Radio Advertising

Review of Industrial Organization 14: 239–256, 1999. © 1999 Kluwer Academic Publishers. Printed in the Netherlands.

239

Is Radio Advertising a Distinct Local Market? An Empirical Analysis ? ROBERT B. EKELUND, JR. Department of Economics, 203 Lowder Business Bldg, Auburn University, Auburn, AL 36849, U.S.A.

GEORGE S. FORD?? MCI Telecommunications,1801 Pennsylvania Ave NW, Washington, DC 20006, U.S.A.

JOHN D. JACKSON Department of Economics, 203 Lowder Business Bldg, Auburn University, Auburn, AL 36849, U.S.A.

Abstract. The recent relaxation of regulations limiting the ability of an entity to own multiple radio stations has led to a dramatic increase in ownership concentration in local radio markets. As a result, the Department of Justice has, on several occasions, required limited divestiture of multiple station owners. Implicit in the agency’s action is that radio advertising constitutes an antitrust market. Using the framework established by the Merger Guidelines, this paper evaluates whether or not radio advertising is a distinct local market by estimating an own-price elasticity of demand for radio advertising. Our results support the assertion that radio advertising is an antitrust market.

I. Introduction The Telecommunications Act of 1996 will have an enormous effect upon the entire communications industry in the United States. Not least will be the impact on the AM and FM radio market, especially at the local levels. Nationally, a single entity may now own an unrestricted number of licenses across the entire country. The strict ownership limitations in local radio markets, where ownership concentration is more controversial, were relaxed but not abandoned. Prior to 1992 a single entity ? The authors would like to thank Jerry B. Duvall, William Drake, David B. Ford, Jennifer Kaiser, Jeena Kim, Kevin Roth, Rick Warren-Boulton and two anonymous referees for helpful comments and suggestions on earlier drafts of this paper. We are also grateful to the economic staff of the FCC’s Mass Media Bureau, Policy Division, who provided numerous helpful comments on an earlier version of this paper. All remaining errors are the sole responsibility of the authors. ?? Affiliated Scholar, Auburn Policy Research Center. The analysis and conclusions of this paper represent those of the author and do not necessarily reflect the views of MCI Telecommunications Corporation, its staff, or its subsidiaries or the Auburn Policy Research Center, its staff, or its sponsors. Comments on the paper can be directed to George S. Ford, 1801 Pennsylvania Ave. NW, Suite 428, Washington, DC, 20006.

240

ROBERT B. EKELUND, JR. ET AL.

could not own more than one radio station of the same service (i.e., FM or AM) in a given market. Now, a single ownership entity may own up to eight stations in a single market, depending upon the total number of radio stations in that market and, to a lesser extent, their relative market shares.1 The relaxation of ownership limitations follows a trend that began some years ago in an era of radio station insolvencies that followed the “openly competitive” environment of the 1980s. Due to Federal Communications Commission (FCC) rule changes, concentration in both national and local markets began to increase in 1992,2 reaching new highs of activity by 1996.3 While the FCC clings to “public interest” ends such as “diversity” and “localism”, concentration at the local level has the potential to thwart such goals. 4 In addition, while issues of diversity are somewhat unique to media markets, the problems arising from high levels of concentration in media industries are not altogether different from those of other industries. In particular, horizontal concentration in a particular geographic market may lead to reduced competition and therefore higher advertising rates, reduced output and constrained consumer choice. The link between the rapid consolidation of the radio industry and the ensuing potential market power has led the Department of Justice (DOJ) to closely scrutinize radio mergers. In August of 1996, the DOJ’s first challenge of a radio merger resulted in the divestiture of a single radio station in Cincinnati by the station’s owner Jacor. The DOJ alleged that Jacor’s ownership of the station (WKRQ-FM) would give it control of more than 50 percent of the sales of radio advertising time in Cincinnati, and could enable the company “to increase prices to advertisers and substantially reduce competition in the $80 million Cincinnati radio advertising market”. Implicit in the DOJ’s action is that it has concluded that radio advertising constitutes an “antitrust market”, existing independently of advertising markets for other local media.5 1 See Telecommunications Act of 1996, Section 202(b). 2 See Notice of Proposed Rulemaking, 6 FCC Rcd 3275 (1991); Report and Order, MM Docket

No. 91-140, 7 FCC Rcd 2755 (1992a); Memorandum Opinion and Order, MM Docket No. 91-140, 7 FCC Rcd 6387 (1992b). 3 Mergers of leading radio group owners included deals amounting to $5.2 billion in the first six months of 1996 alone. The top three firms in 1994 (CBS, Infinity, and Westinghouse) merged (in 1996) to become the largest firm in the market. Clear Channel Communications, thirteenth largest firm in 1994 became through merger the second largest firm in 1996 (Rathbun, 1996, p. 6; Petrozzello and Rathbun 1996). (Other huge media mergers have proceeded apace. The Time-Warner/Turner merger, involving the merger of two movie studios and a dozen cable channels and given the go-ahead in September 1996, raised concerns of the Federal Trade Commission). 4 Diversity and localism are, in effect, goals to promote political and other forms of access to local radio media. Information substitution may, for example, be high for state or national politics, but a real concern may attend the emergence of local market power vis-`a-vis local politics. 5 Justice Department Requires Jacor to Sell Cincinnati Radio Station, DOJ Public Announcement (August 5, 1996). In the absence of data to calculate the own-price and cross-price elasticities of demand, the Merger Guidelines suggests the use of practical indicia such as buyer and seller perceptions and actions towards the timing and costs of shifting products and/or suppliers, price movements and

IS RADIO ADVERTISING A DISTINCT LOCAL MARKET?

241

In contrast to the position held by the DOJ, the National Association of Broadcasters (NAB) asserts that, at the local level, radio advertising does not constitute a separate market from other local advertising. Other broadcast media and print outlets, in addition to other advertising media are supposed to constitute such a market. Kerr (1996, p. 1), on behalf of the NAB, asserts “. . . radio has no category of advertisers who are dependent on the radio medium to distribute their advertising. Radio stations therefore compete not only against other stations, but against myriad other media to which radio advertisers switch when relative prices change”. A central issue regarding the need for antitrust scrutiny of radio mergers, therefore, is whether or not there is a “radio market” for advertising.6 The purpose of this paper, is to evaluate empirically whether a radio advertising market exists independently of other local advertising markets. Using data on 110 separate radio markets (derived from an initial sample of 200) and other local market media – newspapers and television stations – we estimate the own-price elasticity of demand for radio advertising as well as cross-price elasticities between radio and newspapers and television stations. By doing so, we shed light on the issue of whether a separate and distinct market for radio advertising exists. We caution, however, that market definition is only a first step in a complete antitrust analysis and we therefore do not address the issue of whether antitrust should be used. The paper proceeds as follows. In Section II we briefly discuss the methodology of delineating an antitrust market according to the DOJ’s Merger Guidelines. In Section III, we provide an overview of the evolution and development of radio markets, with particular emphasis on the kind of ownership restrictions that the FCC placed in local and national markets between 1980 and the Telecommunications Act of 1996. In Section IV, we outline a procedure, proposed by Kamerschen (1994), for delineating an antitrust market. In Section V we develop empirical estimates of the own-price and cross-price elasticities of demand for radio advertising. Conclusions are provided in the final section. II. Demand Substitutability and Market Definition Under traditional antitrust analysis, the extent of market power is closely linked to the level of concentration in an industry. Naturally in measuring any degree of concentration, market share is important. But prior to establishing an institutional shipment patterns, influence of downstream competition faced by the buyers in their output markets, and so forth. This approach, rather than the direct estimation of own-price and cross-price elasticities of demand has heretofore been the approach (at least, to our knowledge) adopted by the DOJ in its investigation of radio mergers. 6 By “radio market” we mean a market for radio services by advertisers. Radio stations also compete in the market for audience. However, the competition for audience is simply “manufacturing” a product to sell to advertisers and in itself generates no revenue for the station. The real question is therefore whether or not advertisers view radio, as a potential outlet for their messages, as having distinct and non-transferable characteristics that differentiate it from other potential outlets.

242

ROBERT B. EKELUND, JR. ET AL.

analysis of market share, the concept of a market, that is, the definition of a market is essential.7 According to the Justice Department’s Merger Guidelines, “[a] market is defined as a product or group of products and a geographic area in which it is produced or sold such that a hypothetical profit-maximizing firm, not subject to price regulation, that was the only present and future producer or seller of those products in that area likely would impose at least a “small but significant and nontransitory” increase in price, assuming the terms of sale of all other products are held constant”.8 The relevant product or geographic antitrust market is the smallest group of products (the product market) or smallest geographic area (the geographic market) that satisfies this test. More generally, an antitrust market includes the buyers and sellers in a geographic area who significantly influence the price, quality, and quantity of specific commodities. An antitrust market, then, is delineated by determining what firms in what geographic area selling what products would have to belong to a hypothetical monopolist (or cartel) to make it profitable and optimal to raise prices (Kamerschen, 1994, pp. 2–3).9 The Merger Guidelines identify three factors that could constrain the ability of some (hypothetical) monopolist to exercise market power over a particular group of products in a particular area: 1) demand substitutability; 2) supply substitutability; and 3) entry. Demand substitutability refers to the ability of consumers to substitute related products or the same product produced in other areas. Supply substitutability and entry both refer to the ability of producers not currently selling a particular product to begin doing so. The primary distinction between supply substitution and entry is that entry entails significant new investment in production or distribution or requires more than one year to accomplish while supply substitution does not. Under the Merger Guidelines approach, each factor is addressed in a separate step. But, as Kamerschen explains (1994, pp. 3–5), the antitrust market definition focuses solely on demand substitution, where the relevant market encompasses all products which buyers consider to be good substitutes.10 Further, demand substitution is measured by own-price elasticity in a multiple regression containing possible substitutes (see Section IV below) in contrast to the standard measurement of cross-elasticity coefficients for possible substitutes.11 7 See Brown Shoe Co. v. U. S., 370 U. S. 294 (1962). 8 United States Department of Justice and Federal Trade Commission, Horizontal Merger Guide-

lines (April 2, 1992). The distinction between antitrust market and economic market has received substantial attention. Generally, these two market concepts can differ substantially. 9 In most contexts, the DOJ will use a price increase of five percent lasting for the foreseeable future as its definition of a “small but significant and nontransitory” increase in price. Using a price increase that is larger or smaller than five percent is not prohibited. 10 The group of substitute products also includes all of the geographic areas among which buyers are readily able to shift their purchases. 11 In standard economic theory, the measure for determining demand substitutability is the “crossprice elasticity of demand”, defined as the percentage change in the quantity of one product demanded in response to a given percentage change in the price of related products. Where the cross-price elasticity of demand is high, buyers readily shift their purchases among the products

IS RADIO ADVERTISING A DISTINCT LOCAL MARKET?

243

The crux of applying the Merger Guidelines approach to the radio advertising market is to answer the question: Could a hypothetical monopolist in a local radio market find it profitable and optimal to raise price and sustain the raise, ceteris paribus? Kamerschen (1994) has developed a procedure to answer this question by comparing demand substitutability (as measured by the own-price elasticity of demand) with the optimal markup on price (as measured by a Lerner Index). Before turning to a detailed discussion of this procedure and its application to the problem at hand, it is crucial to understand how the change in ownership requirements embodied in the Telecommunications Act have altered the rules of the game for local radio advertising. The following discussion considers this change in historical context. III. Radio Market Regulation The evolution of radio market regulation and its relation to other forms of communications regulation is long and tortuous, beginning (literally) with the sinking of the Titanic.12 The story is familiar: spectrum regulation developed on the basis of public goods status. The appearance of substitutes created an “umbrella” of regulation that came to include television, cable and other forms of communication. While the public good rationale for such regulation has been questioned (Krattenmaker and Power, 1994, pp. 33–55), our concern with radio regulation has it origins in more recent policies. 1.

THE

1980 S

AND THE ENSUING

“ CRISIS ”

The Reagan-inspired deregulation initiatives of the early 1980s had a profound effect on radio regulations. Support for competing electronic media along with expansion of existing media was the order of the day for the FCC. The Commission encouraged the proliferation of both AM and FM stations, including new regional channels, daytime-only stations, and over 700 new FM channels throughout the country (primarily low-powered channels). This growth was accompanied by massive growth in all electronic media (Hunsaker, 1994, pp. 22–23). Media mergers and new investment followed altering the competitive landscape. Such competition, not surprisingly, threw particular radio stations and markets into a period of grave financial insecurity and bankruptcies in 1990.13 The FCC’s policies were reoriented in response to changes in their relative prices. The products and regions are good substitutes, and, hence, are considered to be in the same relevant market (Werden, 1992, p. 132) 12 The Radio Act of 1912 was passed in the wake of the Titanic disaster. The Marconi station in Newfoundland had picked up signals from the distressed ship, but the signals were “stepped on” by amateur (“ham”) radio stations. That legislation established federal government control (under the Department of Commerce) of broadcasting and the allocation of the spectrum (by government) among competing uses with wattages stipulated. 13 Such problems, including a dramatic fall in financial values, were experienced in the main by commercial stations and smaller competitors. Hunsaker (1994, p. 22) notes that “The FCC, seeing

244

ROBERT B. EKELUND, JR. ET AL.

from facilitating competition and “diversity” into a mode of protection in radio markets. This protectionist stance had actually begun in the 1980s. In 1984 and 1986, while led by a FCC charged with returning radio markets to the forces of market competition (at least in a limited way), a number of defensive accommodations were being made. Low-power FM stations were allowed to upgrade to a higherpower class of station without the possibility of losing existing facilities and a new class of FM station was established. Most significantly, perhaps, the FCC modified its radio duopoly rules. These rules were modified so that common ownership of radio facilities was possible under certain technical conditions.14 One major change in ownership rules occurred in the wake of certain “evolving” marketing agreements in radio markets. Agreements brokering a station’s time – called time brokerage agreements (TBAs) or, more broadly, local marketing agreements (LMAs) – permitted operating agreements between two or more radio stations in the same market. Typically, these agreements would allow one station to “rent” 100 percent of another station’s time in return for the provision of all advertising and programming services at a flat monthly fee. Many of these agreements had “buy out” or “first refusal” clauses in them for the protection of the lessors as well as “early termination” clauses to protect the lessee. A blurring of ownership and control might have violated the FCC’s ownership or “duopoly” rules, but in 1990 the Commission ruled that such agreements (termed “network affiliation agreements”) violated no FCC policy. 2.

FROM DIVERSITY TO SURVIVAL : MARKET STRUCTURE ALTERATIONS

Competitive pressures (and an economy-wide recession) worsened the financial situation in radio markets by 1991. At this point the FCC’s stance definitely turned from “diversity” of programming and ownership towards the protection of station survival. These market structure alterations undertaken in 1991 modified the so-called duopoly rule whereby license holders were restricted to ownership limitations. Prior to 1991 national ownership by a single licensee was restricted to 12 AM and 12 FM stations. Local restrictions under the pre-1991 duopoly rule were that that it was presiding over an industry in serious economic trouble, came to realize that its ‘more is better’ and ‘diversity at any cost’ policies, like most panaceas, worked much better in theory than in practice”. It is not clear to us that open competition – in disequilibrium – is not expected to have such consequences. An analogy to the farm program, where the government for many years maintained a commitment to the ideal of the “small farm”, is not inapt here. 14 These technical conditions relate to “primary service contours” and “principal community contours”. Principal community contours for AM stations, for example, are the distance out from the station’s transmitter where the ground-wave radiated energy equals or exceeds five millivolts over one square meter of surface area-5 mV/m. 47 C.F.R.s 73.24(I). AM stations, with certain exceptions, must place a 5 mV/m or greater contour over 80 percent of its community of license. For FM stations, the principal community contour is 3.16 mV/m. 47 C.F.R.s 73.315(a). Terms such as “principal community contour”, “principal city contour” and “city grade contour” are used interchangeably (see Hunsaker, 1994: note 13).

IS RADIO ADVERTISING A DISTINCT LOCAL MARKET?

245

no single owner could own two or more stations of the same class of service (that is, AM or FM) if the “principal community contours” of the stations overlapped. The FCC (in May 1991) issued a Notice of Proposed Rule Making to raise the permissible ownership limits, justifying the increase by noting the increased competition from an increased number of stations and other media in local markets. As Hunsaker (1994, p. 25) emphasizes, the FCC dealt with three factors in it’s proposal: market size, the number of commonly-owned stations in the market, and the audience share resulting from proposed mergers or acquisition. National ownership limits were to be raised from 12 AM and 12 FM stations to 30 of each. In the proposal relating to local markets, a rather complex 4-tier division of radio markets was contemplated and coupled with audience share percentages.15 The market and audience share percentage was calculated by using Arbitron ratings in metro markets with the market definition hinging on the area encompassed by principal community contours of the stations seeking merger. At this point, the FCC also recognized the LMAs by adopting new rules. These rules place station lessors as acquiring a substantial ownership interests if the lessor provided more than fifteen percent of another station’s programming. If the latter occurred, that arrangement would count toward the ownership limits established by the Commission. This complex system was assailed within the Commission itself and by members of Congress as arbitrarily complex. The FCC ultimately modified (and simplified) the new rules on market size in a Reconsideration Order in September 1993. National ownership limits were set at 18 AM and 18 FM stations with a further expansion to 20 each in two years. At the local level, the Commission adopted a simplified two-tier definition of what constitutes permissible ownership in local markets. Once more, however, a market was defined in terms of overlapping principal community contours. Further, the FCC used Arbitron Radio Markets to estimate (audience) market share. Small radio markets were those where there were 14 (or fewer) of such stations. In these markets a licensee could own no more than three stations total (no more than 2 AM or 2 FM) and the combination had to be less than 50 percent of the total number of stations in the market. Licensees in large radio markets – those with 15 or more stations – were constrained to four stations (with no more than 2 in either class of service). Mergers or acquisitions were further disallowed if the total audience share was 25 percent or larger. The FCC’s proposed rules on TBAs and LMAs remained in effect with the “25 percent or larger” rule also pertaining to such leasing agreements. Mergers of only two stations would not be allowed if under a TBA their combined audience share met or exceeded 25 percent. Thus, up to the passage of the Telecommunications Act of 1996, the FCC conscribed markets in such as manner as to define local market concentration. This measure included both the number of competitors and each competitor’s audience share ratings. Numbers of competitors and share ratings were facilitated by Arbi15 For example, in markets with 30–39 radio stations, one licensee would be permitted to own up

to three AM and two FM stations if the combined audience share did not exceed 25 percent.

246

ROBERT B. EKELUND, JR. ET AL.

tron ratings within metro radio markets.16 Within this scenario, radio is considered a distinct and separate “market” in the sense that these rules did not apply to other types of communication.

3.

THE TELECOMMUNICATIONS ACT OF

1996

With the new rules in effect little more than two years, the Telecommunications Act of 1996 has created cataclysmic changes in the ownership rules and market definition in national and local radio markets. There are no national ownership limits for AM or FM licenses and a single entity may now own licenses covering the entire United States. Changes in local market regulations are of particular interest for our study, however. The FCC was directed to revise its regulation so that a single party may own, operate or control up to 8 commercial radio stations (not more than 5 in the same (AM or FM) service) in a radio market with 45 or more commercial radio stations; in markets with 30 to 44 (inclusive) commercial radio stations, the ownership limit is 7 stations, with not more than 4 in the same service; in markets with between 15 and 29 stations, there is an ownership limit of 5, not more than 4 in the same service, and in markets with 14 or fewer commercial radio stations, a single entity may own, operate, or control up to 5 commercial radio stations, not more than 3 in the same service (with a 50 percent total limit on the number of stations) (Huber et al., 1996, p. 202). Also of interest is the revision of broadcast licensing reforms. Broadcast licenses for both TV broadcasters and for radio may now be issued for eight years (the old rule was 5 years for TV and 7 for radio). The FCC, in addition, is limited in its ability to try and discover whether licensing a competitor at renewal time would better serve the public’s interest (“convenience and necessity”) than re-licensing the incumbent (Huber et al., 1996, pp. 67–69).17

IV. Delineating the Radio Market In Section II, we discussed the fundamentals of the Merger Guidelines’ methodology for determining a relevant antitrust market. In practice, the “small but significant and nontransitory” price increase that the hypothetical monopolist could profitable and optimally undertake is typically taken as five percent, leading to its sobriquet “the five percent rule”. We also noted that Kamerschen (1994) has proposed a procedure by which this widely used conceptual definition of an antitrust 16 Small, non-metro markets could also be rated using Arbitron data and the FCC also permitted other forms of data to be used in particular cases. 17 The Telecommunications Act of 1996 also directed the FCC to review cross ownership rules for the media. FCC Chairman Reed Hundt, according to Huber et al. (1996, p. 68), has promised to re-examine the 21-year-old newspaper-broadcast cross-ownership rules away from strict prohibition.

IS RADIO ADVERTISING A DISTINCT LOCAL MARKET?

247

market can be operationalized to “rigorously assess the validity of any antitrust market (1994, p. 2)”.18 Kamerschen’s reasoning is straightforward: microeconomic theory tells us that a profit-maximizing firm will have an incentive to (reduce output and) raise price whenever its marginal cost exceeds its marginal revenue. By substituting the wellknown relationship between marginal revenue, price, and the (own-price) elasticity of demand into this inequality and rearranging terms, it can be seen that (p − c)/p ≤ −1/η

(1)

where p is price, c is marginal cost, and η is the own-price elasticity of demand. On the left of the inequality is the familiar optimal price-cost margin or Lerner Index; the term on the right is the negative of the reciprocal of the own-price elasticity of demand. The link between this relationship and the Merger Guidelines is that the inequality provides the conditions under which a hypothetical monopolist could profitably and optimally raise price and have an incentive to maintain it at the higher level. Kamerschen points out that implementation of these results for the purpose of delineating an antitrust market requires first the computation of actual markups for the product being sold – in our case, radio advertising. Next an estimate of the own-price elasticity of demand for the product is needed. Such an estimate can be obtained from the price coefficient in a double log regression of output on product price, the prices of related goods, income, tastes, and perhaps a time trend variable which picks up a number of factors (Kamerschen, 1994, p. 5). Finally, the two parameters must be compared. If the own-price elasticity of demand for radio advertising turns out to be highly elastic, say η = −6.0, and the markup turns out to be low, say (p − c)/p = 0.33, the implication is that there are too many close substitutes among alternative local advertising outlets to permit local radio stations (hypothetically, acting in concert) to optimally raise price and profitably 18 A number of empirical methods for market delineation have been applied in the antitrust con-

text. The Elzinga-Hogarty test delineates the geographic boundaries of markets on the basis of two percentages – LIFO (“little in from outside”) and LOFI (“little out from inside”). Under the ElzingaHogarty test, a candidate market is deemed to be a market only if the LIFO and LOFI percentages exceed particular thresholds; that is, only if the area has relatively little imports and relatively little exports. A Stigler-Sherwin test attempts to delineate markets by appealing to the law of one-price. This approach is based on price equality and/or similarities of price movements among firms. An increasingly popular methodology for market delineation is the econometric estimation of residual demand elasticities. Residual demand estimation provides an estimate of the elasticity of demand that would be faced by a hypothetical monopolist over a candidate market. The elasticity of demand can be used along with other information to estimate the amount by which the hypothetical monopolist would increase price. Depending on the degree of competition before the merger, estimated residual demand elasticities can be misleading since the amount by which a hypothetical monopolist over the candidate market would raise price is determined not by the demand elasticity at the competitive or prevailing price, but rather by the elasticity of demand at the monopoly price. See Elzinga and Hogarty (1973); Stigler and Sherwin (1985) and Baker and Bresnahan (1988).

248

ROBERT B. EKELUND, JR. ET AL.

keep it there. The negative reciprocal of the demand elasticity is only one half of the optimal markup on price. From the perspective of determining a relevant antitrust market, the implication of the above hypothetical example is that the product group should be broadened, perhaps including TV and newspaper advertising, and the markup and elasticity recalculated for this broader class of products. The object is to determine whether enough product substitution has been endogenized within the broader grouping to make it profitable for that hypothetical monopolist to raise price and sustain the raise. If this turned out to be the case for this broader product class, the inference would be that the relevant antitrust market is local media advertising rather than local radio advertising. On the other hand, had the estimated own-price elasticity of demand been, say −1.5 rather than −6.0, ceteris paribus, the optimal markup on price would have been only one half the reciprocal of the elasticity of demand. This result would indicate that radio stations could (acting in concert) profitably and optimally raise and maintain price. Of course, there are any number of reasons why we might not observe profit maximizing stations raising their advertising prices in this circumstance – the most obvious being that they do not act in concert, rather they vigorously compete with each other for the radio advertising dollar. The appropriate inference in this case is that radio advertising, by itself, constitutes a distinct antitrust market. Before we apply this methodology to the question of whether or not radio advertising is actually a distinct antitrust market, there are two aspects of this procedure that merit clarification. First, be assured, it is the own-price elasticity of demand that is the relevant parameter for delineating an antitrust market. While this may seem strange in light of our previous discussion and of the role that cross-price elasticities have played in some pioneering antitrust cases (e.g. Times-Picayune Publishing Co. v. United States, 345 U.S. 594, 612 n. 31, 1953), it is a result well grounded in demand theory. Based on the result that demand is homogeneous of degree zero in income and prices, it can be shown that the (negative of the) own-price elasticity of demand must equal the sum of the cross-price and income elasticities. This result, written for Hicksian or income compensated demand functions, states that the negative of the (compensated) own-price elasticity is equal to the sum of the (compensated) cross-price elasticities. Thus, whether we estimate compensated or ordinary demand functions, the inference is clear: the own-price elasticity summarizes in one parameter all of the relevant substitution possibilities for a given product. It is also worth noting that delineation of an antitrust market focuses solely on demand substitution factors. While the Horizontal Merger Guidelines establish important roles for supply substitution factors, these relate to determining which firms could (in the short run) participate in the relevant antitrust market given a price increase and to the analysis of entry (Kamerschen, 1994, p. 3), not to the def-

IS RADIO ADVERTISING A DISTINCT LOCAL MARKET?

249

inition of the relevant product market.19 Thus, all that is required for implementing this methodology, besides of course, an accurately computed industry markup, is a reliable estimate of the demand function for the product – in our case, radio advertising. It is to this task that we now turn. V. Empirical Model The approach outlined in Section IV suggests that we can gauge whether or not radio constitutes a separate market for local advertising (as distinct from other local advertising outlets) by estimating the own-price elasticity of a market level demand function for radio advertising. The estimated demand equation is to be a market demand equation. In other words, we aggregate the data on the relevant variables to the radio advertising market level. In that sense, it can be interpreted as the demand function faced by the hypothetical radio monopolist.20 Following Kamerschen, we start from first principles in attempting to specify our model. Specifically, we view 19 Firms that would quickly and easily substitute in supply in response to an attempted exercise

of market power are considered to be competitors in the market and are assigned market shares even though they do not currently sell products in the market. In radio markets supply side entry is extremely difficult given the policies of spectrum allocation currently and historically followed by the FCC. In almost every commercial market, all frequencies allocated to broadcast radio, both AM and FM, have already been assigned (although all are not currently broadcasting). If a frequency is not assigned, obtaining a license to broadcast is a long, tenuous and uncertain process. Thus entry into the production of radio advertising, and to a certain extent, the production of TV and newspaper advertising, is not problematic; it is simply not possible as a practical matter. Supply substitutability is also highly restricted in radio advertising. If we consider the relevant market as local media (print and broadcast) advertising, then increases in advertising rates are not likely to increase the number of players beyond radio, TV, and newspapers, so that the only likely source of supply substitutability is competition among firms within a given product category. With this caveat in mind, it is not unreasonable to view the supply of local media advertising to be perfectly price inelastic. This means that we can concentrate on demand factors to determine whether a distinct radio market exists, but if we wish to consider whether antitrust policy should be applied to such a market we must additionally consider the effect of competition among stations within that market. In this paper we address the former issue, leaving the latter, equally important, for further inquiry. 20 No one, to the best of our knowledge, has attempted to estimate such a model, but interesting related studies deal with other media. Estimating cross-price elasticities between newspapers and eight other media for the period 1971 to 1985, Busterna (1987) concludes that no other media reside in the same product market as newspapers for national advertising (i.e., no cross-price elasticities are statistically significant). In a study of local advertising markets for newspapers in Ireland, Thompson (1984) found that (despite some limitations in the data) there is strong support for the notion that local market structure is important, that concentration did exist and that it had a consistently significant influence on the price of newspaper advertising. Also see Thompson (1989). An older study (Brown 1982) using three data sets from the late 1970s and early 1980s and covering large, small and “other” local markets showed that station revenues and local station trading prices were unaffected by newspaper circulation and the number of television stations in local markets. Bates (1993) addresses both “advertising” and “audience” markets for local television and cable, suggests that the advertising market was separate and more concentrated than local audience markets. Other approaches include Moulton (1991) who calculates a price index for radio services using a hedonic regression specification.

250

ROBERT B. EKELUND, JR. ET AL.

the quantity of radio advertising demanded as determined by the price of radio advertising, the prices charged by other local advertising outlets, advertiser income, advertiser “tastes” for radio advertising, and a time trend.

1.

SPECIFICATION

Estimating a traditional demand function such as this requires, first and foremost, a measure of output – the quantity of advertising sold by radio stations. Unfortunately this data is not publicly available. However, total radio advertising revenue (TR) is available at the designated market area level from BIA’s Master Access database. Using total revenue (or radio advertising expenditures) as the dependent variable to estimate the own, cross, and income elasticities is quite acceptable. Rather than a demand equation, a linear expenditure equation is estimated from which all the properties of the underlying demand curve can be deduced; specifically, the relevant price and income elasticities. In fact, for the double log specification of the expenditure equation (which we employ here), using total revenue instead of quantity as the dependent variable, requires reinterpretation of only one parameter. In the expenditure equation, the own-price elasticity of demand is given by the estimated coefficient on log price minus one. The remainder of the coefficients retain their interpretations as cross-price and income elasticities without adjustment. Price of radio advertising. A common measure or index of price in media advertising studies, and in business practice, is the cost per rating point (CPP). A rating point of 1 implies that one percent of the potential broadcast audience is tuned to a particular station, network, or program. Potential audience is typically measured as the number of households or total population of an area. The CPP, therefore, indicates the dollar cost of advertising exposure to one percentage point of the potential audience. CPP are available for a wide variety of demographic groups, and the CPP used in this study is based on listeners 18 years or older (one of the broadest demographic groups for which the CPP is calculated). One popular market-wide CPP for radio advertising is produced by Spot Quotations and Data, Inc. (SQAD), and we employ (the logarithm of) that particular estimate, or SPARC, as our measure of the price of radio advertising PR .21 Again, in the expenditure equation, the coefficient on this variable is the own-price elasticity of demand plus unity. Thus, if demand is price inelastic, the estimated coefficient could legitimately be positive (but could never legitimately exceed unity). Elastic demand is indicated 21 According to SQAD, SPARC estimates are “based on an ongoing analysis of actual time buys

made available from various agencies and buying services, as well as information drawn from sales reps and other sources” and that their CPPs are “often used as the benchmark when entertaining competitive bids”. For information on the SPARC and SQAD index, see SQAD’s web page at [http://www.sqad.com].

IS RADIO ADVERTISING A DISTINCT LOCAL MARKET?

251

by a negative coefficient. Clearly, the more negative (i.e., the larger in absolute value) this estimate, the less likely it is that radio advertising is a distinct market. The price of advertising in related media outlets. The “five-percent test” proposed by the Merger Guidelines requires that the “terms of sale of all other products are held constant”. Thus, it is important to include the prices of potential substitutes to radio advertising, primarily the prices of television and newspaper advertisements, in any effort to delineate a “radio market”. Consistent with our price measure for radio advertising, we measure the price of television advertising (PT ) as SQAD’s market-wide CPP of a thirty second television spot.22 While the CPP measure of price is employed for both radio and television, a CPP is not available for newspaper advertising. An attempt was made, however, to calculate a newspaper advertising price that was consistent with the price measures of radio and television by making the reasonable assumption that the potential audience of newspaper advertisement can be estimated by the ratio of newspaper circulation to total households in the market. This figure can then be multiplied by the price of a one-inch, black-and-white, one-day advertisement for the major newspapers in the market to proxy a CPP for newspaper advertisements (PN ).23 It is not difficult to imagine situations where newspaper advertising is used in conjunction with, rather than in place of, radio advertisements, making the expected sign of the cross-price elasticity of radio advertising with respect to newspaper advertising prices is unclear.24 The Radio Advertising Bureau (RAB), a nationwide trade association that consults radio stations on advertising strategy, proposes that “. . . a ‘Media Mix’, including both radio and newspaper, will reach more prospects, more often and more effectively than newspaper alone, without an increase in advertising budget” (Kerr, 1996, 8). Thus, radio and newspapers often have complementary as well as substitutable uses. Since radio and newspaper are alternative outlets for advertising, they may reasonably be viewed as substitutes, indicating a positive cross-price elasticity. Alternatively, a negative crossprice elasticity indicates that the complementary nature of radio and newspaper advertisements dominates. Unlike newspaper advertisements, there is little reason to suspect that radio and television advertisements are frequently used as complements. Rather, they are generally described as alternative outlets for advertising, implying a positive cross-price elasticity of radio advertising with respect to television advertising price. However, even in this case, it is not obvious that these two outlets are close substitutes – one cannot watch TV in a car or see a product on the radio. But 22 CPPs also vary by day-parts. The average of the prime time and late night day-parts are used

here. 23 The data source for P is Newspaper Advertising Source (Wilmette, IL, SRDS, June 1996). N 24 For example, radio advertisements may include the expression “see our ad in Sunday’s paper”.

252

ROBERT B. EKELUND, JR. ET AL.

the question of substitutability, complementarity, or independence of radio and television advertising is a question to be addressed by our estimates of the model. Income. In traditional demand analysis, income refers to the total income of consumers of the product. Here the product is radio advertising, the consumers are businesses, and their total income is total retail sales (Y ). This data is available for each Arbitron market from BIA Master Access. If radio advertising is a normal good, we would expect a positive income elasticity. Taste and time trend. If we were computing our demand estimates at the station level, there are myriad variables that could be used to proxy the “tastes” of a business for advertising on one station versus another: listenership (as measured by the station’s Arbitron rating), station age, station format (e.g., country, rock, talk, etc.), and station power to mention but a few. Evaluating market definition, however, requires one to estimate the market demand curve. For the market demand curve for radio advertising, the advertising outlet tradeoff is between radio and TV and newspaper advertising; unfortunately, there is no ready proxy to adequately measure the business’ tastes for one versus the other of these aggregate alternatives.25 Likewise, since our data are cross-sectional, we are unable to take advantage of positing a catchall time trend variable. We therefore do not include these variables in our estimated demand model. We make these omissions with a clear understanding of the potential for omitted variables bias and the consequent need to carefully test for its presence. The simultaneity issue. Microeconomic theory suggests that market price and quantity (and hence total revenue or expenditure) are jointly determined by the forces of supply and demand. The market for radio advertising – if it exists – is no exception. Since our measure of income is retail sales, since successful local radio advertising should increase local retail sales, and since radio advertising is posited to be a normal good, there is good reason to suspect that Y , as well as PR , is determined simultaneously with T R. Failure to account for this joint determination of parameters can lead to biased, inconsistent, and inefficient estimates. We guard against this potentiality by creating instrumental variables (IV) to employ in lieu of PR and Y. Specifically, we estimate two ordinary least squares regressions which can be viewed as the first stage of a two stage least squares estimation of the implied three equation system. Thus we estimate regressions for PR and, Y respectively, as a function of a constant term, PT , PN , total number of stations in the market, market 25 One option that we considered was to measure taste as the ratio of TV and newspaper advertising

expenditures to radio advertising expenditures, or some heavily disguised measure of this proxy. We rejected this measure because it would introduce both multicolinearity (since advertising prices are already on the right) and spurious correlation (since radio advertising revenue is already on the left) into our estimated model.

IS RADIO ADVERTISING A DISTINCT LOCAL MARKET?

253

rank, per capita income, and total number of households. The first three variables are the exogenous variables in our basic expenditure equation for radio advertising; the remaining variables can be viewed as exogenous variables in other, unspecified equations in the system.26 We take the predicted values from these regressions as instruments, PˆR and Yˆ , and we employ them instead of their respective actual values in our subsequent empirical analysis in order to account for the potential for simultaneous equations bias. 2.

ESTIMATION

Due to data limitations, the econometric analysis of the own-price elasticity of radio demand is limited to 110 of the largest 200 Arbitron radio markets that made up the full sample. All data are for the year 1995. Based on our prior observations and data limitations, we posit the following expenditure model which we estimate by IV regression: ln T Ri = β0 + β1 ln PˆR,i + β2 ln PT ,i + β3 ln PN,i + β4 ln Yˆi + i

(2)

where all variables are as defined above and i is a stochastic error term, i = 1, . . ., 110.27 Initial estimates proved heteroskedastic as measured by a White test. We then weighted all variables by the (inverse of the) Arbitron market ranking variable and re-estimated Equation (2).28 The resulting Generalized Least Squares (GLS) estimates not only proved to be homoskedastic but also we were unable to reject the null hypothesis of no specification error at the five percent level based on Ramsey’s RESET test. This last result indicates, among other things, that our double log functional specification is appropriate; that our instrumental variables approach has successfully overcome the simultaneity problem; and our omission of the taste and time trend measures has resulted in no inappropriate estimates of the parameters of the included variables. All of these inferences are correct in the sense that any biases in the coefficients that could have been produced by these problems are statistically insignificant at the five percent level. Thus, even though the model presented below is simplistic, we are reasonably confident in the reliability of our parameter estimates.

26 Data for these latter variables is also available in BIA’s Master Access Database. 27 The descriptive statistics (Mean: Standard Deviation) of the variables are: P (39.2: 55.0); P R T (89.0: 147.5); PN (80.0: 85.0); Y (7,529.1: 11,285.0); and T R (30,275.: 79,371.2). Both Y and T R

are measured in thousands. 28 It is not surprising that heteroskedasticity was associated with market rank. We expect that the variability in the data would be greater in the smaller markets where data is less available and potentially less reliable.

254 3.

ROBERT B. EKELUND, JR. ET AL.

RESULTS AND INFERENCES ON MARKET DELINEATION

The resulting estimated (GLS, IV) model is presented below:29 ln T Ri −2.67 − 1.101 ln PˆR,i +0.297 ln PT ,i + 0.587 ln PN,i + 1.453 ln Yˆi (−0.99) (−2.07) (1.73) (1.91) (4.62) R 2 = 0.77

F = 192.5

(3)

N = 110

The summary statistics indicate that, even though parsimonious, our specification explains 77 percent of the variations in (log) total radio market advertising revenues and that the coefficients are jointly significant at any reasonable level. Evidently, television and newspaper advertising are substitutes for radio advertising. A 10 percent increase in TV advertising prices leads to about a 3 percent increase in the demand for radio advertising, while a 10 percent increase in newspaper advertising leads to a 6 percent increase in demand for radio advertising. Both effects are statistically significant at the ten percent level. Also, radio advertising appears to be a superior good with its income elasticity bordering on 1.5 and statistically significant at the one percent level. Finally, and most importantly, the own-price elasticity of demand is estimated to be −2.101 (= −1.101-1) which is statistically less than zero at the one percent level (t = −3.95). Indeed, the results in (3) indicate that demand is price elastic, since H0 : η ≥ −1 can be rejected at the five percent level (t = −2.07). In order to implement Kamerschen’s criterion for market delineation, we need to know the industry markup for radio advertising. Veronis, Suhler, & Associates in their Communications Industry Report (14th ed., November 1996, p. 79) find that the operating income (before interest and taxes) margin for radio broadcasters was 18.7 percent in 1995 and that operating cash flow margin was 31.3 percent for 1995.30 Based on the estimates presented above, the reciprocal of the own-price elasticity of demand is 0.48. Comparing this to either of the markup figures (0.187 or 0.313), Kamerschen’s criterion indicates that a hypothetical monopolist in the local radio broadcasting market could profitably and optimally increase advertising prices and sustain the increase. An alternative view is even more compelling. Kamerschen’s methodology suggests that a price elasticity of at least −3.2 (= −1/0.313) is required to make the decision concerning market delineation marginal. It turns out that the estimated own-price elasticity of −2.1 is statistically significantly greater than −3.2 29 In alternative regressions, we specified the price variables in terms of ‘price per thousand homes’. The coefficients estimates and their significance levels were not materially different than those presented here. For example, the coefficient on PR , the primary variable of interest, differed by less than 5 percent (falling to −1.15 with a t-statistic of −1.94). 30 The use of accounting measures of profitability is subject to the normal caveats. Arguably, the operating margin of the firm is the closest measure of the price-cost margin contemplated by economic theory. Of course, the conclusions regarding market definition rely heavily on the accuracy of both the estimated own-price demand elasticity and profit margin.

IS RADIO ADVERTISING A DISTINCT LOCAL MARKET?

255

at the five percent level (t = 2.06). Thus, our econometric results imply that radio advertising is a distinct antitrust market at the local level. If a hypothetical monopolist could profitably, optimally, and sustainably raise price in local markets for radio advertising, why do we not observe advertising prices rising? One possible answer, posited earlier, is that the individual radio stations in the local market do not collude in setting rates, but rather aggressively compete for advertising revenue from local businesses. If this is indeed the case, our results coupled with the new station ownership rules of the Telecommunications Act of 1996 suggest a potential need for continued antitrust scrutiny of radio mergers – at least to the extent that increased concentration of station ownership at the local market level facilitates collusion in setting local radio advertising rates.

VI. Conclusion The empirical literature on market segmentation in local media is extremely sparse. We have, in the present paper, attempted to answer a single question: whether, when controls for a number of variables are taken into account, local radio is or is not a separate and definable market? Applying an operational test based on the Merger Guidelines criterion, we find that substitutability within that market, while present, is low. Our results present a contrasting picture to the claims that local-level media are highly substitutable. If market segmentation is the issue, radio markets could be a separate interest for antitrust. Three important caveats are in order. First, our investigation is not a test for market power within local radio markets – a clear direction for future research. It may well be that multi-station ownership within local radio markets and the associated increase in concentration may not be detrimental to economic performance. We have provided evidence suggesting only that radio markets constitute an antitrust market to the extent that there exists a potential for market power to be exercised by a hypothetical monopolist. Second, the empirical approach adopted in this paper is only one of many potential analytical methods by which to evaluate the existence of a radio advertising market. Ideally, the importance of this topic and continued antitrust scrutiny will lead to the availability of a greater quantity and quality of data on radio advertising and related media markets so that further analysis can be conducted on the existence of a radio market and the degree of competition therein. Third, what we have shown is that the average, typical, or representative market of the 110 that composed our sample is possibly a separate antitrust market. We have not shown that all the markets meet that criterion nor have we shown that any particular real world market does. That latter inference can be considered in a straightforward manner by applying the tools employed here to a time series of that particular market – very much along the lines Kamerschen originally suggested. As a practical matter, in terms of specific antitrust cases, this may be the most important of these caveats.

256

ROBERT B. EKELUND, JR. ET AL.

References Baker, Jonathan B. and Timothy F. Bresnahan (1988) ‘Estimating the Residual Demand Curve Facing a Single Firm’, International Journal of Industrial Organization, 6, 283–303. Bates, Benjamin J. (1993) ‘Concentration in Local Television Markets’, Journal of Media Economics, 6, 3–21. Bowman, Gary W. (1976) ‘Demand and Supply of Network Television Advertising’, Bell Journal of Economics, 7, 258–267. Brown, James A. Jr. (1982) Statistical Determinants of Radio Stations’ Revenues and Trading Prices, Office of Plans and Policy. Washington, DC: Federal Communications Commission, August, pp. 1–37. Busterna, John B. (1987) ‘The Cross-Elasticity of Demand For National Newspaper Advertising’, Journalism Quarterly, Summer–Autumn, pp. 346–351. Department of Justice (1996) ‘Justice Department Requires Jacor to Sell Cincinnati Radio Station’, DOJ Public Announcement (August 5). Elzinga, Kenneth G., and Thomas F. Hogarty (1973) ‘The Problem of Geographic Market Delineation in Antimerger Suits’, Antitrust Bulletin, 18, 45–81. Federal Communications Commission (Policy and Rules Division, Mass Media Bureau) (1995) An Overview of the Broadcast Television Industry, August. Federal Communications Commission (1991) Notice of Proposed Rulemaking, 6 FCC Rcd 3275. Federal Communications Commission (1992a) Report and Order, 7 FCC Rcd 2755. Federal Communications Commission (1992b) Memorandum Opinion and Order, 7 FCC Rcd 6387. Federal Communications Commission (1994) Second Memorandum Opinion and Order, 9 FCC Rcd 7183. Huber, Peter W., Michael K. Kellogg and John Thorne (1996) The Telecommunications Act of 1996. Boston: Little, Brown. Hunsaker, David M. (1994) ‘Duopoly Wars: Analysis and Case Studies of the FCC’s Radio Contour Overlap Rules’, CommLaw Conspectus, 2, 21–41. Kamerschen, David R. (1994) ‘Testing for Antitrust Market Definition Under the Federal government Guidelines’, Journal of Legal Economics, Spring, pp. 1–10. Kennedy, Peter (1994) A Guide to Econometrics. Cambridge: MIT Press. Kerr, William (1996) Comments of the National Association of Broadcasters on the Advertising Product Market (May 15, 1996). Krattenmaker, Thomas G., and Lucas A. Power, Jr. (1995) Regulating Broadcast Programming. Cambridge: MIT Press. Moulton, Brent R. (1991) ‘A Bayesian Approach to Regression Selection and Estimation, with Application to a Price Index for Radio Services’, Journal of Econometrics, 49, 169–193. Petrozzello, Donna, and Elizabeth A. Rathbun (1996) ‘Mega-Deal Rocks Radio’, Broadcasting & Cable (June 24). Rathbun, Elizabeth A. (1996) ‘The Reordering of Radio’, Broadcasting & Cable (July 1). Stigler, George J., and Robert A. Sherwin (1985) ‘The Extent of the Market’, Journal of Law and Economics, 28, 555. Thompson, R. S. (1984) ‘Structure and Conduct in Local Advertising Markets: The Case of Irish Provincial Newspapers’, The Journal of Industrial Economics, 33, 241–249. Thompson, R. S. (1989) ‘Circulation Versus Advertiser Appeal in the Newspaper Industry: An Empirical Investigation’, The Journal of Industrial Economics, 37, 259–271. Werden, Gregory J. (1992) ‘The History of Antitrust Market Delineation’, Marquette Law Review, 76, 123–215.