very different prescription for the optimal organization of the media than does the benign view Interestingly, even the Pigouvian economists, who adopt the perspective of a benevolent government when considering other industries and advocate both heavy regulation and nationalization, avoid this position with respect to the media(Henry Simons 1948, w.Art Lewis 1955, Gunnar Myrdal 1953). Coase(1974) points to this hypocrisy of Pigouvian economists: in the very industry where the case for state ownership is theoretically attractive, they shy away from taking this case seriously. Thus, according to Coase: It is hard to believe that the general public is in a better position to evaluate competing views on economic and social policy than to choose between different kinds of food (p. 389). Nonetheless, the assumption of benevolent government often stops at the doorstep of the media, perhaps because economists want to protect their own right to supply information without being subject to regulation These debates notwithstanding, there is precious little evidence on the organization of the media industries in different countries and its consequences. Our paper aims to fill this gap. We collect data on ownership patterns of media firms- newspapers, television, and radio-in 97 countries.Our paper provides a first systematic look at the extent of state and private ownershi of media firms around the world, of the different kinds of private ownership, and of the prevalence of monopoly across countries and segments of the media industry. Our basic finding is that the two dominant forms of ownership of media firms around the world is that by the state and by concentrated private owners, i.e., controlling families Demsetz(1989)and Demsetz and Lehn(1985 )hypothesize that the amenity potential also known as " the private benefits of control"(Grossman and Hart 1988), arising from owning media outlets is extremely high. In other words, the non-financial benefits, such as fame and
3 very different prescription for the optimal organization of the media than does the benign view. Interestingly, even the Pigouvian economists, who adopt the perspective of a benevolent government when considering other industries and advocate both heavy regulation and nationalization, avoid this position with respect to the media (Henry Simons 1948, W. Arthur Lewis 1955, Gunnar Myrdal 1953). Coase (1974) points to this hypocrisy of Pigouvian economists: in the very industry where the case for state ownership is theoretically attractive, they shy away from taking this case seriously. Thus, according to Coase: “It is hard to believe that the general public is in a better position to evaluate competing views on economic and social policy than to choose between different kinds of food (p. 389).” Nonetheless, the assumption of benevolent government often stops at the doorstep of the media, perhaps because economists want to protect their own right to supply information without being subject to regulation.2 These debates notwithstanding, there is precious little evidence on the organization of the media industries in different countries and its consequences. Our paper aims to fill this gap. We collect data on ownership patterns of media firms – newspapers, television, and radio – in 97 countries. Our paper provides a first systematic look at the extent of state and private ownership of media firms around the world, of the different kinds of private ownership, and of the prevalence of monopoly across countries and segments of the media industry. Our basic finding is that the two dominant forms of ownership of media firms around the world is that by the state and by concentrated private owners, i.e., controlling families. Demsetz (1989) and Demsetz and Lehn (1985) hypothesize that the “amenity potential”, also known as “the private benefits of control” (Grossman and Hart 1988), arising from owning media outlets is extremely high. In other words, the non-financial benefits, such as fame and 2 Much of the available discussion deals with the traditional industrial organization aspects of the media industry, such as product variety and market power, rather than on the broader social consequences of media ownership (Spence and
influence, obtained by controlling a newspaper or a television station must be considerably higher than those from controlling a firm of comparable size in, say, the bottling industry. Economic theory then predicts that private control of media firms should be highly concentrated: with no controllers to enjoy the amenity potential, widely held firms are not a stable institutional form. Put differently, the control of widely held firms with a high amenity potential is up-for-grabs(Bebchuk 1999). Our findings are broadly consistent with these predictions Having established the importance of state ownership of the media, we ask first: in which countries is government ownership of the media higher? We find that government ownership of the media is higher in countries that are poorer, have more autocratic regimes, and higher overall state ownership in the economy. These results cast doubt on the proposition that state ownershi of the media serves benevolent ends We then consider the consequences of state ownership of the media, as measured by freedom of the press, development of economic and political markets, and social outcomes. To this end, we run regressions of a variety of outcomes across countries on state ownership of the media, holding constant the level of development, the degree of autocracy, and overall state ownership of the economy We find pervasive evidence of bad"outcomes associated with state ownership of the media(especially the press), holding country characteristics constant. The evidence is inconsistent with the Pigouvian view of state ownership of the media. Still, since we only have a cross-section of countries, we cannot decisively interpret this evidence as causal, i. e as showing that state ownership of the media rather than some omitted country characteristic is responsible for the bad outcomes. We note, however, that the omitted characteristic must be quite closely related to the inclination of the government to control information flows, since we are controlling for a number Owen 1977, Motta and Polo 1997)
4 influence, obtained by controlling a newspaper or a television station must be considerably higher than those from controlling a firm of comparable size in, say, the bottling industry. Economic theory then predicts that private control of media firms should be highly concentrated: with no controllers to enjoy the amenity potential, widely held firms are not a stable institutional form. Put differently, the control of widely held firms with a high amenity potential is up-for-grabs (Bebchuk 1999). Our findings are broadly consistent with these predictions. Having established the importance of state ownership of the media, we ask first: in which countries is government ownership of the media higher? We find that government ownership of the media is higher in countries that are poorer, have more autocratic regimes, and higher overall state ownership in the economy. These results cast doubt on the proposition that state ownership of the media serves benevolent ends. We then consider the consequences of state ownership of the media, as measured by freedom of the press, development of economic and political markets, and social outcomes. To this end, we run regressions of a variety of outcomes across countries on state ownership of the media, holding constant the level of development, the degree of autocracy, and overall state ownership of the economy. We find pervasive evidence of “bad” outcomes associated with state ownership of the media (especially the press), holding country characteristics constant. The evidence is inconsistent with the Pigouvian view of state ownership of the media. Still, since we only have a cross-section of countries, we cannot decisively interpret this evidence as causal, i.e., as showing that state ownership of the media rather than some omitted country characteristic is responsible for the bad outcomes. We note, however, that the omitted characteristic must be quite closely related to the inclination of the government to control information flows, since we are controlling for a number Owen 1977, Motta and Polo 1997)
of dimensions of"badness" in the regressions In addition to discussing media ownership patterns and their consequences, we examine the role of media monopolies. Recall that Lenin and the founders of the bbc insisted on monopoly for reasons of technology and benevolent censorship. But even ignoring this particular argument, one can still wonder whether any government participation in the media is detrimental to freedom or just the state monopoly. Any government ownership may be bad because the government has the power to advantage the media firms that it owns. Alternatively, private etition may assure that alternative views are supplied to voters and consumers, and prevent government firms from distorting the information they supply too heavily. only the data can resolve which one of these theoretically plausible views better describes reality Section II describes our data on ownership of the media. Section Ill examines the economic and political determinants of media ownership. Section IV then focuses on the consequences of state media ownership for freedom of the press, the efficiency of economic and political markets, and a range of social outcomes across countries. Section V addresses the question of whether the effects of government ownership stem from the very existence of such ownership, or from government monopoly. Section VI summarizes the findings and concludes Ownership Data This section focuses on patterns of ownership in the media industry. Because ownershi bestows control( Grossman and Hart 1986), it shapes the information provided to voters and consumers. Ownership, of course, is not the only determinant of media content. In many ountries, even with private ownership, the government regulates the media industry, provides direct subsidies and advertising revenues to media outlets, restricts access to newsprint and
5 of dimensions of “badness” in the regressions. In addition to discussing media ownership patterns and their consequences, we examine the role of media monopolies. Recall that Lenin and the founders of the BBC insisted on monopoly, for reasons of technology and benevolent censorship. But even ignoring this particular argument, one can still wonder whether any government participation in the media is detrimental to freedom or just the state monopoly. Any government ownership may be bad because the government has the power to advantage the media firms that it owns. Alternatively, private competition may assure that alternative views are supplied to voters and consumers, and prevent government firms from distorting the information they supply too heavily. Only the data can resolve which one of these theoretically plausible views better describes reality. Section II describes our data on ownership of the media. Section III examines the economic and political determinants of media ownership. Section IV then focuses on the consequences of state media ownership for freedom of the press, the efficiency of economic and political markets, and a range of social outcomes across countries. Section V addresses the question of whether the effects of government ownership stem from the very existence of such ownership, or from government monopoly. Section VI summarizes the findings and concludes. II. Ownership Data This section focuses on patterns of ownership in the media industry. Because ownership bestows control (Grossman and Hart 1986), it shapes the information provided to voters and consumers. Ownership, of course, is not the only determinant of media content. In many countries, even with private ownership, the government regulates the media industry, provides direct subsidies and advertising revenues to media outlets, restricts access to newsprint and
information collection, and harasses journalists. We discuss these modes of control as well Construction of the database We gather new data on media ownership in 97 countries. We focus on newspapers and television, since these are the primary sources of news on political, economic and social issues Data on radio ownership are limited. Radio reaches a high proportion of the population even in the lowest income and literacy countries, but it largely delivers entertainment. The radio market is also highly regional, which precludes any single station from achieving a large market share. As a crude index, we gather ownership data on the top radio station as measured by peak dult audience, and on an all-news radio station when one exists in a country Our selection of sample countries is driven by data availability. First, we identify the countries for which we have information on control variables. Since we are interested in the consequences of state ownership of the media, we need to make sure that our results are not driven by differences in the levels of economic development, the level of political competition, or of broad state intervention in the economy. To this end, we control for general levels of state ownership in the economy, a measure of autocracy, and GNP per capita. We use the Fraser Institute(2000)index of the involvement of state owned enterprises(SOEs) in the economy, which is based upon the number of SoEs, their prevalence in particular sectors of the economy, and their share of gross domestic output. A total of 133 countries have the SOE index, GNP per capita, and autocracy data for 1999. Of those, we exclude 5 observations because a) the country is in civil war(Democratic Republic of Congo, Sierra Leone), or b) the entity cannot be classified as a country(Hong Kong), or c)no daily newspapers exist(Belize, Tajikistan ). We also exclude 31 countries lacking sufficient data on media ownership. The final sample of 97 countries includes
6 information collection, and harasses journalists. We discuss these modes of control as well. Construction of the Database We gather new data on media ownership in 97 countries. We focus on newspapers and television, since these are the primary sources of news on political, economic and social issues. Data on radio ownership are limited. Radio reaches a high proportion of the population, even in the lowest income and literacy countries, but it largely delivers entertainment. The radio market is also highly regional, which precludes any single station from achieving a large market share. As a crude index, we gather ownership data on the top radio station as measured by peak adult audience, and on an “all-news” radio station when one exists in a country. Our selection of sample countries is driven by data availability. First, we identify the countries for which we have information on control variables. Since we are interested in the consequences of state ownership of the media, we need to make sure that our results are not driven by differences in the levels of economic development, the level of political competition, or of broad state intervention in the economy. To this end, we control for general levels of state ownership in the economy, a measure of autocracy, and GNP per capita. We use the Fraser Institute (2000) index of the involvement of state owned enterprises (SOEs) in the economy, which is based upon the number of SOEs, their prevalence in particular sectors of the economy, and their share of gross domestic output. 3 A total of 133 countries have the SOE index, GNP per capita, and autocracy data for 1999. Of those, we exclude 5 observations because a) the country is in civil war (Democratic Republic of Congo, Sierra Leone), or b) the entity cannot be classified as a country (Hong Kong), or c) no daily newspapers exist (Belize, Tajikistan). We also exclude 31 countries lacking sufficient data on media ownership. The final sample of 97 countries includes
21 in Africa. 9 in the Americas. 17 in asia and the pacific. 7 in Central asia and the Caucasus. 16 in Central and Eastern Europe, 11 in Middle East and North Africa and 16 in Western Europe Within countries we select media outlets on the basis of market share of the audience and provision of local news content for the year 1999. This approach focuses on who controls the majority of information flows on domestic issues to citizens. We exclude entertainment and sport media, as well as foreign media outlets, if they do not provide local news content. We include in our sample the five largest daily newspapers, as measured by share in the total circulation of all dailies, and the five largest television stations, as measured by share of viewing. We consult three primary data sources to selecting these outlets. First, we use Zenith Media Market and Media Fact Book 2000 publications, which are organized by region, including Western Europe, Central and Eastern Europe, Asia Pacific, Middle East and Africa, and the Americas. Zenith Medias ranking of newspapers are checked with the World Association of Newspapers(Wan) World Press Trends 2000 report. Wan data are also used as the source for total newspaper circulation, which is not reported by Zenith Media. Finally, we use the European Institute for the Media Media in the CIS report as a primary source for countries in the former Soviet Union. Alternative sources are sought in two cases: when there is an inconsistency in data reported by primary sources,or when none of the sources covers the country in question. When this occurs, we use local media survey firms, World Bank external affairs offices, U.S. Department of State information offices, and direct contact with the media outlet 3 For 6 countries, we construct this index using World Bank's(2000)data on state enterprises 4 Following the World Association of Newspapers definition, newspapers are considered dailies if they are published at east four times per week. In the initial phase of the data gathering(first 12 countries)we focused on the top 10 media enterprises in the daily newspaper and television markets. We subsequently reduced the sample to five firms per media, for two reasons. First, the difference in market coverage from increasing the sample of companies from five to 10 marginal. In the first 12 countries, the top five newspapers account for an average of 62. 4% of total circulation, and the top 10 for 74. 1%. The correlation between the two is 94.2%. For the sample as a whole, the top five newspapers account for an average of 66.7% of total circulation. Television markets are even more concentrated-on average the top five firms cover 89.5% of total viewing. Second, 20 countries in our sample do not have more than five daily newspapers, and 42
7 21 in Africa, 9 in the Americas, 17 in Asia and the Pacific, 7 in Central Asia and the Caucasus, 16 in Central and Eastern Europe, 11 in Middle East and North Africa and 16 in Western Europe. Within countries, we select media outlets on the basis of market share of the audience and provision of local news content for the year 1999. This approach focuses on who controls the majority of information flows on domestic issues to citizens. We exclude entertainment and sport media, as well as foreign media outlets, if they do not provide local news content. We include in our sample the five largest daily newspapers, as measured by share in the total circulation of all dailies, and the five largest television stations, as measured by share of viewing.4 We consult three primary data sources to selecting these outlets. First, we use Zenith Media Market and Media Fact Book 2000 publications, which are organized by region, including Western Europe, Central and Eastern Europe, Asia Pacific, Middle East and Africa, and the Americas. Zenith Media’s rankings of newspapers are checked with the World Association of Newspapers (WAN) World Press Trends 2000 report. WAN data are also used as the source for total newspaper circulation, which is not reported by Zenith Media. Finally, we use the European Institute for the Media Media in the CIS report as a primary source for countries in the former Soviet Union. Alternative sources are sought in two cases: when there is an inconsistency in data reported by primary sources, or when none of the sources covers the country in question. When this occurs, we use local media survey firms, World Bank external affairs offices, U.S. Department of State information offices, and direct contact with the media outlets. 3 For 6 countries, we construct this index using World Bank’s (2000) data on state enterprises. 4 Following the World Association of Newspapers definition, newspapers are considered dailies if they are published at least four times per week. In the initial phase of the data gathering (first 12 countries) we focused on the top 10 media enterprises in the daily newspaper and television markets. We subsequently reduced the sample to five firms per media, for two reasons. First, the difference in market coverage from increasing the sample of companies from five to 10 was marginal. In the first 12 countries, the top five newspapers account for an average of 62.4% of total circulation, and the top 10 for 74.1%. The correlation between the two is 94.2%. For the sample as a whole, the top five newspapers account for an average of 66.7% of total circulation. Television markets are even more concentrated – on average the top five firms cover 89.5% of total viewing. Second, 20 countries in our sample do not have more than five daily newspapers, and 42