that cable subscribers of the very smallest systems pay similar prices for half the number of channels of the very largest systems. This differential might best be termed the "rural handicap" and is equivalent to a type of penalty or tax based on population density. In short, the cable operators have the upper hand in dealing with program diversity, which involves negotiation with program providers, but they have some difficulties in increasing the price, which involves negotiation with the consumers and the regulators.(3)
This phenomenon is not unique in cable television; rather the same is true throughout the general mass media where prices remain fixed for periods of time and quality changes, for example, for newspapers, magazines, VCR tapes, and theatrical movies. For all these mass media products, while the quality and popularity vary and determine financial success, prices stay fixed. Only by including quality as a dimension of price, therefore, can we truly understand all the dimensions of downward sloping demand curves for those products.
Price Per Channel and Average Diversity
If the differences in price per channel come from the supposed efficiencies gained from vertical integration, it could be assumed that heavily-integrated systems charge less than lightly-integrated systems. To examine the effect of the degree of vertical integration on the price per channel, the integrated group was disaggregated into the individual MSO to which the systems belong. The result is interesting. Within this grouping of affiliated networks, as the number of affiliated programming networks is increased, the price per channel also increased. Among the integrated MSOs, for example, the highest price per channel ($.74) is charged by TCI systems which have ownership ties with 19 cable programming networks (Table 3). Similarly, Time Warner, the second largest MSO integrated into 14 networks, charge the second highest price per channel ($.67), while Viacom, with 9 integrated networks, the third highest ($.61). At the other extreme, the cheapest price was found in Newhouse Broadcasting systems ($.57) with 6 networks. The price rises slightly with Comcast and Cox ($.59) which have financial interests in 5 networks. This implies a very important finding that there is a threshold demarcating the degree of efficient vertical integration. When nonintegrated, the price per channel is the highest ($.83); wWith six integrated networks, the cable operator charges the lowest price ($.57); and then for inframarginal units, the price is increased as the number of affiliated networks is increased. In other words, there is a decreasing return of vertical integration (see Figure 1).
Table 3 The Diminishing Returns of Vertical Integration
MSOs Number of Price per Average Cases Affiliates Channel Diversity Non-integrated 0 .8336 .5710 211 Comcast & Cox 5 .5865 .6583 40 Newhouse 6 .5690 .6216 19 Viacom 9 .6080 .6060 15 Time Warner 14 .6684 .6124 51 TCI 19 .7443 .5814 102 Total Systems 241 .7518 .5855 441
[Figure 1 ILLUSTRATION OMITTED]
A very similar result was found for the diversity measure. Here again, a decreasing return of vertical integration appears to exist as shown graphically in Figure 1. Compared to the integrated group, nonintegrated MSOs provide less diversity. Among the integrated MSOs, however, the greatest average diversity was found in Comcast and Cox (.6583), whereas the least average diversity was discovered in TCI (.5814). In general, absolute diversity and average diversity are decreased as the number of affiliated networks is increased. But again, both measures of absolute diversity and average diversity are highly likely to be confounded with the number of channels operated.
An important implication from this finding is that a high degree of vertical integration tends to constrict an MSO's freedom of program choices. In other words, the natural propensity or economic imperative to favor their affiliates in carriage decisions may lead to the reduction of the absolute and average diversity since more integrated MSOs have fewer number of options in terms of diversity. For example, if TCI were to favor its affiliates, TCI would provide less diversity than moderately integrated or non-integrated MSOs because not every TCI affiliate provides exclusively discrepant types of programming. In contrast, nonintegrated MSOs can have much wider spectrum of programming options and hence presumably fewer duplicated program options.
Therefore, the more integrated the system, the less the diversity. Both absolute diversity and average diversity are decreased, as the degree of vertical integration is increased (p [is less than] .001). The lower absolute diversity and the average diversity of the nonintegrated MSOs are a direct outcome of their fewer channel capacities as indicated by correlation coefficients between those two variables (r = .55 for the absolute diversity, r = .43 for the average diversity). Lack of significant differences in the relative diversity supports this inference.
With these findings on both price per channel and average diversity, the final task is to seek out the effect size and exact apogee of beneficial vertical integration. It is not clear with the primary data whether such apogee is reached at 6 or 7 or 8 or even 9 networks since not only are there no MSOs that have financial interests in 7 or 8 networks, but also the primary data intermingle with various extraneous factors such as the number of subscribers, the number of channels, and other system-specific variables. Thus, finding this threshold requires controlling these extraneous variables and incorporating both dimensions of price and diversity into a single criterion of consumer welfare.
The Diminishing Effect Model of Vertical Integration
Regression analyses were then conducted to control the effect of various system factors on price per channel and average diversity and to statistically uncover the threshold. Considering the economic and technological constraints with regard to cable subscription fees, channel capacity of local systems, and total availability of cable programming options in the industry, it is reasonable to presume that the functional relationship is a type of production function exhibiting the property of diminishing marginal productivity. In other words, the function increases at a decreasing rate. Hence, the desired equation should properly be specified, not by linear regression, but by the log-log model (double log), the log-level model (semilog 1), the level-log model (semilog 2), or reciprocal functional form (See Litman, 1979b). STATA was used to transform the original data into the natural log of the corresponding variable, and to conduct all statistical analyses (Stata Corp., 1995). Since the log-level model for the price per channel and level-log model for average diversity, usually expressed as logY = a + bX and Y = log a + b logX, respectively, provided the best fits, they have been summarized in Table 4.(4)
Table 4 Regression Analyses Regression of log Price per Channel Independent Variables Coefficients Number of Basic Channels -.035559 Number of Subscribers (in hundreds) .000109 Pay-Per-View (binary) .055847 Penetration Rates (%) -.001557 Vertical Integration -.026885 Vertical Integration Squared .001451 Year of Operation -.001637 Constant .780863 Overall F (7, 376) = 119.03, p = .0000; [R.sup.2] 69; Adj. [R.sup.2] = .69 (*) significance at the two-tailed test Regression of Average Diversity Independent Variables Coefficients log Number of Basic Channels .145685 log Number of Subscribers (in hundreds) .009140 log Vertical Integration -.017208 Penetration Rates (%) -.000075 Constant .123571 Overall F (4, 205) = 49.4, p =.0000; [R.sup.2] =.49; Adj. [R.sup.2] =.48 (*) significance at the two-tailed test Regression of log Consumer Welfare Ratio Independent Variables Coefficients Number of Basic Channels .048986 Number of Subscribers (in hundreds) -.000114 Penetration Rates (%) .001644 Vertical Integration .029466 Vertical Integration Squared -.001582 Year of Operation .001489 Constant -1.727596 Overall F (6, 366) = 174.36, p =.0000; [R.sup.2] =.74; Adj. [R.sup.2] = .74 (*) significance at the two-tailed test Regression of log Price per Channel Independent Variables Standard Error Number of Basic Channels .001464 Number of Subscribers (in hundreds) .000032 Pay-Per-View (binary) .027499 Penetration Rates (%) .000680 Vertical Integration .006292 Vertical Integration Squared .000337 Year of Operation .001110 Constant .057488 Overall F (7, 376) = 119.03, p = .0000; [R.sup.2] 69; Adj. [R.sup.2] = .69 (*) significance at the two-tailed test Regression of Average Diversity Independent Variables Standard Error log Number of Basic Channels .022006 log Number of Subscribers (in hundreds) .003676 log Vertical Integration .009281 Penetration Rates (%) .000285 Constant .072259 Overall F (4, 205) = 49.4, p =.0000; [R.sup.2] =.49; Adj. [R.sup.2] =.48 (*) significance at the two-tailed test Regression of log Consumer Welfare Ratio Independent Variables Standard Error Number of Basic Channels .001731 Number of Subscribers (in hundreds) .000039 Penetration Rates (%) .000864 Vertical Integration .007908 Vertical Integration Squared .000419 Year of Operation .001406 Constant .072785 Overall F (6, 366) = 174.36, p =.0000; [R.sup.2] =.74; Adj. [R.sup.2] = .74 (*) significance at the two-tailed test Regression of log Price per Channel Independent Variables t p(*) Number of Basic Channels -24.309 .000 Number of Subscribers (in hundreds) 3.430 .001 Pay-Per-View (binary) 2.031 .043 Penetration Rates (%) -2.288 .023 Vertical Integration -4.273 .000 Vertical Integration Squared 4.301 .000 Year of Operation -1.474 .141 Constant 13.583 .000 Overall F (7, 376) = 119.03, p = .0000; [R.sup.2] = .69; Adj. [R.sup.2] = .69 (*) significance at the two-tailed test Regression of Average Diversity Independent Variables t p(*) log Number of Basic Channels 6.620 .000 log Number of Subscribers (in hundreds) 2.486 .014 log Vertical Integration -1.854 .065 Penetration Rates (%) -0.264 .792 Constant 1.710 .089 Overall F (4, 205) = 49.4, p =.0000; [R.sup.2] =.49; Adj. [R.sup.2] =.48 (*) significance at the two-tailed test Regression of log Consumer Welfare Ratio Independent Variables t p(*) Number of Basic Channels 28.303 .000 Number of Subscribers (in hundreds) -2.900 .004 Penetration Rates (%) 1.902 .058 Vertical Integration 3.726 .000 Vertical Integration Squared -3.778 .000 Year of Operation 1.059 .290 Constant -23.048 .000 Overall F (6, 366) = 174.36, p =.0000; [R.sup.2] =.74; Adj. [R.sup.2] = .74 (*) significance at the two-tailed test
The regression models explained 69% of the total variance in price per channel (adjusted [R.sup.2] = 69%) and 49% of the variance in average diversity (adjusted [R.sup.2] = 48%), a result that is statistically significant (p [is less than] .001). As expected, both price per channel and average diversity were highly dependent upon the number of basic cable channels. In our log-level model for the price per channel, one more basic channel is predicted to decrease price per channel by 3.6% (p [is less than] .001), holding the other factors fixed, whereas 100 additional subscribers are predicted to increase the price by 1.1% (p [is less than] .01). A 1% increase in penetration rates is predicted, ceteris paribus, to decrease price per channel by .16% (p [is less than] .05). The coefficient on pay per view capability, the only dummy variable in the equation, suggests that price per channel charged by cable systems with PPV technology is 6% higher than the price charged by systems without such technology (p [is less than] .05). Even though all the above variables are statistically significant, their effect sizes are not economically very large, with the exception of the number of channels and PPV.
Regression results confirm that vertical integration has tremendous implications for price per channel since it shows diminishing returns as well as relatively large economic effects. It is predicted, ceteris paribus, to decrease price per channel before it reaches a level, but after that level, to increase price per channel. For example, one integrated programming network is predicted to decrease price per channel by 2.4% (-.02688 + 2 x .00145 = -.02398). However, this percentage effect of having more integrated networks changes as the number of networks changes. For those systems with financial interest in fewer than 10 networks, adding one more network returns a beneficial effect on price per channel, resulting in reduced prices for the consumer. With more than 10 networks, however, the partialled-out effect of vertical integration becomes detrimental to consumers (i.e., increase prices), and decreases at an increasing rate as more networks are added. In more specific terms, in our model, the price is decreased by 1.2% with 5 networks, .95% with 6 networks, and .08% with 9 networks. However, it is increased by .2% with 10 networks, the threshold of increasing price, by 1.4% with 14 networks, by 2.8% with 19 networks, and so on. Thus, including the quadratic term of vertical integration in the equation allows us to model the decreasing rate of return, in terms of price per channel, of having more integrated programming networks on the local cable systems. Further, not only is the size of this effect economically large compared to other factors, but both component terms of vertical integration are highly statistically significant (p [is less than] .001). Together, these two components imply that the quadratic term of vertical integration should be included in the model to more precisely estimate the effect of vertical integration. It should also be remembered that this effect is irrespective or independent of other variables, such as the number of subscribers and the number of channels, that are closely related to scale effects.
The regression estimates for average diversity resulted in very similar outcomes. Holding the other factors constant, a 10% increase in the number of channels is predicted to increase average diversity by .015 (p [is less than] .001). The number of subscribers has a very small effect on average diversity, even if it is statistically significant. Since log[(X).sup.2] is an exact linear function of log(X), the squared term of log(vertical integration) cannot be included in the equation. It can be said that a 10% increase in vertical integration is predicted to decrease average diversity by .002, although this effect is not very large, nor is it statistically very significant (p [is less than] .10). it should be noted, however, that the level-log model itself is the representation of the diminishing effect model and the coefficient on vertical integration is negative, meaning vertical integration has an adverse effect on average diversity.
Supposedly, the assumption of both the efficiency model and the foreclosure model is that the relationship between consumer welfare and vertical integration is linear. The efficiency model implies that consumer welfare increases as the degree of vertical integration increases, while the foreclosure model suggests that consumer welfare decreases as the degree of vertical integration increases. Our analysis suggests, however, that the relationship is not a linear, but a quadratic functional form.
The remaining questions are (1) how to incorporate both critical dimensions--price and diversity--into a single criterion of consumer welfare, and (2) at what point this concept of consumer welfare can be maximized with respect to vertical integration. For the first question, we believe that consumer welfare here can be measured by the Consumer Welfare Ratio (CWR) which incorporates both dimensions of consumer welfare by dividing average diversity by price per channel (CWR=average diversity/price per channel). This Consumer Welfare Ratio will increase as its numerator is increased (r = .70 with average diversity) or its denominator is decreased (r = -.81 with price per channel). Thus, it will be maximized when diversity is increased or when price is decreased.
Regression of CWR is also reported in Table 4. With the log-level model in this case, the equation explained an impressive 74% of the variance in the consumer welfare as a single dependent variable. One basic channel is predicted to increase consumer welfare by 5% (p [is less than] .001), while one hundred subscribers are predicted to have a ceteris paribus effect of -.01%. A 1% increase in the penetration rate is predicted to increase consumer welfare by .16% (p [is less than] .10). While the effect sizes of subscribers and penetration rate are statistically significant, they are negligible.
Vertical integration, on the other hand, turned out to have a relatively large effect as well as high significance. The regression analysis here demonstrates a diminishing effect of vertical integration on consumer welfare, as proved by a positive coefficient on vertical integration and a negative coefficient on vertical integration squared
The second question involves finding the turning point of the quadratic equation with vertical integration. The turning point can be computed from the coefficients on vertical integration, using a simple formula (T.P. = [[Beta].sub.1]/2[[Beta].sub.2]) which yields 9.3 (-.029466/-2:001582 [approximately equals, congruent] 9.3). This means that consumer welfare varies positively with vertical integration, but it begins to decrease after reaching the apogee of 9.3 vertically integrated programming networks, other factors held constant.
The monopoly status of local cable systems has been obtained through a declining long-run average cost. It is premised on the assumption that the cost for cable services continuously declines as the number of subscriber increases. When a business enterprise possesses monopoly power at one level of production or distribution, vertical integration may enable it to extend that power to earlier or later levels, or use it for strategic advantages of access or product foreclosure. Such a case arises where a part of the integrated concern's field of operation lies in a segment of the economy that is exempt from the general public policy of competition. The cable business operates in a field which entry has historically been limited by local governments.
The possibility of anticompetitive effects is remote as long as the vertically integrated concern controls only a small fraction of the commerce conducted at each successive level of activity in the field of business in which it operates and so long as its size is not significantly different from the size of its rivals. But when a vertically integrated concern controls a substantial part of the total volume of business at one or more levels of activity, as it extends over several successive levels, there develop special opportunities for the exercise of strategic business power (Edwards, 1953). Hence monopoly power can be exacerbated.
The FCC announced its new "cost-of-service" rules in 1994. In the new rate regulation, it established a uniform rate of return at 11.25%. Along with taxes on the provision of cable service, franchise fees and the costs of meeting franchise requirements, programming costs are also included in external costs that will be exempted from the base for the rate (47 USC 76.922). In this new regulatory scheme, the extensive vertical integration in the cable industry casts another implication. Uninhibited by regulatory constraint, a cable monopolist could be expected to exercise full monopoly power in the exhibition stage and would not have any incentive to integrate backward into the distribution and the production stages. But assuming that a cable system were subjected to effective rate-of-return regulation, a unique incentive for vertical integration into the upstream stage is created-avoidance of rate-of-return regulation, as is often attributed to AT&T during its era of vertical power. While we now appear headed toward a laissez faire period of unfettered competition and convergence across transmission technologies, the pendulum may swing back again toward re-regulation of prices should the dream and promises of full-fledged rivalry not materialize. It is then possible for the firm to circumvent the effect of the regulatory constraint completely by transfer pricing the internally supplied intermediate product above its marginal production cost. The practical difficulties inherent in analyzing separated costs between the various levels of an integrated enterprise make the task of the regulatory body extremely difficult even assuming that it can bring together the records of the company from every level. A number of economists (Adelman, 1949; Stigler, 1951; Edwards, 1953; Kahn, 1971) have pointed out this possibility of avoiding the rate regulation by adopting vertical integration.
Even in this case, however, the consumer welfare should be taken into account as an important standard for the final judgment. Here, we took a look at the performance of cable operators rather than their conduct. Conduct-based research and policy tend to lead to an anthology of competing theories and anecdotes, and fail to make impartial judgment. Performance-based policy making, on the other hand, is much easier and clearer.
The findings of this study confirm neither the efficiency model nor the market power model. Rather, they suggest a compromise between those two competing models. We call it the diminishing effects model of vertical integration. While the integrated MSOs provide the consumer with greater diversity and lower prices than their nonintegrated counterparts, within the integrated group, the consumer benefited more from the moderately-integrated MSOs than from the heavily-integrated MSOs. This means there exists a threshold at which the consumer welfare is maximized from the scale economies and vertical integration in the cable services.
Although our findings do not support complete vertical disintegration, they suggest that there should be partial divestiture of networking from system operation or limits on the extent of vertical integration. In fact, section 11 of the Cable Television Consumer Protection and Competition Act of 1992 requires the FCC to "prescribe rules and regulations establishing reasonable limits on the number of channels on a cable system that can be occupied by a video programmer in which a cable operator has an attributable interest" (47 U.S.C. 533). In 1993, the FCC adopted in its Second Report and Order the channel occupancy rules, which generally prohibit a cable operator from devoting more than 40% of its activated channels to affiliated programming. The current FCC rules include, inter alia, the inclusion of over-the-air broadcast, PEG and leased access channels in its total channel capacity, the application of the limits only to a cable system's first 75 channels, the exemption of local and regional networks from channel occupancy limits, and no exemption of the rules for cable systems facing competition (FCC, 1993).
In general, these rules seem to strike the proper balance between the risks and benefits of vertical integration. Nevertheless, our findings here raise a question on the current channel occupancy limits. Because efficiency gains from vertical integration are not proportionate to channel capacity, "percentage" limitations may not help to achieve the goals of the rules. In other words, there must be at least some absolute number of integrated programming for a cable system to achieve an economic efficiency and eventually for consumer welfare to be materialized. It was about 10 channels in our study. To illustrate the point, let us take two contrary cases. On a system with 20 channels, according to the current rules, only 8 channels could be occupied by integrated programming services. This is not sufficient for a cable operator to fully take advantage of vertical integration, and hence will result in a suboptimal efficiency achievement. On a system with 50 channels, on the other hand, a cable operator is allowed to devote up to 20 channels to integrated programming services. In this case, the benefits of vertical integration will quickly diminish and the risks will significantly arise without any consumer welfare gains.
In addition to these problems, the 40% limit seems to be quite arbitrary. The FCC based its decision only on speculative reasoning. It was a compromise between 20% limits advocated by franchise authorities and citizen groups, and 50% limits favored by the large MSOs. No empirical evidence was cited in making that decision (see FCC, 1993 and 1995).
Perhaps a more precise and better approach, as advocated by the cable industry, would be to implement channel occupancy limits based on bandwidth rather than the number of channels. The limit would then be calculated by counting each 6 MHz segment as a single unit and a cable operator should be allowed to take some amount of total frequencies on the cable system. This approach will encourage cable operators to further invest in new compression technologies while still enhancing the opportunities for them to develop innovative programming services (see FCC, 1993).
Another implication of the present study is found for the Telecommunications Act of 1996 and the recent wave of merger activity by the largest MSOs and the telephone companies. Under the new Act, telephone companies are permitted to offer video services either by distributing programming as a cable television system or by establishing an "open video system" for transport of video programming on a common carrier basis. By letting cable and telephone companies compete and combine more freely, the Act could bring forth some breeze of competition in the long run. In the short run, however, combinations between cable companies are most likely to intensify before full-fledged competition between cable and telephone companies is realized. The largest MSOs will likely prepare for a tumultuous new era of competition through more extensive vertical integration as well as horizontal mergers. Telephone companies' next step in the video programming business would undoubtedly be to enter the distribution and production stages. Mega merger deals by TCI and Time Warner who also are highly vertically integrated suggest poor performance in the area of pricing and diversity since they clearly lie above the optimal threshold level that is beneficial to consumers.
Our regression analyses demonstrate that vertical integration should not be viewed as illegal per se. At the same time, however, it starts being detrimental to the consumer if it reaches beyond the optimal level and limits on horizontal market concentration alone would not be sufficient to cure vertical restraints. Horizontal concentration makes the matter worse, but it is not the sole root of problems. Thus, vertical integration, if accompanied by a competition-suppressing amount of horizontal concentration and if harmful to cost and price reduction, should be seriously questioned. We suggest that MSOs' vertical ownership be limited to fewer than 10 networks for the consumer to benefit from vertical integration in the cable industry. Beyond this threshold lies the area of either economic rent or inefficiencies. The same standard can apply equally to new entrants such as telephone companies and alternative multichannel television distributors. In implementing this policy, however, the FCC might consider a modified rule of reason approach in which the burden of proof would be shifted to those big cable companies to prove that vertical integration above the threshold yields significant efficiencies or synergy. This analysis is based on the economic efficiency role of vertical integration within the technological context of the 1990s. Significant changes in the technology or business organization due to the Telecommunications Act of 1996 may alter the threshold of efficiency. The flexibility of the modified rule of reason approach should provide a means to deal with the rapidly changing circumstances in this industry.
Even if our findings look like complex compromise between the two models, our conclusion is simple and uncompromising. Economies of scale certainly help to achieve efficiencies. So does vertical integration. But the critical point that is forgotten is that it works so only within some limits; it may not be linearly good in all its dimensions.
Faced with increasing activity in, and combination of, vertical integration and horizontal concentration in the cable industry, much more vigilant investigations are warranted than ever. This study could not reveal the effect of rate regulation since the data were gathered before the new rate regulation. it should also be noted that the diversity measure in this study accounts only for the "breadth" of program diversity. Together with "depth" of program diversity, more complete measures of consumer welfare should also encompass "quality" dimensions of cable programming (See Litman, 1 992b). Future studies are expected to do this task.
(1) Due to limited space, a complete list of samples is not reported here. Sampling data and correlation coefficient matrix are available from the authors.
(2) Categorization of Channel Types
Local Channels: Broadcasting network affiliates: ABC, NBC, CBS, FOX; Independent stations: Broadcasting stations (local & superstations); Public broadcasting stations: PBS; Local origination: PEG access and other (weather, time, ad. etc.); National/Regional Services: General interest: USA, TNT, Nick at Nite; Movie emphasis: HBO, Showtime; Arts/cultural emphasis: A&E; Entertainment emphasis: E! TV; Comedy emphasis: Comedy Central; News: CNN, HLN; Sports: ESPN, Prime Ticket, PASS (All regional sports networks); Business/finance: CNBC, BIZNET; Public affairs: CSPAN I & II, Court TV, Public Interest Video Network; Music/videos: MTV, VH-1, Hit Video USA, TNN, CMT, Video jukebox Network/The Box; Nature and science: Discovery; Weather: Weather Channel; Religious programming: ACTS, TBN, VISN, The Inspirational Network, EWTN; Shopping: QVC Networks, HSN I & II; Adult programming: Playboy, Spice; Travel programming: Travel Channel; Foreign language: Galavision, Univision, Jewish Television Network, The Greek Channel, SCOLA, International Channel; Education: Learning Channel, Mind Extension Univ.; Minorities/special interest: BET, America's Disability Channel/Silent Network; Women: Lifetime; Family: Disney, The Family Channel; Children: Nicklodeon; Movies/classic: American Movie Classics, Encore, Nostalgia; Current: Cinemax, The Movie Channel; International: Bravo; Pay Per View: Action PPV, Cable Video Store, Guest Cinema, Request TV, Viewer's Choice I, II, Prism; Other: Cancom, Prevue Network, EPG
(3) Theoretically and legally, diversity is independent of regulatory pressure since the cable operator enjoys complete discretion over programming decision. Franchising authority is prohibited from requiring the cable operator to provide any specific types or channels of programs.
(4) Typically, the log transformation is used for variables with large integer values such as income, number of employees or sales revenues, etc. But here we took logs to make the interpretation of regression coefficients easier and simpler, even if the variables did not have such a characteristic.
Adams, W., & Dirlam, J. B. (1964). Steel imports and vertical oligopoly power. American Economic Review, 14, 626-655.
Adelman, M. A. (1949). Integration and antitrust policy. Harvard Law Review, 63, 27-77.
Blair, R., & Kaserman, D. (1983). Law and economics of vertical integration and control. New York: Academic Press.
Bork, R. (1954). Vertical integration and the Sherman Act: The legal history of an economic misconception. University of Chicago Law Review, 22, 183-84.
Brennan, T. J. (1990). Vertical integration, monopoly and the First amendment. journal of Media Economics, 3, 57-76.
Carlton, D. W. (11979). Vertical integration in competitive markets under uncertainty. Journal of Industrial Economics, 27, 189-209.
Coase, R. (1937). The nature of the firm. Economica, 4, 387-405.
Comanor, W. (1967). Vertical mergers, market powers, and the antitrust laws. American Economic Review, 57, 254-265.
De Chazeau, M., & Kahn, A. (1959). Integration and competition in the petroleum industry. New Haven: Yale University Press.
De Jong, A., & Bates, B. (1991). Channel diversity in cable television. journal of Broadcasting & Electronic Media, 35, 159-166.
Eccles, R., & White, H. (1988). Price and authority in inter-profit center transactions. American journal of Sociology, 94, S17-S51.
Edwards, C. (1953). Vertical integration and the monopoly problem. Journal of Marketing, 17, 404-410.
FCC (1990). Rate deregulation and the commission's policies relating to the provision of cable television service. MM Docket No. 89-600. FCC Record, 5, 4962-5129.
FCC (1993). In the matter of implementation of sections 11 and 13 of the Cable Television Consumer Protection and Competition Act of 1992: Horizontal and vertical ownership limits. FCC Record, 8, 8565-8617.
FCC (1994). In the matter of implementation of sections 19 of the Cable Television Consumer Protection and Competition Act of 1992: Annual assessment of the status of competition in the market for the delivery of video programming. FCC Record, 9, 7442-7637.
FCC (1995). In the matter of implementation of sections 11 (c) of the Cable Television Consumer Protection and Competition Act of 1992: Vertical ownership limits. Memorandum opinion and order on reconsideration of the Second Report and Order. FCC 95-147.
Flaherty, M. (1981). Price versus quantities and vertical financial integration. Bell Journal of Economics, 12, 507-525.
Green, J. (1986). Vertical integration and assurance of markets. In J. Stiglitz & G. F. Mathewson. New developments in the analysis of market structure (pp. 177-207). London: Macmillan.
Higgins, J. (1994, Jan. 10). Nostalgia losing 400,000 subs in NYC. Multichannel News, pp. 2, 8.
Jacobs, R. (1995). Exploring the determinants of cable television subscriber satisfaction. Journal of Broadcasting & Electronic Media, 39, 262-274.
Kahn, A. (1971). The economics of regulation: Principles and institutions. New York: John Wiley & Sons, Inc.
Kerkvliet, J. (1991). Efficiency and vertical integration: The case of mine-mouth electric generating plants. Journal of Industrial Economics, 39, 467-481.
Klein, B. (1989). The competitive consequences of vertical integration in the cable industry. Paper for the National Cable Television Association, Washington, D.C.
Levin, H. (1971). Program duplication, diversity, and effective viewer choices: Some empirical findings. American Economic Review, 61, 81-88.
Levin, H. (1980). Fact and fancy in television regulation. New York: Russel Sage Foundation.
Lieberman, M. (1991). Determinants of vertical integration: An empirical test. Journal of Industrial Economics, 34, 451-467.
Litman, B. (1979a). The vertical structure of the television broadcasting industry: The coalescence of power. East Lansing, MI: Michigan State University.
Litman, B. (1979b). Public interest programming and the Carroll doctrine: A re-examination. Journal of Broadcasting, 23, 51-60.
Litman, B. (1992a). Mass media economics. Unpublished manuscript.
Litman, B. (1992b). Economic aspects of program quality: The case for diversity. Studies of Broadcasting, February, 121-156.
Machlup, F., & Taber, M. (1960). Bilateral monopoly, successive monopoly, and vertical integration. Economica, 27, 101-119.
Malmgren, H. (1961). Information, expectations and the theory of the firm. Quarterly Journal of Economics, 75, 399-421.
McGee, J., & Bassett, L. (1976). Vertical integration revisited. Journal of Law and Economics, 14, 17-38.
National Telecommunications and Information Administration (1988). Video program distribution and cable television: Current policy issues and recommendations. NTIA Report 88-233, June.
Owen, B., & Wildman, S. (1992). Video economics. Cambridge, MA: Harvard University Press.
Owen, B., Beebe, J., & Manning, W. (1974). Television economics. Lexington, MA: Lexington Books.
Perry, M. (1978). Vertical integration: The monopsony case. American Economic Review, 68, 561-570.
Perry, M. (1989). Vertical integration: Determinants and effects. In R. Schmalensee & R. Willig (Eds.), Handbook of industrial organization, Vol.1 (pp. 183-255). Amsterdam, Netherlands: Elsevier Science.
Rotemberg, J. (1991). A theory of inefficient intrafirm transactions. American Economic Review, 81, 191-209.
Salinger, M. (1991). Vertical mergers in multi-product industries and Edgeworth's paradox of taxation. Journal of Industrial Economics, 39, 545-556.
Salinger, M. (1988). Vertical mergers and market foreclosure. Quarterly Journal of Economics, 345-356.
Spengler, J. (1950). Vertical integration and antitrust policy. Journal of Political Economy, 58, 347-352.
Stata Corp. (1995). Stata Statistical Software: Release 4.0, College Station, TX: Stata Corporation.
Stigler, G. (1951). The division of labor is limited by the extent of the market. Journal of Political Economy, 59, 185-193.
United States v. Paramount Pictures (1948). 334 U.S. 131.
Waterman, D. (1993). A model of vertical integration and economies scale in information product distribution. Journal of Media Economics, 6, 23-35.
Waterman, D., & Weiss, A. (1993a). The effects of vertical integration between cable television systems and pay cable networks. Unpublished manuscript.
Waterman, D., & Weiss, A. (1993b). Vertical integration in cable industry, Paper prepared for The American Enterprise Institute for Public Policy Research. Washington D.C.
Williamson, O. (1971). The vertical integration of production: Market failure considerations. American Economic Review, 61, 112-127.
Williamson, O. (1974). The economics of antitrust: Transaction cost considerations. University of Pennsylvania Law Review, 122, 1439-1496.
Williamson, O. (1979). Assessing vertical market restrictions: Antitrust ramifications of the transaction cost approach. University of Pennsylvania Law Review, 127, 953-993.
Williamson, O. (1985). The Economic Institutions of Capitalism. New York: Free Press.
Hoekyun Ahn (M.A., Michigan State University, 1994) is a doctoral student in the Mass Media Ph.D. program at Michigan State University. His research interests include telecommunication economics and policy.
Barry R. Litman (Ph.D., Michigan State University, 1976) is Professor in the Department of Telecommunication at Michigan State University. His research interests include telecommunication and mass media economics. An earlier version of this article was presented to the Association for Education in Journalism and Mass Communication, Washington, D.C., in August 1995. This manuscript was accepted for publication in July 1996.…