The welfare and equity implications of competition in television broadcasting: the role of viewer tastes

J Cult Econ (2006) 30:127 140 DOI 10.1007/s10824-006-9007-6 ORIGINAL ARTICLE The welfare and equity implications of competition in television broadcasting: the role of viewer tastes Yong Liu Daniel S. Putler Charles B. Weinberg C Science + Business Media B.V. 2006 Abstract This paper studies the behavior of commercial television broadcasters in markets where the distribution of viewer tastes varies. Our results show that a highly clustered market enables the broadcaster to offer a program of a popular type but with lower quality (i.e., lower production values) than is the case when viewers have more diffused tastes. We find that viewer equity in the television market (i.e., the percentage of all potential viewers who have at least one program they consider worth watching) and viewer welfare (i.e., total consumer surplus) may not coincide. Depending upon the distribution of viewer tastes and the cost of providing quality programming, the number of broadcasters required to fully cover the market to avoid market failure may be greater than the number of broadcasters that produce greater viewer welfare. The study suggests that regulatory bodies need to pay attention to the distribution pattern of viewer tastes and the broadcasters desire for return on programming investments, since both factors have important implications for competitive outcomes and viewer well-being. Keywords Broadcasting. Television. Competition. Viewer taste. Viewer welfare Y. Liu Assistant Professor at the Martin J. Whitman School of Management, and Fellow at the Center for the Study of Popular Television, S.I. Newhouse School of Public Communication, Syracuse University, Syracuse, NY 13244, USA e-mail: yoliu@syr.edu D. S. Putler Associate Professor, Sauder School of Business, University of British Columbia, Vancouver, BC V6T 1Z2, Canada e-mail: putler@sauder.ubc.ca C. B. Weinberg Presidents of SMEV Professor of Marketing, Sauder School of Business, University of British Columbia, Vancouver, BC V6T 1Z2, Canada e-mail: weinberg@sauder.ubc.ca

128 J Cult Econ (2006) 30:127 140 1. Introduction The broadcast television industry has experienced significant changes over the past two decades. Mainly due to technological advances, notably digital cable and satellite broadcasting, television viewers across the globe have more channels and programs from which to choose. The competition among broadcasters has also intensified as the trend of deregulation continues and as U.S. style advertising-supported networks become increasingly significant worldwide (Vogel, 2001; Walker and Ferguson, 1998; Barwise and Ehrenberg, 1988). However, the effects of competition on viewer well-being are complex. It is not always clear that companies, when faced with increased competition, will necessarily behave in a way that is beneficial to consumers (Stiglitz, 1981; Ma and Burgess, 1993). Commercial broadcasting differs from many businesses in that there are dual clients audience members and advertisers but consumers (audience members) do not pay directly for services (i.e., to watch an individual program). Thus the business of commercial broadcasting can be viewed as producing audience to earn advertising revenue (Owen and Wildman, 1992). A simplified framework of the broadcast television market (Owen et al., 1974) is presented in Figure 1 to illustrate these interactions. Beyond the particular context of television audience and advertisers, researchers have started to examine the economic mechanism of two-sided markets in general (such as the sellers and buyers on ebay), where the existence of two interdependent customer groups drives business strategies and market structure (e.g., Evans and Schmalensee, 2005). This study examines how one important characteristic of television viewers, namely the distribution of viewer tastes, influences market outcomes. The diversity of viewer tastes is self evident (Comstock and Scharrer, 1999), as exemplified by the growth of narrowcasting (such as MTV and CNN) through which networks attempt to profit from viewer diversity. Broadcast television is a very profitable industry. 1 However, from a societal point of view, we are concerned with the issues of viewer equity and viewer welfare, both of which are of critical importance to society and government. Viewer equity relates to the willingness of broadcasters to serve all potential television viewers. Of special concern is whether certain minority groups (who can be represented by the low-density region of the taste distribution) are underserved or even overlooked. Viewer welfare is concerned with the total utility provided by broadcasters to the viewing public. The welfare issue is particularly important given that the public s attitude towards television has gradually declined since the 1960s (Barwise and Ehrenberg, 1988, p. 49). What ultimately drives the level of viewer equity and welfare are the features of television programs offered by broadcasters. In this paper, we focus on two such features, namely the types and quality of TV programs, that interact with viewer preferences to influence how many viewers find the programs worth watching and the amount of utility (which can be informally conceptualized as the level of viewing enjoyment ) they receive. Little published research has examined the implications of the distribution of viewer tastes for public policy. Yet, a few studies show that the distribution pattern of consumer preferences constitutes a significant factor in firms product and pricing strategies (e.g., Neven, 1986; Ansari et al., 1994). In the context of broadcast television, this suggests that the behavior of broadcasters is likely to be influenced by how viewer tastes are distributed 1 Even with the increasingly strong competition from cable and new entertainment/information sources such as Internet services, home-video devices, and computerized-entertainment software (e.g., video games), the broadcasting industry (including television and radio) still has high relative cash flow and pretax margins of over 20%, which few other industries can achieve (Vogel, 2001, p. 196).

J Cult Econ (2006) 30:127 140 129 Fig. 1 The commercial broadcast television market and its main participants. along important dimensions differentiating television programs. Distinctive features of the broadcast television market, especially its dual client structure and its public good character, also need to be considered when the effect of the distribution of viewer tastes is examined. Among recent studies on television and media markets, Dukes and Gal-Or (2003) investigate the conditions under which exclusive advertising contracts benefit both advertisers and the media. Liu et al. (2004) examine television programming strategies under the assumption of uniformly distributed viewer preferences. Zhou (2004) uses a monopoly model to study a television network s decision on the number, length and timing of commercial breaks. By modeling the interaction between advertising and subscription fees, Baye and Morgan (2000) offer theoretical explanations as to why advertising revenue may dominate subscription fees for certain media. In this study, we follow the basic framework of Liu et al. (2004), but relax the important assumption of uniformly-distributed viewer tastes. Furthermore, and different from most of the extant literature, our main concern is on the implications of the distribution of viewer tastes on viewer equity and welfare. 2. Model setup The broadcast television market is influenced by government regulatory bodies (especially the Federal Communications Commission in the U.S.) and three separate entities the television audience, advertisers, and broadcasters themselves. Each of the entities has different objectives and is influenced by the decisions of others. 2.1. The television audience Based on data from thirty-four countries, Screen Digest (February 1992, p. 37) estimates that the average family watches television for about twenty hours per week. How do people choose what to watch? Extant studies on viewing choice find that there is a strong correlation between a viewer s taste and the type of the program she chooses to watch (Goettler and Shachar, 2001; Rust et al., 1992; Lehmann, 1971). For instance, Goettler and Shachar (2001) use an ideal point structure to model the utility that a viewer with a particular taste receives from a particular program type, and find that this structure captures viewing choice well.

130 J Cult Econ (2006) 30:127 140 It is also well recognized that production values (i.e., financial cost) of a program have significant impact on viewing choice. The reason is intuitive things that make programming more expensive for the broadcaster, such as more and better stars, more time spent on editing, the quality of the sound studio used, and more outside locations, are what the typical viewer positively responds to (Barwise and Ehrenberg, 1988). Even though production values is only one of the factors that determine the overall quality of a television program, it is certainly one of the most important and one that can be directly controlled by a broadcaster. Therefore, similar to Liu et al. (2004), we focus on two factors that influence viewing choice the match between a viewer s taste and a specific program type, and the production values of the TV program. We follow Goettler and Shachar (2001) in using an ideal point structure to model the taste component of utility, and combine it with the quality dimension in a linear fashion (see Neven and Thisse, 1990 for similar models in the spatial competition literature). While it is typical in the literature to model the taste dimension and the quality dimension independently, the empirical literature provides support for this approach. For example, in a study about viewing behavior, Lehmann (1971) reports that the dimension of well produced and directed is minimally correlated with taste dimensions such as how humorous it is. This suggests that the dimension of production values and the dimension of viewer taste should be captured separately when formulating a model of viewing utility. For ease of exposition we use the terms production values and quality interchangeably in the rest of the paper, while acknowledging that although production values constitutes a critical component of the total quality of a television program, there are other artistic, social and cultural factors that also influence quality, but they are much less controllable from a managerial point of view. It is assumed that at any given point in time when facing a number of viewing choices, viewer i receives utility u ij from watching program j, u ij = x i y j +s j, (1) where x i is viewer i s taste (or ideal type of program), y j and s j are the type and quality of program j 2. The viewer will choose the program that brings the greatest (and positive) u ij. If two or more programs have the same utility, then the viewer will randomly tune to one of them. If none of the programs brings a positive utility level, then the TV set is turned off and the viewer opts to do something else. The audience share (i.e., ratings) of each broadcaster can then be obtained by aggregating viewer choices. Following the competitive positioning literature initiated by Hotelling (1929), we scale the space of viewer tastes into a linear segment bounded within the unit interval. The symmetric beta distribution is used to model the distribution of viewer tastes. This distribution is very flexible and can be used to represent a continuously varying degree of viewer heterogeneity. In addition, it is bounded within the interval of [0, 1] and thus fits nicely into a Hotelling setup. Furthermore, using a specific density function such as the beta enables us to derive 2 Note that for tractability reasons we make use of a uni-dimensional measure for the taste/type space. This can be conceptualized as either the single most important taste/type factor that differentiates various television programs, or the aggregation of several such factors. What factors should qualify as important is an interesting empirical question, which can be investigated using survey data with multidimensional scaling, or using viewing choice data with a latent-attribute approach. For instance, Goettler and Shachar (2001) find four useful dimensions that distinguish television shows, one of which, for example, is the plot dimension, ranging from shows with intricate, well-developed plot lines to those with simpler, less-developed plots.

J Cult Econ (2006) 30:127 140 131 Fig. 2 Plots of the probability density functions of the symmetric beta distribution. analytical solutions. 3 Ansari et al. (1994) provide an excellent summary of the features of this distribution. The probability density function of the symmetric beta distribution is f (x,β) = Ɣ(2β)[x(1 x)]β 1, x [0, 1],β >0 (2) Ɣ 2 (β) where Ɣ( ) is the Gamma function. The distribution is symmetric around x = 1/2. β is the shape parameter that captures three distinctive patterns greater density in the middle of the linear space, greater density at both ends, and a uniform density across the unit interval. When β>1 the distribution is unimodal with the greatest density at x = 1/2. The larger β, the more concentrated the distribution. When β<1, the distribution is bipolar with the highest density at the two ends. In this case, a smaller β leads to more extreme bi-modality. When β = 1 the beta distribution becomes the uniform distribution that represents an extremely heterogeneous market. Figure 2 plots the probability density function for three particular values of the shape parameter β: β = 2 (unimodal), β = 1 (uniform), and β = 0.5 (bipolar). 2.2. Advertisers and broadcasters Advertisers pay for airtime (typically a 30 second time slot) in which their commercials are placed. A television program that receives higher ratings is able to charge a higher price for its commercial airtime. If broadcaster j receives a ratings level of R j (based on the demand function in Equation (1)), we specify the price it can charge for a (30 second) slot of 3 We derive closed-form solutions whenever possible. In situations where closed-form solutions cannot be derived, we solve the problem numerically using a dense-grid search over the range of parameter values.

132 J Cult Econ (2006) 30:127 140 commercial time to be λr j, where λ is the parameter representing cost-per-thousand (CPM) or cost per point (CPP), both of which are standard industry metrics (Walker and Ferguson, 1998, p. 52). In this study, we will assume that λ is exogenously determined. The notion that ratings drive a broadcaster s revenue and that the broadcast television industry lacks price competition is supported by numerous studies (e.g., Wright, 1994; Goettler and Shachar, 2001; Fournier and Martin, 1983; Fournier, 1985; Litman, 1979). Additional empirical evidence of this can be found in Barwise and Ehrenberg, (1988) and Liu et al. (2004). One important factor leading to a fairly rigid advertising rate per ratings point is the existence of advertisers whose target consumers (thus the target television audience) differ significantly the willingness of diversified groups of advertisers to pay for different audience segments will ultimately drive up the importance of audience size. Another important (exogenous) factor that drives the level of λ is the competition from other advertising media such as newspapers and magazines. 4 Due to the fact that making watchable television is very expensive (Barwise and Ehrenberg, 1988, p. 102), we capture the cost of programming with a convex function (Economides, 1989). To produce a television program with quality s j, broadcaster j needs to incur a cost of cs 2 j. Putting the cost and revenue functions together (and normalizing λ to 1), broadcaster j s operating profit can be written as: π j = R j ms 2 j, (3) where m = c/λ is the normalized cost parameter. Equations (1) to (3) then allow us to derive optimal programming strategies and examine viewer equity and welfare issues accordingly. 3. Analysis In this study we focus on the behavior of a monopoly broadcaster versus that of two duopoly broadcasters. Basic results of the uniform distribution are similar to those of Liu et al. (2004). As a result, we will devote more discussion to non-uniform results and their implications for viewer equity and welfare. 3.1. Monopoly behavior The monopolist makes a quality-type decision by evaluating the tradeoff between the marginal cost of quality provision and the marginal revenue of advertising. The marginal revenue depends directly on the distribution of viewer tastes, which also drives the monopolist s type decision (i.e., where to locate on the taste dimension). 3.1.1. Uniform distribution of viewer tastes If the distribution of viewer tastes is uniform (β = 1), the monopolist does not have any preference in terms of locations, as long as it does not over-cover the market. If it sets a quality level of s and offers a program of type y, then according to Equation (1) the indifferent viewers are located at x L = y s to its left and at x R = y + s to its right. It must ensure 4 Although a few select television programs, such as the Super Bowl in the U.S., can charge a higher rate on a CPM basis, a constant CPM rate is a close approximation for most shows.

J Cult Econ (2006) 30:127 140 133 Table 1 Monopolist s optimal quality for different viewer distributions Note: For 0 <β<1, lower β represents higher viewer density near each of the two end points. For β>1, higher beta represents higher viewer density near the mid-point. Cost parameter Shape of distribution Shape parameter m = 2.70 m = 3.00 β = 0.5 0.236 0.214 Bipolar β = 0.7 0.299 0.266 β = 0.9 0.354 0.314 Uniform β = 1.0 0.370 0.333 β = 1.5 0.343 0.324 Unimodal β = 2.0 0.323 0.309 β = 3.0 0.295 0.285 that the viewer who derives zero utility from watching its program (i.e., who is indifferent between turning the TV on and off) is located on the line segment [0, 1]. That is, y s 0 and y + s 1. The demand equals x R x L = 2s and the profits amount to π = 2s ms 2. Thus the monopolist s optimal quality level is at sunif M = 1/m and it may locate anywhere in [SUnif M, 1 S Unif M ]. Depending upon the cost factor m, it will set s Unif M at 1/m when m > 2, and at 1 / 2 when m 2 (and will locate at the market center, y = 1/2). 3.1.2. Bipolar distribution of viewer tastes If the density of viewer tastes is greater at the two ends of the market than it is in the market center (i.e., where β<1 in the beta distribution), the monopolist will naturally focus on one end of the market unless the cost of serving all viewers is sufficiently low. Specifically, it is optimal for the monopolist to set y = s so that it always covers one end of the market. Putting together Equations (1) to (3), the profit function can be written as π = 2s 0 Ɣ(2β)[x(1 x)] β 1 Ɣ 2 (β) dx ms 2. Since the maximization problem cannot be solved analytically, we derive the optimal sbipo M using numeric search of the parameter space.5 Some representative results are summarized in Table 1. Table 1 shows that optimal quality sbipo M decreases as the distribution becomes more bipolar (i.e., β decreases further below 1). This occurs since the greater concentration of viewers at one end of the market (or more generally, at any point of the market) gives the monopolist an opportunity to offer a program that better matches more viewers tastes, but with lower production values. Put another way, the quality level provided by the monopolist never increases as viewer tastes become more concentrated. 3.1.3. Unimodal distribution of viewer tastes The case where the greatest density of viewers is in the market center (i.e., when β>1in the beta distribution), the monopolist s decision can be analyzed in a similar fashion. In this case the monopolist will always want to locate at the center and decide how much to span out to the two ends of the market. Thus y = 1/2, and π = 1/2+s Ɣ(2β)[x(1 x)] β 1 1/2 s dx ms 2. Ɣ 2 (β) The optimal results for several β values are also provided in Table 1. As the density of viewers increases at the market center (i.e., as β increases), the monopolist takes advantage 5 Which end of the market we deal with is arbitrary. If the right-hand side is chosen (i.e., y = 1 s), the solution is simply the mirror image of that when the left-hand side is chosen (i.e., y = s). All subsequent proofs, derivations and computer programs (written in S-Plus and R) are available from the authors upon request.

134 J Cult Econ (2006) 30:127 140 of the homogeneity of viewer tastes by offering less expensive programming. However, the percentage of viewers served by the broadcaster is always positively correlated with the degree of taste homogeneity. For the specific parameter values provided in Table 1, the monopolist serves the largest number of viewers when β = 3.0. An important feature of the monopoly behavior is that when the market is not fully covered (i.e., s M < 1/2), a higher cost of programming necessarily forces the broadcaster to reduce quality. In the unimodal case, this can be analytically proven since maximizing Ɣ(2β)[x(1 x)] β 1 Ɣ 2 (β) π = 1/2+s 1/2 s SUnim M /[m + 2s Unim M dx ms 2 leads to an implicit function of dsunim M /dm, which equals M2 ρ(β 1)(1/4 SUnim )β 2 ] and is always negative (ρ = Ɣ(2β)/Ɣ 2 (β)). Table 1 illustrates the numerical patterns between m = 2.70 and m = 3.00. If the cost level is not too high, then the monopolist chooses to serve the entire market with optimal quality s M = 1/2. As m becomes sufficiently high, the monopolist will find it optimal to reduce its quality and no longer serve the entire market. A primary consequence of alternative distributions of viewer tastes is to change the marginal revenue of quality provision (MR). MR = 2 in the uniform case. However, the bipolar and unimodal distributions make MR a function of the shape parameter (β) and the quality levels (s). As s increases from low to high, the MR of the bipolar distribution will be first decreasing and then (after s becomes greater than 1 / 4 ) increasing. MR of the unimodal distribution will be always decreasing in s. The optimal quality level is achieved when MR equals the marginal cost of quality provision (MC = 2 ms), a standard result of profit maximization. 3.2. Duopoly behavior With two competitors, we solve the duopoly game assuming that each broadcaster attempts to maximize profits. As they choose locations along the taste/type space, we also assume that a broadcaster will not give up viewers in an uncontested area to fight for a contested area. 6 Two technical issues are worth noting. First, if the cost of providing quality is sufficiently low, the broadcasters will be able to compete by indefinitely increasing quality so that no pure strategy Nash equilibrium will exist. Second, if the cost factor is sufficiently high, optimizing broadcasters will provide such a limited level of quality that each of them will only serve a part of the market, and hence be local monopolists. 3.2.1. Uniform distribution of viewer tastes Suppose broadcasters 1 and 2 are in the market. Without loss of generality broadcaster 1 is assumed to be located to the left of broadcaster 2. When broadcasters 1 and 2 directly compete, the indifferent viewer is located at x = 1/2 + s 1 s 2. Marginal revenue thus equals 1 for both broadcasters. Solving the optimization problem leads to a unique equilibrium that the broadcasters choose S 1 Unif = S2 Unif = S D Unif = 1/4 when 2.67 m < 4, and S1 Unif = S2 Unif = S D Unif = 1/c when m 4 (as local monopolists). A pure-strategy Nash equilibrium does not exist when m < 2.67. 6 This assumption implies that the broadcasters have competitive foresight to avoid self-destructive behavior. For some parameter values it helps prevent one broadcaster from jumping to its competitor s side of the market, resulting in a quality war in which the profits of both firms will be driven away and no pure strategy equilibrium can exist.

J Cult Econ (2006) 30:127 140 135 Table 2 Duopoly equilibrium quality for bipolar distributions when β 0.536 Shape of distribution Cost parameter (m) Quality equilibrium m < 2.67 No pure strategy equilibrium Bipolar 2.67 m < 4ρ4 1 β sbipo D = 1/4 (1 >β 0.536) m 4ρ4 1 β Solved numerically 3.2.2. Non-uniform distributions of viewer tastes In the Appendix we show that in a simultaneous quality-location game broadcaster 1 and broadcaster 2 will always choose the same quality level at equilibrium. This is not surprising given the symmetric cost structure assumed in the model. Bipolar distribution of viewer tastes. The bi-modality of viewer tastes heightens the duopolists tendency to pay more attention to their own hinterlands where the density of viewers is greater than in the market center. When the cost of programming is sufficiently high that the local monopoly situation occurs, the market boundaries will be [0, 2s 1 ] for broadcaster 1 and [1 2s 2, 1] for broadcaster 2. The ratings are R 1 = 2s 1 0 ρ[x(1 x)] β 1 dx and R 2 = 1 1 2s 2 ρ[x(1 x)] β 1 dx. When the cost of programming becomes low enough that direct competition develops between the two broadcasters, broadcaster 1 s ratings is R 1 = 1/2+s1 s 2 0 ρ[x(1 x)] β 1 dx and that of broadcaster 2 is R 2 = 1/2+s 2 s 1 0 ρ[x(1 x)] β 1 dx. Using s 1 = s 2 (see the Appendix) we can solve for the marginal revenue. The marginal cost of programming remains at 2 ms. Combining analytical solutions and numerical results, we derive the equilibria for all possible values of m and β(β <1). The most interesting results happen when 0.536 β<1. 7 Three situations are possible when β is in this range: (1) if m < 2.67, there is no pure-strategy equilibrium, (2) if 2.67 m < 4ρ4 1 β, the duopoly equilibrium quality is at sbipo D = 1/4, and (3) if m 4ρ4 1 β, local monopoly occurs. Table 2 summarizes these results. The location equilibrium follows the quality equilibrium as y 1 = s 1 and y 2 = 1 s 2. Unimodal distribution of viewer tastes. When there is a concentrated mass of viewers in the market center, the duopolists will compete more aggressively for ratings, moving towards the market center if doing so is profitable. However, very aggressive competition through a fierce quality war with its rival is not justified when m is sufficiently high. In that case, the broadcasters find it prudent to locate away from the center. The market boundaries of the two broadcasters overlap at x = 1/2, and the viewers whose ideal point is at x = 1/2 are indifferent between viewing either program. The exact values of the equilibrium quality (sunim D ) are found numerically. For example, when β = 2, a pure strategy Nash equilibrium exists only for m 5.22. When m = 5.50, sunim D equals 0.160. As another example, when β = 3 and m = 8.50, s Unim D = 0.125. Moreover, for the values of m where a pure strategy equilibrium exists for certain values of β, increasing the concentration of viewer tastes (i.e., higher levels of β) tends to lead to lower program quality, and never leads to higher quality levels. 7 If the distribution is too bimodal (i.e., β<0.536) then the two broadcasters are either true local monopolists, or no pure strategy equilibrium exists.

136 J Cult Econ (2006) 30:127 140 3.3. Viewer equity and welfare comparisons As we discussed in the previous section, the television broadcaster s objective is profit maximization. In this subsection we examine the implications of broadcasters behavior on television viewers, specifically as it relates to issues of viewer equity and welfare. Viewer equity is measured as the percentage of the viewing public that is able to find a television program that is worth watching. Specifically, viewer equity (Equ) of the duopoly market equals b 1R b 1L f (x)dx + b 2R b 2L f (x)dx where b 1L and b 1R are the left and right market boundaries of broadcaster 1, b 2L and b 2R are those of broadcaster 2, and f (x) is the probability density function of a particular beta distribution. Equ = b 2R b 1L f (x) dx if the broadcasters market boundaries overlap. Viewer welfare can be measured by the total net consumer surplus the viewing public receives from the broadcasters. That is, viewer welfare (Wel) equals the utility aggregated across all viewers (individual utility is defined in Equation 1), or Wel = { n y j j=1 y j s j [s j (y j x)] f (x)dx + y j +s j y j } [s j (x y j )] f (x)dx, (4) where n = 2 for duopoly and n = 1 for monopoly. From a public policy point of view, regulators would ideally want to identify a market structure either monopoly or duopoly in our case that provides higher values of both viewer equity and welfare. Such cases do exist. For example, if the industry s cost structure is such that m = 5.50 and viewer tastes follow either a bipolar or uniform distribution, then the duopoly market offers both greater viewer equity and greater viewer welfare. However, the industry structure does not always allow this to happen. Still in the case of m = 5.50, if viewer tastes are distributed unimodally, then it is possible that the monopolist provides greater welfare but less equity, as illustrated by the case of β = 2 in Table 3. Based on the analysis of all parameter values, we find that if the distribution of viewer tastes is strongly bipolar (i.e., β<0.895), then for any cost structure for which a pure strategy Nash equilibrium exists, the duopolists will provide both greater equity and greater welfare. This result is not general, however. Given a cost structure represented by m, depending on the distribution of viewer preferences (as long as β 0.895), either the Table 3 Comparison of viewer equity and viewer welfare Distribution Viewer equity (Equ) Viewer welfare (Wel) Monopoly Duopoly Shape Parameter quality quality Monopoly Duopoly Monopoly Duopoly β = 0.90 0.358 0.250 0.704 1.000 0.123 0.122 Bipolar (m = 2.67) β = 0.95 0.367 0.250 0.728 1.000 0.132 0.124 (m = 2.67) Uniform β = 1.00 0.373 0.250 0.746 1.000 0.139 0.125 (m = 2.67) β = 2.00 0.220 0.160 0.617 0.829 0.070 0.068 Unimodal (m = 5.50) β = 3.00 0.172 0.125 0.596 0.793 0.053 0.051 (m = 8.50)

J Cult Econ (2006) 30:127 140 137 Table 4 Ranges of the cost parameter (m) where the duopolists offers greater viewer equity but lower viewer welfare Shape of distribution Shape parameter Range of m β = 0.895 [2.67, 2.671) Bipolar β = 0.900 [2.67, 2.68) β = 0.950 [2.67, 2.75) Uniform β = 1.000 [2.67, 2.83) β = 2.000 [5.22, 5.80) Unimodal β = 3.000 [8.45, 9.39) β = 4.000 [11.68, 12.98) monopoly or the duopoly market could offer a greater level of viewer welfare, while viewer equity is never lower in the duopoly market relative to the monopoly market. In other words, the number of broadcasters required to provide more coverage (enhancing viewer equity) may be different from the number required to produce greater viewer welfare. Table 3 provides several examples of these situations for all the three possible patterns of viewer tastes. For instance, when the distribution is bipolar at β = 0.95 and the cost parameter is m = 2.67, the duopolists provide greater viewer equity than a monopolist by serving all potential television viewers. However, it is the monopolist that provides the greater amount of viewer welfare (0.132) although fewer people (72.8%) would watch its program. Table 4 further summarizes, for additional values of β, the overall range of m that induces the duopolists to provide greater market equity but less efficiency than the monopolist. (As noted above, such a competitive outcome is possible whenever β 0.895.) For example, if the distribution is moderately unimodal at β = 2.0, the monopolist provides greater efficiency but less equity for any values of m within the range of [5.22, 5.80). 4. Conclusion This paper presents an analysis of the broadcast television market based on the economics literature of spatial competition and positioning. Different from the previous studies in this literature, we focus on the different patterns of viewer tastes and examine how they influence viewer equity and viewer welfare, two market outcomes that are of critical public policy value. Two primary results are derived. First, the distribution pattern of viewer tastes, especially how clustered they are, directly influences the broadcasters programming strategies. The existence of highly clustered tastes enables broadcasters to cater to popular taste (by offering a specific type of program) and at the same time reduce the production values of that program. A higher cost of programming also induces broadcasters to offer lower-quality programs, but with a greater emphasis towards serving local tastes (i.e., narrowcasting ). Second, it may not be feasible to expect a single market structure to offer both greater viewer equity and greater viewer welfare. In many cases the duopoly market provides a greater amount of viewer equity but less viewer welfare, compared with the monopoly market. This is found to be true under a wide variety of viewer taste distributions, but not if the distribution is highly bipolar. In terms of public policy and social welfare, our results indicate the importance of understanding viewer preferences and incorporating the distribution of viewer tastes into regulatory considerations. Since viewer equity and welfare goals may not coincide with each other, regulatory agencies need to have clear priorities when prescribing regulatory measures and pay special attention to the distribution pattern of viewer tastes and the broadcasters

138 J Cult Econ (2006) 30:127 140 desire for return on programming investments. While there has been much discussion about program diversity, surprisingly little attention has been paid to the production values aspect of programming. This is particularly unfortunate since production values directly affect the bottom line of a broadcaster s operation and, as this study suggests, have significant impacts on market outcomes. Taking a broader perspective, theoretical models of market competition often assume a uniform distribution of consumer tastes in order to concentrate on what are presumably the more central topics. Our analysis shows that the nature of the distribution of tastes has an effect on such public policy relevant issues as viewer equity and welfare, and should be treated with care. A few assumptions of our basic modeling framework can be further relaxed in future research. First, the utility function in Equation (1) uses a linear structure, and assumes that viewers are homogeneous in the way they perceive quality as well as in the weight given to the quality dimension. While these are common assumptions in the literature, other functional forms and a more heterogeneous taste for quality should be considered. Second, the economics literature on collective passion and informational cascade has produced a useful knowledge base for viewer behavior that is interdependent (Bikhchandani, Hirshleifer and Welch, 1998; Grilo, Shy and Thisse, 2001). The impact of such network externalities on viewing behavior has not been explored in the television market. Third, we examine the case where each broadcaster has only one channel. The implications of media mergers and multiple-outlet ownership are interesting issues that deserve further research attention. Lastly, we have limited our attention to an exogenously determined number of broadcasters. Extensions may be made to examine entry/exit decisions by new networks or channels that are enabled by digital technologies such as cable and satellite. We have focused on the monopoly versus duopoly situations in this study. While doing so offers a useful framework to generate valuable implications for the interactions between key variables, it would certainly be interesting to further examine to what extent the results would be applicable to a multiple-firm multiple-modal setting. For similar reasons, we note that empirical research regarding the distribution of viewer tastes, either generally or in a particular regional market, is limited. It is our belief that such research would produce valuable insights for both network operators and regulators. Appendix In this technical appendix we prove that for non-uniform distributions of viewer tastes, two broadcasters will always set the same level of quality at equilibrium, if a pure strategy equilibrium exists. We will prove that s1 D < s 2 D cannot be a pure strategy Nash equilibrium. The case of s1 D > s 2 D can be proven in a symmetric fashion. In the bipolar case the broadcasters start from their hinterlands to search for an optimal location and quality combination, since there is no competition but only greater viewer density. We further assume that the broadcasters do not forgo uncontested demand in the hinterland in order to aggressively compete for contested demand. When the broadcasters are local monopolists, it will be the case that s1 D = s 2 D since they have the same cost structure. When they directly compete, let us assume 1/4 < s1 D < s 2 D < 1/2 without loss of generality. Then it must be the case that π 2 s2=s2 D >π 2 s2=s2 D ε, where ε is an infinitesimally small positive amount, otherwise (s1 D, s 2 D ) will not be a possible equilibrium. This means that starting with a level of s2 D ε, the additional viewers broadcaster 2 attracts in the market center (where density is the lowest) more than compensate for the additional cost of quality

J Cult Econ (2006) 30:127 140 139 increase. Since the broadcasters share the same cost parameter, broadcaster 1 will also find it profitable to increase its quality level by ε to capture the same audience level. Thus broadcaster 1 will not stay at s1 D.Ifs 2 D 1/2 and s 1 D < s 2 D, broadcaster 1 will not attract any audience, thus it will not stay at s1 D either. In summary, s1 D < s 2 D cannot be a pure strategy Nash equilibrium in the bipolar case. Similar to the bipolar case, and under similar assumptions, there is no equilibrium if s2 D 1/2 and s 1 D < s 2 D for the unimodal distribution. We only need to consider s2 D < 1/2. Suppose the broadcasters are located at y 1 D < y 2 D. The indifferent viewers are at x = [s1 D s 2 D + y 1 D y 2 D]/2. Since s 1 D < s 2 D broadcaster 2 always wants, and is able to, locate at the market center. Thus y2 D = 1/2. Given broadcaster 2 s choice of {s 2 D, y 2 D }, it must be the case that {s1 D, y 1 D} is more profitable for broadcaster 1 than {s 1 D ε, y 1 D }. That is, the additional audience broadcaster 1 attracts by increasing quality from s1 D ε to s 1 D (while staying at y1 D) results in greater profits. This incremental demand is in the amount of R 1 = B 1L (ε) + B 1R ( x x 0 ), where x 0 is the location of the indifferent viewers when s 1 = s1 D ε, B 1L(ε) is the cumulative beta density to the left of y1 D for a distance of ε, and B 1R( x x 0 ) is that to the right of broadcaster 1 for a distance of x x 0. Thus R 1 = B 1L (ε) + B 1R (ε/2). Given {s1 D, y 1 D} and {s 2 D, y 2 D} we now consider an increase of s 2 D by ε. The additional ratings broadcaster 2 would receive will be R 2 = B 2R (ε) + B 2L (ε/2). Since the distribution is unimodal and there is greater density in the market center (where broadcaster 2 locates), it must be the case that B 2L (ε/2) = B 1R (ε/2) but B 2R (ε) > B 1L (ε). Thus R 2 > R 1. Since the additional audience of R 1 is profitable for broadcaster 1, R 2 must also be profitable for broadcaster 2. Therefore, broadcaster 2 will not stay at s2 D, and s 1 D < s 2 D cannot be a pure strategy equilibrium for the unimodal case. Acknowledgements Yong Liu thanks the Center for the Study of Popular Television at Syracuse University for a grant supporting this research. Charles B. Weinberg gratefully acknowledges the support of the Social Sciences and Humanities Research Council of Canada. References Ansari, A., Economides, N., & Ghosh, A. (1994). Competitive positioning in markets with nonuniform preferences. Marketing Science, 13(3), 248 273. Barwise, P., & Ehrenberg, A. (1988). Television and its audience. SAGE publications Ltd. Baye, M., & Morgan, J. (2000). A simple model of advertising and subscription fees. Economics Letters, 69, 345 351. Bikhchandani, S., Hirshleifer, D., & Welch, I. (1998). Learning from the behavior of others: conformity, fads and information cascades. Journal of Economic Perspectives, 12(3), 151 170. Comstock, G., & Scharrer, E. (1999). Television: what s on, who s watching, and what it means. Academic Press. Dukes, A., & Gal-Or, E. (2003). Negotiations and exclusivity contracts for advertising. Marketing Science, 22(2), 222 245. Economides, N. (1989). Quality variations and maximal variety differentiation. Regional Science and Urban Economics, 19(1), 21 29. Evans, D., & Schmalensee, R. (August 2005). The industrial organization of markets with two-sided platforms. Working paper: http://ssrn.com/abstract=786627. Fournier, G. (1985). Nonprice competition and the dissipation of rents from television regulation. Southern Economic Journal, January, 754 765. Fournier, G., & Martin, D. (1983). Does government-restricted entry produce market power?: New evidence from the market for television advertising. Bell Journal of Economics, Spring, 44 56. Goettler, R., & Shachar, R. (2001). Spatial competition in the network television industry. Rand Journal of Economics, 32(4), 624 656.

140 J Cult Econ (2006) 30:127 140 Grilo, I., Shy, O., & Thisse, J.-F. (2001). Price competition when consumer behavior is characterized by conformity or vanity. Journal of Public Economics, 80(3), 385 408. Hotelling, H. (1929). Stability in competition. The Economic Journal, 39, 41 57. Lehmann, D. (1971). Television Show preference: application of a choice model. Journal of Marketing Research, 87(February), 47 55. Litman, B. (1979). The television networks, competition, and program diversity. Journal of Broadcasting, 23(4), 393 409. Liu, Y., Putler, D., & Weinberg, C. (2004). Is having more channels really better? A model of competition among commercial television broadcasters. Marketing Science, 23(1), 120 133. Ma, C., & Burgess, J. (1993). Quality competition, welfare, and regulation. Journal of Economics, 58(2), 153 173. Neven, D. (1986). On hotelling s competition with non-uniform customer distributions. Economic Letters, 21, 121 126. Neven, D., & Thisse, J.F. (1990). On quality and variety competition. In J.Gabszewicz, J. Richard and Wolsey, F. (eds), Economic Decision Making: Games, Econometrics and Optimization, Amsterdam: North-Holland, 175 199. Owen, B., Beebe, J., & Manning, W. (1974). Television economics. Lexington, MA: Heath. Owen, B., & Wildman, S. (1992). Video economics. Cambridge, MA: Harvard University Press. Rust, R., Kamakura, W., & Alpert, M. (1992). Viewer preference segmentation and viewing choice models for network television. Journal of Advertising, 21(1): 1 18. Stiglitz, J. (1981). Potential competition may reduce welfare. American Economic Review, 71(2): 184 190. Vogel, H. (2001). Entertainment industry economics: a guide for financial analysis, Cambridge, UK: University of Cambridge Press. Walker, J., & Ferguson, D. (1998). The Broadcast Television Industry, Needham Heights, MA: Allyn & Bacon. Wright, D. (1994). Television advertising regulation and program quality. The Economic Record, 70(2), 361 367. Zhou, W. (2004). The choice of commercial breaks in television programs: the number, length and timing. The Journal of Industrial Economics, LII(3), 315 326.