![MEDIA CONCENTRATION POLICY IN THE<br />EUROPEAN UNION AND THE PUBLIC INTEREST<br />PETROS IOSIFIDIS<br />A thesis submitted in partial fulfillment of the <br />requirements of the University of Westminster <br />for the degree of Doctor of Philosophy<br />Centre for Communication and Information Studies<br />University of Westminster, London<br />June 1996<br />CHAPTER 4<br />4.1INTRODUCTION<br />The subject of this chapter is the exploration and analysis of the different methodologies of measuring media concentration. The appropriate measure depends on the objective of the measurement. The crucial question that should be asked beforehand is what is the purpose of the measurement: the examination of economic power or an assessment of whether market structure might restrict diversity in the media industry? So far media academics have borrowed measures that were developed by economists. Regarding the examination of economic power, economists have used companies' market share, shares of assets, value-added, sales, advertising revenue or even number of employees in forming an opinion of their bulk in the economy. These measures are more appropriate for industrial structure and manufacturing sector. In the media, because of their nature and their significant role in culture, society and politics measures examining the media firms' economic power alone seem to be inadequate. The special social significance attached to the media's role in disseminating information requires an investigation of whether a concentrated media market restricts the free flow of information.<br />To overcome this limitation of economic-based measures a few media analysts have proposed a number of media concentration measures which take into account their importance to the public. The view that has emerged from the recent debate on media concentration at European (initiated by the EU 1992 Green Paper) and national (Arthur Andersen's 1994 study UK Media Concentration) levels, is that it is possible to measure the 'influence' exerted by the media by applying audience-based criteria. It has been put forth that while financial units are close to the traditional systems of concentration measurement which permit assessment of media market concentration or even the existence of a dominant position (concentration of resources), audience-based methods are coherent with the cultural/political standpoint and can be held to be most effective for the measurement of pluralism and influence in the market-place for ideas. Nevertheless, influence over the audience cannot be assessed by using audience-based criteria, whether that is readership, audience reach, viewing or listenership share, etc. Audience exposure to mass media is certainly not the same as influence over the audience. In the end, these so-called new measures are nothing but refinements of measures of market power. They measure market power, although in a more sophisticated way. They are a form of market share measurement, which is a classic economic measurement.<br />This chapter examines the problems associated with both types of concentration measures and assesses their appropriateness in the broad context of media concentration's impact on the public interest. It will be argued that both financial and audience-based units apply, since economic power and diversity (especially in the range of material offered) are closely linked. A thorough definition of the markets in which market shares would be calculated is firstly provided, followed by a discussion on the measurement of economic power. In discussing the measuring of shares in the political/cultural market, the notion of the 'media controller' is firstly analysed, followed by an extensive examination of the various units that have been said to capture a media operator's ability to influence public opinion. At the end of this chapter alternative proposals are put forward in an attempt to develop a better measure which takes into account a large number of public interest issues.<br />4.2DEFINITION OF MARKETS<br />4.2.1 Defining the Geographical Market<br />When focusing on levels of concentration, one has to define the relevant market for which the market shares would be calculated. Relevant markets are defined firstly in terms of geography and secondly in terms of products or services (the good in question). The definition of the market-place can have an impact on what is perceived as its structure. The geographical dimension of market definition determines the scope of the market, that is, whether markets are defined as being local, regional, national or even international. The issue of the relevant geographic market being examined is particularly important, since the adaptation of either narrow or wide market definitions may lead to totally different results in measuring levels of concentration. <br />Most media economic studies have adopted fairly narrow definitions of markets. Regarding the newspaper market structure, those studies were often restricted to local markets for dailies. US mass communication researchers have investigated newspaper concentration for decades assuming that the city was the appropriate territory. Nixon (1968) talked about competition as a city phenomenon while Busterna (1988b, 1988c) analysed levels of concentration in cities. Picard (1988) moved away from the traditional way of describing markets and examined daily competition in metropolitan areas. After considering the national newspaper market and thirty markets from the largest 100 metropolitan areas, he reached the conclusion that newspaper concentration increased as the geographical size of the market decreased. Finally, Lacy and Davenport (1994) examined the degree of concentration and competition in the US newspaper industry based on the county. They argued that although daily newspaper markets showed higher concentration and lower competition than other US industries, these markets were less concentrated and had a greater potential for competition in 1983 to 1988 than suggested by other studies using cities as the geographic market. <br />It is obvious then that using the city as the geographic market yields different descriptions of market structure than using the county as the geographic market. The same could be said about national and local or regional markets. The Arthur Andersen study, for instance, adopted a national perspective in examining UK media concentration. As we shall see below, in order to measure company shares in the national cultural market, the consultancy used the time that the public actually spends on the media. It is easily understood that, under such interpretation, regional channel Carlton's share will be smaller in relation to media with national coverage (i.e. BBC). In fact, the above analysis found that BBC's share in the national market is 44.1% while Carlton's share in the same market is just 6.9%. However, it is meaningless to calculate Carlton's national market share since people living outside the London area do not have access to that channel.<br />Broadly speaking, some markets such as the financial ones may be international, while others such as that of magazines may be regional or national, and others such as cable or the sale of advertising may still be very local. Local advertisers, for example, are interested primarily in the local market and thus normally restrict themselves to consideration of local markets. So there are several definitions of the geographical market depending on the product in question. One thus should be careful when establishing the sphere of competition and trying to determine the extent and effect of competition on pluralism. Insufficient competition and therefore excessive influence may arise at a number of levels - local, national, regional, or international. The regulator must ensure reasonable diversity and balance in the information and entertainment available at all these levels. Referring to the UK media market, Robinson et. al (1995: 57) argued that 'the main worry, from a pluralism viewpoint, is that a company might acquire excessive influence in a local market through its ownership of both a local newspaper and the local commercial TV franchise. Yet, if such a company operated only in one English region, its share of the national media markets would be so small that it would comfortably pass under all the proposed thresholds'.<br />The above discussion illustrates that cross-ownership between a local TV channel and a local newspaper (or a local radio station) would lead to elimination of competition and therefore increased concentration in a media market, only if they are based in the same area. This is generally true since different newspaper groups or broadcasting channels operating in different regions do not compete for the same audiences. Yet they might be in direct competition for advertising revenue. Referring to the newspaper business, Veljanovski (1990: 18) stated that 'national newspapers are seeking a greater share of the classified advertising business dominated by the regional and local press, through regional editions and inserts, and the regional character of their readership. Regional newspaper companies which each own many local and regional newspapers, now offer attractive rates for advertising simultaneously in all their newspapers'. From a pragmatic viewpoint thus, it cannot always be assumed that regional and national media outlets are not in direct competition for advertising revenue.<br />4.2.2 Defining the Product Market<br />Besides being geographically-defined, markets are also product (or service)-defined. 'The products chosen must form part of a group with characteristics which separate them in an economically sensible way from all other products' (Frazer, 1992: 14). The economic criteria for market definition are based on the notion of substitutability across products as measured by their price elasticities. According to competition policy, two products belong to the same market when their producers' price-setting behaviour is interdependent. There must be substantial interchangeability (if not homogeneity) between these products and others. According to Frazer (1992: 14-16), the degree of interchangeability may be assessed from both producers' and consumers' viewpoints. The first method of assessment - supply-side analysis - seeks to identify those firms which are either supplying the products, or which will be willing and able to supply the products if there were an increase in their price. The second method of assessment - demand-side analysis - measures the willingness of consumers to substitute one product for another. If there is a high degree of substitution between the two products, then they will be considered to be part of the same market. If, for example, consumers are willing to substitute free-access television (whether financed by advertising or licence fee) for pay-TV, then these will form part of the same market.<br />The current technological developments in the media sector have made it difficult to reach a proper product market definition. Indeed, in the current 'multimedia' age (i.e. proliferation of new media, convergence of technologies), the question of what products are part of the same market is becoming hard to answer. The above described studies concerning the US local newspaper industry, narrow the market by ignoring the impact of non-local dailies and of print sources other than dailies (other non-print media). Given that a meaningful definition of a media market must indicate the range of alternative sources of supply open to readers, listeners and viewers, conventional ideas about control of the market by publishing or broadcasting companies must be open to criticism (Peacock, foreword in C. Veljanovski, 1990: 6). Cento Veljanovski (1990: 16) illustrates this by giving the example of one firm which produces 100 per cent of the domestic output of white handkerchiefs. This may indicate that the company is the only domestic producer, but it is not evidence of complete monopoly and does not imply the ability to influence price unilaterally. Such a manufacturer faces competition from imports, substitute products (including handkerchiefs of another colour or tissues), and from potential entrants. Similar considerations, the argument follows, apply to the media industry. Television companies compete for audiences not only between themselves but also with VCRs, cable and satellite TV, and for advertising revenue with all these and also with radio, newspapers and magazines. Competition among media sectors thus must be taken into account when defining the relevant product market.<br />The fact that different media are seen as substitutes is shown if someone looks at local news which could be found on television, radio and local papers, and at national news which could be found in all these and also in cable TV, regional radio outlets, and so on. The same or similar information can be delivered via print, on-line, teletext or audiotext and there are products and services which combine different communication modes: images, text and audio. What is the definition of the broadcasting market then? If we take as an example the release of a film which can be viewed firstly in a cinema, then as a TV programme and lastly in a home video, then the market for this product should include all these sub-markets. A soap opera, on the other hand, may only be used as a TV programme and therefore the market for this product is television. Thus the market for soap operas and films is defined differently. Further, should the definition of the broadcasting market cover electronic publishing? However, if electronic papers are defined as televised media, then they should be covered by the very same rules that applied to the broadcasting sector (many would strongly oppose that). As stated by Jonathan Davis (1994: 7) 'it will be necessary to adopt a much more fungible approach to market definition than either competition authorities are used to or is currently applied to the media'.<br />As the distinction between both electronic and non-electronic media and terrestrial broadcast TV and telecommunications services becomes more and more blurred, so does the definition of the relevant market. The presence of close substitutes can expand the market, resulting in lower levels of concentration. Therefore, several alternative market definitions need to be considered simultaneously. Levy and Setzer (1984), for example, considered concentration levels in the US national video market-place under alternative definitions of the video market-place-broadcast only, broadcast plus cable, etc. There have also been many studies that, by trying to identify the user of the medium while measuring concentration, draw a distinction between audience markets and advertising markets. Mc Fadyen et al. (1980) noted the difference between market for revenue (through time sales to advertisers) and market for audiences and measured concentration in Canadian broadcast markets separately for the two groups. Larson (1980) not only considered concentration in the US television industry at both national and local levels but he also looked at levels of concentration in broadcasting and among broadcast advertisers. Levels of seller concentration (TV stations) were measured by station net weekly circulation, estimated TV station income, network advertising revenues and by national and regional advertising revenues. Buyer concentration was treated both by advertising agencies, who purchase time in the industry for television advertisers and by the advertisers themselves, who are the ultimate consumers of the television time. These studies make it clear that much of the media serve two general markets - that of consumers and that of advertisers - and further distinctions within these markets based on specific types of advertising and/or region have to be drawn.<br />The main point so far has been that media today can be seen as close substitutes because of the convergence of technologies. However, it is not only the substitutability in distribution networks that matters but also substitutability in content. We indeed have convergence in distribution today but that does not necessarily mean that what is on TV is substitutable for what we read in newspapers, or listening to the radio is substitutable for watching television. Newspapers may be read through computers today (i.e. Internet) but the content that is provided to Personal Computers (PCs) is the same. In economic literature, substitutability relates to the question of whether a range of goods or services in a market are essentially, as far as the consumer is concerned, choices between one or the other in terms of quality or price; or are they provisions of alternative goods, in which case the choice between them will depend on the consumer's criteria of utility. People do make choices between going to the cinema or going to restaurant, but these do not compete in the same market (they are not substitutes). On the other hand, going to a restaurant and take-away from a restaurant are probably substitutes. <br />When it comes to the media industry, the question not only is whether television is a substitute for newspapers but also whether advertising in television is a substitute for advertising in newspapers; or news and opinion output (where pluralism matters greatly) provided by newspapers is a substitute for news provided by TV. From a consumer point of view, does it make any difference if he/she orders a video tape from a telephone company or takes it from a video store? Surely such decision has nothing to do with content at all; it is just the convenience of delivery which is competitive and that makes a big difference on policy grounds. Referring to the television sector, Bates (1993: 7) argues that 'audiences are concerned primarily with video signals, no matter the delivery channel. In this sense, audiences may not make much of a distinction between regular broadcast television, cable and alternative delivery systems such as home satellite dishes, low power TV, multi-point delivery systems, etc., at least to the extent that such alternatives are available'. The definition of substitutability thus has to be drawn very carefully, encompassing both substitutability in distribution networks and substitutability in content.<br />These distinctions suggest that considerations of the print and broadcast market structure need to be based on multiple definitions of markets, with separate considerations of local, regional, national and international markets, of audience and advertising markets and of potential substitutes. Apart from this aspect of the problem however, one has to find an appropriate standard for the measurement of market share. There follows an assessment of the relative merits of both revenue and audience related criteria firstly for the measurement of economic power and secondly for the measurement of diversity and pluralism. It will be argued that both financial and audience-based units apply, since economic power and diversity (especially in the range of material offered) are closely linked.<br />4.3MEASUREMENT OF ECONOMIC POWER<br />One of the arguments about levels of concentration concerns the measurement of the magnitudes. The most usual method consists of the presentation of absolute numbers: rankings according to turnover, profits, number of employees in the companies or, in the media industry specifically, number of TV stations or newspaper titles, and so on (Seymore-Ure, 1991). Shepherd (1979) argued that one can use firms' shares of assets, value-added, employees, or profits in forming an impression of their bulk in the economy. Nevertheless, he also noted that each measure has a slightly different meaning. Assets may be the best indicator of the firms' economic power while value-added shows their role in the country's production process. The large companies' share of workers reflects their influence over jobs or even votes whereas profits reveal their ability to gain market advantages compared to the rest. All these 'technical' problems in measuring corporations' scope (size) indicate that there is no single best measure of 'aggregate concentration' in an industry. 'Bigness' therefore is something everyone understands but no one can measure definitely. <br />Apart from these problems there is the question of which number of large firms is the 'best' one for defining the share of large firms. As indicated elsewhere, measuring concentration in media markets requires both a measure of the market (identification of the market) and a measure of concentration. A number of indexes have been developed since the 1950s in the USA for both purposes. The Bureau of Census and scholars of American industries have presented concentration data and ratios as a percentage of control of a measured value by the largest 4, largest 8, and largest 20 ownership units in an industry (see Larson, 1980: 25). The 4, 8 and 20 criteria were used by Caves (1962, 1966) in his airlines and in his water industry studies, and by Bain (1968) in his studies of business concentration. One thus can slice the large-firm share at any of a number of levels. Shepherd (1970, 1979) however argued that the four-firm concentration index is the most commonly used measure of concentration in general industry studies. The advantages of a concentration ratio are firstly that it shows the degree of concentration of few-firm leadership directly and correctly and secondly that it shows changes in market power with special clarity. The defects though are that concentration ratios are often too inclusive, embracing scores of disparate products and distinct regional sub-markets. The national ratio of, let say, 10 per cent for newspapers for instance, may conceal very high local concentration. In addition, concentration ratios may be too narrow. In some cases imports, not included in the ratios, cause concentration in total sales to be lower than the ratios state.<br />The 4, 8 and 20 criteria were also used by scholars including Nixon and Hahn (1971) who specifically treated concentration in the newspaper industry. A more recent analysis carried out by Picard (1988) used four ways to measure concentration in the US daily newspaper industry: the one-firm concentration, the two-firm concentration, the three-firm concentration and the Herfindahl-Hirschman index (HHI). The first three concentration measures are the percentage of circulation controlled by the largest circulation newspaper, the two largest circulation papers and the three largest circulation titles respectively. The HHI is a way of placing concentration on a continuum from 0 to 10,000, the latter representing a monopoly. The index is determined by summing the squared value of the market shares of all of the firms in a market. The HHI uses shares ranging from 0 to 10,000. But at what point is the market considered concentrated? Stigler (1983), one of the supporters of using the HHI for concentration measurement purposes, argued that one can roughly identify markets with HHI>1,500 as tight oligopolies and those with HHI<1,000 as loose oligopolies. If the index falls well below 1,000 then the market is unconcentrated and when it falls above 1,800 then the market is highly concentrated.<br />As regards the television market, in particular, the evidence shows that little variation can be found in the most commonly used concentration ratios (2, 3, 4, 8 , or even 20-firm). Larson's 1980 study referred to above, measured levels of concentration in the US TV industry using a 3, 8 and 20 breakout and concluded that all the used measures revealed that the TV station industry was not highly concentrated. Such conclusions may not be true, not least because television broadcasting is not an ordinary, typical industry. The television market is distinctive in a number of ways: a) in that the question of economic power has manifested itself in two distinctive policy arenas. On the one hand, there is concern over the impact of concentration on economic conduct and performance, pricing behaviours, barriers to entry and limitation of output (Ferguson, 1983; Bates, 1993). On the other hand, there is the question of the social performance of the market (pluralism, diversity, and so forth); b) TV stations are operating in several markets simultaneously. In one market they trade programming for audiences and in another market they sell audience to advertisers (Bates, 1987, 1993); and c) television stations operate in a number of geographical levels (Compaine, 1981, 1985). Therefore, according to Bates (1993), the strict barriers to entry and the domination of affiliates of the three networks make the concentration ratios of little importance in measuring concentration in the US local TV markets.<br />4.4MEASURING SHARES IN THE POLITICAL AND CULTURAL MARKET<br />Large companies' sales and turnover may be the best indicator of their economic power and reveal their ability to gain market advantages compared to the rest. In other words, very powerful firms can influence economic conduct, performance and pricing behaviours and have an impact on barriers to entry and limitation of output. Therefore, when the purpose is the traditional examination of market power then a high revenue company share may provide a quite useful guide. When it comes to the media however, the concern not only is over the impact of concentration on economic aspects but there is also the question of the social performance of the market (i.e. pluralism and diversity). Are measures tailored to measure economic concentration good enough to capture concentration levels in the political and cultural market, the so-called 'market-place for ideas'?<br />The view which has emerged from the recent debate on media concentration at European (EU 1992 Green Paper) and national (Arthur Andersen's 1994 study UK Media Concentration) levels, is that it is possible to identify a sort of relevant 'market for ideas', which does not coincide with the economic definition of relevant market; and that de facto restrictions of pluralism and diversity are the results of an abuse of power in such market (abuse of political and cultural power). There are three problems associated with such approach. Firstly, there are substantial difficulties in defining a suitable notion of relevant market in the political and cultural sense. As the relevant product tends to extend across different media, the cultural/political notion of the relevant market may be significantly broader than the economic one. The problem is bound to be exacerbated as multimedia conglomerates expand their activities further, and ownership of complex transnational media chains becomes widespread. To illustrate, how does one assess the effective combined share of, say, News International in the broader market for information, culture and political opinion, comprising newspapers and TV outlets in several countries? Secondly, the exact nature of the potential abuse is not clear and explicable and cannot be specified in the same way as abuses of economic market may be specified. What then counts as an abuse of power in the political/ cultural market? Beyond the general assumption that all media exercise some form of political and cultural influence on the public, there have emerged no satisfactory criteria so far for the definition of a broad political and cultural market in which spheres of influence by a single controller could be assessed.<br />The most serious reservation concerning this approach though has to do with the selection of the criteria for measuring diversity in the market place for ideas. It has been put forth that while financial units are close to the traditional systems of concentration measurement which permit assessment of media market concentration or even the existence of a dominant position (concentration of resources), audience-based methods are coherent with the cultural/political standpoint and can be held to be most effective for the measurement of pluralism and influence in the market-place for ideas. Nevertheless, influence over the audience cannot be assessed by using audience-based criteria, whether that is readership, audience reach, viewing or listenership share, etc. Audience exposure to mass media is certainly not the same as influence over the audience. In the end, these so-called new measures are nothing but refinements of measures of market power. They measure market power, although in a more sophisticated way. They are a form of market share measurement, which is a classic economic measurement. Audience-based units are the equivalent of, say, measuring sales - i.e. market share - which is a classic economic measure of market power.<br />In any case, political/cultural diversity and economic power are closely linked. It might be worth at this point spelling out the arguments about the relationship between economic power and the range of material offered. There is a clear relationship between economic measures of media power and influence/pluralism because economic power determines the control over choices offered. In fact, in terms of the public interest and debates about regulation and concentration of media ownership, there are two wide-spread arguments. On the one hand, there is the argument saying that a highly concentrated market structure in the media sector is of concern not only for the possibility that it may lead to abuses of economic market power, but also for the potential effects on pluralism. A large media player who controls a substantial portion of at least one media sector (for example daily press or TV) has the potential for forcing his/her views across a range of products (political/cultural bias), and thus for restricting the choice of products available to the public in political and cultural terms. In this sense, a competition policy decision aimed at curbing an abuse of economic market power (excessive pricing, creation of barriers to entry, etc.) may also increase pluralism, at least in the sense of reducing bias.<br />On the other hand, there is the argument saying that increased competition may lead to less pluralism in the market. Increasing the number of firms in an industry does not necessarily imply greater diversity in the quality and variety of products on offer - especially where price competition is weak. If firms compete on price, product differentiation provides a device for softening the intensity of competition: in a simplified world with only two companies, they will have an incentive to locate themselves as far as possible from each other on the product line (i.e. to offer as diverse a product as possible in terms of product variety and quality). Proximity of location would mean that prices are gradually eroded as the companies compete for each other's business. However, if there is no explicit interaction in the firms' pricing decisions, the opposite result obtains: the firms will locate as close as possible to one another, as the 'market share effect' (the incentive to be where demand is, or to increase one's market share given the market structure) prevails over the 'strategic effect' (the interdependence of the two firms' pricing decisions). Thus the incentive to differentiate products is weaker when companies are able to operate in the near-absence of price competition. The tendency to converge on tried-and-tested formulae poses a potential danger to welfare in terms of the variety of products offered by the market. Hotelling (1929: 41), who originally discussed this effect, talked of 'an undue tendency for competitors to imitate each other'. Therefore, a more fragmented industry structure in the media sector may not necessarily deliver the socially desirable level of product differentiation because it may be more profitable for the companies to locate 'where demand is' (i.e. stick to the middle ground in order to catch the widest audience).<br />A further important question relates to the possibility that too much competition might display a bias in favour of certain types of products and neglect others. The particular bundle of commodities that are actually produced in the media market (the type of programmes/titles available) might be sub optimal from a social welfare point of view. When demand for products in a particular category is generally inelastic, the products which are being actually offered may end up positioning too close to each other (sub optimal product diversity); and those products for which the elasticity is comparatively lower may not be produced at all. The implication could be that some segments of tastes and preferences might systematically not be catered for, although there might be a large number of different media products. So, strictly from the point of view of pluralism, there might be no automatic advantage to be gained from a more diverse media structure. On the other hand, so the argument runs, a very concentrated industry structure might lead to great diversity, if the dominant firm(s) seeks to prevent entry in the market by filling all gaps in product space.<br />Having provided, to some extent, an argument that economic power affects the range of material offered, and having spelled out the arguments as to whether concentration or a fragmented industry can deliver best the desired diversity, we now turn to the question of which criterion is the 'best' one for measuring concentration levels for media pluralism purposes. The close relationship between economic power and pluralism/diversity indicates that criteria that are being used for the measurement of market power can also be used, at least in principle, for the measurement of media influence and vice versa. Financial criteria, for instance, a long-established method for measuring market power, could also be adopted for measuring 'influence' (audience exposure to the mass media); and audience figures, supposed to be more efficient for measuring diversity in the market place for ideas, could also be a measure of economic power, especially as they are sold to advertisers. There are still problems though, since different criteria might give a rather different picture in terms of the participants' individual shares and may lead to disparate inferences on their roles and importance. The rest of this chapter examines the problems that are inherent in defining an appropriate measure of concentration in the information market. <br />4.4.1 Criteria Based on the No of Licences Controlled by a Single Operator<br />Media concentration rules in the audio-visual sector have traditionally been placed on the number of licences or franchises a single operator can hold. Most Western European countries have developed such restrictions with the aim of preserving pluralism in the market. For Frohlinger (1995), head of Unit E/5 'Media, Commercial Communication and Unfair Competition' of DG XV, there were two reasons for this: the scarcity of available frequencies, which required the prevention of one person or body from using all or most of the spectrum; and the lack of reliable audience measurement. However, technological change together with the proliferation of channels and the convergence of different forms of distribution imply that it is no longer necessary (or even suitable) for governments to limit the number of licenses. Further, the progress in audience measurement enables the calculation of reliable figures on the basis of which thresholds for the access to media ownership can be fixed. This is one of the reasons why the European Commission wants the replacement of numerical restrictions on the number of licences by thresholds based on audience share. As is explored below, the Commission is of the opinion that the audience share criterion would allow more flexibility, growth, innovation and increased diversity in the media market. It would facilitate, in particular, the launch of multi-channel packages (packages of specialised or thematic channels) which under existing laws limiting the number of channels are difficult to develop.<br />Rules placing limits on the number of licences or channels a person may control are indeed not appropriate by themselves as measures of plurality in the media market, since they pay no attention to media revenue, coverage or audience. These rules may measure diversity of media ownership but are crude as measures of the existence in a society of a number of different points of view. To illustrate, in the UK some Channel 3 licences are much larger than others in terms of advertising revenue and audience, so that rules limiting the number of franchises one can own but taking no account of revenue and audience size cannot prevent undue influence. It is imperative thus to take total revenues and population coverage into account, in addition to the number of outlets owned by an individual or a group, when constructing rules on media ownership. Lessons from press regulation are useful here. The press has never been subject to rules counting the number of titles one may hold. Traditionally provisions aimed at limiting press concentrations have fixed thresholds based on circulation figures and not on the number of titles that can be owned by a single publisher. In most European countries a merger qualifies for investigation on the basis of circulation, and not on the number of titles controlled.<br />4.4.2 The criterion of the Media Controller<br />Diversity of media ownership could thus be measured by using the criterion of the number of media carriers (TV channels, newspaper titles or radio outlets) owned by a single 'controller'. The purpose is to find a definition that makes it possible, when a new channel or newspaper is launched by a group or an individual (or when he/she acquires control of an existing channel or newspaper), to estimate the number of media that are already controlled by the group or individual and broadcast or distributed over the same area as that of the new scheme. Determining who controls what media has a distinct advantage. This sort of knowledge undoubtedly makes it possible to provide the legal certainty for investment and sufficient effectiveness for avoiding circumvention. In addition, the concept of media controller is more accurate compared with the concept of owner or majority shareholder in the media company, since there may be several shareholders with the same proportion of ownership. The problem is that it is not always easy to identify clearly who is the controller with decisive influence. Who is the actual controller of a media outlet in the case, say, where there are three shareholders each having 33 per cent share of the capital of the medium?<br />There is thus a technical issue of the relationship between levels of ownership and control. Before discussing that however, it seems to us that a preliminary issue arises, and that is the question of whether owners are controllers at all. In other words, can we simply equate ownership and control? Does ownership actually translate into control? Undoubtedly the answer is yes in cases where the founder of a media company holds the majority of the shares and his/her family the rest. Traditional press company structures have been formed along these lines. Today things are different though. The formation of large vertically-integrated and diversified units has led to a corporate structure in which shares are normally dispersed, in which financial institutions and industrial corporations have significant shareholdings, and in which highly qualified managers make key decisions. In situations of dispersed share ownership, who can be said to control the firm in question? When ownership is dispersed, does that mean that the share owners still control it? <br />The measurement of control in terms of ownership has weaknesses because it does not take into account other ways and means by which influence can be exerted over a firm's decision-making centres. Managers, for example, have been appointed in key positions in diversified enterprises (and often given the opportunity to hold minority stakes) because their specialised knowledge and expertise has been considered essential for the smooth running of the modern media enterprise. This, however, creates an agency problem, i.e. it poses the problem of the relationship between owners/shareholders and management, often putting the editorial control with the management. An agency problem is always evident in the presence of delegation, that is, in all cases where one individual is contracted to carry out a task on behalf of another and the parties' incentives do not coincide a priori. A typical case is the shareholder/manager relationship, where the manager's decisions reflect his/her own preferences (which might include greater power or increased prestige) rather than the shareholder's goal of profit or value maximisation. The instrument available to the shareholder for aligning the manager's incentives with his/her own is the contractual arrangement: by making the manager's compensation also a function of his/her same objective (firm's profit or stock market value) the efficiency loss from delegation might be reduced.<br />Some economists stated that the above argument has great validity in cases where there are few shareholders but it might be open to question where there are many minority holdings. They pointed out that solving the agency problem and placing an effective control on management is relatively easy in cases where there is one or few shareholders. Nevertheless, it is far less likely to be successful where ownership is significantly fragmented across a large shareholder base. Dispersed share ownership, according to them, has important implications for the effectiveness of shareholder monitoring of management. In the media sector, the implication of having fragmented ownership due to strict ownership rules, might be that the editorial control would rest entirely with the management. To the extent that management might also pursue a particular agenda, ownership restrictions would then simply replace one form of potential threat to pluralism (that from the owner) with another (that from the management). Therefore, while the dominance of a single individual or shareholder group might be prevented, the management might in turn pursue objectives which are not less damaging in terms of pluralism and diversity. Within a company thus there is clearly a question of who actually controls the company's policy - the owners or the managers. It seems to us that a case-by-case analysis is required to assess who makes the key decisions which influence the editorial content.<br />It is not only the emergence of the new elite of managers that has raised questions over the relationship between ownership and control. It is also the links between advertising industries and the media that have masked the above separation. There have been two trends in the advertising industry that justify such assumption. Firstly, the progressive concentration of the advertising mediators (agencies and buying houses) and the separation between the creative activities (agencies) and the management of accounts (buying houses) have changed a key stage in the relationship between the media and the advertisers: the determination of the prices of insertion and other contractual conditions of advertising space and time. Secondly, new advertising formulae for advertising and commercial promotion such as bartering, product placement and teleshopping represent a means of integration between the advertisers and programme production industries and undoubtedly weaken the barriers between production, programming and advertising. Despite the economic advantages of bartering, for example, this formulae for advertising poses the risk of loss of editorial control by broadcasters, as does teleshopping and to a lesser extent product placement.<br />The increased involvement of other concerns such as financial institutions and industrial corporations in media share ownership is another trend that needs consideration. Marxists like Murdock and Golding argued as early as 1977 that 'shareholdings in the large corporations have become progressively concentrated in the hands of the dominant financial institutions and of other large corporations' (Murdock and Golding, 1977: 30). They added that 'compared with the general passivity of most private shareholders, institutional investors are likely to pursue a much more \"
interventionist\"
policy with respect to the operations of companies in which they have holdings' (ibid: 30). For Murdock and Golding though, such interconnection of the major communication firms with the dominant financial and industrial interests prove the fact that the ownership has not actually been separated from control. 'Control over the key processes of resource allocation is still significantly tied to ownership' (ibid: 32). They supported that argument by saying that both interests - communication and industrial/financial ones - share similar views. In their words, 'the owning group continue to constitute an identifiable capitalist class with recognisable interests in common' (ibid: 33). Negrine (1994b: 63-4) however challenged that statement on the grounds that it is 'too general to be meaningful', it is a general assumption 'pertaining to a concern for success and survival'. For Negrine the media exist in a competitive market place and have to employ whatever means possible to survive in that market place. So the general validity of the argument concerning the allocative power of ownership is in no doubt. What is in doubt for him, in view of lack of evidence, is the assumption that economic aims constitute the basis of pursuing a particular ideology ('conservatism') in the media.<br />The question of whether ownership is equated with control (and the search for a proper interpretation of 'control') is still with us today and forms a significant part of the current debate over establishing common ownership rules at European level. The Commission has initiated a study on this subject in order to reach a definition of 'controller' which will reflect experience at national level and will be in line with the various definitions of control already used in competition law and company law. The study, carried out by the European Institute for the Media (EIM), tried to identify who is genuinely in control of a press or broadcasting enterprise. After reviewing several criteria for defining the controller in a number of European countries, it identified five means of influencing a firm's decision-making centre: direct links, indirect links, family connections, external financial or contractual relationships and unofficial agreements. Despite the fact that direct links with the enterprise appear at first glance to be the most influential means, it would be wrong, according to the analysis, to underestimate the others. Owning a stake in a company either indirectly or through a relative (Kirch, for instance), could be just as effective a means of wielding influence as being a majority shareholder. Likewise, appointing the management (Murdoch's strategy) could be as powerful a means of exerting influence as being appointed to a management position. The study also identified four ways of wielding influence: influence through ownership links (capital), influence through links with staff, influence through financial links (debts) and influence through contractual links (supplies, distribution). It concluded that ownership links are the most effective way of gaining influence over the board of directors, although the other three - financial, staff and contractual links - should also be taken into consideration.<br /> <br />On the basis of these criteria it is then possible, with respect to each player directly or indirectly involved in the operation of the media in question, to evaluate by which ways and means he/she may exert a decisive influence on the operation. Margot Frohlinger of the European Commission (1995) argued that such a qualitative definition of media control seems both to be better adapted to the complexity of current ownership and control/dependency structures and respond to various attempts at circumvention of media ownership rules. Indeed, the EIM's analysis suggests that 'control' must be interpreted as widely as possible to include any means of exerting influence over the actions, editorial content, strategic direction and so on of the media organisation in question. Apparently though the Commission is concerned that the four criteria of wielding influence defined in the study are not clear enough to ensure fair and impartial application throughout the EU. This concern is expressed in its Questionnaire No III, questions E.1-E.8 (see Appendix XIII) where it seeks opinions on which of those criteria are most important and which have subsidiary status. According to National Economic Research Associates (NERA, 1995), it does not seem feasible to put together so many different variables for monitoring media control. Their conclusion is that the only satisfactory approach is to abandon hope of achieving absolute legal transparency, and to adopt a discretionary test based on an ownership criterion such as that provided by the newspaper provisions of the British Fair Trading Act (i.e. control applies where one person owns a certain per cent of the voting rights of the media enterprise).<br />It is indeed impractical to determine theoretically whether one particular criterion is of more importance than another. According to British Broadcasting Corporation (The BBC's Views: Pluralism and Media Concentration in the Internal Market, 1995), there are two reasons for that: a) the relative importance of criteria is changing over time; and b) in some cases a 'weak' link can be used to exert more influence than a 'strong' link in others - for example, non-voting shares offer less opportunity to influence editorial control than friendship with the editor. However, any definition of the controller based on ownership links (or the voting rights of a media company) should be complemented by at least one of the other criteria identified in the EIM study, that of contractual links. The inclusion of the criterion of contractual links is particularly important because it makes it possible to cover cases of vertical integration. The EIM analysis has the merit of demonstrating how growing vertical integration and how concentration on so-called 'intermediate markets' (i.e. programme supply and distribution) have an increasing effect on media pluralism. In particular, the study examines the possible consequences of the introduction of new technology and concludes that both upstream and downstream industries of the media enterprise are developing in a way which is likely to reverse the balance of power against the broadcasters. According to the European Broadcasting Union (EBU, 1995), it is not so much vertical integration as such that poses a problem for media pluralism but concentration in intermediate markets at different stages of the audio-visual chain. Indeed, there is nothing intrinsically wrong with vertical integration. The fact that a broadcaster has its own network of correspondents, produces a great part of its programmes and operates its own transmitter network helps it to maintain its independence and probably contribute to media pluralism. Vertical links become a threat to media pluralism only if they are coupled with dominant positions in intermediate markets, thus undermining the independence and development of the broadcasters (either by imposing or by withholding certain programme material).<br />Downstream activities such as distribution are increasingly developing into industries in their own right, separating their activities from those of content provision, and facing increasing competition as new technologies facilitate the entry of new players. This means that their bargaining power in terms of contractual links will take on an importance not seen in any of the traditional media. The ownership of conditional access systems by broadcasters is an example of this trend. The provider of the conditional access system that consumers need to decode encrypted digital broadcasts becomes the 'gatekeeper' of the digital broadcast, able to limit the flow of broadcasts and the access of broadcasters to his/her system. If the conditional access company is owned by a broadcaster, that broadcaster can restrict competition in the market (BBC, 1995). Thus ownership of this form of distribution system should be included among the list of criteria for identifying the nature of the media controller and the exercise of that control. Contractual rights to content should also be added to forms of control that can be exercised. Certainly, control can be exercised by buying the rights to be in the popular markets of sports and films. This is what is happening with BSkyB imposing conditions on cable channels with the right to carry its programme. Media groups should not be in a position to control the preceding stage of the audiovisual chain to an extent which would allow them to prevent broadcasting services not belonging to the same group from having access to film catalogues, sports transmission rights, television advertising, or other resources.<br />Having setting up the arguments and explained, to some extent, that there is a difference between ownership and control and therefore diversity of ownership might not lead to the desired levels of pluralism in the media market unless we extend the notion of 'control' to encompass influences exerted by managers, advertisers, companies upstream or downstream, etc., we now turn to the question of how control can be defined technically. A controlling interest (over 50%) or the ownership of a certain percent of the voting rights of the company should be enough to imply control of the enterprise in question. The problem arises where there are two or more large share-holdings in the same medium (for example, two persons having over 20% stake). The simplest way is to treat only the biggest shareholder as the controller of the medium in question. Although this way does not cause any problems in numerical terms, it has two major drawbacks. Firstly, the largest shareholder is not always the effective controller of an enterprise. Secondly, this method gives incentives for small shareholdings across different media.<br />Another approach would be to attribute owners influence according to their share-holding level. There is an on-going debate in Europe, and in Germany in particular, on whether owners above a certain level should be treated as controllers of a media unit. The German regulators suggested in September 1994 that an audience-share model to control TV ownership should be adopted (a reorientation of the debate is now under way, with consideration of a new model based on commercial revenue). The setting of a limit for each investor's holdings at 25% of total viewing share, could define the controller(s). Bertelsmann and the Kirch Group argued that it is unfair to take into account the whole audience share of a channel in determining whether an investor's overall share breaches the limit; instead, a proportion of the share, equivalent to the investor's stake, should be considered (see Table 4.1). Although the companies asserted that pro rating each person according to their stake is an impartial method that provides for fair company treatment within a media sector, we will agree with the German authorities that this model is complicated and may also be used to conceal an owner's real influence in a channel.<br />Table 4.1<br />How German TV Groups Would be Affected by the Media Authorities' Plan - and by their Own Counter-plans<br /> Audience shares to count towards companies' combined audience share Groups and channelsPercentage stake %Audience share %Authorities model %Bertelsman model %Kirch model %BertelsmannRTL37.117.817.86.66.6RTL 27.83.43.4-0.3Vox24.91.91.90.50.5Premiere*37.5----Total23.17.17.4Kirch GroupSat 1**43.014.614.66.36.3DSF24.51.21.20.30.3Premiere*25.0----Total15.86.66.6If Th. Kirch chnnels inclPro 747.59.29.24.44.4Kabelkanal45.02.02.00.90.9Overall Kirch total27.011.911.9CLTRTL TV47.917.817.88.58.5RTL 215.03.43.4-0.5Total21.28.59.0<br />Source: Satellite TV Finance, 29 September 1994<br />Note: The authorities recommended that an investor's stake of 5 per cent or more attracts 100% of a channel's overall German audience share, and that an investor's combined share should not exceed 25 per cent. Bertelsmann argued that only investor stakes of 20 per cent or more should be taken into account, and that only an equivalent proportion of the audience share should be allocated. Kirch also mentioned that proportionate audience shares should be allocated, but there should be no lower stake limit. An investor should be allowed a combined audience share of 35 per cent.<br />* Shareholdings in Premiere are not relevant (a separate structure was to be developed for pay-TV)<br />** No account is taken of Kirch's stake in Sat 1 shareholder Springer<br />A more reasonable and feasible approach would be to treat all those with a shareholding above some threshold as controllers of the firm in question. Each of the partners above a certain level of control should be considered as a 100% controllers in terms of their position in the market, and should be subject to the same constraints and responsibilities as if they were the sole controller. There are still two problems with this approach. Firstly, it is likely to lead to totals summing to more than 100%. Pearson, for instance, owns the Financial Times and the Economist, and it has a 14% stake in 'Yorkshire/Tyne Tees TV', to name but a few. The second problem is to fix the threshold above which an operator would be regarded as the controller of the medium in question. NERA (1995) proposed that a 20% stake should be considered as the minimum threshold that a media operator should have in order to be regarded as controller. The analysis assumed that ownership below 20% does not confer control. The 20% level was chosen arbitrarily using three reference points, namely the 20% limit on cross holdings and accumulation of holdings in the British 1990 Broadcasting Act, the 25% limit for newspapers mergers, and the 15% test level used by the Director General of Fair Trading (DGFT) in judging the likelihood of material interest. The 20% level seems reasonable and close to that used in other European countries' legislation such as Germany (25%) and Greece (25%). One point that should be born in mind though is that in the United States a shareholder can be deemed to be in full control of a company once he/she has a stake of as little as 5% of its capital.<br />Any restriction setting a ceiling on the maximum share of the market which may be held by one controller (individual or interest group) runs the danger of being not only arbitrary but also easy to bypass. Choosing the admissible ceiling is a delicate decision, as a simple example may reveal. Suppose it is established that no individual may hold more than 5% of the media sector as a whole, and assume that this includes one television station which accounts for 9% of the overall media audience and one newspaper which accounts for 7%. The implication is that the rule makes it equally possible for the individual to hold 55% of the TV channel (which implies full control), or, for example, 22% of the TV channel plus 43% of the newspaper (in both situations he/she holds under 5% of the total media market). Yet if the ceiling is raised, say from 5% to 8.2% of the media market, then it is possible for the operator to control both the newspaper and the TV station (8.16% of the market accommodates 51% of both). If, on the other hand, the ceiling is reduced even slightly to 4%, the operator will not have even the formal control of the TV channel (and indeed nobody will: with a 4% ceiling, any one party may control at the most 44% of the television channel) (see Table 4.2). It is obvious then that setting too 'high' a level may defeat the purpose of having diversified media ownership.<br />Table 4.2<br />Thresholds of Ownership (%)<br />1. Share of total media market taken by each mediumTV Station9Newspaper7Other84Total100Scenario A2. Share of medium3. Share of market (=2x1)554.954.95Scenario B4. Share of medium5. Share of market (=4x1)221.98433.014.99Scenario C6. Share of medium7. Share of market (=6x1)514.59513.578.16Scenario D8. Share of medium9. Share of market (=8x1)443.963.96<br />Source: London Economics, April 1995 <br />The key point is that restriction rules based on universal thresholds may have different results depending on the underlying distribution of the shares of the various component parts of the market - TV channels, radio stations, press titles, etc. The danger is that a pan-European rule established at the Commission level would have a significantly different impact in different national contexts. The fixing of a limit on the maximum share of the market which may be held by one operator is a question of media policy and, above all, a political decision. Given the great heterogeneity of the national media ownership regulatory frameworks though, it is doubtful whether there will be a consensus on this sensitive issue. In addition, there is the issue of how media ownership rules may be actually enforced. Most rules setting a ceiling on ownership can be de facto bypassed, for example through interlocking shareholdings or the formation of stable syndicates and interest groups. Thus where an owner is forced to divest from part of his/her holdings, there are ways in which his/her original editorial control might still be maintained, for example through transfer of shares to sympathetic partners, or business allies, or even to family members (i.e. Berlusconi in Italy).<br />The crucial question however should not only be who controls a media company but also how much influence does he wield over the diversity of opinions. To answer this question, the European Commission (through both its 1992 Green Paper on Pluralism and Media Concentration and the 1994 Follow-Up to the Green Paper) and various economic consultants stated that one should have a clear picture of the diversity of media ownership by measuring the audience reached by media controllers. They rejected revenue-based criteria on the grounds that they are only suitable for the measurement of media economic market power and not for assessing plurality. Audience-based measures (daily reach, time spent reading, viewing or listening and so forth), they suggested, are preferable for evaluating plurality because they reveal directly the impact of media controllers on the citizens. What should become clear however, is that neither criteria reflecting media consumption nor financial data will tell us how much influence is wielded over the diversity of opinions. What would measure influence is a very difficult question to answer and no satisfactory criteria have emerged so far. In the absence of a direct measure of influence, crude numerical (counting) measures which attempt to proxy for influence can be used as a substitute. We first examine market shares measured in terms of revenue to the media enterprise. We then move on to assess the potential use of audience-based measures.<br />4.4.3 The Criterion of Revenue<br />The most obvious criterion for the calculation of the individual controllers' shares, which is the analogue to the standard calculation of market shares in any industry, is the reference to total revenues. The general rationale for using the revenue criterion is that if a company produces a large proportion of an industry's total output or turnover, its actions will have a greater impact on that industry's supply, and therefore on the level of prices. In this classic sense, a high revenue share provides an indicator of the firm's market power. Is revenue a good unit for the assessment of the level of concentration in the market place for ideas?<br />To answer this question, it might be helpful to distinguish between two rather different categories of revenue for the media companies, that is advertising revenue and consumer expenditures on the media. One can link these distinctive markets with the two basic arenas of policy concern: the concern over economic equity has focused on the advertising market, while the concern over the market-place for ideas and diversity places grater emphasis on the market for audiences and the programming that attracts them. Advertising revenue reflects advertisers or advertising agencies' willingness to pay in order to reach a particular audience. Advertising then is not considered as an expenditure by families to obtain information; instead, it is considered as an expenditure by companies to acquire audiences. Arthur Andersen's 1994 study pointed out that, while in assessing media firms' economic power advertising revenue may indicate their strength in a market, in measuring media influence it seems to be a poor guide. The reason is that, in the latter situation, the crucial issue is the audience's use of media products. Hence, consumer expenditures on the mass media, whether that is subscriptions to TV channels or revenue from newspaper sales, seem at first glance to be more suitable for measuring concentration in the political and cultural market.<br />By focusing on consumer expenditures on the media however, we will not be able to determine plurality in a market. The money that the public spends on various media does not reveal either how the public uses the media products or how people's beliefs are influenced by the media they choose. In any case, both advertising revenue and direct consumer expenditures on the media have a major defect as a way of measuring concentration of influence. Some media such as pay-per-view channels or most of public service broadcasters across Europe obtain no advertising so that, by the advertising revenue standard, they would gain no share in the market-place. Likewise, other media such as the free press and commercial broadcasters depend entirely on advertising revenue and consequently, by the consumer expenditure criterion, they would gain no share in the market-place.<br />Another indicator of a media operator's influence might be his/her share of the total expenditures on creating programmes or publishing. Many media as well as disseminating information produce much of the material they distribute. Other media spend huge amounts in acquiring programmes and bidding for rights and talents. It is logical thus to raise the question of whether expenditures could be a more suitable criterion for assessing a media owner's influence on the public. The obvious drawback when using this standard is that it yields rather narrow interpretation of market power, as given by the channel's share of the industry's spending. There are two additional disadvantages. Firstly, there is the problem of data availability since few companies publish expenditure figures. Secondly, the expenditure standard tends to emphasise the weight of public service broadcasting organisations where expenditures represent a larger proportion of revenues. Since the vast majority of public channels are not profit-maximisation oriented and cater for creating innovative programmes, the expenditure criterion would grant them untrue influence.<br />Another hypothesis is to calculate a media proprietor's influence on the basis of cross-subsidisation, that is, an operator's ability to finance his loss-making sector with revenues extracted from other businesses he may possess. Rupert Murdoch, for instance, decided in 1993 to cut the price of his three newspapers in the UK - The Times, the Daily Mail and the Sun - at a figure that his rivals could not match, even, in the case of The Times, at a loss (he recently put them up again). Of course, he was able to finance this loss-making paper through his other businesses (BSkyB and other assets abroad). The emerging newspaper price-cutting war is most likely to lead to huge losses for newspaper companies and even closure of some titles. The plurality of British newspapers is apparently at stake and readers face less choice. Another example is the use of the substantial profits of Robert Maxwell's Mirror Group Newspapers (before 49 per cent of the company was floated in the early 1990s) to invest in other private side firms (see Financial Times, 17 October 1995, p9, 20). Once again however, the selection of the cross-subsidisation criterion bears rather narrow interpretation of market power.<br />Measures based on revenue have been rejected by the Commission through the 1994 Follow-up to the Green Paper on Pluralism and Media Concentration in the Internal Market (COM(94) 353 final) on the grounds that they can only apply to the measurement of economic market power and not to assess influence. As stated in the Follow-up to the Green Paper, the disadvantage of the market share standard is that 'it duplicates the efforts of competition law, which uses this criterion to assess the behaviour of economic operators among themselves (taking account, in particular, of earnings) and not to evaluate as such the impact of media control on the 'citizen consumer\"
(COM(94) 353: 42). Indeed, as Robinson et al. (1995: 40) argues, if the European Commission were to legislate in this way, it would be prejudging work that is elsewhere left to the empirical investigations of the Merger Task Force. There would therefore be a potential conflict between one part of the law and another. Revenue-based measures have also been dismissed as inappropriate in the Andersen's study which also pointed out that revenue can only be used to evaluate a company's economic power.<br />Nevertheless, it may not be right to reject revenue as a measure of pluralism so easily. A significant revenue-based market share is related to economic strength and normally (although not necessarily) reflects plurality in the market indirectly. Revenue may serve as a useful element to be considered because it captures power in the market place to acquire rights, for example, in a way that other criteria (such as audience share) fail to do. Of course, mapping the breakdown of media revenues could not be considered as the only criterion for measuring pluralism, not least because of the complexity of definition of pluralism. However, it seems useful to consider the possibility of collecting revenue data for regulatory purposes. The advantage, as Lange (1995: 8) reminds us, is that this kind of system could be based on hard data (i.e. company turnover) already available and published in the general framework of general company laws. In addition, revenue-based measures have the advantage of providing an appropriate unit for measuring concentration across media. In this way, they are suitable for aggregation and, combined with direct measures of audience (as it is argued below), they could capture in principle an operator's ability to influence public opinion. <br />4.4.4 The Criterion of the Media Consumer<br />As stated in the 1994 Follow-Up to the Green Paper on Pluralism and Concentration in the Internal Market, in addition to the question of who actually controls a media company, one has to find out the actual number of consumers reached by a media company or 'controller': 'when it comes to the question of pluralism in the mass media, it is necessary to determine the effects of concentrations on the \"
citizen consumer\"
. We summarise here the arguments listed in the 1994 Follow-Up to the Green Paper to support the use of the audience standard: 'compared with the criteria based on company share holdings, the media consumer criterion does not duplicate the efforts of competition law, which uses the above standard to assess the behaviour of proprietors among themselves. Moreover, compared with the criterion of the number of channels that can be controlled by an individual, the consumer unit is more suited to the expected development of digital compression technology which will lead to a proliferation of channels. Whether a media company controls two or a hundred outlets will then no longer be decisive in itself in determining the granting of licences. The important point will be how many consumers are already being reached (within a relevant market) by these different outlets (TV, radio or the press). In this respect, this criterion has also the advantage of covering all branches of the media, whether that be newspapers, television, magazines, radio etc., thereby being a good indicator of evaluating multimedia concentrations. Furthermore, it could be economically more sensitive and less harsh in the sense that it could facilitate the creation of a large number of specialised channels which normally do not attract large audiences. The criterion of the number of channels controlled by a single person could prevent this and therefore limit the level of pluralism in a particular area. Likewise, the audience standard may be more effective as regards preserving pluralism since a single channel with a large audience could also frustrate pluralism' (COM(94) 353: 42).<br />Despite these arguments, the criterion of audience share has met a lot of scepticism by both the media industry and some academics. In their response to the Commission's complementary questionnaire asking specifically about the possible use of the audience criterion, the majority of media firms and media representatives expressed reservations about its use. Those reservations regarded both the adequacy and the feasibility of such a criterion. The European Broadcasting Union (EBU), for example, stated that it would not be possible to apply the same audience thresholds in different contexts (i.e. different kinds of media and different distribution areas). Indeed, in smaller countries and restricted language areas there may be room for one or two generalist channels, whereas in large markets an audience share of, say, 25% would already appear very high. In addition, it mentioned that pluralism must be safeguarded within each relevant media sector. Even within one and the same media sector, for example television, further distinctions would have to be made between pay-TV and television free at the point of reception, between generalist and thematic channels, and between different kinds of thematic channels (news, sports, culture, etc.). In view of such differences, EBU argued that, in case 'audience' was selected, thresholds should distinguish between media and be fixed at the national/regional level only in accordance with the principle of subsidiarity. This is linked to the question of the proper definition of the market, to which we have previously referred.<br />Many media organisations argued that the use of the audience criterion would punish growth and penalise the success of a proprietor. The EPC (European Publishers Council) and ENPA (Association Europeenne des editeurs de journaux), for instance, pointed out that if audiences grow in response to a company's commercial success, then the effect would be to restrict that success, something which seems pointless when the aim of any new regulation should be to stimulate the market and give media firms more opportunities in the Single market. Others challenged the Commission's argument that the use of the audience criterion will encourage the establishment of thematic channels. EGTA (European Group of Television Advertising) and ITCA (Independent Television Companies Association) stressed that there is a strong possibility that the opposite effect will result, as a preliminary question must be asked once again about the definition of the market. What is the position of, say, a specialist channel taking 50% of the viewers in its niche market but an insignificant overall share? Even those in favour of a possible use of the audience standard (i.e. Independent Television Commission) questioned its robustness and its capability of applying equally in different circumstances and between different media, or argued that it must be accompanied by a 'harder' unit such as revenue.<br />There has also been opposition to the application of audience shares for assessing media concentrations from the academic side. In an article in The Independent, Barnett (1995a: 15) expressed strong disapproval of the 'share of voice' argument and criticised any kind of proposals that attempt to shift the emphasis from ownership. Ownership for him still matters, for different owners have different versions of the world and different attitudes to news. It is then wrong to replace existing rules based on the number of outlets an operator can possess by rules based on newspaper circulation, television viewing and radio listening. In a later paper in Public Policy Review, Barnett (1995b: 1-5) proposed that regulators should allow owners to possess media outlets in just one sector and bar completely cross-ownership. In his words, 'we simply need to maintain a clear distinction between them [media sectors], and to insist that those who wish to own newspapers should continue to be debarred from owning television channels and vice versa' (ibid: 5). Similar views had been expressed earlier on by Curran and Seaton (1985: 342-3). Lange (1995: 7-8) criticised the pertinence of the audience standard from both a technical point of view and from a conceptual point of view. The technical problems he saw are firstly that there are important margins of errors of audience and circulation measurements, and secondly that these measurements are meaningless unless they are done on different geographical scales. From a conceptual point of view, he challenged the assumption that audience measures are more appropriate for evaluating pluralism. In fact, the 'audience' proposal analyses the question of pluralism in terms of consumption of information and not in terms of offer (i.e. in terms of possibilities to access a great diversity of information). Audience measurement is a quantitative analysis of consumption by readers, listeners or viewers, developed for market analysis purposes. Pluralism means basically 'diversity of information'. However, this is a qualitative objective both in terms of offer of information and in terms of reception of that information. Audience research, the argument follows, does not measure either the diversity of offer or the quality of reception.<br />The audience criterion thus has raised a number of serious questions. This has prompted the Commission to launch a feasibility study examining the comparability and compatibility of the different national systems of audience measurement, the types of audience measurement that can be used and the possibility of using a single composite criterion or of combining several types of measurement depending on the configuration of the media in question. The study distinguished between monomedia audience maps (where only one medium is under review) and multimedia ones (where TV, radio and newspapers are combined). In constructing monomedia audience maps, it recommended the use of audience share for television, circulation share for newspapers and listenership share for radio measurement purposes. In the case of multimedia maps, it recommended using the daily average contacts as a criterion that satisfies the requirement that the measures are equivalent across media. The analysis concluded that, although there were disparities among national audience measurement systems, the use of audience measures to examine pluralism across the EU is both valid and practically feasible. It takes the consumer as the point of reference and will therefore be of maximum effectiveness in relation to the objective sought, namely that of serving the interests of the media consumer. One has to remember though that the study was commissioned by DG XV which would very much want a definite statement on the application of audience measures as indices of pluralism. It comes as no surprise that the GAH Group's findings are in line with the DG XV's wishes, as many times although not necessarily the employee wishes to satisfy the employer.<br />In fact, there has been much criticism of the GAH Group's work. Leighton Andrews (1995: 26), head of Worldwide Corporate Affairs BBC, stated that the study 'simply demonstrates that comparable audience data is available across the EU, it does nothing to make the argument as to why audience share is the most suitable measure of dominance'. As mentioned above, the EC's consultants came to the conclusion that despite the differences in national audience measurement systems the measures are comparable and compatible and can be the basis of uniform thresholds to be applied throughout Europe. However, it left a number of questions unresolved. In particular, the question of weighting the audience of different types of media such as TV, radio and the press and the question of where to fix the thresholds. Of course the latter is a matter of media policy but, due to the great heterogeneity of the current European media map, it will be very difficult (if not impossible) to find a political consensus on such a threshold at a European level.<br />The first problem - the establishment of market shares based on the number of people reached by each medium - is equally difficult to solve. How is it possible to create an overall market share of, say, a company holding 20 per cent of the TV market, 10 per cent of the national newspaper market and 5 per cent of the radio sector? As Robinson et. al. (1995: 59) argue, 'we need to find some robust and objective method of constructing a weighted average of these three market shares to arrive at a single figure for the media market as a whole'. The absence of a weighting will lead to false results as the GAH Group's analysis reveals. The EC's consultants recommended the use of daily reach as an indicator of influence. As demonstrated by the National Economic Research Associates the use of daily reach in the UK media market gives a high weighting to television and radio and a low weighting to the press (see Table 4.3). According to NERA, this is because most people have ready access to all TV and radio stations every day, and can switch between them in deciding what to watch or listen to. Nevertheless, they generally choose one or two newspapers a day and cannot pick and choose content from one or another. This results in a much smaller number of daily contacts for the press, which is not necessarily a measure of relative influence.<br />Table 4.3<br />Use of Reach Weighting in the UK Media Market (1994)<br />SectorWeightingTotal Daily Reach ContactsTelevision55.3%133,815,268Radio18.9%45,714,000National Press15.6%37,578,186Regional Press10.2%24,715,286Total100%241,822,740<br />Source: NERA, based on data supplied by BARB (television), RAJAR (radio) and Pressad/JICREG, Newspaper Society, ABC, NRS (newspapers)<br />Note: TV reach is based on people over the age of 4; other reach figures are based on adults over the age of 15<br />Undoubtedly it will be difficult to develop a suitable and generally accepted method of ascertaining these figures. One such method, taking into account the different sizes of, and power wielded by, different media, has been proposed by the British Media Industry Group, a consortium of four big newspaper groups (Associated Newspapers, the Guardian Media Group, Pearson and the Telegraph Group) set up in August 1993. Their study uses the well-established audience figures to calculate market shares for television and radio, and circulation figures to assess market shares for newspapers. In calculating overall share of voice in the UK media market, they recommend that radio listening be down-weighted by 50% compared with newspaper circulation and TV viewing. BMIG justify these weightings on the grounds that radio has lower impact on diversity of view issues, either due to the prevalence of music-based radio formats, or because radio is normally consumed in parallel with other activities. <br />According to the study thus 1% of newspaper circulation or TV viewing is deemed to have more impact on diversity of view issues than 1% of radio listening. So their suggested weighting is 28.6% for television, 28.6% for national press, 28.6% for regional press and 14.3% for radio. Their findings, shown in Table 4.4, give BBC the highest audience share (19.7%). The next top two places are occupied by News International on 10.6% and ITV network combined on 9.4%. The top 8 media firms have 57.4% of national share of voice. The main advantage of the BMIG approach is that it treats television, national newspapers and regional titles equally. The inclusion of regional papers, in particular, is a positive step towards realising that it is not only the national market that matters but local/regional monopolies should also be tackled. Further, the selection of the most appropriate units to measure market shares for each medium (circulation for national and regional papers, share of viewing for terrestrial TV, cable and satellite, and share of listening for regional and local radio) provides a more or less accurate picture of market shares within each media sector.<br />Table 4.4<br />Media Concentration in the UK Under Audience Share <br />Analysis (1993/94)<br />Media Groups% share of audienceBBC19.7News International plc10.6(ITV Network)*9.4Daily Mail & General Trust plc 7.8Mirror Group Newspapers Ltd7.6United Newspapers plc5.7Carlton Communications plc3.1Channel 42.9Total as a % of papers/TV/radio 57.4<br />Source: BMIG (February 1995), Robinson et al. [London Economics (LE)], 1995<br />Note: Radio down-weighted by 50%<br />* LE estimate using BMIG methodology<br />This last point however becomes a disadvantage when it comes to measuring aggregate shares. Robinson et. al. (1995: 59) argue that it is inconsistent to measure shares in the printed media market on the basis of circulation (which is essentially a household based measure) while using audience shares (which count individual viewers) for the electronic media. As the EC's consultants showed, it is more consistent to use readership figures when it comes to comparing different media. Another problem with the study is that it does not give any weight to consumer periodicals. A further criticism to the share of voice approach is that it, including the BBC and Channel 4, underestimates the way the rest of Britain's media ownership is concentrated in so few private hands (Peak and Fisher, 1995: 227). The most serious problem with the BMIG approach though is their unreasonable decision to down-weight radio listening by 50% compared with the other media. It is true that radio lost much of its appeal after the appearance of television. It is also true that radio is mostly an entertainment medium and there has been a proliferation of music-based radio formats. It is finally true that radio can be listened to while people are engaged in other activities. All these points make radio seem as a 'secondary medium'. But they do not prove that radio is less influential and of less importance to diversity of news. The influence of news-based radio formats, broadcasting news, current affairs and other programmes devoted to commentary and opinions is considerable. Even music-dominated radio stations have regular news bulletins normally announced every hour. No matter whether one accepts that the influence of radio is less than that of television or the written press, BMIG's proposal to down-weight its influence to 50% is both arbitrary and gives no objective justification.<br />There has been no objective basis so far for the choice of weights. It has proved difficult to identify satisfactory criteria for a measure which successfully aggregates media with very different patterns of consumption and use. This difficulty is inherent in another study, the Arthur Andersen's analysis, which uses a single unit of measurement for the cultural market. In order to measure company shares in that market, Shew used the time that the public actually spends on media. Here, the potential of the mass media to influence public opinion is assessed best by examining how people distribute their time over them. Table 4.5 shows the results of applying this methodology. What is interesting is that the amount of time spent watching BBC and ITV is nearly twenty five times as great as that spent reading News International's newspapers. In fact, there is an obvious exaggeration of the impact of TV compared to the press, since no other press groups appear in the top seven.<br />Table 4.5<br />Media Concentration in the UK Under Time-use Analysis (1993)<br />Media Groups% of time spent 'consuming'BBC44.1(ITV Network)*25.4Carlton Communications plc6.9Channel 46.2Granada Group plc4.1Capital Radio plc3.4News International plc3.4Total as % of the press TV & radio93.5<br />Source: Shew (Arthur Andersen) 1994, Robinson et al. (LE) 1995<br />* LE estimate using Arthur Andersen methodology<br />In the Andersen's study each controller's individual share of the total media market is calculated as the share of his/her medium in the total time spent by the public consuming media. This definition may be both too vague and too limited. It is vague because it does not tackle the nature of the media consumed. For instance, how much of the time watching TV is spent viewing news and how much sports? Moreover, one could not treat equally time spent reading The Guardian and time spend reading The Racing Post (both national newspapers). A greater value should be added to the former title due to in-depth coverage of current affairs. Regarding the different media, it may also be false to use the time standard for assessing their impact on the public. One hour TV watching (which is normally passive) may not be comparable to one hour newspaper reading (which requires greater effort). It seems safe to argue that the influence of the latter is greater. Equally, it is hard to compare the influence of a 15-minute news bulletin with that of a newspaper headline on the same topic which is on display the whole day and can be read by thousands or even millions of passers-by. In the latter case, the item is most probably being watched for a fleeting moment, but its influence lies in the fact that it is being watched by many people. The analysis is also too limited. Although it includes reading periodicals, it excludes viewing of pre-recorded videos, which are a direct substitute to television viewing.<br />These fundamental problems aside, Shew argues that where there is cross-ownership of different media by the same controller, his/her share is arrived at simply by summing each of his/her channels' or titles' shares in the total time spent by the public consuming media. Therefore, if owner A controls both channel X and paper Y, which take up 20 per cent and 10 per cent respectively of the public's total time, then A's share in the hypothetical total cultural market is said to be 30 per cent. Similarly, in case he/she controls only a fraction of a channel or title, his/her share is calculated as that same fraction of the time devoted by the public to that particular channel or title, relative to the total time spent consuming media: if controller B has 25 per cent share in channel Z, which is found to account for 40 per cent of the public's total 'consuming media' time, then B is said to control 10 per cent of the total market for ideas.<br />This methodology is troublesome and biased towards both television relative to the press and national channels and titles relative to regional/local ones. It artificially inflates the market share of television and particularly of national channels, while it understates the relative importance of the press and of all regional/local channels and titles. This is because the average weekly hours spent watching TV are nearly seven times as great as those reading the press (according to the study, it is 24.1 hours for TV against 3.5 for papers), so the apparent market 'influence' of a TV controller will be vastly more significant than that of a newspaper publisher if measured in this way. If, in addition, the issue of the effective geographical coverage of a channel or title is ignored, it is hardly surprising that the total time spent with a regional/local channel by all viewers in a particular area will be small in relation to the aggregated time spent by a national audience watching a national channel. To illustrate, watching Carlton television is not an option for those who live outside the London area, and it is therefore meaningless to calculate Carlton's national market share as the time spent by anyone watching Carlton over the total time spent by the public consuming all media nation-wide. It is obvious that Carlton's share will be dwarfed relative to media with national coverage such as the BBC.<br />A partial refinement of the Arthur Andersen approach has been provided by London Economics. That study attempted to weight the respective UK media shares by some measure of the inherent impact of the medium. The analysis proxies this by the amount that advertisers are prepared to pay to reach a consumer audience through a particular medium. It formed the following hypothesis: if advertisers pay according to the relative impact they believe a particular medium has, then the relative price per consumption minute across media provides a market-based approach to cross-media weighting. This is shown in Table 4.6, where the analysis estimated the relative cost of advertising across media in Britain using relative advertising cost per viewer minutes spent on the media. The advertising-based weightings suggest that the printed media are significantly more important, and radio far less significant, than in the Arthur Andersen paper.<br />Table 4.6<br />Relative Advertising Costs in the UK Media Market (1994)<br />Medium£ per hour* per capitaTelevision1.16Radio0.22Newspapers2.52Cinema2.66Magazines2.82<br />Source: Advertising Association handbook, 1994; Arthur Andersen, 1994; London Economics analysis, 1995 <br />* of viewing, listening to or reading advertisements. London Economics analysis makes the assumption that people spend the same time on advertisements for each hour of reading newspapers and magazines, as for each hour of viewing or listening to commercial television or radio<br />The London Economics study re-calculated the Arthur Andersen estimates of audience share using a methodology in which the time devoted to each media sector is given a weight which reflects its value to advertisers (as shown in the above table). The effect, as presented in Table 4.7, is to enhance the market share attributed to the press at the expense of television and radio. In particular, the BBC's share falls from 44.1% to 28.3%, Capital Radio's share falls from 3.4% to 0.3%, whereas News International's share rises from 3.4% to 12.5%. These results demonstrate how sensitive the final outcome is to input assumptions. This methodology however, as London Economics itself admits, remains subject to the basic objections made above to Arthur Andersen approach, namely the problems with product and market definitions. For example, the figure of Capital Radio would rise if the study took the London metropolitan area as the relevant market - which would be appropriate for Capital Radio, as metropolitan London is the limit of its broadcast range.<br />Table 4.7<br />Alternative Measures of UK Media Concentration<br />Media GroupSimple time use*Weighted time use**BBC44.1%28.3%Carlton Communications plc6.9%9.8%Channel 46.2%5.5%Granada Group plc4.1%6.9%Capital Radio plc3.4%0.3%News International plc3.4%12.5%<br />Source: London Economics, 1995<br />* Arthur Andersen analysis<br />** Using time data from Arthur Andersen paper modified to reflect the different value of time to advertisers in different media<br />4.4.5 Critique of all Audience-based Measures <br />Mono-media market measures<br />Overall, there are a number of aspects that should be considered in methodologies which set out to examine mono-media or multi-media cultural power in terms of audience. The relative influence of a communication vehicle depends on a number of different factors. It obviously depends on the number of people it can reach - a newspaper sold to one million people will tend to have more influence than one which reaches only fifty thousand people. However, the fact that a newspaper is bought (something that can be measured) does not necessarily mean that it is read, or read by only one person (readership is much harder to measure than circulation). Furthermore, a title that is read carefully will have more influence than one which is merely skimmed, and whereas we can use surveys to find out how many people read each newspaper, it is impossible to know how carefully they read them, or what they read in them. Serious news, current affairs and political comment compete for the readers' attention with sport and the TV listings. Moreover, it is difficult to establish whether publications with more editorial content count for more influence. Some newspapers (particularly the Sunday titles) are fat because they offer readers a choice of specialist material which is not opinion-forming. Finally, there is the problem with the weekly and Sunday press. Should the latter be given one sixth the weight of the daily press, in proportion to the average daily reach of each? There are two problems with this assumption: first, weekly papers tend to have a longer read period; second, multiplying the circulation figures for daily newspapers by six to produce a weekly circulation figure and putting them together in terms of a weekly market (as the BMIG did), presupposes that daily and Sunday papers are entirely substitutable for readers. This however is a strong assumption to make. The Sunday Times, for example, dominates the UK Sunday quality newspaper market but its sister paper The Times has a much lower market share of the quality newspaper daily market (see ABC figures, June 1996).<br /> <br />Similarly for the TV programme to have any influence, the set must be switched on and tuned to the right channel, which can be measured electronically. In some countries, like the Netherlands, Germany and Britain, there are also surveys which establish who is watching it and for how long. But the degree of attention with which it is being watched is almost impossible to establish. Another dimension to the problem concerns who is in the audience. A programme or newspaper article that reaches a single person may have a lot more influence than one which reaches a million if the single person happens to be the President of the State. Programmes and articles which regularly reach opinion formers and policy makers such as members of the parliament, editors and journalists clearly have influence which cannot be measured just by counting heads. Thus what counts is not simply how many people watch (and read and listen) as the BMIG assumed, but who watches and what they watch. <br />'Reach' in general is less attractive as a measure of the influence of broadcast media. People who contact a broadcaster may rapidly switch away from it, in contrast to the press. People typically contact several TV channels or radio stations each day to find out what is on. That results in a measure that shows little difference from channel to channel. In addition, people spend more time with some channels than others in proportions that are not reflected in all national daily reach statistics. On the other hand, people have less opportunity to flick between press publications than between broadcast channels. Therefore, the total number of daily contacts for television and radio is much greater than the total number of daily contacts with the press. This does not reflect relative influence. What does reflect is the much shorter decision period the audience has in deciding where to spend its time in the different media.<br />Time spent consuming media is intuitively attractive as a measure of influence. The distinct advantage of this approach is that time use is a common denominator across all media. Thus if there is any validity in the basic proposition that the influence of a medium of communication is proportional to the time taken to absorb the message, the time use measure can be used to assess concentration across all media. Time can be used as a measure of market share in each medium, and by simply summing the time use across all media we can calculate an overall market share. Audience time may be suitable for the measurement of the broadcast media influence because programmes are viewed at the same speed by everyone, so that time may be a good indicator of content taken in. The problem is that the time that people spend to watch television or listen to the radio is certainly not the most suitable indicator to establish impact over them. Time spent watching or listening to the media does not necessarily reflects influence. Apart from this problem, which is common in all audience-based criteria, there is the issue that television and radio, in particular, can be a backdrop to household activity. High audience hours may arise as a consequence of the broadcast being used as background to other activities.<br />The methodological weaknesses of audience time measures are more apparent in the press for a number of reasons. Firstly, reading speed is variable, so that reading time becomes less attractive as a measure of the influence of the print media. Secondly, the newspaper reader can focus on what is of interest and value for him/her and ignore the rest. According to how much 'filling' the reader chooses to edit out, similar levels of 'influence' might be achieved in very different lengths of time spent with the publication. Thirdly, and according to Arthur Andersen study described above, the amount of time spent reading national newspapers is nearly seven times less than time spent watching TV. This figure may substantially exaggerate the impact of television compared to the printed medium. However, national newspapers have an impact that lasts longer than the time spent reading them. Newspaper headlines are seen by, and influence, millions of people who may not buy them. But it is the newspaper headlines that largely determine the current affairs agenda at the start of each day, and that influence persists even though news sources of information are obtained from television in the evening. There is indeed extensive discussion of press opinions in other media, perhaps because the press is permitted to give opinion whereas broadcasters (particularly public service broadcasters) are constrained to treat news 'impartially'.<br />Having explored, to some extent, the pros and cons of the audience-based criteria, our impression is that time measures are preferable for broadcast media than reach measures. The nature of TV and radio means that reach or contact with a station can happen almost incidentally as people switch between channels searching for one that captures their attention. Data for television viewing is readily available and reliable, whereas radio listening data are adequate for the majority of the stations in all European Union countries (see ESOMAR, 1995). Concerning the print media, daily reach might better reflect the influence of the press than time spent reading, as for the most part people buy newspapers with the intention of reading them. Also data on time spent reading is either non-existent or unreliable. The choice between circulation or readership as the reach measure is difficult. Arguably, the influence of newspapers depends more on the number of readers than on the number of copies sold, just as the influence of TV programmes depends more on the number of viewers than on the number of sets tuned. Therefore newspaper readership would be a theoretically better measure of influence of the printed media. Readership data though is not available for Portugal, Greece and Northern Ireland and is not collected on a regular basis for many regional titles. This would create insoluble problems in case we want to establish a Europe-wide regulation based on this criterion. By contrast, circulation data is the most robust and comprehensive audience measure for the press. It is thus a more established currency in the industry. In practice though, and as the GAH (1994), BMIG (1995) and NERA (1995) studies found, percentage share of newspaper circulation is not very different from percentage share of readership.<br />The aggregation of media market measures<br />Once a satisfactory method of calculating audience shares within each media sector has been devised, some method is required to compare audience shares across media. A likely policy aim of deregulating media ownership rules at national or even European levels is to allow cross-media ownership in a manner which is proportionate. Put it another way, the permitted holdings in one sector of the media would be subject to the size of holdings in other sectors of the media. However, there is no obvious solution to the problem of creating an overall market share. If a particular company has 15 per cent of the television market, 20 per cent of the press market and 5 per cent in radio, what is its share of the total media market? We have to devise a method of calculating an average market share of the whole media. If shares in each sector are measured in the same dimension, for example a consistent reach statistic (as recommended by the GAH Group), then there is a natural method of aggregation by adding all the shares. If, however, market shares within the media are measured in ways which are not directly comparable (and this would be wiser since, as we explained above, some measures are more suitable for the press and others more preferable for broadcast media), then a weighted average of all market shares must be constructed in order to arrive at a single figure for the media market as a whole.<br />A weighting can be based on some objective data which, one might assume, compares the relative influence shares of the media on an aggregate basis. A weighting could also be constructed on policy grounds, reflecting a policy view of the relative influence of the media, which may or may not be based on some objective information. As shown in great detail above, the BMIG falls down on the arbitrary nature of its weighting system. Their approach uses audience figures to calculate market shares for radio and television and circulation figures to calculate the market share for newspapers. They then construct an average share of the overall media market by assigning weights of one to television, national and regional newspapers, and 0.5 to radio. However, there is no objective justification for the arbitrary figure of 0.5, especially as it is far from obvious that radio has a low level of news and current affairs compared to other media. The London-based Capital Radio FM and the nation-wide Virgin Radio, for example, both music-based radio formats, have regular news bulletins as well as other programmes devoted to commentary and opinions. This is a pattern followed by most radio stations. The other problem with the approach is that it is not consistent to measure shares in the printed media market on the basis of circulation (which is basically a household-based measure) while using audience figures (which count individual viewers) for TV and radio. <br />As the European Commission's consultants stated, it would have been more consistent to use newspaper readership, rather than circulation, as the basis of comparison. So they recommended the use of daily average contacts as a measure of the overall market share. This method however results in a much smaller number of daily contacts for the press (see Table 4.3 above), which is not necessarily a measure of relative influence. A weighting could also be based upon the hours of use or audience time of the media (Arthur Andersen's approach). The idea of using hours of use as the basis of a weighting appears intuitively appealing until one sees what figures it implies (see Table 4.5 above). People spend very much more time with television and radio than with the press. Most people would not find it credible that the press has such small influence in relation to other media. In addition, the data for the average time spent reading the press is well known to be unreliable. But the major problem associated with this approach is that it seems intuitively unlikely that time spent consuming media directly proxies the influence of a particular medium. The influence deriving from the intrinsic impact of the medium and the influence due to the number of minutes for which the medium is consumed are being conflated.<br />It seems to us that a much more appropriate way of measuring overall market shares and possibly influence across media would be to weight the respective audience shares by advertising expenditure. By examining the behaviour of advertisers, that is, by examining how much advertisers are prepared to spend on which medium, we will have a more thorough picture of the relative influence of the media. Advertisers pay according to the relative impact they believe a particular medium has. They buy space in national and regional newspapers and time on television because of beliefs regarding the influence of those media. They also recognise that it is not just the number of viewers, listeners and readers that matters; it is who they are. Therefore, advertising expenditure on the media provides a reasonable market-based approach to cross-media weightings. This can be shown in practice. London Economics analysis, for instance, attempted to weight the respective time spent media shares (Arthur Andersen's approach) by the amount that advertisers pay to reach consumers through a particular medium. The chief merit of the London Economics approach, the results of which are shown in Table 4.7 above, is that it finds that the printed media are significantly more important than in the Arthur Andersen analysis. The time devoted to each media sector is given a weight which reflects its value to advertisers and the effect is the enhancement of market shares attributed to the press at the expense of the electronic media.<br />Data robustness<br />A necessary condition for using audience measures to assess concentration levels and pluralism within local, national or supra-national markets also depends on the quality and comprehensivity of the surveys in use. In an ideal world, surveys would measure every single television and radio channel, and every single newspaper title uniformly and with equal precision and accuracy across the European Union. The existence of regular surveys is hardly a problem as all EU Member States (with the exception of Luxembourg) possess their own national surveys for TV, press and radio. There is also a certain level of continuity of data collection in the EU, as all current TV surveys collect audience data on all days of the year, and the majority of press and radio surveys collect data across at least ten months of the year (see ESOMAR, 1994, 1995). But it is also essential that audience surveys are comprehensive in their coverage of the media. A specific limitation of television audience measurement is the restriction of almost all national surveys to measuring viewing in the home, although some (such as Italy and Sweden) measure guest viewing and treat it as a substitute for out-of-home television viewing in other TV homes. Regarding press readership surveys, their main drawback is their limited coverage of titles. In the case of newspapers, readership surveys invariably cover all the national distributed titles, but there are some big gaps in the coverage of regional titles in France (see ESOMAR, 1995). By contrast, circulation data are quoted for almost every newspaper and virtually all are audited. Regarding radio surveys, the one instance where a radio survey appeared to give well below full coverage of stations was the Dutch Intomart survey, which excluded the non-commercial local stations (ibid.).<br />There is also the issue with the universality of coverage. Almost all national surveys measure media consumption across national universes. However, the Greek AGB Hellas (television) covers urban areas only, whereas the Portuguese Bareme (press and radio) covers mainland Portugal. Geographical restrictions, as between urban versus rural areas, could have a significant effect on measures of total audience, where significant variations exist in media availability that coincide with the geographic restrictions. The problem really only arises for some media in Greece and Portugal. In addition, there is the issue of harmonisation of measures. It is important for supra-national framework legislation that audience measures are comparable across borders. The GAH Group report concluded that comparability is not a significant obstacle for the construction of audience maps. The European Commission's consultants found that national surveys within each media category all employ the same fundamental concepts of media consumption, even where the methodologies vary (as with 'Recent Reading' and 'First Read Yesterday' measures of reading). And although TV surveys are divided between those which treat viewing as presence in room with TV set switched on and those which require individuals to be present and watching, the end differences are almost certainly negligible. So cross-border comparisons are broadly feasible.<br />In future though, with the expected increase in the number of media outlets, it is possible that the quality of the data will decline unless the industry spends more money on increasing the size of survey samples. Perhaps more crucially, as new opinion-formers operate through the Internet, it will become increasingly difficult to measure their 'audience'. While it is possible to calculate the number of 'hits' on an individual Web site or bulletin board, it would be quite unrealistic to know how long each hit lasted, or what percentage of them came from outside the country. As technology develops, cross-media measurement based on audience share will become more and more difficult to realise unless further research is conducted. There are some encouraging signs though. The Audit Bureau of Circulations initiated a plan in January 1996 to measure the readership figures for UK on-line titles and the results of its trials should come out later this year. The plan is to provide advertising agencies with a certificate showing how many people have registered for the site and how many have visited it in the last month. A bit of software, running on the same computer as the electronic publication, will make random checks and e-mail some of those who have registered to make sure they exist (see the Guardian, Monday February 5 1996, pp. 12-13). It remains to be seen whether the trials will succeed.<br />4.5CONCLUSION - ALTERNATIVE PROPOSALS<br />In this chapter, we tried to analyse critically different methods of measuring media concentration and influence. The relevance of the existing counting criteria (i.e. rules based on the number of channels controlled by a single operator) was briefly reviewed, followed by an analysis of the audience and revenue-based units. Although it became evident that numerical rules are not appropriate on their own to measure levels of concentration in the multimedia era, it might not be right to abandon them altogether. Many, including the European Commission, argue that today it does not matter how many outlets one owns, but how much revenue or audience he/she reaches. However, we will agree with Steven Barnett that ownership still matters. It seems to us that it is not acceptable, from a pluralism point of view, for one owner having the majority of a country's opinion-forming TV channels broadcasting news bulletins or current affairs, even when these channels reach small audiences and obtain negligible amount of commercial revenue. This is because, under these conditions, the proprietor will be free to exercise excessive power to advance his/her own political and socio-cultural views. The controller of a large number of titles and/or channels disseminating political commentary can easily pursue the values and political beliefs he/she espouses. He/she may also be in a position to influence the outcome of an election, as the Greek example described above clearly indicates. In addition, he/she may cater only to bourgeois tastes and not to others so as the latter cannot gain public exposure.<br />Such influence may not only be dictated by political and socio-cultural considerations but also by commercial or other considerations. Monopoly or excessive control of a number of channels may allow proprietors to promote a particular economic philosophy (foster, for instance, positive attitudes towards free enterprise and unbridled competition). Although the question of whether media controllers' influence over what people think is much disputed (precisely because influence is hard to measure), another example may help us to form a clearer opinion of the matter. In 1989 in Britain, the Broadcasting Research Unit (BRU) examined some opinion data collected by the then Independent Broadcasting Authority (now Independent Television Commission) on whether people thought the licence fee was the proper way of paying for the BBC. Murdoch's newspapers (and The Sun in particular) opposed the BBC and its licensing system. They did so because Murdoch was preparing to launch his satellite channel BSkyB at that time and wanted as few competitors as possible in order to provide a major boost to News International's UK television interests. The BRU found big differences in opinion between readers of News International newspapers and non-readers: 37% of News International's readers approved of the BBC's funding system while 56% disapproved; but amongst non-readers, the position was reversed [nearly half approved (49%), whereas 44% disapproved]. According to Barnett (1993: 17) one cannot be absolutely sure that the result was due to Murdoch papers' editorial line which developed a general feeling of opposition to the licence fee between their readers. Barnett however adds that such enormous differences can scarcely be coincidental.<br />Media concentration in political coverage and commentary (where pluralism matters greatly), but also in philosophy and opera criticism, science and business, entertainment, etc. is thus not tolerable. Performances in the media affect such varied aspects of national life as dress codes, political opinions, religious beliefs and so on. At least potentially, undue concentration in cross-media ownership might lead to an over-enthusiasm for particular parties, or to excesses of approval or disapproval of particular forms of dress, among other things. Television, radio and the press have a unique role in the free expression of ideas and opinion, and thus in the democratic process. The main objective must therefore be to secure a plurality of sources of information and opinion and a plurality of editorial control over them. Public policy has to ensure that viewers, listeners and readers benefit from the multi-media revolution and that a provision of a number of opinions from different sources are available. Ownership restrictions should thus be maintained. It is imperative though that they are supplemented with audience and/or revenue-based measures, since using the number of channels or titles as a sole criterion is meaningless without some consideration of coverage or financial strength.<br />The two different sets of methods illustrated above (audience and revenue -based) are said to correspond to two levels of measurement of concentration in the information market: the political/cultural or pluralism and the economic or concentration of resources. It has been put forth that revenue-based methods are close to the traditional systems of concentration measurement which permit assessment of the existence of a dominant position (concentration of resources), whereas audience-based methods are coherent with the cultural/political standpoint and can be held to be most effective for measurement of pluralism. However, due to the close relationship between economic power and pluralism, audience figures could also measure market power. In fact, audience-based measures are a form of market share measurement, which is a classic economic measurement. 'Audience' are the equivalent of measuring sales (i.e. market share), which is a classic economic measure of power. Therefore, the distinction between economic measures and cultural/political measures is irrelevant. Both sets of media market measurement assess market power. In the absence of a direct way of establishing 'impact', crude measures based on market power (i.e. criteria about market structure) are used instead. And what the audience and revenue-based methods are doing is in fact that - they evaluate market power.<br />The search for the 'best' criterion for measuring concentration levels in the media sector has been centred on structure and not on content. Neither financial data nor audience figures and certainly not the number of media outlets controlled by the same entity are fully reliable indicators of the controller's impact on public opinion. Such an impact depends to a large extent on the content of the channel/title, which is not taken into account by either of these methods. Ideally, pluralism would be best assessed if account was taken of the editorial characteristics of the medium in question (type of programme or column, editorial opinions, etc.) and studies were conducted to evaluate whether the citizen-consumer genuinely benefits from such diversity. However, it would not be possible to assess the content of different media and consequently their impact on pluralism/diversity. Studies that measure content quantitatively and qualitatively in order to assess whether content is balanced and whether it is pluralistic or biased are not readily available. Given the complexity of these analyses, it would be a very costly and time-consuming exercise (let alone subjective) to carry them out for the special use of the media authorities. Disaggregating media concentration by content area (i.e. classifying individual programmes by type or subject matter) is also impractical since most individual media vehicles present content covering a broad range of issues. Thus, in the absence of such studies, pragmatic solutions based on already available data have to be found and enforced.<br />Having made that clear that neither end-use based indicators nor revenue-based ones can directly measure media firms' impact on public opinion, the question is whether it is possible to isolate one of these measures and use it as a single unit that would best determine media companies' market power (and possibly reveal their impact on public opinion indirectly). The currently favoured audience share measure is not a subtle measure of cultural power. The GAH study prepared for the Commission is flawed because, although it offers a useful summary of the availability of audience measures across the European Union, it does not assess the suitability of audience share data as a measure of 'influence'. The problems with the approach of setting uniform Community-wide thresholds based on the as yet untested new concept of audience share are with the definitions of the relevant market and with market share. Regarding the relevant market, audience share cannot effectively compare media with entirely different intentions and market segments such as a racing magazine and a political commentary. In addition, audience share cannot extend to all substitutable products - time spent watching pre-recorded videos can be as important as time spent viewing scheduled television. Finally, audience share cannot compare media with differing geographical reaches - measured on a local level the market share of a local TV company can dwarf the local share of a national TV channel.<br />Regarding market share across media, it is very difficult to compare the relative significance of many shares in several different media or different media organisations with high shares of a smaller number. It might be the case that an individual has a high aggregate market share which is spread across many media. This however will cause no concern on behalf of the competition or media authorities, since he/she does not have formal control over any of the separate media. Secondly, media with relatively low audience shares could have higher impact than media with high audience shares. The London Financial Times, for instance, with a national daily circulation of just under 300,000, is regarded as more influential than, say, The Sun, with a national daily circulation of more than 4,000,000 (Audit Bureau of Circulations, January 1995). Thirdly, audience shares are constantly changing, and therefore no single limit could be applied with any degree of accuracy in practice (BBC, 1995).<br />Thus audience shares criteria are not a priori better indicators for measuring media concentration than revenue-based ones. The use of revenue market shares of whatever kind (advertising, subscriptions, sales, grants, licence fees, etc.) seems to be less problematic since it avoids the need to find practical solutions to the various problems arising from the use of audience data. Audience data are neither fully harmonised nor equally reliable across Europe (see ESOMAR'S 1994 and 1995 reports). On the other hand, common accounting principles already exist. Financial data for different media sectors are easily comparable. A system based on financial data would make it easier to achieve closer co-operation and synergy effects between the authorities responsible for competition and those in charge of the media. Instead of building up two completely separate systems, this approach would at least make it easier to collect data and to enforce transparency rules. It should also be mentioned that financial data have the advantage of reflecting media consumption - the higher the audience ratings, the greater the revenue from advertising; the larger the number of subscribers, the greater the revenue from pay-TV fees.<br />The main advantages of applying a revenue-based measure are firstly that it is a long established and tested method for measuring market concentration, and secondly that it provides a common currency of measurement across media. However, the NERA 1995 study pointed out a number of reasons why revenue may be an inappropriate measure: a) excludability. The capability of an enterprise to earn revenue depends upon the degree of control it can exercise over those who consume it. Most radio and TV broadcasts do not require the audience to pay, but rather rely on advertising income or taxation, thus not obtaining most of the consumer value of their product. The available technology for making broadcast audiences pay is at an early stage of development and is still quite expensive. The problem of excludability is not present in printed publications, since people usually pay a small amount to buy a newspaper or magazine; b) advertising potential. Broadcasts and publications which cover a wide area, or are targeted at a general audience, earn less income per unit audience than narrow area and special interest broadcasts, because they are less well targeted. This distorts the comparison between different types of supplier in all media. TV advertising revenue is determined, in part, by limits on the number of channels permitted to supply advertising time and by limits on the amount of advertising per hour; c) cost structure. Different media have different underlying costs, and unless supply is heavily constrained (as in the case of television) revenues will, as a broad approximation, tend towards costs through the process of competition; d) choice of influence over income. A proprietor may decide deliberately to forgo income in order to obtain greater influence. A media product which is less suited to high value entertainment might earn less, despite having much influence; and e) relative wealth. A publication aimed at wealthier classes or provided in a wealthier area of a country might earn more than others, without reflecting any greater influence (NERA, 1995: 13-4).<br />Establishing a method of measuring multimedia concentration on the basis of a single unit seems impossible. Some indicators (end-use based) come a bit closer to the concept of pluralism since they focus on the citizen consumer, while others (revenue-based) correspond more with the ways of concentration used in other sectors. But a single audience or financial criterion is hardly conceivable. Apart from the problems that each criterion pose on its own due, this is mainly because none of them has been tested yet for measuring influence. It might be 'safer' then if we took both criteria into account at this stage. The difficulty in finding a system that is applicable to all information services with their different characteristics, leads us to the conclusion that a mix of the two methods described above could provide a valuable alternative. Combining the two types of measurement depending on the configuration of the media in question might provide a quite valid method. A combined test involving advertising and/or subscription revenues and audience shares (together, of course, with ownership restrictions) could effectively measure market power and possibly reveal influence on total media consumption. Regarding commercial broadcasting, for example, the main measurement of market share is by advertising revenue. However, the audience share should also be used as an additional measure, since the revenue by advertising and the prices for advertising spots almost entirely depend on the share of audience for a channel or a certain programme.<br />That both methods are needed for an overall and accurate picture of concentration in the market-place for ideas could be shown through the following example. Let us consider the radio medium whose production and distribution costs are obviously lower than those of television. Assuming that both media vehicles attract the same audience, the first method, based on end-use (audience), will attribute the same degree of market power to both media and consequently the radio contact will have identical value to television contact. The second method, based on monetary measurement, will attribute varying levels of market power to media having the same number of contacts (radio contacts will have a lower value than television ones). The two different methods thus come up with different measurements of the market position of the media. The combination of the two methods of measurement (the 'middle way' as we may call it) would eliminate this discrepancy and provide an avenue to arrive at a more or less accurate picture of concentration levels in the mass media field.<br />The fact that a monetary-based measurement is by itself insufficient to reveal the exact level of media market power can be shown through another example. Revenues, which is the unit of measurement most heavily used in competition policy, is clearly related to economic power. A company which has a large share of revenues is likely to invest a larger portion of its resources in improving the quality of its product in order to stay ahead of its competitors. If the investment is successful, then the improvement in product quality can result into higher market share. In the media domain, a firm must be capable of retaining and therefore influencing its audiences, in addition to commanding its resources. However, the relationship between revenues and influence is far from straightforward and varies across different media. Consider a case of a newspaper which cuts its cover price. It will increase its circulation and hence its influence, but may reduce its revenue (depending both on the elasticity of sales and the importance of advertising revenue with respect to cover price). A powerful newspaper proprietor with deep pockets, like Rupert Murdoch, who is determined to get his message across at any cost, could simultaneously reduce his market share and increase his audience by reducing the cover price of his newspapers. Measuring Murdoch's shares only in terms of revenue will probably not reveal the exact strength of his market position. But by adding another dimension (i.e. measuring his audience share), a more accurate picture of his market position will come up.<br />Marrying time use data with advertising expenditure data provides a more concrete example of our 'middle way' proposals. Advertising expenditure is the amount that advertisers pay for advertising, rather than the amount the media receive from it, since the former reflects the value to purchaser. The difference includes agency commission and the costs of producing advertising copy. As Arthur Andersen analysis indicated, time spent on television is over seven times more than time spent on newspapers. However, advertisers in Britain in 1993 were prepared to spend £1.2 billion on national newspapers and £2.6 billion on television advertisements (Advertising Association, Advertising Statistics Yearbook 1994). So although TV is much more important than newspapers in terms of time use, advertisers only regard it as roughly twice as important. Advertisers also consider it essential who is in the audience, in addition to how many watch, read or listen. The Independent, for instance, is a serious newspaper read by many influential people in business, the City, Whitehall and Westminster. So although space in the Independent is cheaper than space in the Sun (because there are far fewer readers), the cost per reader of advertising in the Independent is much higher than that of advertising in the Sun. This is illustrated by the advertising rate cards shown in Table 4.8. The amount that advertisers are prepared to pay to reach the highly influential Financial Times reader would buy access to no less than fourteen Sun readers.<br />Table 4.8<br />Advertising Ratecard Costs of Selected UK National Newspapers<br />Newspaper£/page£/000 adult readersThe Sun28,0002.84Daily Mail22,6804.80Independent14,00012.20Daily Telegraph34,50012.71Sunday Times47,00013.28Financial Times29,56839.69<br />Source: BRAD, 1994; London Economics, 1995.<br />These considerations reinforce the attractions of combining different measures for assessing media concentration levels and possibly establishing impact. The relationship between money and influence is certainly imperfect; however, monetary-based measures do at least take into account, in a way that head counts cannot, the fact that the influence of any medium depends on who it reaches in addition to how many it reaches. Advertising expenditure is thus an important candidate dimension in order to calculate influence shares. Nevertheless, although it gives us some indication of the influence of radio and TV compared to the press, it is hard to construct a consistent statistical measure of influence. The exercise is heroic because print is bought by the column inch or page part while radio and television is bought by the second. Another point that must be borne in mind when making such attempt is that television (and therefore advertising in television) may be more influential at prime-time (usually between 8 to 11 pm), when the audiences are huge, rather than in day time. And space in newspapers may be more influential on some pages than others. We also have to make the assumption (as the London Economics study did) that people spend the same time on advertisements for each hour of reading the press as for each hour of viewing or listening to commercial television or radio. All these factors must be taken into account by authorities wishing to adopt this model.<br />Another possibility in relation to complex combinations of media would be to set a percentage of market share in terms of revenue as a threshold for further examination of the position. This approach is taken in UK monopoly legislation in relation to newspaper ownership. Should a media controller breach the pre-determined revenue threshold, the argument follows, an audience share test will apply to determine whether he/she attracts a high percentage of consumers as well. This undoubtedly gives more weight to the economic criterion of revenue. The reason why it is considered here is that the sources of bias and error are potentially so many in all audience-based measures of shares in the aggregate media sector, that policy decisions based on such measures alone should be made extremely carefully. A long established, crude method for measuring market concentration such as revenue may moderate potential inaccuracies of the newly-conceived audience-based methods.<br />Finally, a system which can combine different measures such as numerical criteria, revenue, audience share and audience time spent consuming a medium has been proposed by the British Broadcasting Corporation (BBC, 1995; Andrew, 1995). According to BBC, a weighted points system which combines several measurement criteria may represent the most reasonable approach. It should have the following characteristics: a) the points system could award a certain number of points per channel or title owned according to the particular relevant market; b) the weighting by points would allow channel/title ownership to be informed by other criteria, rather than being applied as a sole criterion; c) the number of points would reflect the revenue, audience share and audience time spent which channels/titles in a given market typically command; d) it would be important for the points system to be adjustable over time in order to take account of the changing impact of different media; e) points thresholds should be set to provide an indication of whether organisations are moving into positions from which they might have the potential to abuse pluralism, and to instigate an in-depth review; and f) the thresholds should be designed so as to limit the share of commercial revenues for single media (30% of revenues) together with a lower limit (20% of revenues across the relevant sectors) for cross media ownership. BBC stated that, used in this way, the lack of sophistication of the measure used is less of a concern for regulators, since more detailed information can be taken into account in the review itself.<br />In any case, we are still a bit away from constructing a single indicator capable of assessing effectively multi-media concentration. Both the Commission and several European countries which are in the process of changing their national media ownership rules (i.e. Germany, Great Britain) are aware of the need to examine more thoroughly certain questions concerning content before any initiative is presented. Germany's example, where until recently there seemed to be a clear preference for a new system based on audience data while there is now a consideration of a model based on commercial revenue, is an indication of the confusion over the 'best' possible measure of media concentration and influence. There are many possible different ways of measuring share of the market. Each one of these measures conveys some important information about the markets and the degree of influence of each firm within them; however, none is so perfect to assess the exact level of influence shares in the media industry. This is precisely the reason why we propose the use of a combination of measures. An over-reliance on one of the measures may lead to flawed policies. The media regulator must gather as much information as he/she can about a company's market share; he/she then must compute all of the different measures (or those he/she deems necessary to build up a complete picture of the market) and examine them carefully. By this way, he/she will be able to come to a definite conclusion about a potential excessive influence of the company in question.<br />Finally, the regulator must ensure that the framework for media ownership regulation is clearly set out and not affected by day to day fluctuations in companies' shares. Audience shares, circulation, shares of revenue and so forth all very materially from day to day or month to month, something that may present difficulties. A firm may find itself in line with a proposed threshold one month but it may exceed it next month, simply owing to its own success relative to its competitors. It would not be right if media controllers were required to divest because, owing to their own market success, they have crossed some line, quite possibly temporarily. One approach to such a problem would be to have delays built in to the rules. If an enterprise crosses some pre-determined line, action would only be taken if it were still over the line some period later. This would mean that action was not taken on temporary fluctuations. It would also give media controllers the ability to take corrective action of their own during the delay period. A second approach would be to have two limits, a lower one which may trigger a discretionary power, and an upper one which automatically triggers action or divestment. A third approach, suggested by NERA (1995), is to establish that the rules as to what was allowed would only come into play at times when mergers and acquisitions are proposed, and when licences are applied or re-applied for. So, for instance, if a media owner wished to acquire another newspaper, this would be prevented if it took him/her over the limit. If a TV licence owner went over the limit during the course of the licence, he/she would be prevented from reapplying.<br />](https://image.slidesharecdn.com/iosifidisphd-ch4-110307023651-phpapp01/85/Iosifidis_PhDch4_1996-1-320.jpg)
Iosifidis_PhDch4_1996
- 1. MEDIA CONCENTRATION POLICY IN THE<br />EUROPEAN UNION AND THE PUBLIC INTEREST<br />PETROS IOSIFIDIS<br />A thesis submitted in partial fulfillment of the <br />requirements of the University of Westminster <br />for the degree of Doctor of Philosophy<br />Centre for Communication and Information Studies<br />University of Westminster, London<br />June 1996<br />CHAPTER 4<br />4.1INTRODUCTION<br />The subject of this chapter is the exploration and analysis of the different methodologies of measuring media concentration. The appropriate measure depends on the objective of the measurement. The crucial question that should be asked beforehand is what is the purpose of the measurement: the examination of economic power or an assessment of whether market structure might restrict diversity in the media industry? So far media academics have borrowed measures that were developed by economists. Regarding the examination of economic power, economists have used companies' market share, shares of assets, value-added, sales, advertising revenue or even number of employees in forming an opinion of their bulk in the economy. These measures are more appropriate for industrial structure and manufacturing sector. In the media, because of their nature and their significant role in culture, society and politics measures examining the media firms' economic power alone seem to be inadequate. The special social significance attached to the media's role in disseminating information requires an investigation of whether a concentrated media market restricts the free flow of information.<br />To overcome this limitation of economic-based measures a few media analysts have proposed a number of media concentration measures which take into account their importance to the public. The view that has emerged from the recent debate on media concentration at European (initiated by the EU 1992 Green Paper) and national (Arthur Andersen's 1994 study UK Media Concentration) levels, is that it is possible to measure the 'influence' exerted by the media by applying audience-based criteria. It has been put forth that while financial units are close to the traditional systems of concentration measurement which permit assessment of media market concentration or even the existence of a dominant position (concentration of resources), audience-based methods are coherent with the cultural/political standpoint and can be held to be most effective for the measurement of pluralism and influence in the market-place for ideas. Nevertheless, influence over the audience cannot be assessed by using audience-based criteria, whether that is readership, audience reach, viewing or listenership share, etc. Audience exposure to mass media is certainly not the same as influence over the audience. In the end, these so-called new measures are nothing but refinements of measures of market power. They measure market power, although in a more sophisticated way. They are a form of market share measurement, which is a classic economic measurement.<br />This chapter examines the problems associated with both types of concentration measures and assesses their appropriateness in the broad context of media concentration's impact on the public interest. It will be argued that both financial and audience-based units apply, since economic power and diversity (especially in the range of material offered) are closely linked. A thorough definition of the markets in which market shares would be calculated is firstly provided, followed by a discussion on the measurement of economic power. In discussing the measuring of shares in the political/cultural market, the notion of the 'media controller' is firstly analysed, followed by an extensive examination of the various units that have been said to capture a media operator's ability to influence public opinion. At the end of this chapter alternative proposals are put forward in an attempt to develop a better measure which takes into account a large number of public interest issues.<br />4.2DEFINITION OF MARKETS<br />4.2.1 Defining the Geographical Market<br />When focusing on levels of concentration, one has to define the relevant market for which the market shares would be calculated. Relevant markets are defined firstly in terms of geography and secondly in terms of products or services (the good in question). The definition of the market-place can have an impact on what is perceived as its structure. The geographical dimension of market definition determines the scope of the market, that is, whether markets are defined as being local, regional, national or even international. The issue of the relevant geographic market being examined is particularly important, since the adaptation of either narrow or wide market definitions may lead to totally different results in measuring levels of concentration. <br />Most media economic studies have adopted fairly narrow definitions of markets. Regarding the newspaper market structure, those studies were often restricted to local markets for dailies. US mass communication researchers have investigated newspaper concentration for decades assuming that the city was the appropriate territory. Nixon (1968) talked about competition as a city phenomenon while Busterna (1988b, 1988c) analysed levels of concentration in cities. Picard (1988) moved away from the traditional way of describing markets and examined daily competition in metropolitan areas. After considering the national newspaper market and thirty markets from the largest 100 metropolitan areas, he reached the conclusion that newspaper concentration increased as the geographical size of the market decreased. Finally, Lacy and Davenport (1994) examined the degree of concentration and competition in the US newspaper industry based on the county. They argued that although daily newspaper markets showed higher concentration and lower competition than other US industries, these markets were less concentrated and had a greater potential for competition in 1983 to 1988 than suggested by other studies using cities as the geographic market. <br />It is obvious then that using the city as the geographic market yields different descriptions of market structure than using the county as the geographic market. The same could be said about national and local or regional markets. The Arthur Andersen study, for instance, adopted a national perspective in examining UK media concentration. As we shall see below, in order to measure company shares in the national cultural market, the consultancy used the time that the public actually spends on the media. It is easily understood that, under such interpretation, regional channel Carlton's share will be smaller in relation to media with national coverage (i.e. BBC). In fact, the above analysis found that BBC's share in the national market is 44.1% while Carlton's share in the same market is just 6.9%. However, it is meaningless to calculate Carlton's national market share since people living outside the London area do not have access to that channel.<br />Broadly speaking, some markets such as the financial ones may be international, while others such as that of magazines may be regional or national, and others such as cable or the sale of advertising may still be very local. Local advertisers, for example, are interested primarily in the local market and thus normally restrict themselves to consideration of local markets. So there are several definitions of the geographical market depending on the product in question. One thus should be careful when establishing the sphere of competition and trying to determine the extent and effect of competition on pluralism. Insufficient competition and therefore excessive influence may arise at a number of levels - local, national, regional, or international. The regulator must ensure reasonable diversity and balance in the information and entertainment available at all these levels. Referring to the UK media market, Robinson et. al (1995: 57) argued that 'the main worry, from a pluralism viewpoint, is that a company might acquire excessive influence in a local market through its ownership of both a local newspaper and the local commercial TV franchise. Yet, if such a company operated only in one English region, its share of the national media markets would be so small that it would comfortably pass under all the proposed thresholds'.<br />The above discussion illustrates that cross-ownership between a local TV channel and a local newspaper (or a local radio station) would lead to elimination of competition and therefore increased concentration in a media market, only if they are based in the same area. This is generally true since different newspaper groups or broadcasting channels operating in different regions do not compete for the same audiences. Yet they might be in direct competition for advertising revenue. Referring to the newspaper business, Veljanovski (1990: 18) stated that 'national newspapers are seeking a greater share of the classified advertising business dominated by the regional and local press, through regional editions and inserts, and the regional character of their readership. Regional newspaper companies which each own many local and regional newspapers, now offer attractive rates for advertising simultaneously in all their newspapers'. From a pragmatic viewpoint thus, it cannot always be assumed that regional and national media outlets are not in direct competition for advertising revenue.<br />4.2.2 Defining the Product Market<br />Besides being geographically-defined, markets are also product (or service)-defined. 'The products chosen must form part of a group with characteristics which separate them in an economically sensible way from all other products' (Frazer, 1992: 14). The economic criteria for market definition are based on the notion of substitutability across products as measured by their price elasticities. According to competition policy, two products belong to the same market when their producers' price-setting behaviour is interdependent. There must be substantial interchangeability (if not homogeneity) between these products and others. According to Frazer (1992: 14-16), the degree of interchangeability may be assessed from both producers' and consumers' viewpoints. The first method of assessment - supply-side analysis - seeks to identify those firms which are either supplying the products, or which will be willing and able to supply the products if there were an increase in their price. The second method of assessment - demand-side analysis - measures the willingness of consumers to substitute one product for another. If there is a high degree of substitution between the two products, then they will be considered to be part of the same market. If, for example, consumers are willing to substitute free-access television (whether financed by advertising or licence fee) for pay-TV, then these will form part of the same market.<br />The current technological developments in the media sector have made it difficult to reach a proper product market definition. Indeed, in the current 'multimedia' age (i.e. proliferation of new media, convergence of technologies), the question of what products are part of the same market is becoming hard to answer. The above described studies concerning the US local newspaper industry, narrow the market by ignoring the impact of non-local dailies and of print sources other than dailies (other non-print media). Given that a meaningful definition of a media market must indicate the range of alternative sources of supply open to readers, listeners and viewers, conventional ideas about control of the market by publishing or broadcasting companies must be open to criticism (Peacock, foreword in C. Veljanovski, 1990: 6). Cento Veljanovski (1990: 16) illustrates this by giving the example of one firm which produces 100 per cent of the domestic output of white handkerchiefs. This may indicate that the company is the only domestic producer, but it is not evidence of complete monopoly and does not imply the ability to influence price unilaterally. Such a manufacturer faces competition from imports, substitute products (including handkerchiefs of another colour or tissues), and from potential entrants. Similar considerations, the argument follows, apply to the media industry. Television companies compete for audiences not only between themselves but also with VCRs, cable and satellite TV, and for advertising revenue with all these and also with radio, newspapers and magazines. Competition among media sectors thus must be taken into account when defining the relevant product market.<br />The fact that different media are seen as substitutes is shown if someone looks at local news which could be found on television, radio and local papers, and at national news which could be found in all these and also in cable TV, regional radio outlets, and so on. The same or similar information can be delivered via print, on-line, teletext or audiotext and there are products and services which combine different communication modes: images, text and audio. What is the definition of the broadcasting market then? If we take as an example the release of a film which can be viewed firstly in a cinema, then as a TV programme and lastly in a home video, then the market for this product should include all these sub-markets. A soap opera, on the other hand, may only be used as a TV programme and therefore the market for this product is television. Thus the market for soap operas and films is defined differently. Further, should the definition of the broadcasting market cover electronic publishing? However, if electronic papers are defined as televised media, then they should be covered by the very same rules that applied to the broadcasting sector (many would strongly oppose that). As stated by Jonathan Davis (1994: 7) 'it will be necessary to adopt a much more fungible approach to market definition than either competition authorities are used to or is currently applied to the media'.<br />As the distinction between both electronic and non-electronic media and terrestrial broadcast TV and telecommunications services becomes more and more blurred, so does the definition of the relevant market. The presence of close substitutes can expand the market, resulting in lower levels of concentration. Therefore, several alternative market definitions need to be considered simultaneously. Levy and Setzer (1984), for example, considered concentration levels in the US national video market-place under alternative definitions of the video market-place-broadcast only, broadcast plus cable, etc. There have also been many studies that, by trying to identify the user of the medium while measuring concentration, draw a distinction between audience markets and advertising markets. Mc Fadyen et al. (1980) noted the difference between market for revenue (through time sales to advertisers) and market for audiences and measured concentration in Canadian broadcast markets separately for the two groups. Larson (1980) not only considered concentration in the US television industry at both national and local levels but he also looked at levels of concentration in broadcasting and among broadcast advertisers. Levels of seller concentration (TV stations) were measured by station net weekly circulation, estimated TV station income, network advertising revenues and by national and regional advertising revenues. Buyer concentration was treated both by advertising agencies, who purchase time in the industry for television advertisers and by the advertisers themselves, who are the ultimate consumers of the television time. These studies make it clear that much of the media serve two general markets - that of consumers and that of advertisers - and further distinctions within these markets based on specific types of advertising and/or region have to be drawn.<br />The main point so far has been that media today can be seen as close substitutes because of the convergence of technologies. However, it is not only the substitutability in distribution networks that matters but also substitutability in content. We indeed have convergence in distribution today but that does not necessarily mean that what is on TV is substitutable for what we read in newspapers, or listening to the radio is substitutable for watching television. Newspapers may be read through computers today (i.e. Internet) but the content that is provided to Personal Computers (PCs) is the same. In economic literature, substitutability relates to the question of whether a range of goods or services in a market are essentially, as far as the consumer is concerned, choices between one or the other in terms of quality or price; or are they provisions of alternative goods, in which case the choice between them will depend on the consumer's criteria of utility. People do make choices between going to the cinema or going to restaurant, but these do not compete in the same market (they are not substitutes). On the other hand, going to a restaurant and take-away from a restaurant are probably substitutes. <br />When it comes to the media industry, the question not only is whether television is a substitute for newspapers but also whether advertising in television is a substitute for advertising in newspapers; or news and opinion output (where pluralism matters greatly) provided by newspapers is a substitute for news provided by TV. From a consumer point of view, does it make any difference if he/she orders a video tape from a telephone company or takes it from a video store? Surely such decision has nothing to do with content at all; it is just the convenience of delivery which is competitive and that makes a big difference on policy grounds. Referring to the television sector, Bates (1993: 7) argues that 'audiences are concerned primarily with video signals, no matter the delivery channel. In this sense, audiences may not make much of a distinction between regular broadcast television, cable and alternative delivery systems such as home satellite dishes, low power TV, multi-point delivery systems, etc., at least to the extent that such alternatives are available'. The definition of substitutability thus has to be drawn very carefully, encompassing both substitutability in distribution networks and substitutability in content.<br />These distinctions suggest that considerations of the print and broadcast market structure need to be based on multiple definitions of markets, with separate considerations of local, regional, national and international markets, of audience and advertising markets and of potential substitutes. Apart from this aspect of the problem however, one has to find an appropriate standard for the measurement of market share. There follows an assessment of the relative merits of both revenue and audience related criteria firstly for the measurement of economic power and secondly for the measurement of diversity and pluralism. It will be argued that both financial and audience-based units apply, since economic power and diversity (especially in the range of material offered) are closely linked.<br />4.3MEASUREMENT OF ECONOMIC POWER<br />One of the arguments about levels of concentration concerns the measurement of the magnitudes. The most usual method consists of the presentation of absolute numbers: rankings according to turnover, profits, number of employees in the companies or, in the media industry specifically, number of TV stations or newspaper titles, and so on (Seymore-Ure, 1991). Shepherd (1979) argued that one can use firms' shares of assets, value-added, employees, or profits in forming an impression of their bulk in the economy. Nevertheless, he also noted that each measure has a slightly different meaning. Assets may be the best indicator of the firms' economic power while value-added shows their role in the country's production process. The large companies' share of workers reflects their influence over jobs or even votes whereas profits reveal their ability to gain market advantages compared to the rest. All these 'technical' problems in measuring corporations' scope (size) indicate that there is no single best measure of 'aggregate concentration' in an industry. 'Bigness' therefore is something everyone understands but no one can measure definitely. <br />Apart from these problems there is the question of which number of large firms is the 'best' one for defining the share of large firms. As indicated elsewhere, measuring concentration in media markets requires both a measure of the market (identification of the market) and a measure of concentration. A number of indexes have been developed since the 1950s in the USA for both purposes. The Bureau of Census and scholars of American industries have presented concentration data and ratios as a percentage of control of a measured value by the largest 4, largest 8, and largest 20 ownership units in an industry (see Larson, 1980: 25). The 4, 8 and 20 criteria were used by Caves (1962, 1966) in his airlines and in his water industry studies, and by Bain (1968) in his studies of business concentration. One thus can slice the large-firm share at any of a number of levels. Shepherd (1970, 1979) however argued that the four-firm concentration index is the most commonly used measure of concentration in general industry studies. The advantages of a concentration ratio are firstly that it shows the degree of concentration of few-firm leadership directly and correctly and secondly that it shows changes in market power with special clarity. The defects though are that concentration ratios are often too inclusive, embracing scores of disparate products and distinct regional sub-markets. The national ratio of, let say, 10 per cent for newspapers for instance, may conceal very high local concentration. In addition, concentration ratios may be too narrow. In some cases imports, not included in the ratios, cause concentration in total sales to be lower than the ratios state.<br />The 4, 8 and 20 criteria were also used by scholars including Nixon and Hahn (1971) who specifically treated concentration in the newspaper industry. A more recent analysis carried out by Picard (1988) used four ways to measure concentration in the US daily newspaper industry: the one-firm concentration, the two-firm concentration, the three-firm concentration and the Herfindahl-Hirschman index (HHI). The first three concentration measures are the percentage of circulation controlled by the largest circulation newspaper, the two largest circulation papers and the three largest circulation titles respectively. The HHI is a way of placing concentration on a continuum from 0 to 10,000, the latter representing a monopoly. The index is determined by summing the squared value of the market shares of all of the firms in a market. The HHI uses shares ranging from 0 to 10,000. But at what point is the market considered concentrated? Stigler (1983), one of the supporters of using the HHI for concentration measurement purposes, argued that one can roughly identify markets with HHI>1,500 as tight oligopolies and those with HHI<1,000 as loose oligopolies. If the index falls well below 1,000 then the market is unconcentrated and when it falls above 1,800 then the market is highly concentrated.<br />As regards the television market, in particular, the evidence shows that little variation can be found in the most commonly used concentration ratios (2, 3, 4, 8 , or even 20-firm). Larson's 1980 study referred to above, measured levels of concentration in the US TV industry using a 3, 8 and 20 breakout and concluded that all the used measures revealed that the TV station industry was not highly concentrated. Such conclusions may not be true, not least because television broadcasting is not an ordinary, typical industry. The television market is distinctive in a number of ways: a) in that the question of economic power has manifested itself in two distinctive policy arenas. On the one hand, there is concern over the impact of concentration on economic conduct and performance, pricing behaviours, barriers to entry and limitation of output (Ferguson, 1983; Bates, 1993). On the other hand, there is the question of the social performance of the market (pluralism, diversity, and so forth); b) TV stations are operating in several markets simultaneously. In one market they trade programming for audiences and in another market they sell audience to advertisers (Bates, 1987, 1993); and c) television stations operate in a number of geographical levels (Compaine, 1981, 1985). Therefore, according to Bates (1993), the strict barriers to entry and the domination of affiliates of the three networks make the concentration ratios of little importance in measuring concentration in the US local TV markets.<br />4.4MEASURING SHARES IN THE POLITICAL AND CULTURAL MARKET<br />Large companies' sales and turnover may be the best indicator of their economic power and reveal their ability to gain market advantages compared to the rest. In other words, very powerful firms can influence economic conduct, performance and pricing behaviours and have an impact on barriers to entry and limitation of output. Therefore, when the purpose is the traditional examination of market power then a high revenue company share may provide a quite useful guide. When it comes to the media however, the concern not only is over the impact of concentration on economic aspects but there is also the question of the social performance of the market (i.e. pluralism and diversity). Are measures tailored to measure economic concentration good enough to capture concentration levels in the political and cultural market, the so-called 'market-place for ideas'?<br />The view which has emerged from the recent debate on media concentration at European (EU 1992 Green Paper) and national (Arthur Andersen's 1994 study UK Media Concentration) levels, is that it is possible to identify a sort of relevant 'market for ideas', which does not coincide with the economic definition of relevant market; and that de facto restrictions of pluralism and diversity are the results of an abuse of power in such market (abuse of political and cultural power). There are three problems associated with such approach. Firstly, there are substantial difficulties in defining a suitable notion of relevant market in the political and cultural sense. As the relevant product tends to extend across different media, the cultural/political notion of the relevant market may be significantly broader than the economic one. The problem is bound to be exacerbated as multimedia conglomerates expand their activities further, and ownership of complex transnational media chains becomes widespread. To illustrate, how does one assess the effective combined share of, say, News International in the broader market for information, culture and political opinion, comprising newspapers and TV outlets in several countries? Secondly, the exact nature of the potential abuse is not clear and explicable and cannot be specified in the same way as abuses of economic market may be specified. What then counts as an abuse of power in the political/ cultural market? Beyond the general assumption that all media exercise some form of political and cultural influence on the public, there have emerged no satisfactory criteria so far for the definition of a broad political and cultural market in which spheres of influence by a single controller could be assessed.<br />The most serious reservation concerning this approach though has to do with the selection of the criteria for measuring diversity in the market place for ideas. It has been put forth that while financial units are close to the traditional systems of concentration measurement which permit assessment of media market concentration or even the existence of a dominant position (concentration of resources), audience-based methods are coherent with the cultural/political standpoint and can be held to be most effective for the measurement of pluralism and influence in the market-place for ideas. Nevertheless, influence over the audience cannot be assessed by using audience-based criteria, whether that is readership, audience reach, viewing or listenership share, etc. Audience exposure to mass media is certainly not the same as influence over the audience. In the end, these so-called new measures are nothing but refinements of measures of market power. They measure market power, although in a more sophisticated way. They are a form of market share measurement, which is a classic economic measurement. Audience-based units are the equivalent of, say, measuring sales - i.e. market share - which is a classic economic measure of market power.<br />In any case, political/cultural diversity and economic power are closely linked. It might be worth at this point spelling out the arguments about the relationship between economic power and the range of material offered. There is a clear relationship between economic measures of media power and influence/pluralism because economic power determines the control over choices offered. In fact, in terms of the public interest and debates about regulation and concentration of media ownership, there are two wide-spread arguments. On the one hand, there is the argument saying that a highly concentrated market structure in the media sector is of concern not only for the possibility that it may lead to abuses of economic market power, but also for the potential effects on pluralism. A large media player who controls a substantial portion of at least one media sector (for example daily press or TV) has the potential for forcing his/her views across a range of products (political/cultural bias), and thus for restricting the choice of products available to the public in political and cultural terms. In this sense, a competition policy decision aimed at curbing an abuse of economic market power (excessive pricing, creation of barriers to entry, etc.) may also increase pluralism, at least in the sense of reducing bias.<br />On the other hand, there is the argument saying that increased competition may lead to less pluralism in the market. Increasing the number of firms in an industry does not necessarily imply greater diversity in the quality and variety of products on offer - especially where price competition is weak. If firms compete on price, product differentiation provides a device for softening the intensity of competition: in a simplified world with only two companies, they will have an incentive to locate themselves as far as possible from each other on the product line (i.e. to offer as diverse a product as possible in terms of product variety and quality). Proximity of location would mean that prices are gradually eroded as the companies compete for each other's business. However, if there is no explicit interaction in the firms' pricing decisions, the opposite result obtains: the firms will locate as close as possible to one another, as the 'market share effect' (the incentive to be where demand is, or to increase one's market share given the market structure) prevails over the 'strategic effect' (the interdependence of the two firms' pricing decisions). Thus the incentive to differentiate products is weaker when companies are able to operate in the near-absence of price competition. The tendency to converge on tried-and-tested formulae poses a potential danger to welfare in terms of the variety of products offered by the market. Hotelling (1929: 41), who originally discussed this effect, talked of 'an undue tendency for competitors to imitate each other'. Therefore, a more fragmented industry structure in the media sector may not necessarily deliver the socially desirable level of product differentiation because it may be more profitable for the companies to locate 'where demand is' (i.e. stick to the middle ground in order to catch the widest audience).<br />A further important question relates to the possibility that too much competition might display a bias in favour of certain types of products and neglect others. The particular bundle of commodities that are actually produced in the media market (the type of programmes/titles available) might be sub optimal from a social welfare point of view. When demand for products in a particular category is generally inelastic, the products which are being actually offered may end up positioning too close to each other (sub optimal product diversity); and those products for which the elasticity is comparatively lower may not be produced at all. The implication could be that some segments of tastes and preferences might systematically not be catered for, although there might be a large number of different media products. So, strictly from the point of view of pluralism, there might be no automatic advantage to be gained from a more diverse media structure. On the other hand, so the argument runs, a very concentrated industry structure might lead to great diversity, if the dominant firm(s) seeks to prevent entry in the market by filling all gaps in product space.<br />Having provided, to some extent, an argument that economic power affects the range of material offered, and having spelled out the arguments as to whether concentration or a fragmented industry can deliver best the desired diversity, we now turn to the question of which criterion is the 'best' one for measuring concentration levels for media pluralism purposes. The close relationship between economic power and pluralism/diversity indicates that criteria that are being used for the measurement of market power can also be used, at least in principle, for the measurement of media influence and vice versa. Financial criteria, for instance, a long-established method for measuring market power, could also be adopted for measuring 'influence' (audience exposure to the mass media); and audience figures, supposed to be more efficient for measuring diversity in the market place for ideas, could also be a measure of economic power, especially as they are sold to advertisers. There are still problems though, since different criteria might give a rather different picture in terms of the participants' individual shares and may lead to disparate inferences on their roles and importance. The rest of this chapter examines the problems that are inherent in defining an appropriate measure of concentration in the information market. <br />4.4.1 Criteria Based on the No of Licences Controlled by a Single Operator<br />Media concentration rules in the audio-visual sector have traditionally been placed on the number of licences or franchises a single operator can hold. Most Western European countries have developed such restrictions with the aim of preserving pluralism in the market. For Frohlinger (1995), head of Unit E/5 'Media, Commercial Communication and Unfair Competition' of DG XV, there were two reasons for this: the scarcity of available frequencies, which required the prevention of one person or body from using all or most of the spectrum; and the lack of reliable audience measurement. However, technological change together with the proliferation of channels and the convergence of different forms of distribution imply that it is no longer necessary (or even suitable) for governments to limit the number of licenses. Further, the progress in audience measurement enables the calculation of reliable figures on the basis of which thresholds for the access to media ownership can be fixed. This is one of the reasons why the European Commission wants the replacement of numerical restrictions on the number of licences by thresholds based on audience share. As is explored below, the Commission is of the opinion that the audience share criterion would allow more flexibility, growth, innovation and increased diversity in the media market. It would facilitate, in particular, the launch of multi-channel packages (packages of specialised or thematic channels) which under existing laws limiting the number of channels are difficult to develop.<br />Rules placing limits on the number of licences or channels a person may control are indeed not appropriate by themselves as measures of plurality in the media market, since they pay no attention to media revenue, coverage or audience. These rules may measure diversity of media ownership but are crude as measures of the existence in a society of a number of different points of view. To illustrate, in the UK some Channel 3 licences are much larger than others in terms of advertising revenue and audience, so that rules limiting the number of franchises one can own but taking no account of revenue and audience size cannot prevent undue influence. It is imperative thus to take total revenues and population coverage into account, in addition to the number of outlets owned by an individual or a group, when constructing rules on media ownership. Lessons from press regulation are useful here. The press has never been subject to rules counting the number of titles one may hold. Traditionally provisions aimed at limiting press concentrations have fixed thresholds based on circulation figures and not on the number of titles that can be owned by a single publisher. In most European countries a merger qualifies for investigation on the basis of circulation, and not on the number of titles controlled.<br />4.4.2 The criterion of the Media Controller<br />Diversity of media ownership could thus be measured by using the criterion of the number of media carriers (TV channels, newspaper titles or radio outlets) owned by a single 'controller'. The purpose is to find a definition that makes it possible, when a new channel or newspaper is launched by a group or an individual (or when he/she acquires control of an existing channel or newspaper), to estimate the number of media that are already controlled by the group or individual and broadcast or distributed over the same area as that of the new scheme. Determining who controls what media has a distinct advantage. This sort of knowledge undoubtedly makes it possible to provide the legal certainty for investment and sufficient effectiveness for avoiding circumvention. In addition, the concept of media controller is more accurate compared with the concept of owner or majority shareholder in the media company, since there may be several shareholders with the same proportion of ownership. The problem is that it is not always easy to identify clearly who is the controller with decisive influence. Who is the actual controller of a media outlet in the case, say, where there are three shareholders each having 33 per cent share of the capital of the medium?<br />There is thus a technical issue of the relationship between levels of ownership and control. Before discussing that however, it seems to us that a preliminary issue arises, and that is the question of whether owners are controllers at all. In other words, can we simply equate ownership and control? Does ownership actually translate into control? Undoubtedly the answer is yes in cases where the founder of a media company holds the majority of the shares and his/her family the rest. Traditional press company structures have been formed along these lines. Today things are different though. The formation of large vertically-integrated and diversified units has led to a corporate structure in which shares are normally dispersed, in which financial institutions and industrial corporations have significant shareholdings, and in which highly qualified managers make key decisions. In situations of dispersed share ownership, who can be said to control the firm in question? When ownership is dispersed, does that mean that the share owners still control it? <br />The measurement of control in terms of ownership has weaknesses because it does not take into account other ways and means by which influence can be exerted over a firm's decision-making centres. Managers, for example, have been appointed in key positions in diversified enterprises (and often given the opportunity to hold minority stakes) because their specialised knowledge and expertise has been considered essential for the smooth running of the modern media enterprise. This, however, creates an agency problem, i.e. it poses the problem of the relationship between owners/shareholders and management, often putting the editorial control with the management. An agency problem is always evident in the presence of delegation, that is, in all cases where one individual is contracted to carry out a task on behalf of another and the parties' incentives do not coincide a priori. A typical case is the shareholder/manager relationship, where the manager's decisions reflect his/her own preferences (which might include greater power or increased prestige) rather than the shareholder's goal of profit or value maximisation. The instrument available to the shareholder for aligning the manager's incentives with his/her own is the contractual arrangement: by making the manager's compensation also a function of his/her same objective (firm's profit or stock market value) the efficiency loss from delegation might be reduced.<br />Some economists stated that the above argument has great validity in cases where there are few shareholders but it might be open to question where there are many minority holdings. They pointed out that solving the agency problem and placing an effective control on management is relatively easy in cases where there is one or few shareholders. Nevertheless, it is far less likely to be successful where ownership is significantly fragmented across a large shareholder base. Dispersed share ownership, according to them, has important implications for the effectiveness of shareholder monitoring of management. In the media sector, the implication of having fragmented ownership due to strict ownership rules, might be that the editorial control would rest entirely with the management. To the extent that management might also pursue a particular agenda, ownership restrictions would then simply replace one form of potential threat to pluralism (that from the owner) with another (that from the management). Therefore, while the dominance of a single individual or shareholder group might be prevented, the management might in turn pursue objectives which are not less damaging in terms of pluralism and diversity. Within a company thus there is clearly a question of who actually controls the company's policy - the owners or the managers. It seems to us that a case-by-case analysis is required to assess who makes the key decisions which influence the editorial content.<br />It is not only the emergence of the new elite of managers that has raised questions over the relationship between ownership and control. It is also the links between advertising industries and the media that have masked the above separation. There have been two trends in the advertising industry that justify such assumption. Firstly, the progressive concentration of the advertising mediators (agencies and buying houses) and the separation between the creative activities (agencies) and the management of accounts (buying houses) have changed a key stage in the relationship between the media and the advertisers: the determination of the prices of insertion and other contractual conditions of advertising space and time. Secondly, new advertising formulae for advertising and commercial promotion such as bartering, product placement and teleshopping represent a means of integration between the advertisers and programme production industries and undoubtedly weaken the barriers between production, programming and advertising. Despite the economic advantages of bartering, for example, this formulae for advertising poses the risk of loss of editorial control by broadcasters, as does teleshopping and to a lesser extent product placement.<br />The increased involvement of other concerns such as financial institutions and industrial corporations in media share ownership is another trend that needs consideration. Marxists like Murdock and Golding argued as early as 1977 that 'shareholdings in the large corporations have become progressively concentrated in the hands of the dominant financial institutions and of other large corporations' (Murdock and Golding, 1977: 30). They added that 'compared with the general passivity of most private shareholders, institutional investors are likely to pursue a much more \" interventionist\" policy with respect to the operations of companies in which they have holdings' (ibid: 30). For Murdock and Golding though, such interconnection of the major communication firms with the dominant financial and industrial interests prove the fact that the ownership has not actually been separated from control. 'Control over the key processes of resource allocation is still significantly tied to ownership' (ibid: 32). They supported that argument by saying that both interests - communication and industrial/financial ones - share similar views. In their words, 'the owning group continue to constitute an identifiable capitalist class with recognisable interests in common' (ibid: 33). Negrine (1994b: 63-4) however challenged that statement on the grounds that it is 'too general to be meaningful', it is a general assumption 'pertaining to a concern for success and survival'. For Negrine the media exist in a competitive market place and have to employ whatever means possible to survive in that market place. So the general validity of the argument concerning the allocative power of ownership is in no doubt. What is in doubt for him, in view of lack of evidence, is the assumption that economic aims constitute the basis of pursuing a particular ideology ('conservatism') in the media.<br />The question of whether ownership is equated with control (and the search for a proper interpretation of 'control') is still with us today and forms a significant part of the current debate over establishing common ownership rules at European level. The Commission has initiated a study on this subject in order to reach a definition of 'controller' which will reflect experience at national level and will be in line with the various definitions of control already used in competition law and company law. The study, carried out by the European Institute for the Media (EIM), tried to identify who is genuinely in control of a press or broadcasting enterprise. After reviewing several criteria for defining the controller in a number of European countries, it identified five means of influencing a firm's decision-making centre: direct links, indirect links, family connections, external financial or contractual relationships and unofficial agreements. Despite the fact that direct links with the enterprise appear at first glance to be the most influential means, it would be wrong, according to the analysis, to underestimate the others. Owning a stake in a company either indirectly or through a relative (Kirch, for instance), could be just as effective a means of wielding influence as being a majority shareholder. Likewise, appointing the management (Murdoch's strategy) could be as powerful a means of exerting influence as being appointed to a management position. The study also identified four ways of wielding influence: influence through ownership links (capital), influence through links with staff, influence through financial links (debts) and influence through contractual links (supplies, distribution). It concluded that ownership links are the most effective way of gaining influence over the board of directors, although the other three - financial, staff and contractual links - should also be taken into consideration.<br /> <br />On the basis of these criteria it is then possible, with respect to each player directly or indirectly involved in the operation of the media in question, to evaluate by which ways and means he/she may exert a decisive influence on the operation. Margot Frohlinger of the European Commission (1995) argued that such a qualitative definition of media control seems both to be better adapted to the complexity of current ownership and control/dependency structures and respond to various attempts at circumvention of media ownership rules. Indeed, the EIM's analysis suggests that 'control' must be interpreted as widely as possible to include any means of exerting influence over the actions, editorial content, strategic direction and so on of the media organisation in question. Apparently though the Commission is concerned that the four criteria of wielding influence defined in the study are not clear enough to ensure fair and impartial application throughout the EU. This concern is expressed in its Questionnaire No III, questions E.1-E.8 (see Appendix XIII) where it seeks opinions on which of those criteria are most important and which have subsidiary status. According to National Economic Research Associates (NERA, 1995), it does not seem feasible to put together so many different variables for monitoring media control. Their conclusion is that the only satisfactory approach is to abandon hope of achieving absolute legal transparency, and to adopt a discretionary test based on an ownership criterion such as that provided by the newspaper provisions of the British Fair Trading Act (i.e. control applies where one person owns a certain per cent of the voting rights of the media enterprise).<br />It is indeed impractical to determine theoretically whether one particular criterion is of more importance than another. According to British Broadcasting Corporation (The BBC's Views: Pluralism and Media Concentration in the Internal Market, 1995), there are two reasons for that: a) the relative importance of criteria is changing over time; and b) in some cases a 'weak' link can be used to exert more influence than a 'strong' link in others - for example, non-voting shares offer less opportunity to influence editorial control than friendship with the editor. However, any definition of the controller based on ownership links (or the voting rights of a media company) should be complemented by at least one of the other criteria identified in the EIM study, that of contractual links. The inclusion of the criterion of contractual links is particularly important because it makes it possible to cover cases of vertical integration. The EIM analysis has the merit of demonstrating how growing vertical integration and how concentration on so-called 'intermediate markets' (i.e. programme supply and distribution) have an increasing effect on media pluralism. In particular, the study examines the possible consequences of the introduction of new technology and concludes that both upstream and downstream industries of the media enterprise are developing in a way which is likely to reverse the balance of power against the broadcasters. According to the European Broadcasting Union (EBU, 1995), it is not so much vertical integration as such that poses a problem for media pluralism but concentration in intermediate markets at different stages of the audio-visual chain. Indeed, there is nothing intrinsically wrong with vertical integration. The fact that a broadcaster has its own network of correspondents, produces a great part of its programmes and operates its own transmitter network helps it to maintain its independence and probably contribute to media pluralism. Vertical links become a threat to media pluralism only if they are coupled with dominant positions in intermediate markets, thus undermining the independence and development of the broadcasters (either by imposing or by withholding certain programme material).<br />Downstream activities such as distribution are increasingly developing into industries in their own right, separating their activities from those of content provision, and facing increasing competition as new technologies facilitate the entry of new players. This means that their bargaining power in terms of contractual links will take on an importance not seen in any of the traditional media. The ownership of conditional access systems by broadcasters is an example of this trend. The provider of the conditional access system that consumers need to decode encrypted digital broadcasts becomes the 'gatekeeper' of the digital broadcast, able to limit the flow of broadcasts and the access of broadcasters to his/her system. If the conditional access company is owned by a broadcaster, that broadcaster can restrict competition in the market (BBC, 1995). Thus ownership of this form of distribution system should be included among the list of criteria for identifying the nature of the media controller and the exercise of that control. Contractual rights to content should also be added to forms of control that can be exercised. Certainly, control can be exercised by buying the rights to be in the popular markets of sports and films. This is what is happening with BSkyB imposing conditions on cable channels with the right to carry its programme. Media groups should not be in a position to control the preceding stage of the audiovisual chain to an extent which would allow them to prevent broadcasting services not belonging to the same group from having access to film catalogues, sports transmission rights, television advertising, or other resources.<br />Having setting up the arguments and explained, to some extent, that there is a difference between ownership and control and therefore diversity of ownership might not lead to the desired levels of pluralism in the media market unless we extend the notion of 'control' to encompass influences exerted by managers, advertisers, companies upstream or downstream, etc., we now turn to the question of how control can be defined technically. A controlling interest (over 50%) or the ownership of a certain percent of the voting rights of the company should be enough to imply control of the enterprise in question. The problem arises where there are two or more large share-holdings in the same medium (for example, two persons having over 20% stake). The simplest way is to treat only the biggest shareholder as the controller of the medium in question. Although this way does not cause any problems in numerical terms, it has two major drawbacks. Firstly, the largest shareholder is not always the effective controller of an enterprise. Secondly, this method gives incentives for small shareholdings across different media.<br />Another approach would be to attribute owners influence according to their share-holding level. There is an on-going debate in Europe, and in Germany in particular, on whether owners above a certain level should be treated as controllers of a media unit. The German regulators suggested in September 1994 that an audience-share model to control TV ownership should be adopted (a reorientation of the debate is now under way, with consideration of a new model based on commercial revenue). The setting of a limit for each investor's holdings at 25% of total viewing share, could define the controller(s). Bertelsmann and the Kirch Group argued that it is unfair to take into account the whole audience share of a channel in determining whether an investor's overall share breaches the limit; instead, a proportion of the share, equivalent to the investor's stake, should be considered (see Table 4.1). Although the companies asserted that pro rating each person according to their stake is an impartial method that provides for fair company treatment within a media sector, we will agree with the German authorities that this model is complicated and may also be used to conceal an owner's real influence in a channel.<br />Table 4.1<br />How German TV Groups Would be Affected by the Media Authorities' Plan - and by their Own Counter-plans<br /> Audience shares to count towards companies' combined audience share Groups and channelsPercentage stake %Audience share %Authorities model %Bertelsman model %Kirch model %BertelsmannRTL37.117.817.86.66.6RTL 27.83.43.4-0.3Vox24.91.91.90.50.5Premiere*37.5----Total23.17.17.4Kirch GroupSat 1**43.014.614.66.36.3DSF24.51.21.20.30.3Premiere*25.0----Total15.86.66.6If Th. Kirch chnnels inclPro 747.59.29.24.44.4Kabelkanal45.02.02.00.90.9Overall Kirch total27.011.911.9CLTRTL TV47.917.817.88.58.5RTL 215.03.43.4-0.5Total21.28.59.0<br />Source: Satellite TV Finance, 29 September 1994<br />Note: The authorities recommended that an investor's stake of 5 per cent or more attracts 100% of a channel's overall German audience share, and that an investor's combined share should not exceed 25 per cent. Bertelsmann argued that only investor stakes of 20 per cent or more should be taken into account, and that only an equivalent proportion of the audience share should be allocated. Kirch also mentioned that proportionate audience shares should be allocated, but there should be no lower stake limit. An investor should be allowed a combined audience share of 35 per cent.<br />* Shareholdings in Premiere are not relevant (a separate structure was to be developed for pay-TV)<br />** No account is taken of Kirch's stake in Sat 1 shareholder Springer<br />A more reasonable and feasible approach would be to treat all those with a shareholding above some threshold as controllers of the firm in question. Each of the partners above a certain level of control should be considered as a 100% controllers in terms of their position in the market, and should be subject to the same constraints and responsibilities as if they were the sole controller. There are still two problems with this approach. Firstly, it is likely to lead to totals summing to more than 100%. Pearson, for instance, owns the Financial Times and the Economist, and it has a 14% stake in 'Yorkshire/Tyne Tees TV', to name but a few. The second problem is to fix the threshold above which an operator would be regarded as the controller of the medium in question. NERA (1995) proposed that a 20% stake should be considered as the minimum threshold that a media operator should have in order to be regarded as controller. The analysis assumed that ownership below 20% does not confer control. The 20% level was chosen arbitrarily using three reference points, namely the 20% limit on cross holdings and accumulation of holdings in the British 1990 Broadcasting Act, the 25% limit for newspapers mergers, and the 15% test level used by the Director General of Fair Trading (DGFT) in judging the likelihood of material interest. The 20% level seems reasonable and close to that used in other European countries' legislation such as Germany (25%) and Greece (25%). One point that should be born in mind though is that in the United States a shareholder can be deemed to be in full control of a company once he/she has a stake of as little as 5% of its capital.<br />Any restriction setting a ceiling on the maximum share of the market which may be held by one controller (individual or interest group) runs the danger of being not only arbitrary but also easy to bypass. Choosing the admissible ceiling is a delicate decision, as a simple example may reveal. Suppose it is established that no individual may hold more than 5% of the media sector as a whole, and assume that this includes one television station which accounts for 9% of the overall media audience and one newspaper which accounts for 7%. The implication is that the rule makes it equally possible for the individual to hold 55% of the TV channel (which implies full control), or, for example, 22% of the TV channel plus 43% of the newspaper (in both situations he/she holds under 5% of the total media market). Yet if the ceiling is raised, say from 5% to 8.2% of the media market, then it is possible for the operator to control both the newspaper and the TV station (8.16% of the market accommodates 51% of both). If, on the other hand, the ceiling is reduced even slightly to 4%, the operator will not have even the formal control of the TV channel (and indeed nobody will: with a 4% ceiling, any one party may control at the most 44% of the television channel) (see Table 4.2). It is obvious then that setting too 'high' a level may defeat the purpose of having diversified media ownership.<br />Table 4.2<br />Thresholds of Ownership (%)<br />1. Share of total media market taken by each mediumTV Station9Newspaper7Other84Total100Scenario A2. Share of medium3. Share of market (=2x1)554.954.95Scenario B4. Share of medium5. Share of market (=4x1)221.98433.014.99Scenario C6. Share of medium7. Share of market (=6x1)514.59513.578.16Scenario D8. Share of medium9. Share of market (=8x1)443.963.96<br />Source: London Economics, April 1995 <br />The key point is that restriction rules based on universal thresholds may have different results depending on the underlying distribution of the shares of the various component parts of the market - TV channels, radio stations, press titles, etc. The danger is that a pan-European rule established at the Commission level would have a significantly different impact in different national contexts. The fixing of a limit on the maximum share of the market which may be held by one operator is a question of media policy and, above all, a political decision. Given the great heterogeneity of the national media ownership regulatory frameworks though, it is doubtful whether there will be a consensus on this sensitive issue. In addition, there is the issue of how media ownership rules may be actually enforced. Most rules setting a ceiling on ownership can be de facto bypassed, for example through interlocking shareholdings or the formation of stable syndicates and interest groups. Thus where an owner is forced to divest from part of his/her holdings, there are ways in which his/her original editorial control might still be maintained, for example through transfer of shares to sympathetic partners, or business allies, or even to family members (i.e. Berlusconi in Italy).<br />The crucial question however should not only be who controls a media company but also how much influence does he wield over the diversity of opinions. To answer this question, the European Commission (through both its 1992 Green Paper on Pluralism and Media Concentration and the 1994 Follow-Up to the Green Paper) and various economic consultants stated that one should have a clear picture of the diversity of media ownership by measuring the audience reached by media controllers. They rejected revenue-based criteria on the grounds that they are only suitable for the measurement of media economic market power and not for assessing plurality. Audience-based measures (daily reach, time spent reading, viewing or listening and so forth), they suggested, are preferable for evaluating plurality because they reveal directly the impact of media controllers on the citizens. What should become clear however, is that neither criteria reflecting media consumption nor financial data will tell us how much influence is wielded over the diversity of opinions. What would measure influence is a very difficult question to answer and no satisfactory criteria have emerged so far. In the absence of a direct measure of influence, crude numerical (counting) measures which attempt to proxy for influence can be used as a substitute. We first examine market shares measured in terms of revenue to the media enterprise. We then move on to assess the potential use of audience-based measures.<br />4.4.3 The Criterion of Revenue<br />The most obvious criterion for the calculation of the individual controllers' shares, which is the analogue to the standard calculation of market shares in any industry, is the reference to total revenues. The general rationale for using the revenue criterion is that if a company produces a large proportion of an industry's total output or turnover, its actions will have a greater impact on that industry's supply, and therefore on the level of prices. In this classic sense, a high revenue share provides an indicator of the firm's market power. Is revenue a good unit for the assessment of the level of concentration in the market place for ideas?<br />To answer this question, it might be helpful to distinguish between two rather different categories of revenue for the media companies, that is advertising revenue and consumer expenditures on the media. One can link these distinctive markets with the two basic arenas of policy concern: the concern over economic equity has focused on the advertising market, while the concern over the market-place for ideas and diversity places grater emphasis on the market for audiences and the programming that attracts them. Advertising revenue reflects advertisers or advertising agencies' willingness to pay in order to reach a particular audience. Advertising then is not considered as an expenditure by families to obtain information; instead, it is considered as an expenditure by companies to acquire audiences. Arthur Andersen's 1994 study pointed out that, while in assessing media firms' economic power advertising revenue may indicate their strength in a market, in measuring media influence it seems to be a poor guide. The reason is that, in the latter situation, the crucial issue is the audience's use of media products. Hence, consumer expenditures on the mass media, whether that is subscriptions to TV channels or revenue from newspaper sales, seem at first glance to be more suitable for measuring concentration in the political and cultural market.<br />By focusing on consumer expenditures on the media however, we will not be able to determine plurality in a market. The money that the public spends on various media does not reveal either how the public uses the media products or how people's beliefs are influenced by the media they choose. In any case, both advertising revenue and direct consumer expenditures on the media have a major defect as a way of measuring concentration of influence. Some media such as pay-per-view channels or most of public service broadcasters across Europe obtain no advertising so that, by the advertising revenue standard, they would gain no share in the market-place. Likewise, other media such as the free press and commercial broadcasters depend entirely on advertising revenue and consequently, by the consumer expenditure criterion, they would gain no share in the market-place.<br />Another indicator of a media operator's influence might be his/her share of the total expenditures on creating programmes or publishing. Many media as well as disseminating information produce much of the material they distribute. Other media spend huge amounts in acquiring programmes and bidding for rights and talents. It is logical thus to raise the question of whether expenditures could be a more suitable criterion for assessing a media owner's influence on the public. The obvious drawback when using this standard is that it yields rather narrow interpretation of market power, as given by the channel's share of the industry's spending. There are two additional disadvantages. Firstly, there is the problem of data availability since few companies publish expenditure figures. Secondly, the expenditure standard tends to emphasise the weight of public service broadcasting organisations where expenditures represent a larger proportion of revenues. Since the vast majority of public channels are not profit-maximisation oriented and cater for creating innovative programmes, the expenditure criterion would grant them untrue influence.<br />Another hypothesis is to calculate a media proprietor's influence on the basis of cross-subsidisation, that is, an operator's ability to finance his loss-making sector with revenues extracted from other businesses he may possess. Rupert Murdoch, for instance, decided in 1993 to cut the price of his three newspapers in the UK - The Times, the Daily Mail and the Sun - at a figure that his rivals could not match, even, in the case of The Times, at a loss (he recently put them up again). Of course, he was able to finance this loss-making paper through his other businesses (BSkyB and other assets abroad). The emerging newspaper price-cutting war is most likely to lead to huge losses for newspaper companies and even closure of some titles. The plurality of British newspapers is apparently at stake and readers face less choice. Another example is the use of the substantial profits of Robert Maxwell's Mirror Group Newspapers (before 49 per cent of the company was floated in the early 1990s) to invest in other private side firms (see Financial Times, 17 October 1995, p9, 20). Once again however, the selection of the cross-subsidisation criterion bears rather narrow interpretation of market power.<br />Measures based on revenue have been rejected by the Commission through the 1994 Follow-up to the Green Paper on Pluralism and Media Concentration in the Internal Market (COM(94) 353 final) on the grounds that they can only apply to the measurement of economic market power and not to assess influence. As stated in the Follow-up to the Green Paper, the disadvantage of the market share standard is that 'it duplicates the efforts of competition law, which uses this criterion to assess the behaviour of economic operators among themselves (taking account, in particular, of earnings) and not to evaluate as such the impact of media control on the 'citizen consumer\" (COM(94) 353: 42). Indeed, as Robinson et al. (1995: 40) argues, if the European Commission were to legislate in this way, it would be prejudging work that is elsewhere left to the empirical investigations of the Merger Task Force. There would therefore be a potential conflict between one part of the law and another. Revenue-based measures have also been dismissed as inappropriate in the Andersen's study which also pointed out that revenue can only be used to evaluate a company's economic power.<br />Nevertheless, it may not be right to reject revenue as a measure of pluralism so easily. A significant revenue-based market share is related to economic strength and normally (although not necessarily) reflects plurality in the market indirectly. Revenue may serve as a useful element to be considered because it captures power in the market place to acquire rights, for example, in a way that other criteria (such as audience share) fail to do. Of course, mapping the breakdown of media revenues could not be considered as the only criterion for measuring pluralism, not least because of the complexity of definition of pluralism. However, it seems useful to consider the possibility of collecting revenue data for regulatory purposes. The advantage, as Lange (1995: 8) reminds us, is that this kind of system could be based on hard data (i.e. company turnover) already available and published in the general framework of general company laws. In addition, revenue-based measures have the advantage of providing an appropriate unit for measuring concentration across media. In this way, they are suitable for aggregation and, combined with direct measures of audience (as it is argued below), they could capture in principle an operator's ability to influence public opinion. <br />4.4.4 The Criterion of the Media Consumer<br />As stated in the 1994 Follow-Up to the Green Paper on Pluralism and Concentration in the Internal Market, in addition to the question of who actually controls a media company, one has to find out the actual number of consumers reached by a media company or 'controller': 'when it comes to the question of pluralism in the mass media, it is necessary to determine the effects of concentrations on the \" citizen consumer\" . We summarise here the arguments listed in the 1994 Follow-Up to the Green Paper to support the use of the audience standard: 'compared with the criteria based on company share holdings, the media consumer criterion does not duplicate the efforts of competition law, which uses the above standard to assess the behaviour of proprietors among themselves. Moreover, compared with the criterion of the number of channels that can be controlled by an individual, the consumer unit is more suited to the expected development of digital compression technology which will lead to a proliferation of channels. Whether a media company controls two or a hundred outlets will then no longer be decisive in itself in determining the granting of licences. The important point will be how many consumers are already being reached (within a relevant market) by these different outlets (TV, radio or the press). In this respect, this criterion has also the advantage of covering all branches of the media, whether that be newspapers, television, magazines, radio etc., thereby being a good indicator of evaluating multimedia concentrations. Furthermore, it could be economically more sensitive and less harsh in the sense that it could facilitate the creation of a large number of specialised channels which normally do not attract large audiences. The criterion of the number of channels controlled by a single person could prevent this and therefore limit the level of pluralism in a particular area. Likewise, the audience standard may be more effective as regards preserving pluralism since a single channel with a large audience could also frustrate pluralism' (COM(94) 353: 42).<br />Despite these arguments, the criterion of audience share has met a lot of scepticism by both the media industry and some academics. In their response to the Commission's complementary questionnaire asking specifically about the possible use of the audience criterion, the majority of media firms and media representatives expressed reservations about its use. Those reservations regarded both the adequacy and the feasibility of such a criterion. The European Broadcasting Union (EBU), for example, stated that it would not be possible to apply the same audience thresholds in different contexts (i.e. different kinds of media and different distribution areas). Indeed, in smaller countries and restricted language areas there may be room for one or two generalist channels, whereas in large markets an audience share of, say, 25% would already appear very high. In addition, it mentioned that pluralism must be safeguarded within each relevant media sector. Even within one and the same media sector, for example television, further distinctions would have to be made between pay-TV and television free at the point of reception, between generalist and thematic channels, and between different kinds of thematic channels (news, sports, culture, etc.). In view of such differences, EBU argued that, in case 'audience' was selected, thresholds should distinguish between media and be fixed at the national/regional level only in accordance with the principle of subsidiarity. This is linked to the question of the proper definition of the market, to which we have previously referred.<br />Many media organisations argued that the use of the audience criterion would punish growth and penalise the success of a proprietor. The EPC (European Publishers Council) and ENPA (Association Europeenne des editeurs de journaux), for instance, pointed out that if audiences grow in response to a company's commercial success, then the effect would be to restrict that success, something which seems pointless when the aim of any new regulation should be to stimulate the market and give media firms more opportunities in the Single market. Others challenged the Commission's argument that the use of the audience criterion will encourage the establishment of thematic channels. EGTA (European Group of Television Advertising) and ITCA (Independent Television Companies Association) stressed that there is a strong possibility that the opposite effect will result, as a preliminary question must be asked once again about the definition of the market. What is the position of, say, a specialist channel taking 50% of the viewers in its niche market but an insignificant overall share? Even those in favour of a possible use of the audience standard (i.e. Independent Television Commission) questioned its robustness and its capability of applying equally in different circumstances and between different media, or argued that it must be accompanied by a 'harder' unit such as revenue.<br />There has also been opposition to the application of audience shares for assessing media concentrations from the academic side. In an article in The Independent, Barnett (1995a: 15) expressed strong disapproval of the 'share of voice' argument and criticised any kind of proposals that attempt to shift the emphasis from ownership. Ownership for him still matters, for different owners have different versions of the world and different attitudes to news. It is then wrong to replace existing rules based on the number of outlets an operator can possess by rules based on newspaper circulation, television viewing and radio listening. In a later paper in Public Policy Review, Barnett (1995b: 1-5) proposed that regulators should allow owners to possess media outlets in just one sector and bar completely cross-ownership. In his words, 'we simply need to maintain a clear distinction between them [media sectors], and to insist that those who wish to own newspapers should continue to be debarred from owning television channels and vice versa' (ibid: 5). Similar views had been expressed earlier on by Curran and Seaton (1985: 342-3). Lange (1995: 7-8) criticised the pertinence of the audience standard from both a technical point of view and from a conceptual point of view. The technical problems he saw are firstly that there are important margins of errors of audience and circulation measurements, and secondly that these measurements are meaningless unless they are done on different geographical scales. From a conceptual point of view, he challenged the assumption that audience measures are more appropriate for evaluating pluralism. In fact, the 'audience' proposal analyses the question of pluralism in terms of consumption of information and not in terms of offer (i.e. in terms of possibilities to access a great diversity of information). Audience measurement is a quantitative analysis of consumption by readers, listeners or viewers, developed for market analysis purposes. Pluralism means basically 'diversity of information'. However, this is a qualitative objective both in terms of offer of information and in terms of reception of that information. Audience research, the argument follows, does not measure either the diversity of offer or the quality of reception.<br />The audience criterion thus has raised a number of serious questions. This has prompted the Commission to launch a feasibility study examining the comparability and compatibility of the different national systems of audience measurement, the types of audience measurement that can be used and the possibility of using a single composite criterion or of combining several types of measurement depending on the configuration of the media in question. The study distinguished between monomedia audience maps (where only one medium is under review) and multimedia ones (where TV, radio and newspapers are combined). In constructing monomedia audience maps, it recommended the use of audience share for television, circulation share for newspapers and listenership share for radio measurement purposes. In the case of multimedia maps, it recommended using the daily average contacts as a criterion that satisfies the requirement that the measures are equivalent across media. The analysis concluded that, although there were disparities among national audience measurement systems, the use of audience measures to examine pluralism across the EU is both valid and practically feasible. It takes the consumer as the point of reference and will therefore be of maximum effectiveness in relation to the objective sought, namely that of serving the interests of the media consumer. One has to remember though that the study was commissioned by DG XV which would very much want a definite statement on the application of audience measures as indices of pluralism. It comes as no surprise that the GAH Group's findings are in line with the DG XV's wishes, as many times although not necessarily the employee wishes to satisfy the employer.<br />In fact, there has been much criticism of the GAH Group's work. Leighton Andrews (1995: 26), head of Worldwide Corporate Affairs BBC, stated that the study 'simply demonstrates that comparable audience data is available across the EU, it does nothing to make the argument as to why audience share is the most suitable measure of dominance'. As mentioned above, the EC's consultants came to the conclusion that despite the differences in national audience measurement systems the measures are comparable and compatible and can be the basis of uniform thresholds to be applied throughout Europe. However, it left a number of questions unresolved. In particular, the question of weighting the audience of different types of media such as TV, radio and the press and the question of where to fix the thresholds. Of course the latter is a matter of media policy but, due to the great heterogeneity of the current European media map, it will be very difficult (if not impossible) to find a political consensus on such a threshold at a European level.<br />The first problem - the establishment of market shares based on the number of people reached by each medium - is equally difficult to solve. How is it possible to create an overall market share of, say, a company holding 20 per cent of the TV market, 10 per cent of the national newspaper market and 5 per cent of the radio sector? As Robinson et. al. (1995: 59) argue, 'we need to find some robust and objective method of constructing a weighted average of these three market shares to arrive at a single figure for the media market as a whole'. The absence of a weighting will lead to false results as the GAH Group's analysis reveals. The EC's consultants recommended the use of daily reach as an indicator of influence. As demonstrated by the National Economic Research Associates the use of daily reach in the UK media market gives a high weighting to television and radio and a low weighting to the press (see Table 4.3). According to NERA, this is because most people have ready access to all TV and radio stations every day, and can switch between them in deciding what to watch or listen to. Nevertheless, they generally choose one or two newspapers a day and cannot pick and choose content from one or another. This results in a much smaller number of daily contacts for the press, which is not necessarily a measure of relative influence.<br />Table 4.3<br />Use of Reach Weighting in the UK Media Market (1994)<br />SectorWeightingTotal Daily Reach ContactsTelevision55.3%133,815,268Radio18.9%45,714,000National Press15.6%37,578,186Regional Press10.2%24,715,286Total100%241,822,740<br />Source: NERA, based on data supplied by BARB (television), RAJAR (radio) and Pressad/JICREG, Newspaper Society, ABC, NRS (newspapers)<br />Note: TV reach is based on people over the age of 4; other reach figures are based on adults over the age of 15<br />Undoubtedly it will be difficult to develop a suitable and generally accepted method of ascertaining these figures. One such method, taking into account the different sizes of, and power wielded by, different media, has been proposed by the British Media Industry Group, a consortium of four big newspaper groups (Associated Newspapers, the Guardian Media Group, Pearson and the Telegraph Group) set up in August 1993. Their study uses the well-established audience figures to calculate market shares for television and radio, and circulation figures to assess market shares for newspapers. In calculating overall share of voice in the UK media market, they recommend that radio listening be down-weighted by 50% compared with newspaper circulation and TV viewing. BMIG justify these weightings on the grounds that radio has lower impact on diversity of view issues, either due to the prevalence of music-based radio formats, or because radio is normally consumed in parallel with other activities. <br />According to the study thus 1% of newspaper circulation or TV viewing is deemed to have more impact on diversity of view issues than 1% of radio listening. So their suggested weighting is 28.6% for television, 28.6% for national press, 28.6% for regional press and 14.3% for radio. Their findings, shown in Table 4.4, give BBC the highest audience share (19.7%). The next top two places are occupied by News International on 10.6% and ITV network combined on 9.4%. The top 8 media firms have 57.4% of national share of voice. The main advantage of the BMIG approach is that it treats television, national newspapers and regional titles equally. The inclusion of regional papers, in particular, is a positive step towards realising that it is not only the national market that matters but local/regional monopolies should also be tackled. Further, the selection of the most appropriate units to measure market shares for each medium (circulation for national and regional papers, share of viewing for terrestrial TV, cable and satellite, and share of listening for regional and local radio) provides a more or less accurate picture of market shares within each media sector.<br />Table 4.4<br />Media Concentration in the UK Under Audience Share <br />Analysis (1993/94)<br />Media Groups% share of audienceBBC19.7News International plc10.6(ITV Network)*9.4Daily Mail & General Trust plc 7.8Mirror Group Newspapers Ltd7.6United Newspapers plc5.7Carlton Communications plc3.1Channel 42.9Total as a % of papers/TV/radio 57.4<br />Source: BMIG (February 1995), Robinson et al. [London Economics (LE)], 1995<br />Note: Radio down-weighted by 50%<br />* LE estimate using BMIG methodology<br />This last point however becomes a disadvantage when it comes to measuring aggregate shares. Robinson et. al. (1995: 59) argue that it is inconsistent to measure shares in the printed media market on the basis of circulation (which is essentially a household based measure) while using audience shares (which count individual viewers) for the electronic media. As the EC's consultants showed, it is more consistent to use readership figures when it comes to comparing different media. Another problem with the study is that it does not give any weight to consumer periodicals. A further criticism to the share of voice approach is that it, including the BBC and Channel 4, underestimates the way the rest of Britain's media ownership is concentrated in so few private hands (Peak and Fisher, 1995: 227). The most serious problem with the BMIG approach though is their unreasonable decision to down-weight radio listening by 50% compared with the other media. It is true that radio lost much of its appeal after the appearance of television. It is also true that radio is mostly an entertainment medium and there has been a proliferation of music-based radio formats. It is finally true that radio can be listened to while people are engaged in other activities. All these points make radio seem as a 'secondary medium'. But they do not prove that radio is less influential and of less importance to diversity of news. The influence of news-based radio formats, broadcasting news, current affairs and other programmes devoted to commentary and opinions is considerable. Even music-dominated radio stations have regular news bulletins normally announced every hour. No matter whether one accepts that the influence of radio is less than that of television or the written press, BMIG's proposal to down-weight its influence to 50% is both arbitrary and gives no objective justification.<br />There has been no objective basis so far for the choice of weights. It has proved difficult to identify satisfactory criteria for a measure which successfully aggregates media with very different patterns of consumption and use. This difficulty is inherent in another study, the Arthur Andersen's analysis, which uses a single unit of measurement for the cultural market. In order to measure company shares in that market, Shew used the time that the public actually spends on media. Here, the potential of the mass media to influence public opinion is assessed best by examining how people distribute their time over them. Table 4.5 shows the results of applying this methodology. What is interesting is that the amount of time spent watching BBC and ITV is nearly twenty five times as great as that spent reading News International's newspapers. In fact, there is an obvious exaggeration of the impact of TV compared to the press, since no other press groups appear in the top seven.<br />Table 4.5<br />Media Concentration in the UK Under Time-use Analysis (1993)<br />Media Groups% of time spent 'consuming'BBC44.1(ITV Network)*25.4Carlton Communications plc6.9Channel 46.2Granada Group plc4.1Capital Radio plc3.4News International plc3.4Total as % of the press TV & radio93.5<br />Source: Shew (Arthur Andersen) 1994, Robinson et al. (LE) 1995<br />* LE estimate using Arthur Andersen methodology<br />In the Andersen's study each controller's individual share of the total media market is calculated as the share of his/her medium in the total time spent by the public consuming media. This definition may be both too vague and too limited. It is vague because it does not tackle the nature of the media consumed. For instance, how much of the time watching TV is spent viewing news and how much sports? Moreover, one could not treat equally time spent reading The Guardian and time spend reading The Racing Post (both national newspapers). A greater value should be added to the former title due to in-depth coverage of current affairs. Regarding the different media, it may also be false to use the time standard for assessing their impact on the public. One hour TV watching (which is normally passive) may not be comparable to one hour newspaper reading (which requires greater effort). It seems safe to argue that the influence of the latter is greater. Equally, it is hard to compare the influence of a 15-minute news bulletin with that of a newspaper headline on the same topic which is on display the whole day and can be read by thousands or even millions of passers-by. In the latter case, the item is most probably being watched for a fleeting moment, but its influence lies in the fact that it is being watched by many people. The analysis is also too limited. Although it includes reading periodicals, it excludes viewing of pre-recorded videos, which are a direct substitute to television viewing.<br />These fundamental problems aside, Shew argues that where there is cross-ownership of different media by the same controller, his/her share is arrived at simply by summing each of his/her channels' or titles' shares in the total time spent by the public consuming media. Therefore, if owner A controls both channel X and paper Y, which take up 20 per cent and 10 per cent respectively of the public's total time, then A's share in the hypothetical total cultural market is said to be 30 per cent. Similarly, in case he/she controls only a fraction of a channel or title, his/her share is calculated as that same fraction of the time devoted by the public to that particular channel or title, relative to the total time spent consuming media: if controller B has 25 per cent share in channel Z, which is found to account for 40 per cent of the public's total 'consuming media' time, then B is said to control 10 per cent of the total market for ideas.<br />This methodology is troublesome and biased towards both television relative to the press and national channels and titles relative to regional/local ones. It artificially inflates the market share of television and particularly of national channels, while it understates the relative importance of the press and of all regional/local channels and titles. This is because the average weekly hours spent watching TV are nearly seven times as great as those reading the press (according to the study, it is 24.1 hours for TV against 3.5 for papers), so the apparent market 'influence' of a TV controller will be vastly more significant than that of a newspaper publisher if measured in this way. If, in addition, the issue of the effective geographical coverage of a channel or title is ignored, it is hardly surprising that the total time spent with a regional/local channel by all viewers in a particular area will be small in relation to the aggregated time spent by a national audience watching a national channel. To illustrate, watching Carlton television is not an option for those who live outside the London area, and it is therefore meaningless to calculate Carlton's national market share as the time spent by anyone watching Carlton over the total time spent by the public consuming all media nation-wide. It is obvious that Carlton's share will be dwarfed relative to media with national coverage such as the BBC.<br />A partial refinement of the Arthur Andersen approach has been provided by London Economics. That study attempted to weight the respective UK media shares by some measure of the inherent impact of the medium. The analysis proxies this by the amount that advertisers are prepared to pay to reach a consumer audience through a particular medium. It formed the following hypothesis: if advertisers pay according to the relative impact they believe a particular medium has, then the relative price per consumption minute across media provides a market-based approach to cross-media weighting. This is shown in Table 4.6, where the analysis estimated the relative cost of advertising across media in Britain using relative advertising cost per viewer minutes spent on the media. The advertising-based weightings suggest that the printed media are significantly more important, and radio far less significant, than in the Arthur Andersen paper.<br />Table 4.6<br />Relative Advertising Costs in the UK Media Market (1994)<br />Medium£ per hour* per capitaTelevision1.16Radio0.22Newspapers2.52Cinema2.66Magazines2.82<br />Source: Advertising Association handbook, 1994; Arthur Andersen, 1994; London Economics analysis, 1995 <br />* of viewing, listening to or reading advertisements. London Economics analysis makes the assumption that people spend the same time on advertisements for each hour of reading newspapers and magazines, as for each hour of viewing or listening to commercial television or radio<br />The London Economics study re-calculated the Arthur Andersen estimates of audience share using a methodology in which the time devoted to each media sector is given a weight which reflects its value to advertisers (as shown in the above table). The effect, as presented in Table 4.7, is to enhance the market share attributed to the press at the expense of television and radio. In particular, the BBC's share falls from 44.1% to 28.3%, Capital Radio's share falls from 3.4% to 0.3%, whereas News International's share rises from 3.4% to 12.5%. These results demonstrate how sensitive the final outcome is to input assumptions. This methodology however, as London Economics itself admits, remains subject to the basic objections made above to Arthur Andersen approach, namely the problems with product and market definitions. For example, the figure of Capital Radio would rise if the study took the London metropolitan area as the relevant market - which would be appropriate for Capital Radio, as metropolitan London is the limit of its broadcast range.<br />Table 4.7<br />Alternative Measures of UK Media Concentration<br />Media GroupSimple time use*Weighted time use**BBC44.1%28.3%Carlton Communications plc6.9%9.8%Channel 46.2%5.5%Granada Group plc4.1%6.9%Capital Radio plc3.4%0.3%News International plc3.4%12.5%<br />Source: London Economics, 1995<br />* Arthur Andersen analysis<br />** Using time data from Arthur Andersen paper modified to reflect the different value of time to advertisers in different media<br />4.4.5 Critique of all Audience-based Measures <br />Mono-media market measures<br />Overall, there are a number of aspects that should be considered in methodologies which set out to examine mono-media or multi-media cultural power in terms of audience. The relative influence of a communication vehicle depends on a number of different factors. It obviously depends on the number of people it can reach - a newspaper sold to one million people will tend to have more influence than one which reaches only fifty thousand people. However, the fact that a newspaper is bought (something that can be measured) does not necessarily mean that it is read, or read by only one person (readership is much harder to measure than circulation). Furthermore, a title that is read carefully will have more influence than one which is merely skimmed, and whereas we can use surveys to find out how many people read each newspaper, it is impossible to know how carefully they read them, or what they read in them. Serious news, current affairs and political comment compete for the readers' attention with sport and the TV listings. Moreover, it is difficult to establish whether publications with more editorial content count for more influence. Some newspapers (particularly the Sunday titles) are fat because they offer readers a choice of specialist material which is not opinion-forming. Finally, there is the problem with the weekly and Sunday press. Should the latter be given one sixth the weight of the daily press, in proportion to the average daily reach of each? There are two problems with this assumption: first, weekly papers tend to have a longer read period; second, multiplying the circulation figures for daily newspapers by six to produce a weekly circulation figure and putting them together in terms of a weekly market (as the BMIG did), presupposes that daily and Sunday papers are entirely substitutable for readers. This however is a strong assumption to make. The Sunday Times, for example, dominates the UK Sunday quality newspaper market but its sister paper The Times has a much lower market share of the quality newspaper daily market (see ABC figures, June 1996).<br /> <br />Similarly for the TV programme to have any influence, the set must be switched on and tuned to the right channel, which can be measured electronically. In some countries, like the Netherlands, Germany and Britain, there are also surveys which establish who is watching it and for how long. But the degree of attention with which it is being watched is almost impossible to establish. Another dimension to the problem concerns who is in the audience. A programme or newspaper article that reaches a single person may have a lot more influence than one which reaches a million if the single person happens to be the President of the State. Programmes and articles which regularly reach opinion formers and policy makers such as members of the parliament, editors and journalists clearly have influence which cannot be measured just by counting heads. Thus what counts is not simply how many people watch (and read and listen) as the BMIG assumed, but who watches and what they watch. <br />'Reach' in general is less attractive as a measure of the influence of broadcast media. People who contact a broadcaster may rapidly switch away from it, in contrast to the press. People typically contact several TV channels or radio stations each day to find out what is on. That results in a measure that shows little difference from channel to channel. In addition, people spend more time with some channels than others in proportions that are not reflected in all national daily reach statistics. On the other hand, people have less opportunity to flick between press publications than between broadcast channels. Therefore, the total number of daily contacts for television and radio is much greater than the total number of daily contacts with the press. This does not reflect relative influence. What does reflect is the much shorter decision period the audience has in deciding where to spend its time in the different media.<br />Time spent consuming media is intuitively attractive as a measure of influence. The distinct advantage of this approach is that time use is a common denominator across all media. Thus if there is any validity in the basic proposition that the influence of a medium of communication is proportional to the time taken to absorb the message, the time use measure can be used to assess concentration across all media. Time can be used as a measure of market share in each medium, and by simply summing the time use across all media we can calculate an overall market share. Audience time may be suitable for the measurement of the broadcast media influence because programmes are viewed at the same speed by everyone, so that time may be a good indicator of content taken in. The problem is that the time that people spend to watch television or listen to the radio is certainly not the most suitable indicator to establish impact over them. Time spent watching or listening to the media does not necessarily reflects influence. Apart from this problem, which is common in all audience-based criteria, there is the issue that television and radio, in particular, can be a backdrop to household activity. High audience hours may arise as a consequence of the broadcast being used as background to other activities.<br />The methodological weaknesses of audience time measures are more apparent in the press for a number of reasons. Firstly, reading speed is variable, so that reading time becomes less attractive as a measure of the influence of the print media. Secondly, the newspaper reader can focus on what is of interest and value for him/her and ignore the rest. According to how much 'filling' the reader chooses to edit out, similar levels of 'influence' might be achieved in very different lengths of time spent with the publication. Thirdly, and according to Arthur Andersen study described above, the amount of time spent reading national newspapers is nearly seven times less than time spent watching TV. This figure may substantially exaggerate the impact of television compared to the printed medium. However, national newspapers have an impact that lasts longer than the time spent reading them. Newspaper headlines are seen by, and influence, millions of people who may not buy them. But it is the newspaper headlines that largely determine the current affairs agenda at the start of each day, and that influence persists even though news sources of information are obtained from television in the evening. There is indeed extensive discussion of press opinions in other media, perhaps because the press is permitted to give opinion whereas broadcasters (particularly public service broadcasters) are constrained to treat news 'impartially'.<br />Having explored, to some extent, the pros and cons of the audience-based criteria, our impression is that time measures are preferable for broadcast media than reach measures. The nature of TV and radio means that reach or contact with a station can happen almost incidentally as people switch between channels searching for one that captures their attention. Data for television viewing is readily available and reliable, whereas radio listening data are adequate for the majority of the stations in all European Union countries (see ESOMAR, 1995). Concerning the print media, daily reach might better reflect the influence of the press than time spent reading, as for the most part people buy newspapers with the intention of reading them. Also data on time spent reading is either non-existent or unreliable. The choice between circulation or readership as the reach measure is difficult. Arguably, the influence of newspapers depends more on the number of readers than on the number of copies sold, just as the influence of TV programmes depends more on the number of viewers than on the number of sets tuned. Therefore newspaper readership would be a theoretically better measure of influence of the printed media. Readership data though is not available for Portugal, Greece and Northern Ireland and is not collected on a regular basis for many regional titles. This would create insoluble problems in case we want to establish a Europe-wide regulation based on this criterion. By contrast, circulation data is the most robust and comprehensive audience measure for the press. It is thus a more established currency in the industry. In practice though, and as the GAH (1994), BMIG (1995) and NERA (1995) studies found, percentage share of newspaper circulation is not very different from percentage share of readership.<br />The aggregation of media market measures<br />Once a satisfactory method of calculating audience shares within each media sector has been devised, some method is required to compare audience shares across media. A likely policy aim of deregulating media ownership rules at national or even European levels is to allow cross-media ownership in a manner which is proportionate. Put it another way, the permitted holdings in one sector of the media would be subject to the size of holdings in other sectors of the media. However, there is no obvious solution to the problem of creating an overall market share. If a particular company has 15 per cent of the television market, 20 per cent of the press market and 5 per cent in radio, what is its share of the total media market? We have to devise a method of calculating an average market share of the whole media. If shares in each sector are measured in the same dimension, for example a consistent reach statistic (as recommended by the GAH Group), then there is a natural method of aggregation by adding all the shares. If, however, market shares within the media are measured in ways which are not directly comparable (and this would be wiser since, as we explained above, some measures are more suitable for the press and others more preferable for broadcast media), then a weighted average of all market shares must be constructed in order to arrive at a single figure for the media market as a whole.<br />A weighting can be based on some objective data which, one might assume, compares the relative influence shares of the media on an aggregate basis. A weighting could also be constructed on policy grounds, reflecting a policy view of the relative influence of the media, which may or may not be based on some objective information. As shown in great detail above, the BMIG falls down on the arbitrary nature of its weighting system. Their approach uses audience figures to calculate market shares for radio and television and circulation figures to calculate the market share for newspapers. They then construct an average share of the overall media market by assigning weights of one to television, national and regional newspapers, and 0.5 to radio. However, there is no objective justification for the arbitrary figure of 0.5, especially as it is far from obvious that radio has a low level of news and current affairs compared to other media. The London-based Capital Radio FM and the nation-wide Virgin Radio, for example, both music-based radio formats, have regular news bulletins as well as other programmes devoted to commentary and opinions. This is a pattern followed by most radio stations. The other problem with the approach is that it is not consistent to measure shares in the printed media market on the basis of circulation (which is basically a household-based measure) while using audience figures (which count individual viewers) for TV and radio. <br />As the European Commission's consultants stated, it would have been more consistent to use newspaper readership, rather than circulation, as the basis of comparison. So they recommended the use of daily average contacts as a measure of the overall market share. This method however results in a much smaller number of daily contacts for the press (see Table 4.3 above), which is not necessarily a measure of relative influence. A weighting could also be based upon the hours of use or audience time of the media (Arthur Andersen's approach). The idea of using hours of use as the basis of a weighting appears intuitively appealing until one sees what figures it implies (see Table 4.5 above). People spend very much more time with television and radio than with the press. Most people would not find it credible that the press has such small influence in relation to other media. In addition, the data for the average time spent reading the press is well known to be unreliable. But the major problem associated with this approach is that it seems intuitively unlikely that time spent consuming media directly proxies the influence of a particular medium. The influence deriving from the intrinsic impact of the medium and the influence due to the number of minutes for which the medium is consumed are being conflated.<br />It seems to us that a much more appropriate way of measuring overall market shares and possibly influence across media would be to weight the respective audience shares by advertising expenditure. By examining the behaviour of advertisers, that is, by examining how much advertisers are prepared to spend on which medium, we will have a more thorough picture of the relative influence of the media. Advertisers pay according to the relative impact they believe a particular medium has. They buy space in national and regional newspapers and time on television because of beliefs regarding the influence of those media. They also recognise that it is not just the number of viewers, listeners and readers that matters; it is who they are. Therefore, advertising expenditure on the media provides a reasonable market-based approach to cross-media weightings. This can be shown in practice. London Economics analysis, for instance, attempted to weight the respective time spent media shares (Arthur Andersen's approach) by the amount that advertisers pay to reach consumers through a particular medium. The chief merit of the London Economics approach, the results of which are shown in Table 4.7 above, is that it finds that the printed media are significantly more important than in the Arthur Andersen analysis. The time devoted to each media sector is given a weight which reflects its value to advertisers and the effect is the enhancement of market shares attributed to the press at the expense of the electronic media.<br />Data robustness<br />A necessary condition for using audience measures to assess concentration levels and pluralism within local, national or supra-national markets also depends on the quality and comprehensivity of the surveys in use. In an ideal world, surveys would measure every single television and radio channel, and every single newspaper title uniformly and with equal precision and accuracy across the European Union. The existence of regular surveys is hardly a problem as all EU Member States (with the exception of Luxembourg) possess their own national surveys for TV, press and radio. There is also a certain level of continuity of data collection in the EU, as all current TV surveys collect audience data on all days of the year, and the majority of press and radio surveys collect data across at least ten months of the year (see ESOMAR, 1994, 1995). But it is also essential that audience surveys are comprehensive in their coverage of the media. A specific limitation of television audience measurement is the restriction of almost all national surveys to measuring viewing in the home, although some (such as Italy and Sweden) measure guest viewing and treat it as a substitute for out-of-home television viewing in other TV homes. Regarding press readership surveys, their main drawback is their limited coverage of titles. In the case of newspapers, readership surveys invariably cover all the national distributed titles, but there are some big gaps in the coverage of regional titles in France (see ESOMAR, 1995). By contrast, circulation data are quoted for almost every newspaper and virtually all are audited. Regarding radio surveys, the one instance where a radio survey appeared to give well below full coverage of stations was the Dutch Intomart survey, which excluded the non-commercial local stations (ibid.).<br />There is also the issue with the universality of coverage. Almost all national surveys measure media consumption across national universes. However, the Greek AGB Hellas (television) covers urban areas only, whereas the Portuguese Bareme (press and radio) covers mainland Portugal. Geographical restrictions, as between urban versus rural areas, could have a significant effect on measures of total audience, where significant variations exist in media availability that coincide with the geographic restrictions. The problem really only arises for some media in Greece and Portugal. In addition, there is the issue of harmonisation of measures. It is important for supra-national framework legislation that audience measures are comparable across borders. The GAH Group report concluded that comparability is not a significant obstacle for the construction of audience maps. The European Commission's consultants found that national surveys within each media category all employ the same fundamental concepts of media consumption, even where the methodologies vary (as with 'Recent Reading' and 'First Read Yesterday' measures of reading). And although TV surveys are divided between those which treat viewing as presence in room with TV set switched on and those which require individuals to be present and watching, the end differences are almost certainly negligible. So cross-border comparisons are broadly feasible.<br />In future though, with the expected increase in the number of media outlets, it is possible that the quality of the data will decline unless the industry spends more money on increasing the size of survey samples. Perhaps more crucially, as new opinion-formers operate through the Internet, it will become increasingly difficult to measure their 'audience'. While it is possible to calculate the number of 'hits' on an individual Web site or bulletin board, it would be quite unrealistic to know how long each hit lasted, or what percentage of them came from outside the country. As technology develops, cross-media measurement based on audience share will become more and more difficult to realise unless further research is conducted. There are some encouraging signs though. The Audit Bureau of Circulations initiated a plan in January 1996 to measure the readership figures for UK on-line titles and the results of its trials should come out later this year. The plan is to provide advertising agencies with a certificate showing how many people have registered for the site and how many have visited it in the last month. A bit of software, running on the same computer as the electronic publication, will make random checks and e-mail some of those who have registered to make sure they exist (see the Guardian, Monday February 5 1996, pp. 12-13). It remains to be seen whether the trials will succeed.<br />4.5CONCLUSION - ALTERNATIVE PROPOSALS<br />In this chapter, we tried to analyse critically different methods of measuring media concentration and influence. The relevance of the existing counting criteria (i.e. rules based on the number of channels controlled by a single operator) was briefly reviewed, followed by an analysis of the audience and revenue-based units. Although it became evident that numerical rules are not appropriate on their own to measure levels of concentration in the multimedia era, it might not be right to abandon them altogether. Many, including the European Commission, argue that today it does not matter how many outlets one owns, but how much revenue or audience he/she reaches. However, we will agree with Steven Barnett that ownership still matters. It seems to us that it is not acceptable, from a pluralism point of view, for one owner having the majority of a country's opinion-forming TV channels broadcasting news bulletins or current affairs, even when these channels reach small audiences and obtain negligible amount of commercial revenue. This is because, under these conditions, the proprietor will be free to exercise excessive power to advance his/her own political and socio-cultural views. The controller of a large number of titles and/or channels disseminating political commentary can easily pursue the values and political beliefs he/she espouses. He/she may also be in a position to influence the outcome of an election, as the Greek example described above clearly indicates. In addition, he/she may cater only to bourgeois tastes and not to others so as the latter cannot gain public exposure.<br />Such influence may not only be dictated by political and socio-cultural considerations but also by commercial or other considerations. Monopoly or excessive control of a number of channels may allow proprietors to promote a particular economic philosophy (foster, for instance, positive attitudes towards free enterprise and unbridled competition). Although the question of whether media controllers' influence over what people think is much disputed (precisely because influence is hard to measure), another example may help us to form a clearer opinion of the matter. In 1989 in Britain, the Broadcasting Research Unit (BRU) examined some opinion data collected by the then Independent Broadcasting Authority (now Independent Television Commission) on whether people thought the licence fee was the proper way of paying for the BBC. Murdoch's newspapers (and The Sun in particular) opposed the BBC and its licensing system. They did so because Murdoch was preparing to launch his satellite channel BSkyB at that time and wanted as few competitors as possible in order to provide a major boost to News International's UK television interests. The BRU found big differences in opinion between readers of News International newspapers and non-readers: 37% of News International's readers approved of the BBC's funding system while 56% disapproved; but amongst non-readers, the position was reversed [nearly half approved (49%), whereas 44% disapproved]. According to Barnett (1993: 17) one cannot be absolutely sure that the result was due to Murdoch papers' editorial line which developed a general feeling of opposition to the licence fee between their readers. Barnett however adds that such enormous differences can scarcely be coincidental.<br />Media concentration in political coverage and commentary (where pluralism matters greatly), but also in philosophy and opera criticism, science and business, entertainment, etc. is thus not tolerable. Performances in the media affect such varied aspects of national life as dress codes, political opinions, religious beliefs and so on. At least potentially, undue concentration in cross-media ownership might lead to an over-enthusiasm for particular parties, or to excesses of approval or disapproval of particular forms of dress, among other things. Television, radio and the press have a unique role in the free expression of ideas and opinion, and thus in the democratic process. The main objective must therefore be to secure a plurality of sources of information and opinion and a plurality of editorial control over them. Public policy has to ensure that viewers, listeners and readers benefit from the multi-media revolution and that a provision of a number of opinions from different sources are available. Ownership restrictions should thus be maintained. It is imperative though that they are supplemented with audience and/or revenue-based measures, since using the number of channels or titles as a sole criterion is meaningless without some consideration of coverage or financial strength.<br />The two different sets of methods illustrated above (audience and revenue -based) are said to correspond to two levels of measurement of concentration in the information market: the political/cultural or pluralism and the economic or concentration of resources. It has been put forth that revenue-based methods are close to the traditional systems of concentration measurement which permit assessment of the existence of a dominant position (concentration of resources), whereas audience-based methods are coherent with the cultural/political standpoint and can be held to be most effective for measurement of pluralism. However, due to the close relationship between economic power and pluralism, audience figures could also measure market power. In fact, audience-based measures are a form of market share measurement, which is a classic economic measurement. 'Audience' are the equivalent of measuring sales (i.e. market share), which is a classic economic measure of power. Therefore, the distinction between economic measures and cultural/political measures is irrelevant. Both sets of media market measurement assess market power. In the absence of a direct way of establishing 'impact', crude measures based on market power (i.e. criteria about market structure) are used instead. And what the audience and revenue-based methods are doing is in fact that - they evaluate market power.<br />The search for the 'best' criterion for measuring concentration levels in the media sector has been centred on structure and not on content. Neither financial data nor audience figures and certainly not the number of media outlets controlled by the same entity are fully reliable indicators of the controller's impact on public opinion. Such an impact depends to a large extent on the content of the channel/title, which is not taken into account by either of these methods. Ideally, pluralism would be best assessed if account was taken of the editorial characteristics of the medium in question (type of programme or column, editorial opinions, etc.) and studies were conducted to evaluate whether the citizen-consumer genuinely benefits from such diversity. However, it would not be possible to assess the content of different media and consequently their impact on pluralism/diversity. Studies that measure content quantitatively and qualitatively in order to assess whether content is balanced and whether it is pluralistic or biased are not readily available. Given the complexity of these analyses, it would be a very costly and time-consuming exercise (let alone subjective) to carry them out for the special use of the media authorities. Disaggregating media concentration by content area (i.e. classifying individual programmes by type or subject matter) is also impractical since most individual media vehicles present content covering a broad range of issues. Thus, in the absence of such studies, pragmatic solutions based on already available data have to be found and enforced.<br />Having made that clear that neither end-use based indicators nor revenue-based ones can directly measure media firms' impact on public opinion, the question is whether it is possible to isolate one of these measures and use it as a single unit that would best determine media companies' market power (and possibly reveal their impact on public opinion indirectly). The currently favoured audience share measure is not a subtle measure of cultural power. The GAH study prepared for the Commission is flawed because, although it offers a useful summary of the availability of audience measures across the European Union, it does not assess the suitability of audience share data as a measure of 'influence'. The problems with the approach of setting uniform Community-wide thresholds based on the as yet untested new concept of audience share are with the definitions of the relevant market and with market share. Regarding the relevant market, audience share cannot effectively compare media with entirely different intentions and market segments such as a racing magazine and a political commentary. In addition, audience share cannot extend to all substitutable products - time spent watching pre-recorded videos can be as important as time spent viewing scheduled television. Finally, audience share cannot compare media with differing geographical reaches - measured on a local level the market share of a local TV company can dwarf the local share of a national TV channel.<br />Regarding market share across media, it is very difficult to compare the relative significance of many shares in several different media or different media organisations with high shares of a smaller number. It might be the case that an individual has a high aggregate market share which is spread across many media. This however will cause no concern on behalf of the competition or media authorities, since he/she does not have formal control over any of the separate media. Secondly, media with relatively low audience shares could have higher impact than media with high audience shares. The London Financial Times, for instance, with a national daily circulation of just under 300,000, is regarded as more influential than, say, The Sun, with a national daily circulation of more than 4,000,000 (Audit Bureau of Circulations, January 1995). Thirdly, audience shares are constantly changing, and therefore no single limit could be applied with any degree of accuracy in practice (BBC, 1995).<br />Thus audience shares criteria are not a priori better indicators for measuring media concentration than revenue-based ones. The use of revenue market shares of whatever kind (advertising, subscriptions, sales, grants, licence fees, etc.) seems to be less problematic since it avoids the need to find practical solutions to the various problems arising from the use of audience data. Audience data are neither fully harmonised nor equally reliable across Europe (see ESOMAR'S 1994 and 1995 reports). On the other hand, common accounting principles already exist. Financial data for different media sectors are easily comparable. A system based on financial data would make it easier to achieve closer co-operation and synergy effects between the authorities responsible for competition and those in charge of the media. Instead of building up two completely separate systems, this approach would at least make it easier to collect data and to enforce transparency rules. It should also be mentioned that financial data have the advantage of reflecting media consumption - the higher the audience ratings, the greater the revenue from advertising; the larger the number of subscribers, the greater the revenue from pay-TV fees.<br />The main advantages of applying a revenue-based measure are firstly that it is a long established and tested method for measuring market concentration, and secondly that it provides a common currency of measurement across media. However, the NERA 1995 study pointed out a number of reasons why revenue may be an inappropriate measure: a) excludability. The capability of an enterprise to earn revenue depends upon the degree of control it can exercise over those who consume it. Most radio and TV broadcasts do not require the audience to pay, but rather rely on advertising income or taxation, thus not obtaining most of the consumer value of their product. The available technology for making broadcast audiences pay is at an early stage of development and is still quite expensive. The problem of excludability is not present in printed publications, since people usually pay a small amount to buy a newspaper or magazine; b) advertising potential. Broadcasts and publications which cover a wide area, or are targeted at a general audience, earn less income per unit audience than narrow area and special interest broadcasts, because they are less well targeted. This distorts the comparison between different types of supplier in all media. TV advertising revenue is determined, in part, by limits on the number of channels permitted to supply advertising time and by limits on the amount of advertising per hour; c) cost structure. Different media have different underlying costs, and unless supply is heavily constrained (as in the case of television) revenues will, as a broad approximation, tend towards costs through the process of competition; d) choice of influence over income. A proprietor may decide deliberately to forgo income in order to obtain greater influence. A media product which is less suited to high value entertainment might earn less, despite having much influence; and e) relative wealth. A publication aimed at wealthier classes or provided in a wealthier area of a country might earn more than others, without reflecting any greater influence (NERA, 1995: 13-4).<br />Establishing a method of measuring multimedia concentration on the basis of a single unit seems impossible. Some indicators (end-use based) come a bit closer to the concept of pluralism since they focus on the citizen consumer, while others (revenue-based) correspond more with the ways of concentration used in other sectors. But a single audience or financial criterion is hardly conceivable. Apart from the problems that each criterion pose on its own due, this is mainly because none of them has been tested yet for measuring influence. It might be 'safer' then if we took both criteria into account at this stage. The difficulty in finding a system that is applicable to all information services with their different characteristics, leads us to the conclusion that a mix of the two methods described above could provide a valuable alternative. Combining the two types of measurement depending on the configuration of the media in question might provide a quite valid method. A combined test involving advertising and/or subscription revenues and audience shares (together, of course, with ownership restrictions) could effectively measure market power and possibly reveal influence on total media consumption. Regarding commercial broadcasting, for example, the main measurement of market share is by advertising revenue. However, the audience share should also be used as an additional measure, since the revenue by advertising and the prices for advertising spots almost entirely depend on the share of audience for a channel or a certain programme.<br />That both methods are needed for an overall and accurate picture of concentration in the market-place for ideas could be shown through the following example. Let us consider the radio medium whose production and distribution costs are obviously lower than those of television. Assuming that both media vehicles attract the same audience, the first method, based on end-use (audience), will attribute the same degree of market power to both media and consequently the radio contact will have identical value to television contact. The second method, based on monetary measurement, will attribute varying levels of market power to media having the same number of contacts (radio contacts will have a lower value than television ones). The two different methods thus come up with different measurements of the market position of the media. The combination of the two methods of measurement (the 'middle way' as we may call it) would eliminate this discrepancy and provide an avenue to arrive at a more or less accurate picture of concentration levels in the mass media field.<br />The fact that a monetary-based measurement is by itself insufficient to reveal the exact level of media market power can be shown through another example. Revenues, which is the unit of measurement most heavily used in competition policy, is clearly related to economic power. A company which has a large share of revenues is likely to invest a larger portion of its resources in improving the quality of its product in order to stay ahead of its competitors. If the investment is successful, then the improvement in product quality can result into higher market share. In the media domain, a firm must be capable of retaining and therefore influencing its audiences, in addition to commanding its resources. However, the relationship between revenues and influence is far from straightforward and varies across different media. Consider a case of a newspaper which cuts its cover price. It will increase its circulation and hence its influence, but may reduce its revenue (depending both on the elasticity of sales and the importance of advertising revenue with respect to cover price). A powerful newspaper proprietor with deep pockets, like Rupert Murdoch, who is determined to get his message across at any cost, could simultaneously reduce his market share and increase his audience by reducing the cover price of his newspapers. Measuring Murdoch's shares only in terms of revenue will probably not reveal the exact strength of his market position. But by adding another dimension (i.e. measuring his audience share), a more accurate picture of his market position will come up.<br />Marrying time use data with advertising expenditure data provides a more concrete example of our 'middle way' proposals. Advertising expenditure is the amount that advertisers pay for advertising, rather than the amount the media receive from it, since the former reflects the value to purchaser. The difference includes agency commission and the costs of producing advertising copy. As Arthur Andersen analysis indicated, time spent on television is over seven times more than time spent on newspapers. However, advertisers in Britain in 1993 were prepared to spend £1.2 billion on national newspapers and £2.6 billion on television advertisements (Advertising Association, Advertising Statistics Yearbook 1994). So although TV is much more important than newspapers in terms of time use, advertisers only regard it as roughly twice as important. Advertisers also consider it essential who is in the audience, in addition to how many watch, read or listen. The Independent, for instance, is a serious newspaper read by many influential people in business, the City, Whitehall and Westminster. So although space in the Independent is cheaper than space in the Sun (because there are far fewer readers), the cost per reader of advertising in the Independent is much higher than that of advertising in the Sun. This is illustrated by the advertising rate cards shown in Table 4.8. The amount that advertisers are prepared to pay to reach the highly influential Financial Times reader would buy access to no less than fourteen Sun readers.<br />Table 4.8<br />Advertising Ratecard Costs of Selected UK National Newspapers<br />Newspaper£/page£/000 adult readersThe Sun28,0002.84Daily Mail22,6804.80Independent14,00012.20Daily Telegraph34,50012.71Sunday Times47,00013.28Financial Times29,56839.69<br />Source: BRAD, 1994; London Economics, 1995.<br />These considerations reinforce the attractions of combining different measures for assessing media concentration levels and possibly establishing impact. The relationship between money and influence is certainly imperfect; however, monetary-based measures do at least take into account, in a way that head counts cannot, the fact that the influence of any medium depends on who it reaches in addition to how many it reaches. Advertising expenditure is thus an important candidate dimension in order to calculate influence shares. Nevertheless, although it gives us some indication of the influence of radio and TV compared to the press, it is hard to construct a consistent statistical measure of influence. The exercise is heroic because print is bought by the column inch or page part while radio and television is bought by the second. Another point that must be borne in mind when making such attempt is that television (and therefore advertising in television) may be more influential at prime-time (usually between 8 to 11 pm), when the audiences are huge, rather than in day time. And space in newspapers may be more influential on some pages than others. We also have to make the assumption (as the London Economics study did) that people spend the same time on advertisements for each hour of reading the press as for each hour of viewing or listening to commercial television or radio. All these factors must be taken into account by authorities wishing to adopt this model.<br />Another possibility in relation to complex combinations of media would be to set a percentage of market share in terms of revenue as a threshold for further examination of the position. This approach is taken in UK monopoly legislation in relation to newspaper ownership. Should a media controller breach the pre-determined revenue threshold, the argument follows, an audience share test will apply to determine whether he/she attracts a high percentage of consumers as well. This undoubtedly gives more weight to the economic criterion of revenue. The reason why it is considered here is that the sources of bias and error are potentially so many in all audience-based measures of shares in the aggregate media sector, that policy decisions based on such measures alone should be made extremely carefully. A long established, crude method for measuring market concentration such as revenue may moderate potential inaccuracies of the newly-conceived audience-based methods.<br />Finally, a system which can combine different measures such as numerical criteria, revenue, audience share and audience time spent consuming a medium has been proposed by the British Broadcasting Corporation (BBC, 1995; Andrew, 1995). According to BBC, a weighted points system which combines several measurement criteria may represent the most reasonable approach. It should have the following characteristics: a) the points system could award a certain number of points per channel or title owned according to the particular relevant market; b) the weighting by points would allow channel/title ownership to be informed by other criteria, rather than being applied as a sole criterion; c) the number of points would reflect the revenue, audience share and audience time spent which channels/titles in a given market typically command; d) it would be important for the points system to be adjustable over time in order to take account of the changing impact of different media; e) points thresholds should be set to provide an indication of whether organisations are moving into positions from which they might have the potential to abuse pluralism, and to instigate an in-depth review; and f) the thresholds should be designed so as to limit the share of commercial revenues for single media (30% of revenues) together with a lower limit (20% of revenues across the relevant sectors) for cross media ownership. BBC stated that, used in this way, the lack of sophistication of the measure used is less of a concern for regulators, since more detailed information can be taken into account in the review itself.<br />In any case, we are still a bit away from constructing a single indicator capable of assessing effectively multi-media concentration. Both the Commission and several European countries which are in the process of changing their national media ownership rules (i.e. Germany, Great Britain) are aware of the need to examine more thoroughly certain questions concerning content before any initiative is presented. Germany's example, where until recently there seemed to be a clear preference for a new system based on audience data while there is now a consideration of a model based on commercial revenue, is an indication of the confusion over the 'best' possible measure of media concentration and influence. There are many possible different ways of measuring share of the market. Each one of these measures conveys some important information about the markets and the degree of influence of each firm within them; however, none is so perfect to assess the exact level of influence shares in the media industry. This is precisely the reason why we propose the use of a combination of measures. An over-reliance on one of the measures may lead to flawed policies. The media regulator must gather as much information as he/she can about a company's market share; he/she then must compute all of the different measures (or those he/she deems necessary to build up a complete picture of the market) and examine them carefully. By this way, he/she will be able to come to a definite conclusion about a potential excessive influence of the company in question.<br />Finally, the regulator must ensure that the framework for media ownership regulation is clearly set out and not affected by day to day fluctuations in companies' shares. Audience shares, circulation, shares of revenue and so forth all very materially from day to day or month to month, something that may present difficulties. A firm may find itself in line with a proposed threshold one month but it may exceed it next month, simply owing to its own success relative to its competitors. It would not be right if media controllers were required to divest because, owing to their own market success, they have crossed some line, quite possibly temporarily. One approach to such a problem would be to have delays built in to the rules. If an enterprise crosses some pre-determined line, action would only be taken if it were still over the line some period later. This would mean that action was not taken on temporary fluctuations. It would also give media controllers the ability to take corrective action of their own during the delay period. A second approach would be to have two limits, a lower one which may trigger a discretionary power, and an upper one which automatically triggers action or divestment. A third approach, suggested by NERA (1995), is to establish that the rules as to what was allowed would only come into play at times when mergers and acquisitions are proposed, and when licences are applied or re-applied for. So, for instance, if a media owner wished to acquire another newspaper, this would be prevented if it took him/her over the limit. If a TV licence owner went over the limit during the course of the licence, he/she would be prevented from reapplying.<br />