Policy Research Working Paper 8984 Does Media Stimulate Reform Efforts? Rita Ramalho Valentina Saltane Development Economics Global Indicators Group August 2019 Policy Research Working Paper 8984 Abstract This paper investigates to what extent media impacts achieve this objective, the study put together a comprehen- political decisions. A viable practical approach to test the sive data set that encompasses country-specific local media relationship between mass media and political actions is coverage of the Doing Business report in 190 economies. through the use of the World Bank’s Doing Business data, The study finds that local media coverage of Doing Business specifically, by assessing local media coverage of Doing Busi- has a significant influence on regulators’ actions. First, the ness and implementation of business regulatory reforms. analysis shows that the number of local media articles tends The tested hypothesis is that countries with higher media to increase the probability of whether a country does any coverage of Doing Business tend to carry out more busi- reform. Second, countries with greater media coverage of ness regulatory reforms, assuming one- and two-year lags Doing Business indicators tend to have higher numbers of between media coverage and reform implementation. To implemented reforms. This paper is a product of the Global Indicators Group, Development Economics. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at rramalho@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Does Media Stimulate Reform Efforts? Rita Ramalho1 Valentina Saltane2 JEL: O1, O2, O4 Keywords: Business reforms, reform implementation, regulation, media, development, doing business, political action 1  Senior Manager, Development Economics and Chief Economist, DECIG  2  Senior Private Sector Development Specialist, Development Economics and Chief Economist, DECDB  I. Introduction Effective regulation serves as a backbone of a thriving private sector in any economy. Policy makers understand the importance of increasing their countries’ competitiveness, striving to achieve more advantageous positions in the global economy. Mass media is one of the factors that has a potential to influence some of the reform decisions. In some cases, mass media could even veer regulatory and public actions towards areas of much needed reforms. For example, inadequate business regulation that is heavily criticized in mass media can become hard to ignore. Increased access to information from various media sources addresses the fundamental economic problem of information asymmetry. Media serves as a conduit of information among the public, businesses and policy makers. As a powerful channel of information, media usually shapes the context for political discourse and allows citizens and businesses to make better informed decisions. Also, mass media conveys public and private sectors’ sentiments to policy makers. The media pays close attention to the World Bank’s research that analyzes effective policies, better regulations and economic growth. Doing Business is a research project translated into annual publications which measure, analyze and compare regulatory practices that directly affect the private sector in 190 countries. The reports identify and acknowledge regulatory reforms that de facto affect the ease of Doing Business in the analyzed countries. Numerous countries have developed a momentum for launching reforms to improve their Ease of Doing Business rankings. Largely motivated by the global coverage of the report in mass media, governments often strive to gain positive media coverage by addressing local regulatory shortcomings. The report has been subject to considerable media attention in the past decade. Over the years worldwide media coverage of Doing Business has been increasing exponentially, reaching more than 1,800 media citations and close to 1.5 million website views in the first four months after the launch of the Doing Business 2018 report. Overall, from 2004 to 2018, Doing Business has been highlighted by countries’ local press more than 17,400 times.3 Moreover, Doing Business findings are widely discussed by international newspapers, including The Economist, The Wall Street Journal and The New York Times. Media research presented in this paper demonstrates that to date, the OECD high-income group has the highest media coverage of Doing Business, followed by the East Asia and the Pacific region. Conversely, in the Middle East and North Africa and Sub-Saharan Africa media give Doing Business the least attention compared to the rest of the world. However, as demonstrated by our research, this tendency correlates with the media presence in the regions as well as with reform implementation efforts in the subsequent two years. At the same time, among all income groups, media across lower-middle income countries has the highest number of Doing Business citations. This paper aims to investigate to what extent (if at all) media impacts political decisions. A practical approach to test the relationship between mass media and political actions is through                                                              3  Primary data collection on media coverage of Doing Business has been done through the use of the Factiva database. Factiva is an online comprehensive database primarily designed to provide data to business communities, researchers and specialists, aggregating content from both private and public sources. It pulls data from more than 32,000 information sources, including newspapers, journals, magazines, television and radio transcripts. Factiva has an almost universal reach, allowing to filter data by local and international sources.   2    the use of the Doing Business data, specifically, local media coverage of Doing Business and implementation of business regulatory reforms. Our refined hypothesis is that countries with higher media coverage of Doing Business tend to carry out more business regulatory reforms, assuming one- or two-year lags between media coverage and actual reform implementation. The study adopts an empirical approach to analyze this relationship. To achieve this objective, we have put together a comprehensive panel data set that encompasses country-specific local media coverage of the Doing Business report in 190 economies. The data set goes back to 2004, when the first Doing Business report was published. We find that local media coverage of Doing Business does seem to have a potent influence on regulators’ actions. First, the analysis shows that the number of local media articles tends to increase the probability of whether a country reforms or not. Second, countries with greater media attention to Doing Business indicators do tend to have higher number of implemented reforms within both one- and two-year timeframes after the articles are published. The paper is organized as follows: sections II and III provide an overview of the relevant existing research on the subject of media, good governance and reform implementation; section IV presents descriptive statistics and global data trends, section V outlines the conceptual framework of the analysis, explaining the adopted research methodology; while section VI discusses the analytical results. Section VII concludes. 3    II. Importance of the media To date, there is little research specifically on media and business regulations or business reforms. However, the growing role of mass media in governance effectiveness, political accountability and economic development at large has drawn attention of some researchers and international organizations. Adding to this discourse, a World Bank publication (2009) explores a relationship between media coverage and governance reforms, focusing primarily on subjects including democracy, freedom of speech, elections and government accountability. This study concludes that more diverse and independent media tends to foster good governance. In the same vein, Besley and Burgess (2000, 2001) suggest that free and independent press working in conjunction with democratic institutions can make governments more responsive to the needs of citizens. Building the analysis on the cross-country data with a specific focus on India, the authors provide evidence of strong linkages between media and electoral processes as well as media and government responsiveness. Leeson and Coyne (2007) analyze a relationship between media freedom, foreign aid and economic development in post-socialist transition countries. Their research demonstrates that politically autonomous, independent mass media serves as an effective law enforcement mechanism, enabling political commitment to reform implementation. And that mass media is an important factor in determining successes and/or failures of reforms in transitional economies. Research results show that media’s freedom is linked to an increase in the level of economic development, while foreign aid is associated with higher economic development only if a country has relatively free media. Egorov et al. (2009) explore the relationship between media and governance in the context of natural resource endowment and the ‘resource curse’ problem. The authors analyze factors that determine media freedom in dictatorships. They employ a model predicting that free media improves dictators’ government quality by providing incentives to bureaucrats to gather and transmit information from across different sectors to the higher government levels. However, the extent of media freedom that a dictator chooses varies with resource endowments of a country. The findings show that media is less free in resource-rich economies, with stronger effect in nondemocratic regimes. Further, Garcia-Sanchez et al. (2016) use the World Bank Governance Indicators data for 2002–2008 to analyze the relationship between free media and government effectiveness. The authors demonstrate that government effectiveness level can be determined by media, political characteristics and organizational environment, considering the level of economic development of an economy. Arguably, one of the key roles of media in promoting democratic principles and good governance lies in encouraging public activism and enabling better informed political and economic debates in a society. A sizable body of research examines the effects of media in the context of political discourse and expression of public opinion. McCombs and Shaw (1972) argue about the importance of media in setting the political agenda and shaping political reality by not only informing the audience about certain issues, but also determining the level of their importance. Likewise, Stromberg (2001) in his analysis of the impact of news media on regulatory policy shows that media, due to its special role in transmitting information to mass audiences, increases political influence of dispersed large groups at the expense of small interest or minority groups, thus strongly influencing policy setting. Furthermore, one of the 4    conclusions drawn by Leeson (2008) is that low media freedom is associated with low political participation and poor political knowledge. And countries with less-regulated and more privately-owned media have larger politically active and knowledgeable population. Furthermore, existing media research pays particular attention to the power of media to encourage political accountability. For instance, Voltmer (2009) discusses how the role played by media in both recently established and older democracies can be an indicator of a democracy’s strength, especially with regards to encouraging government accountability and public awareness. Results of this research demonstrate that although media is often criticized for having close ties to governments, it can serve as an effective watchdog for political accountability and good governance. Similarly, Francken et al. (2009) study the relationship between media and capture of public expenditures on education, using 2002-2003 survey data on the level of reach of the public spending on education to local schools. The study demonstrates that media – conditional on the population characteristics and their levels of literacy – can have an effect on the extent of corruption in a country. In turn, Olper and Swinnen (2013) focus on the influence of mass media competition on agricultural policy. These researchers find a significant relationship between the increase in the media diffusion and the changes in agricultural policies in both developed and less developed economies. Furthermore, the authors demonstrate that the larger the share of an economy’s population that is affected by a certain policy, the larger is the mass media coverage of this policy. Notably, substantial research explores the theme of media ownership and free press. Particular attention to this subject is drawn by the research of Djankov et al. (2001), investigating the relationship between media ownership and press freedom focusing on political, economic and social outcomes. The authors find that countries where more media is state-owned have less free press, lower level of governance and market development as well as fewer political rights for citizens and substantially lower outcomes in the areas of education and health. Moreover, the authors argue that newspapers bear even stronger adverse effects on political freedom than television. Furthermore, Brunetti and Weder (2003) demonstrate a negative association between free press and levels of corruption, describing media as a “platform for the private sector to voice complaints” about detrimental effects of extortive corruption on business. Media is a powerful channel of information and influence with an ability to impact political decisions and shape public opinion. By drawing attention to the most important and contentious issues, media fosters government accountability and facilitates better regulation, essential for economic growth and private sector development. However, the degree of its influence on regulatory processes in an economy is conditional on factors such as strength of democratic institutions, freedom of speech, independence of media, geographical coverage of press and access to information. Furthermore, both the breadth of media attention and its impact on policy making often depend on strategic interests of political actors. In the past decades, there has been a growing interest in the analysis of the role of media in policy-making processes. And numerous studies assessed the impact of media on politics and governance. However, limited research has been carried out on media’s potential influence on business regulatory reforms. Furthermore, media’s impact (if any) on the private sector development and growth has not yet been explored. This paper aims to contribute to the existing body of media research with the primary goal of analyzing the relationship between media coverage of Doing Business and implementation of regulatory reforms. 5    III. Business reforms It is important to have a healthy private sector for an economy to thrive. And in order to allow private businesses to flourish, governments put in place laws and regulations that support entrepreneurship, allowing businesses to operate smoothly and expand. Therefore, regulators continuously strive to improve local business climate. Reforms aimed at fostering private sector development and growth often entail features such as cutting the red tape, aligning domestic laws with international best practices and introducing modern ways of doing business (i.e. use of electronic platforms). Reforms implemented in the areas measured by Doing Business have proved to have favorable impacts on local economies. Some of the core benefits of Doing Business’ regulatory changes include increase in entrepreneurial activity and employment rates as well as upsurge in competition and innovation.4 A substantial body of research shows the significance of reforms aimed at enabling favorable regulatory environment for entrepreneurship and hence firm creation. Klapper et al. (2006) highlight that costly regulations impede the establishment of new firms, especially small ones and particularly in the industries that are normally characterized by high entry propensity. Furthermore, cumbersome regulations mostly favor large firms, hindering growth of incumbent smaller businesses. Similarly, Klapper and Love (2010) explore how ease of registering a business and relevant regulatory reforms affect registrations of new firms. The authors demonstrate that costs, number of days and procedures required to start a business are the key factors impacting new firm registration.   Research also shows that improvements in business regulation can drive additional job creation, which is prompted by increased firm registration. Bruhn (2011) analyzed reforms that simplified business entry regulation in Mexico. The author finds that these reforms encouraged wage earners to open their own businesses, increasing the number of registered firms by 5%, while at the same time increasing wage employment by 2.2%. In the same vein, Branstetter et al. (2010), using Portuguese micro level data, investigated economic effects of a regulatory reform in Portugal. Their study found that the reform brought about an increase in the number of firms created as well as in the overall employment level. The authors note that the increase occurred predominantly among the firms that were at the highest risk of being deterred by heavy entry requirements. These were typically small firms, owned by entrepreneurs with relatively poor education level. And such firms commonly operated in agriculture, retail trade and construction sectors. The link between regulation and firm creation is further researched by Bripi (2016). Using data on time and cost to start a business and entry rates by industry in the period of 2005-2007, the author analyzes the impact of entry regulation on firm creation in Italy. The cross-sectional analysis of the data shows that the length and costliness of procedures contribute to lower entry in sectors with commonly high entry. Economic theory suggests that employment as a reliable source of income can ultimately lift people out of poverty and reduce income inequality. Economies with poor quality business regulation have, on average, higher levels of income inequality. Furthermore, when business regulation is overly cumbersome, posing severe obstacles to companies’ daily operations, businesses become more prone to corruption and bribery. Paunov (2016) demonstrates that the                                                              4 Bruhn 2013; Branstetter et al 2010; Klapper and Love 2010; Paunov 2016. 6    likelihood of corruption is higher when regulatory processes are complex and/or unclear. Furthermore, in such cases, firms are pushed out of the formal sector and start operating in the informal economy. Djankov et al. (2002) analyze firm entry regulation in 85 countries and find that in the economies with more heavily regulated business entry the level of corruption and the size of unofficial economy are considerably higher, while the quality of public and private goods is lower. The authors also conclude that opaque regulation of entry is beneficial for bureaucrats and politicians, as it can serve them as a rent extraction mechanism. Conversely, relatively light regulation of entry is commonly observed in the economies with more democratic governments. Focusing on a similar subject, Ulyssea (2010) argues that high costs of entry are linked with higher level of informality, worse labor market performance and lower welfare of an economy. Hence, reduction in the cost of entry to the formal sector is a key element for shrinking the informal economy. Yet, firms operating informally have different reasons and incentives for doing so with regards to regulatory reforms. For instance, while some are prompted to resort to informality due to unwieldly regulations, others remain informal for a limited time - until owners find wage jobs. Moreover, Ulyssea (2018) employs data on Brazilian firms to investigate the response of informal firms to formalization policies as well as the effects of such policies on economic performance and development. The author highlights the importance of differentiation between the two margins of informality – non-payment of registration fees and payment of informal wages to workers – for policy formulation and analysis. This research demonstrates a negative relationship between enforcement and both informality and welfare creation. Entry costs reduction in the formal sector has weaker effect on the decrease of informality, yet strong positive effect on welfare gains, GDP growth and wages. Similarly, Maican and Orth (2018) analyze Swedish retail food stores data and establish that, by enhancing competition, lower entry costs increase welfare in differentiated product markets. According to Bruhn (2013), different types of business owners in the informal economy react to reforms in different ways. Analyzing the effects of a business registration reform in Mexico, the author explains why this reform did not impact all the informal business owners in the same way. The results demonstrate that the owners of informal businesses were more likely to register them if prior to the reform they possessed similar personal characteristics to those of formal business owners. Such characteristics included gender, age, marital status, education as well as being a head of a household and/or a migrant. In turn, owners of informal businesses with traits similar to those of wage workers, were more likely to become wage workers rather than formalize their businesses even when regulations became more conducive. At the same time, Demenet (2016) finds that by moving from informal economy to formal registration, firms get greater access to new equipment and can increase their scale of operations, resulting in higher competitiveness and productivity. Curtis (2016) takes the example of China to examine the impact of business reforms on the country’s total factor productivity growth. Building a model that captures both the private and state sectors, the author finds that entrepreneurship-friendly business reforms lead to expansion of a private sector and exit of least productive state firms from the market, hence increasing total factor productivity gains. Furthermore, Djankov et al. (2018) analyze panel data for 189 economies from 2005 to 2013 to suggest that effective business regulations can contribute to poverty reduction; mostly due to stimulation of job creation and promotion of entrepreneurship. The 7    authors find that the measures of the Doing Business indicators of getting credit and contracts enforcement are positively associated with poverty reduction.  Similarly, Ciccone and Papaioannou (2007) show that in economies where new business registration time is lower, more new firms enter industries undergoing technology shifts and hence benefit from increasing global demand for their products. Darnihamedani et al. (2018) add to the ‘start-up costs - taxes - entry rate’ equation the firms’ propensity to innovate. Using the data on entrepreneurs from 53 countries, the authors find that corporate taxes have a negative relationship with innovative capacity of firms. They also show that higher start-up costs have a significant positive relationship with innovative entrepreneurship, arguing that reduction of entry costs may increase the number of firms. Jerbashian and Kochanova (2016) use industry-level data and World Bank’s Doing Business indicators for 14 OECD countries to demonstrate that high costs of starting a business and registering property decrease investments in information and communication technologies, while the strength of legal rights and of minority investor protection tend to increase such investments.  Bertrand and Kramarz (2002) draw a similar conclusion, finding that cumbersome business entry regulation for the retail sector in France hampered employment in the industry. The authors demonstrate that a burdensome requirement of an approval from a regional zoning board, when either creating or extending any large retail store in the country, decreased retail employment by more than 10%. Bailey and Thomas (2017) run fixed effects regressions using the U.S. data on firm employment and the extent of federal regulations, by industry and from 1998 to 2011. Their research demonstrates that fewer new firms were created and employment growth was slower in industries with heavier regulation. Also, Cacciatore and Fiori (2016) show that business regulation affects firm entry costs, workers’ unemployment benefits and firing restrictions. They use panel VAR estimates for OECD countries to demonstrate the recessionary short-run and expansionary long-run effects of reforms. The authors, thus, prove a strong effect of market deregulation on business cycle fluctuations’ efficiency.   In the context of construction permitting, another area measured by Doing Business, Poel et al. (2014) find that higher administrative burdens adversely affect economic growth. Using a fixed effect regression analysis of data from 182 economies, the authors show a clear correlation between the number of days and procedures required to obtain a construction permit and economic growth. Plus, a similar relationship exists between economic growth and days and procedures to start a business as well as to pay taxes. Likewise, in the area of getting electricity, the analysis of Abotsi (2016), using the Enterprise Surveys data for Africa, demonstrates that the number of power outages experienced in a given month negatively impacts firms’ production efficiency. Several other studies emphasize the importance of effective tax regulation on firm formalization. Rocha et al. (2018) analyzing firm-level administrative and individual panel data in Brazil show that the costs of remaining formal in the case of taxes – and not entry costs – pose a serious obstacle to formalization. Specifically, reducing taxes after the registration costs have already been removed lowers firm informality rates, while registration costs reduction per se has no evident effect on informality. At the same time, lower taxes do not necessarily impact creation of new formal firms.   Analyzing regulation of property registration, Berkowitz et al. (2015) show that a property law enactment, which grants more rights to creditors over the assets that are used for loans to private firms leads to a substantial increase in firm value. And the impact is stronger on firms that are less politically connected, have lower internal cash flows and have more opportunities for 8    growth. A substantial body of research has also explored the importance of the availability of credit information. In a study of the finance industry in the United States, Doblas-Madrid and Minetti (2014) demonstrate that when lenders join a credit bureau and partake in information sharing, borrower performance considerably improves, as evidenced by the reduction of overdue contract duration. As to reforms in the sphere of contract enforcement, Nunn (2007) tested whether an ability to enforce written contracts determines a country’s comparative advantage. The author proves the hypothesis that in economies with better contract enforcement under-investment tends to be lower. Furthermore, Chemin (2009) suggests that having slow judiciaries may result in discouraged entrepreneurship and lower incentives to start businesses due to deteriorated security of property rights. Chemin’s research, using as a case study Pakistan's judicial reform of 2002, concludes that speeding up judiciaries is essential for economic growth. Furthermore, substantial attention has been given by researchers to the importance of reforms in trade facilitation. Iwanow and Kirkpatrick (2009) show that manufacturing export performance in Africa can be enhanced through trade facilitation reforms, as well as regulatory changes aimed at improving the quality of basic transport and communications infrastructure. Clark et al. (2004) in the study of determinants of shipping costs from Latin America to the United States argue that improvements in port efficiency tend to increase bilateral trade volume. Port inefficiencies substantially add to high transportation costs that, for most exporting economies, create even greater barriers to trade than import tariffs. And it is excessive regulation that is one of the main causes of port inefficiency. 9    IV. Methodology Doing Business measures and compares 11 areas of business regulations, capturing their impact on firm establishment and operations. The reports attract a wide audience, including governments, private sector, NGOs, civil society organizations as well as academic and research communities. Media is one of the main avenues through which Doing Business reaches its audience. To a certain degree, media coverage of the reports captures and shapes public opinion about domestic business climates, fostering a broader public debate about Doing Business and its direct impact on local entrepreneurs and business climate. Media outlets can either veraciously portray Doing Business data or distort the reports’ messages and findings in pursuit of a particular political agenda. The main objective of this paper, however, is to investigate whether local media coverage of Doing Business spurs business regulatory reforms in the measured economies. In order to test the set hypothesis, we first collected primary data on the global coverage of Doing Business in the local media. Given the previous absence of such data, media-focused research primarily used freedom of press, state ownership of press and journalist abuse of Freedom House as well as relevant questions from World Value Surveys as the main input variables. Two search engines were selected for the collection of comparable media coverage data: Factiva and Google advanced search. These two domains have proven to meet the desired research criteria. The key search engine selection criteria included: access to a large comparable global database encompassing articles from a substantial number of local media sources; availability of reliable media article archives dating back to 2003; presence of search filers allowing to specify countries and time frames (months and years); ability to search for articles in local languages, including major local dialects; possibility to segregate results between the ones published by local and international media and existence of a user-friendly interface. Although more industrious, the concurrent use of two search engines was a deliberate undertaking. Having two similar novel data sets allowed for extra robustness checks, ensuring better accuracy of the novel data. Google advanced search effectively caters information and news to the general public. Operating since 1997 and processing over three billion searches a day, Google has been the internet’s most used search engine with around 86% market share among all the web search platforms as of July 2018.5 It almost instantaneously generates search results reflecting the content already published on the internet. Factiva has been selected as a second prominent data collection and validation source. It is primarily designed to provide data to business communities and specialists, aggregating content from both private and public domains. Factiva provides access to more than 36,000 information sources, including newspapers, journals, magazines, television and radio transcripts. Similar to Google, Factiva has an almost universal reach, allowing to perform search in 28 languages in 200 countries. Factiva’s media database has the ability to filter search by frequency of publication, type of coverage, language, category of primary sources, geography and even industry group. Importantly, what differentiates Factiva from Google, is the ability to segregate results by local and international sources.                                                              5  Statista 2018: https://www.statista.com/statistics/216573/worldwide-market-share-of-search-engines/   10    Apart from Factiva and Google, other specialized search engines and databases that were considered and tested included PressReader, LexisNexis, EBS Co-host, Global Newsstream, ProQuest and Newspaper Source. However, due to certain limitations that failed to meet the fundamental established criteria, these tools have not been employed for our data collection purposes. For example, PressReader was not selected primarily due to its insufficient historical media coverage. LexisNexis is relatively expensive and provides only a limited global media coverage with most of the sources originating from the United Kingdom. Similarly, EBS Co- host and Newspaper Source databases do not clearly differentiate between global and local media outlets. Information on mentions of Doing Business in local media has been collected using specific word search combinations designed to fit particularities of every economy, such as news’ distribution systems and local languages. Each word combination had to include two essential phrases: (1) the name of the project and (2) the name of the organization – either in English or in a local language or both: “Doing Business” + “The World Bank.” Each search entry had to entail a specific geographic location of the search and the selected time parameters. The time span for every search covered the period between (a) the press release date of the Doing Business report for a specific year and (b) the day prior to the press release date of the Doing Business report of the subsequent year. For instance, for Doing Business 2016, the time period of the search is between October 27, 2015 (the day of the launch of Doing Business 2016) and October 24, 2016 (one day prior to the release of the Doing Business 2017). Unavoidably, our methodological approach of collecting the primary data has a number of limitations, predominantly related to the technical properties of Google and Factiva. The first limitation is that the two search engines are generally designed to produce articles that mention the terms “Doing Business” and “the World Bank” in the same paragraph. Such setting signifies that articles mentioning these terms in different paragraphs are less likely to be captured during the data collection process. However, this limitation is slightly offset by using complimentary word search combinations and by cross-comparing the results between the two sources. The second limitation, albeit applicable only to Google, is that it does not distinguish between local and international results. This limitation is mitigated by complementary data collected through Factiva, which has the technical ability to distinguish between the two categories. Other limitations stem from the very nature of the media data collection project. Although the produced results give a relatively accurate representation of the mentions of Doing Business project by the local press, there is always a margin of omitted articles, varying in magnitude from country to country. This margin exists because not all articles are published online, and some that are published online are taken down from the web shortly after publication. Another reason for the omitted margin is that some of the articles are available only in non-represented local languages and dialects. In addition to English, data collection has been conducted in over 100 global languages with the aid of Google translate. Moreover, team members conducting research had native fluency in Spanish, Russian, Arabic, Portuguese and French, which allowed for accurate data collection in major foreign languages. However, searches in some local dialects and small language groups have not been feasible. 11    V. Descriptive statistics Since its inception, Doing Business has been steadily gaining substantial political clout. Given the reports’ power to shape regulatory agendas and inspire business regulatory reforms worldwide, the project’s global media coverage has been growing exponentially (see figure 1). From 2004 to 2018, Doing Business has been highlighted by countries’ local media more than 17,400 times.6 Countries with the highest media coverage of Doing Business during that time frame include Pakistan (1,337 articles; 97 media outlets7), India (1,028 articles; 720 media outlets), Mexico (965 articles; 141 media outlets), and Russian Federation (959 articles; 400 media outlets). Not surprisingly, these four economies have also been active reformers on the Doing Business indicators. Since 2004, Pakistan implemented 19 business regulatory reforms, India – 37, Mexico – 26, and the Russian Federation – 36. Figure 1: Doing Business media coverage has been steadily growing over the years  3,000 Number of articles, Factiva local press  2,500  2,000  1,500  1,000  500  ‐ 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Doing Business year Source: Factiva Note: The time frame for Factiva article count is determined by the interval between (a) the press release date of the Doing Business report for a specific year and (b) the day prior to the press release date of the Doing Business report for the subsequent year. For instance, for Doing Business 2017, the time period for counting Factiva articles is between 10/27/2015 (the day of the launch of Doing Business 2016) and 10/24/2016 (the day prior to the release of the Doing Business 2017). Most of the media articles that discuss Doing Business, especially in India and Russian Federation, tend to highlight increases in countries’ rankings, comparing them either with the preceding years or with those of neighboring economies. Many articles also focus on recent accomplishments of incumbent governments. India’s media, for example, discusses the ambitious plans and strategies of the government to secure a position on the Doing Business’ top-performers list. In Mexico, Doing Business related media coverage often outlines recent reforms and investment attractiveness of the economy, while also highlighting remaining                                                              6  Primary data collection on media coverage of Doing Business has been done through the use of the Factiva database. Factiva is an online comprehensive database primarily designed to provide data to business communities, researchers and specialists, aggregating content from both private and public sources. It pulls data from more than 32,000 information sources, including newspapers, journals, magazines, television and radio transcripts. Factiva has an almost universal reach, allowing to filter data by local and international sources.   7  Media outlet is considered to be a written publication either in print or electronic. If the same outlet  produces both print and online outputs, only one is counted.  12    much-needed improvements in the business regulatory environment. Pakistani media mostly discusses the project’s data in the context of the country’s economic challenges and calls for the need to receive further financial assistance from international donors. In general, the key trending issues covered by the local media citing Doing Business reports include economy rankings, regional benchmarks, business regulatory obstacles faced by SMEs as well as political and economic performance of current governments. It is less common, however, for media to explicitly criticize governments in the countries with comparatively low ease of doing business rankings. The OECD high-income group has the highest media coverage of Doing Business, with over 4,250 articles citing the project in the period from 2004 to 2017. Similarly, media across East Asia and the Pacific region offers Doing Business prominent attention, discussing the project in almost 3,000 articles during the same time-frame. Conversely, Middle East and North Africa and Sub-Saharan Africa are the regions with relatively low overall media coverage of Doing Business, with about 1,240 and 1,540 news articles mentioning Doing Business from 2004 to 2017 (see figure 2). The latter two are also the regions where prominent Doing Business media coverage tends to concentrate in only a limited number of economies. For example, across Sub- Saharan Africa, in nine countries Doing Business has been mentioned at least 50 times over the past 15 years; in 16 economies less than 10 times, while media in eight Sub-Saharan countries never mentioned Doing Business (as per Factiva research). However, such media coverage trends do not indicate that governments in Africa or Middle East do not pay close attention to Doing Business. For instance, despite low Doing Business media coverage, economies in Sub-Saharan Africa carried out 107 business regulatory reforms (as measured by Doing Business) in 2017/18 alone and numerous regulators on Africa’s continent are closely working with the World Bank Group on Doing Business driven reform design and implementation. And Sub-Saharan Africa is the leading region in terms of an overall number of business regulatory reforms implemented since 2004. Moreover, the regions with relatively low Doing Business media coverage also have comparatively low number of media outlets, with an average of nine main media outlets per country operating in Sub-Saharan African economies and 28 in the Middle East and North African ones (see figure 2). In comparison, the average number of media outlets in OECD high- income countries is 447. Figure 2: Doing Business media coverage is the highest in OECD high income economies Media articles   Media outlets  Region  (number, Factiva 2004‐17)  (number)  Middle East & North Africa                           1,240   28  Sub‐Saharan Africa                           1,548   9  Latin America & Caribbean                           1,885   39  Europe & Central Asia                           2,679   48  South Asia                           2,892   113  East Asia & Pacific                           2,966   109  High income: OECD                           4,266   447  13    Yet, it is important to note that when we assess media articles as percentage of total media outlets in each economy, East Asia and the Pacific stands out as the region with the highest proportion of media articles covering Doing Business. Inversely, Latin America and the Caribbean, OECD high income and Middle East and North African countries have the lowest percentage of media articles in total media outlets (see figure 3a). Furthermore, as numbers of media articles discussing Doing Business can substantially fluctuate from year to year, this pattern varies with each Doing Business publication cycle. For example, if we isolate the period of 2016/17 and perform the same analysis, a different pattern emerges (see figure 3b). Figure 3: East Asia and the Pacific has the highest proportion of media articles on Doing Business in total media outlets Figure 3a Figure 3b Media articles as % of outlets Media articles as % of outlets, latest year only 62.31 150 60 125.86 116.60 100 39.40 40 % 70.34 % 26.90 24.88 50 20 35.49 32.02 11.74 11.40 8.84 13.49 8.38 0 0 D ia a an a a fic si ri c ric EC As ia sia D a an fic a ci be lA ric ric Af Af EC Pa As ci O be ib h lA tra Af Af Pa ut th an ar O e: ib & h tra So en or ut C an th r om ar ia e: & ha N So en C or & As C r om a c ha Sa & N & C si a in & ic st c st Sa A b- pe t& & h a in er Ea Ea ig Su ic st ro b- pe s Am h H er Ea Ea ig Eu Su e ro Am dl H tin Eu e id dl La M tin id La M Source: Factiva. Across all income groups (see figure 4), lower-middle income economies had the highest Doing Business media coverage in 2016/17, averaging at 24 articles per country. Among them, press in Pakistan, India, Philippines and Nigeria cited the report the most, which positively correlates with these countries’ active reform implementation efforts given regulators’ ambitious reform agendas. Among the low-income countries, media outlets in Zimbabwe, Nepal, Uganda and Rwanda publicize Doing Business the most. Notably, Rwanda is the regional and this income group’s leader in terms of reforming its business regulatory environment, with more than 50 reforms implemented from 2004 to 2017. Most of these reforms impacted the areas of dealing with construction permits, getting credit and trading across borders. Media outlets in Nepal and Uganda were among the first ones within the low- income group of countries to cite Doing Business, in 2008 and 2009 respectively. By contrast, Zimbabwe’s press, despite the country’s active implementation of business regulatory reforms, commenced discussing the report’s rankings only five years ago.  Interestingly, data show that there is more Doing Business media coverage in countries with less free press. One explanation could be that after successful reform implementation, non- democratic governments encourage local media to emphasize and promulgate positive changes. Media in four of the top 10 countries, in terms of press coverage of the Doing 14    Business report, is assessed by the Freedom House’s Freedom of the Press rating as not free (Pakistan, Russian Federation, Malaysia, United Arab Emirates)8; in five as only partially free (India, Mexico, Philippines, Italy, Nigeria) and in only one – Spain – as completely free. As further discussed in the analysis section, to eliminate the bias of authoritarian governments’ control over media, we perform panel data analysis excluding authoritarian and semi- authoritarian governments. Figure 4: Doing Business media coverage by income group and reforms implemented in 2004-17 Number of  Income group  Media articles  (number, Factiva 2004‐17)  implemented reforms   (2004‐18)  Lower middle income  6,530  972  High income  5,684  839  Upper middle income  4,759  882  Low income  503  495  Variable selection and summary statistics One of the key outcome variables used to test the set hypothesis is “number of Doing Business reforms” in a given country in a given year. The number of implemented Doing Business reforms is a simple count of reforms carried out by each country per Doing Business cycle. The sum of this variable can fluctuate from 0 (no reforms) to 10 reforms,9 which is the maximum possible number of reforms, assuming a country reformed on each of the Doing Business indicators. Aside from the reform count, Doing Business presents data in terms of two main aggregate measures: the Doing Business score and the ease of doing business ranking, which is based on the Doing Business score. The ease of doing business ranking compares economies with one another; the Doing Business score benchmarks economies with respect to regulatory best practice, showing the absolute distance to the best performance on each Doing Business indicator. When compared across years, the Doing Business score shows how much the regulatory environment for local entrepreneurs in an economy has changed over time in absolute terms, while the ease of doing business ranking can show only how much the regulatory environment has changed relative to that in other economies. Therefore, for the purpose of this paper, we mostly use the Doing Business score as one of the main outcome variables. In the historical data set the minimum Doing Business score is 19.98 percentage points (the country is far removed from the perfect score of a100), while the highest Doing Business score is 90.41 (see Table 1 in the Annex).                                                              8 Average of the overall grades of the level of press freedom by year from 2004 to 2016. Source: Freedom House. 9  In general terms, a reform is defined as a legal or regulatory change that makes it easier for local firms to do  business. Reforms can have different impact magnitudes and Doing Business counts only one reform per topic  per country in any given year.   15    As the primary explanatory variable, we employ the “number of local media articles” mentioning Doing Business. This country-specific variable captures the total number of local press articles discussing economies’ performance on the Doing Business indicators over each Doing Business publication year (for more information see the Methodology chapter). With the clearly identified dependent and independent variables put in place, we selected a number of controls imperative for the analysis. First and foremost, the model controls for income differences across countries. To this end, it employs the classical control variable of log income per capita in current USD of the World Development Indicators, the World Bank Group. In addition, using Factiva search engine, we collected the data on the number of media outlets per country. As historical data on media outlets is not available, we use the data for the most recent years – 2017/18. The data show that high income OECD countries have on average over 400 media outlets per economy, compared to only 9 in Sub-Saharan Africa. Another prominent control included in the analysis is freedom of the press variable of the Freedom House. This variable was selected due to its comprehensive global coverage and availability of gap-free historical data. As discussed in the literature section, Freedom House data, and freedom of press variable specifically, are widely used in the existing literature on drivers of reforms. Other sets of variables that we considered and selected as controls only in simple non-panel OLS models include World Justice Project’s rule of law index, Polity IV data on democracy and autocracy, Transparency International’s corruption perception index, the Economist’s democracy classification score and the Worldwide Governance Indicators of the World Bank Group. Prior to testing the aforementioned hypothesis, we first run a number of simple correlations and perform data robustness checks. First, we correlate year on year Factiva and Google data, which yields an average correlation coefficient of 0.6. This is not surprising as Factiva’s data only captures local media, while Google does not differentiate between foreign and local press coverage. Therefore, in the subsequent analysis the two data sets are used as complementary independent variables rather than substitutes. However, we give more prominence to Factiva’s data due to our interest in targeted local media that specifically focuses on domestic politics and local economic development. We then check all the variables for outliers and isolate exorbitant cases. For example, in some of the OLS regressions we limit the number of media articles per year to either 100 or 200 as few of the countries with 200+ articles considerably skew the relationships, thus exaggerating coefficients. 16    VI. Discussion of the results Analysis section The primary theory that we are testing in this paper is whether mass media has the ability to influence political actions. We use the Doing Business data as a proxy to test the set hypothesis. Specifically, we analyze whether media coverage of Doing Business serves as an impetus for the implementation of business regulatory reforms. An innocuous argument could be that countries that are active reformers on Doing Business indicators naturally tend to have higher media coverage. But can media coverage actually spur reforms and turn regulations into actionable outcomes? To answer the posed question, we collected primary data on media coverage for 190 countries worldwide. Using Factiva and Google databases we were able to compile a comprehensive panel data set, perfectly matching Doing Business historical time series on the number of implemented reforms as well as the Doing Business scores and rankings (see section on methodology outlining the data collection process and limitations). To explore strengths and directions of relevant correlations, we first run a number of simple OLS regressions with the latest data at hand (Doing Business 2018). In line with the set hypothesis, the primary dependent variable - “Reforms” - comprises absolute numbers of implemented Doing Business reforms by each of the sampled 190 economies over a given period of time. The set parameters for our applied OLS regression model are the following: Model 1: _ β The explanatory variable “local_media” captures country-specific Doing Business media coverage in terms of numbers of local media articles addressing Doing Business in a given year (spanning over the time period between the publication of two consecutive Doing Business reports). Our logical assumption is that media coverage’s impact on reform implementation is not immediate. A certain period of time needs to elapse between media coverage of Doing Business and possible subsequent regulatory changes. Therefore, we construct the regression model with both one and two-year time lags between the dependent and independent variables, with media coverage variable lagging behind. Also, for this model we employ a standard set of controls represented as Xj. The controls include income per capita, degree of free press and levels of Doing Business scores. Subsequently, we test a logit regression model (Model 2) with a binary outcome variable being whether a country reformed on Doing Business indicator(s) over a set period of time or not. The explanatory variable in this model remains the number of local Factiva articles – “local_media” (with both one and two-year lags) and we use the same standard set of controls, represented by Xj. In addition, in Model 2 we control for the number of total media outlets in each country as coefficients could be inflated in economies with high media outlet concentration. Model 2: Pr reformer 1 γ γ local_media In the following OLS regression model (Model 3) we use year on year deltas of Doing Business scores (∆ DTF as an outcome variable since countries’ performance can either improve or deteriorate over time. Improvements on the Doing Business measure and the magnitude of such improvements are commonly driven by the pace and impact of reforms. In Model 3, as in the previous two models, number of local media articles covering Doing Business serves as an 17    independent variable, “local_media” (with both one and two-year lags), and log income per capital and number of media outlets are used as controls, represented by Xj. Model 3: ∆DTF δ δ local_media Non-panel OLS regression results show a strong positive relationship between the outcome variable of total number of Doing Business reforms implemented by the measured economies and number of local media articles mentioning Doing Business. The relationship is significant at the 1 percent level using both one- and two-year lags on the explanatory variable (see Annex A, Tables A1 and A2). The correlation remains robust with and without the use of the standard set of controls: income per capita, level of Doing Business, free press and number of media outlets. The difference in coefficients between one- and two-year lags on the explanatory variable is minimal, indicating that regulators actively reform in the period of two years after Doing Business’ media exposure. Also, this relationship holds true when employing “Google articles” as a predictor variable. Therefore, at this stage in the analysis, without claiming any causal interaction, we can cautiously suspect that the set forth hypothesis could be true. There is a strong possibility that local or even local and international media coverage of Doing Business could potentially motivate domestic reform efforts. To complement the core OLS model, we replicate the analysis with splitting reforms into two separate categories. The first category includes reforms that focus on complexity and cost of business regulatory processes. Doing Business indicators pertaining to this category are starting a business, obtaining a building permit, connecting to electricity, transferring property, paying taxes and trading across borders. The second category of reforms is aimed at strengthening legal institutions and improving countries’ legal frameworks. Doing Business indicators relevant to the latter group include getting credit, protecting minority investors, enforcing contracts and resolving insolvency. Reforms that pertain to the first category are relatively easy to implement, compared to the legal reforms that can take years to yield desired outcomes. Legal reforms can also coincide with re-election cycles, thus stifling governments’ efforts to change the existing laws to avoid any negative publicity and/or unwelcome criticisms. The results remain significant for the two separate reform categories (see Table 2 in the Annex). However, as expected, the coefficients for Factiva media coverage are even higher for the category of reforms addressing complexity and cost of business regulatory processes. This can be explained by the fact that reforms designed to cut the red tape are relatively easier to implement and can yield measurable positive changes much faster compared to regulatory changes that strive to improve functioning of legal institutions. Results stemming from Model 2, where we employ logit regression due to the binary nature of the selected outcome variable (whether a country is a reformer on Doing Business indicators or not), fail to indicate significant outcomes. Yet, all the controls come out statistically significant with at least a 5 percent level of significance. The lack of significant coefficients of the explanatory variable could be explained by the fact that over 70 percent of the sampled 190 economies in 2016/17 did reform at least on some of the Doing Business indicators. And countries reformed at least one business regulatory area regardless of the magnitude of local media coverage of Doing Business. However, in the case of local media coverage, what matters most is the intensity of undergoing reform efforts in each economy rather than whether a country is reforming at all or not. Given the previous findings (see Model 1), it is not imprudent to assume that countries where media gives prominent attention to Doing Business results tend 18    to have more comprehensive active reform agendas compared to the ones with weak or absent media coverage of Doing Business. The third simple OLS regression model with difference in Doing Business scores as an endogenous variable yields significant results with both one and two-year reform implementation lags (see Annex A, Table A3). The results hold after controlling for income per capita and number of media outlets in each country. This positive and robust correlation supports our initial assumption that Doing Business’ local press coverage could potentially have an impact on the scope and magnitude of reform efforts carried out by regulators. In the next section we further test these assumptions using several panel regression models. Econometric model – Panel data To test whether Doing Business media publicity has any impact on business regulatory reforms, we establish the following specifications: Model 1: ∑ In this panel regression t stands for “year” and j for “country.” Within the parameters of this model, reforms can assume several connotations: (i) total number of reforms in a given country and year; (ii) delta in Doing Business score for a given country and year; and (iii) binary mode for whether a country reformed or not in a given year. Similarly, media variable can be expressed as (i) number of media articles mentioning Doing Business in country “j” and year “t” and (ii) number of media articles mentioning Doing Business as percentage of number of media outlets covered in country j and year t. In turn, Xjt stands for a series of controls, including income per capita, countries’ Doing Business scores, levels of democracy and freedom of press. We hypothesize that local press coverage of Doing Business motivates domestic reform efforts, assuming one and two-year lags (specified as “p”) in reform implementation. The second model is similar to the first one with the main difference that it entails an analysis with an interaction term (mediajt*freepressjt), where “mediajt” is number of media articles and “freepressjt” level of free press. The introduced interaction term comprises an explanatory variable on the number of Doing Business local media articles and the level of free press variable. This model tests whether media has more or less impact in countries with free press. The “free press” variable is also included as one of the controls in the ∑ cohort of the standard controls. We expect the relationship between the interaction term and the outcome variable to be negative, as the interaction term captures whether the impact of media on reforms becomes weaker or stronger when countries have more free press Model 2: ∗ ∑ With Model 3 we test a supposition that countries’ substantial changes in rankings can lead to more prominent local media coverage of Doing Business results, which subsequently could prompt regulators to implement reforms. Regulators and media oftentimes focus more on Doing Business ranking and changes in ranks from year to year, which is a relative measure, 19    rather than on the Doing Business score, which is an absolute measure. With this objective in mind, we develop the following specifications for the panel regression model: Model 3: ∑ Model 3 is similar to Models 1 and 2 discussed above. In Model 3, however, the outcome variable is number Doing Business related media articles in a given country in a given Doing Business publication year and “Δ ranking” represents year on year changes between published rankings. For the delta ranking variable we apply both one and two-year lags, to allow enough time for reform implementation processes to take effect. Model 3 also controls for free press, levels of Doing Business scores and income per capita, expressed as Xjt. To limit the omitted variable bias, we introduce fixed effects regressions to allow us exploit within country variations over time. We deem the fixed effects model to be an appropriate specification as there is substantial year-on-year country-specific variation of both reform implementation and media coverage. As Model 4 equations below indicate, we apply fixed effects to both country and time variables. In Model 4 below and represent country and time fixed effects respectively, while Xjt represents the standard set of controls. We expect results of the fixed effect models to be similar to those of OLS regressions. Model 4: The panel regression results support our original hypothesis that local media coverage of Doing Business does tend to influence regulators’ reform agendas, especially its magnitude. As evidenced by the regression output, there is a strong and positive relationship between the number of implemented reforms on Doing Business indicators and local media coverage of Doing Business, applying both one and two-year lags on press releases (see Annex B, Table B1, Model 1.1a, which refers to Model 1 above). Therefore, one could argue that in countries where media is most vocal about Doing Business, regulators tend to pay closer attention to their economies’ performance on the Doing Business indicators, placing business regulatory reforms high on political agendas. The results are more robust when using Factiva data as an explanatory variable as opposed to the Google data. This is most likely because Google variable includes international press in addition to local one, which dilutes the domestic focus. However, we do get significant results with Google media in specifications 1 and 3 of the model but the coefficient becomes weak and insignificant when we add a control for level of Doing Business score. In the case of Factiva’s data, the relationship is significant at the 1% level in all three specifications, controlling for free press, level of Doing Business score and income per capita. Direction of coefficients on the income controls also indicate that poorer countries tend to reform more than the richer ones. This message is clearly aligned with the findings of annual Doing Business reports since most developed economies already follow international best practices across Doing Business topics and have less need for exponential business regulatory changes. A curious, albeit not entirely surprising, finding is that freedom of press control variable shows positive and highly significant coefficients across all the models. This finding suggests that countries with less free press reform more and that there is higher Doing Business media coverage in such economies. One explanation could be that most countries with state- 20    controlled media are still underdeveloped and hence need to reform more than the developed economies. Hence, local media in authoritarian or semi-authoritarian regimes gives more prominence to regulatory reform coverage. Another reason could be that prior to implementing business regulatory changes, non-democratic administrations encourage local media to emphasize and promulgate upcoming positive reforms. To ensure that our results are not driven by the authoritarian regimes with the tendencies of media manipulation, we re-run all the models isolating authoritarian and semi-authoritarian governments from the sample. After introducing such specifications, all the results still hold with the same levels of significance. When we test the same model with whether a country reformed or not in a given year as a dependent parameter, we find that the number of local media articles tends to increase the probability of whether a country reforms or not (Table B2, Model 1.2b). And as stipulated in the previous model, freedom and press and higher income per capita are associate with a lower probability that a country reforms in a particular year. Table B3: Model 2.1a in Annex B shows the results after we expand the specifications of the previous models to include an interaction term with number of media articles and free press. With the addition of the interaction term, the number of media articles loses significance when the outcome variable is the number of reform count but remains significant in the case when the outcome variable is whether an economy is a Doing Business reformer or not, although the impact is very small. The coefficient of the interaction term is close to zero, albeit significant. When analyzed separately, the correlation coefficient between the number of media articles and free press is 0.6. Moreover, in this model we introduce a complementary independent variable – media articles as a percentage of media outlets (in each country). As media outlets considerably vary from country to country and from region to region, thus potentially either inflating or deflating the number of articles discussing Doing Business, we wanted to have a somewhat more objective estimate of Doing Business coverage. The results with this new independent variable turn out to be significant with both outcome variables: Doing Business reform count and Doing Business reformer, further substantiating our hypothesis (these results apply to both models 1 and 2). The logit specification (see Table B4, Model 2.2a in Annex B) further indicates that countries with broader Doing Business media coverage are more likely to reform on Doing Business indicators – aka be Doing Business reformers. The results show that an increase in the number of media articles by 1, increases the probability of being a Doing Business reformer by 1.3%. The coefficients of all the control variables remain highly significant and maintain their directions, in line with the previous models. These results hold with both one and two-year lags applied to the independent variable of local media articles. Model 3a (see Table B5 in Annex B) tests a different supposition in regard to inverse relationship. It employs number of media articles as an outcome variable and year-on-year change in Doing Business rankings as an explanatory variable, applying both one and two-year lags as well as the standard set of controls. Given this inverse relationship, the year-on-year change in the ranking of an economy is not significant in both specifications, indicating that the previous year’s change in the ranking has no impact on the media coverage of Doing Business reforms. Therefore, our initial assumption that relatively substantive changes in rankings tend to attract more media attention is not supported by data findings. However, a 21    higher Doing Business score is associated with a higher number of articles mentioning Doing Business. When using a fixed-effect regression model to account for within country and within year variation as well as to eliminate the impact of time-invariant characteristics, the results come out statistically significant at a 1% level when using media articles as an independent variable and at a 10% level with media articles as percent of media outlets (see Table B6: Model 4 in Annex B). The results hold with and without controls. This supports the findings of the previously tested models. With the fixed effects model the intraclass correlation is about 30% (variance due to difference across panels). VII. Conclusion Based on the observed results, local media, if used prudently, can serve as a powerful reform impetus. Every year Doing Business prepares regional and country specific media briefings as well as gives numerous interviews to local press across the globe. When regulators receive well-informed press releases on the status of Doing Business indicators in their spheres of influence, they are likely to act on this information. Therefore, the annual upward media coverage trajectory of Doing Business is expected to continue, inspiring positive change and enticing rule makers to further improve the regulatory environment for local businesses. These data findings also allow us to cautiously draw a more general conclusion, presuming that media does have the power to impact political actions. To further substantiate our claims, more similar studies and research are warranted. Continuing to use Doing Business as a proxy, it would be interesting to investigate the quality and relevance of the implemented reforms. One could also explore whether media influences the actual process of reform selection and identification. For instance, in cases when media gives more prominence to reforms tackling bureaucratic processes and red tape, do regulators tend to focus reform efforts primarily in this area or do they go beyond what is discussed by the press? Another viable supposition is that countries care about their performance on Doing Business, or any other indicators, vis a vis their neighbors. There is commonly a race to be the best among the peers. Hence, further analysis could be carried out to measure the magnitude of neighbors’ influence on local reform agendas. Given the data at hand, we used a somewhat narrow scope (Doing Business data) to test a broader hypothesis. However, it would be beneficial to test the same assumptions in other important areas such as health, education, environment and social services. An example of a similar question to ask could be – does media coverage/criticism of poorly functioning education systems lead to reforms aimed at improving the quality of schools? Alternatively, a similar question to pose is - does media attention to environmental degradation spur regulators to adopt more environmentally friendly policies? Researchers, policy makers and media outlets would all greatly benefit from finding robust answers to these and similar questions. 22    Annex Table 1: Summary statistics Variable  Obs.  Mean  Std. Dev.  Min  Max  Reform count  Source: Doing Business  2,660  1.20  1.35  0  8  Doing Business reformer (yes/no)  Source: Doing Business  2,660  0.61  0.49  0  1  Doing Business (Global) score  Source: Doing Business  1,688  60.09  12.95  19.98  90.41  Doing Business rank  Source: Doing Business   2,560  92.14  53.15  1  190  Number of local Doing Business  articles  Source: Factiva  2,470  6.34  19.32  0.00  305.00  Number of local Doing Business  articles  Source: Google  2,458  3.22  13.96  0.00  266  Freedom of the Press  Source: Freedom House  2,435  47.23  23.35  0.00  97.00  Income per capita, log in current  USD  Source: World Development  Indicators  2,703  8.23  1.57  4.47  11.54  Table 2: OLS regressions with two different reform categories Doing Business reform count  Doing Business reform count  Variable (bureaucracy reforms) (legal reforms) Lag 1 year Lag 2 years Lag 1 year Lag 2 years Media articles (Factiva) 0.006*** 0.005*** 0.003** 0.002* [0.002] [0.001] [0.001] [0.001] Log income per capita, current USD 0 0 0 0 [0.000] [0.000] [0.000] [0.000] Level of distance to frontier 0.020*** 0.020*** 0.011* 0.011* [0.006] [0.007] [0.006] [0.006] Free press  0.012*** 0.013*** 0.009*** 0.010*** [0.003] [0.003] [0.004] [0.004] Number of media outlets ‐0.000** ‐0.000** 0 0 [0.000] [0.000] [0.000] [0.000] Observations 184 184 184 184 R‐squared 0.177 0.164 0.098 0.094 Constant ‐1.047** ‐1.111** ‐0.401 ‐0.432 [0.451] [0.457] [0.458] [0.456] Robust standard errors in brackets *** p<0.01, ** p<0.05, * p<0.1 23    Annex A – Cross-section data Table A1: Model 1 - OLS regression model with “total Doing Business reforms” as an outcome variable and the standard set of controls Variable  Doing Business reform count   (0‐8)    lag 1 year  lag 2 years  Media articles (Factiva)  0.010***  0.009***    [0.003]  [0.002]  Log income per capita, current  ‐0.000***  ‐0.000***  USD    [0.000]  [0.000]  Level of Doing Business score  0.041***  0.042***    [0.009]  [0.009]  Free press   0.021***  0.023***    [0.005]  [0.005]  Number of media outlets  ‐0.000**  ‐0.000***    [0.000]  [0.000]  Observations  184  184  R‐squared  0.247  0.241  Constant  ‐1.881***  ‐1.967***    [0.632]  [0.637]  Robust standard errors in      brackets  *** p<0.01, ** p<0.05, * p<0.1      Note: Most annex tables present results with the Factiva data set only, as Factiva data capture local media coverage with better precision than Google. Hence, results are more robust when employing the Factiva variable. All the results hold when we take the log media articles, which applies to all the relevant tables. (See methodology section for more details.) Table A2: Model 2 - Logit regression model with “Doing Business reformer (yes/no)” as an outcome variable and the standard set of controls Variable  Doing Business reformer (yes/no)  (0‐8)    lag 1 year  lag 2 years  Media articles (Factiva)  0.045  0.05    [0.042]  [0.031]  Log income per capita, current USD  ‐0.000**  ‐0.000**    [0.000]  [0.000]  Level of Doing Business score  0.081***  0.079***    [0.025]  [0.023]  Free press   0.034***  0.035***    [0.012]  [0.011]  Number of media outlets  ‐0.002**  ‐0.002***    [0.001]  [0.001]  Constant  ‐5.039***  ‐5.007***    [1.637]  [1.516]  Observations  184  184  Robust standard errors in brackets      *** p<0.01, ** p<0.05, * p<0.1      24    Table A3: Model 3 - OLS regression with year on year deltas of Doing Business scores as an outcome variable Difference in distance to frontier scores (2017‐16) Variable lag 1 year lag 2 years Media articles (Factiva) 0.006*** 0.004* [0.002] [0.002] Log income per capita, current USD ‐0.000** ‐0.000*** [0.000] [0.000] Number of media outlets 0 0 [0.000] [0.000] Observations 118 111 R‐squared 0.066 0.058 Constant 0.668*** 0.740*** [0.152] [0.154] Robust standard errors in brackets *** p<0.01, ** p<0.05, * p<0.1 25    Annex B – Panel data Table B1: Model 1.1a – panel regression output with controls and a two-year reform implementation lag   Doing Business reform count (0‐8)  Variable              (1)  (2)  (3)  (4)  (5)  (6)  Media articles (google, lag 2)  0.013***    0.008*    0.005      [0.005]    [0.005]    [0.005]    Media articles (Factiva, lag 2)    0.015***    0.012***    0.009***      [0.002]    [0.002]    [0.002]  Free press       0.009***  0.007***  0.013***  0.012***        [0.002]  [0.002]  [0.002]  [0.002]  Level of Doing Business score          0.042***  0.039***            [0.004]  [0.004]  Log income per capita,  ‐0.078***  ‐0.089***  ‐0.003  ‐0.023  ‐0.261***  ‐0.264***  current USD    [0.018]  [0.018]  [0.022]  [0.022]  [0.035]  [0.035]  Observations  2,218  2,229  2,013  2,023  1,481  1,488  R‐squared  0.015  0.044  0.023  0.04  0.097  0.108  Constant  1.906***  1.940***  0.889***  1.065***  0.355  0.577**    [0.156]  [0.153]  [0.240]  [0.235]  [0.298]  [0.293]  Robust standard errors in              brackets  *** p<0.01, ** p<0.05, * <0.1              Table B2: Model 1.2b – panel regression output with controls and a two-year reform implementation lag   Doing Business reformer (yes "0"/no "1")  Variable      (1)  (2)        Media articles (google, lag 2)  0.001      [0.002]    Media articles (Factiva, lag 2)    0.002***      [0.001]  Free press   0.003***  0.003***    [0.001]  [0.001]  Level of Doing Business score  0.013***  0.012***    [0.001]  [0.001]  Log income per capita, current USD  ‐0.075***  ‐0.076***    [0.012]  [0.012]  Observations  1,481  1,488  R‐squared  0.062  0.07  Constant  0.354***  0.405***    [0.097]  [0.098]  Robust standard errors in brackets      *** p<0.01, ** p<0.05, * p<0.1      26    Table B3: Model 2.1a – panel regressions with an interaction term Variable  Doing Business reform count   Doing Business reformer   (0‐8)  (yes "0"/no "1")  Media articles   0.002    0.001***    (Factiva, lag 2)    [0.003]    [0.001]    Media articles as % of media outlets  0.002**    0.001***  (Factiva, lag 2)      [0.001]    [0.000]  Interaction term   0.00014***  0.0001***  0.0001**  0.0001***  (number of media articles*free press)    [0.000]  [0.000]  [0.000]  [0.000]  Free press   0.010***  0.010***  0.003***  0.002***    [0.002]  [0.002]  [0.001]  [0.001]  Level of Doing Business score  0.039***  0.038***  0.012***  0.012***    [0.004]  [0.004]  [0.001]  [0.001]  Log income per capita, current USD  ‐0.267***  ‐0.267***  ‐0.076***  ‐0.074***    [0.034]  [0.035]  [0.012]  [0.012]  Observations  1,488  1,442  1,488  1,442  R‐squared  0.119  0.124  0.072  0.079  Constant  0.678**  0.711**  0.419***  0.413***    [0.292]  [0.293]  [0.099]  [0.100]  Robust standard errors in brackets          *** p<0.01, ** p<0.05, * p<0.1          Note: As the data on media outlets is only available for the most recent year, we used it as a proxy for all the previous years, assuming that media outlets data does not fluctuate substantially from year to year. Factiva database has been used to obtain the media outlets data. Table B4: Model 2.2a – Logit panel regression with interaction   Doing Business  reformer   (yes "0"/no "1")  Variable    (1)  Media articles (Factiva, lag 2)  0.013*    [0.007]  Interaction term   0  (number of media articles*free press)    [0.000]  Free press   0.013***    [0.003]  Level of Doing Business score  0.056***    [0.007]  Log income per capita, current USD  ‐0.356***    [0.059]  Observations  1488  Constant  ‐0.401    [0.498]  Robust standard errors in brackets    *** p<0.01, ** p<0.05, * p<0.1    27    Table B5: Model 3 – panel regressions with number of media articles as an outcome variable and year on year change in Doing Business rankings as an explanatory variable; one and two- year lag a) One-year lag on delta rankings Variable  Media articles  Media articles  (Factiva)  (Google)        (1)  (2)  Year on year change in rankings  ‐0.033  ‐0.039    [0.060]  [0.045]  Free press   0.195***  0.053***    [0.030]  [0.018]  Level of Doing Business score  0.321***  0.216***    [0.052]  [0.050]  Log income per capita, current  0.866*  0.546*  USD    [0.520]  [0.312]  Observations  1,461  1,454  R‐squared  0.04  0.03  Constant  ‐26.705***  ‐14.762***    [4.951]  [3.477]  Robust standard errors in      brackets  *** p<0.01, ** p<0.05, * p<0.1      b) Two-year lag on delta rankings Variable  Media articles  Media articles  (Factiva)  (Google)        (1)  (2)  Year on year change in  0.057  ‐0.009  rankings    [0.073]  [0.059]  Free press   0.201***  0.056***    [0.031]  [0.018]  Level of Doing Business score  0.333***  0.218***    [0.052]  [0.046]  Log income per capita, current  0.908*  0.583*  USD    [0.523]  [0.313]  Observations  1,447  1,440  R‐squared  0.041  0.029  Constant  ‐28.062***  ‐15.302***    [5.062]  [3.399]  Robust standard errors in      brackets  *** p<0.01, ** p<0.05, * p<0.1      Note: as the changes in rankings can be either positive or negative (or zero), yielding a concave up curve, we performed the same analysis as in Model 3 using absolute changes in rankings. Although when using the absolute changes, the coefficients have slightly improved the relationships still remained insignificant. 28    Table B6: Model 4 – panel regressions with fixed effects to control for within year and within country variation   Doing Business reform count (0‐8)  Variable      (1)  (2)  Media articles (Factiva), lag 1 year  0.007***      [0.002]    Media articles as % of media outlets, lag 1    0.002*  year      [0.001]  Free press   ‐0.005  ‐0.004    [0.010]  [0.010]  Level of Doing Business score  0.049***  0.048***    [0.011]  [0.011]  Log income per capita, current USD  ‐0.470**  ‐0.471**    [0.234]  [0.240]  Doing Business publication cycle 2011  ‐0.416***  ‐0.414***    [0.119]  [0.122]  Doing Business publication cycle 2012  ‐0.314***  ‐0.311**    [0.121]  [0.124]  Doing Business publication cycle 2013  ‐0.548***  ‐0.529***    [0.127]  [0.129]  Doing Business publication cycle 2014  ‐0.337**  ‐0.295**    [0.132]  [0.134]  Doing Business publication cycle 2015  ‐0.325**  ‐0.304**    [0.139]  [0.142]  Doing Business publication cycle 2016  ‐0.338**  ‐0.305**    [0.144]  [0.146]  Doing Business publication cycle 2017  ‐0.124  ‐0.084    [0.142]  [0.146]  Economy coefficients  not presented due to large number of entries  Observations  1,488  1,442  R‐squared  0.046  0.043  Number of countries  189  183  Constant  2.871  2.858    [2.110]  [2.167]  Robust standard errors in brackets      *** p<0.01, ** p<0.05, * p<0.1      Note: this table presents the lag of one year due to more robust results. As the data on media outlets is only available for the most recent year, we used it as a proxy for all the previous years, assuming that media outlets data does not fluctuate substantially from year to year. Factiva database has been used to obtain the media outlets data. 29    Annex C – Variable definitions table Variable name  Variable Description  Global ease of doing business rank compares economies with  one another. It measures an economy’s relative position with  respect to other sampled countries and ranges from 1 to 190.  Doing Business rank  The rank is determined by sorting the aggregate ease of doing  business scores, rounded to two decimals. Source: Doing  Business.  Doing Business captures the gap between an economy’s  performance and a measure of best practice across the entire  sample of 41 indicators for 10 Doing Business topics. The score  Doing Business Doing Business score  ranges from 0 to 100, where 0 represents the worst regulatory  performance and 100 the best regulatory performance. Source:  Doing Business  Doing Business reform count  Number of total reforms implemented across 10 Doing  Business indicators in a given year. Source: Doing Business  A binary variable (yes/no) indicating whether a country  Doing Business reformer  reformed or not on any of the Doing Business indicators in a  given year. Source: Doing Business  A total number of local media articles discussing Doing Business  in more than one paragraph over a relevant report publication  year. Each year’s data covers the period between (a) the press  Number of media articles (Factiva)  release date of the Doing Business report for a specific year and  (b) the day prior to the press release date of the Doing Business  report of the subsequent year. Source: Factiva.  A total number of local media articles discussing Doing Business  in more than one paragraph over a relevant report publication  year. Each year’s data covers the period between (a) the press  Number of media articles (Google)  release date of the Doing Business report for a specific year and  (b) the day prior to the press release date of the Doing Business  report of the subsequent year. Unlike Factiva, Google variable  also includes articles from international press. Source: Google.  Year‐on‐year change in Doing Business  Represents a simple difference in ranking between two  rankings  consecutive years. Source: Doing Business  Freedom of the Press (FOTP) Data: Editions 2004–2016. The  Free press  scores are assigned as follows: Free (F): 0–30; Partly Free (PF):  31–60; Not Free (NF): 61–100. Source: Freedom House  Income per capita  GNI per capita, Atlas method (current US$). Source: the World  Bank Group, Open Data  Physical number of media outlets per country, based on the  Number of media outlets  latest data available (2017 for most economies). Source:  Factiva  30    References Abotsi, A. K. 2016. “Power Outages and Production Efficiency of Firms in Africa.” International Journal of Energy Economics and Policy, 6 (1): 98-104. Bailey, James B. and Diana W. Thomas. 2017. "Regulating away competition: the effect of regulation on entrepreneurship and employment." Journal of Regulatory Economics, Springer, 52 (3): 237-254. Berkowitz, Daniel, Chen Lin and Yue Ma. 2015. “Do property rights matter? Evidence from a property law enactment.” Journal of Financial Economics, 116 (3): 583-593. Besley, Timothy and Robin Burgess. 2000. “Land Reform, Poverty Reduction, and Growth: Evidence from India.” The Quarterly Journal of Economics, 115 (2): 389-430. Branstetter, Lee G., Francisco Lima, Lowell J. Taylor and Ana Venâncio. 2013. “Do Entry Regulations Deter Entrepreneurship and Job Creation? Evidence from Recent Reforms in Portugal.” The Economic Journal, 124 (577). Bripi, Francesco. 2016. “The Role of Regulation on Entry: Evidence from the Italian Provinces.” Published by Oxford University Press on behalf of the World Bank. Bruhn, Miriam. 2011. “License to Sell: The Effect of Business Registration Reform on Entrepreneurial Activity in Mexico.” Review of Economics and Statistics, 93 (1): 382–86. Bruhn, Miriam. 2013. “A Tale of Two Species: Revisiting the Effect of Registration Reform on Informal Business Owners in Mexico”. Journal of Development Economics, 103: 275-83. Bruhn, Miriam and Jan Loeprick. 2016. “Small business tax policy and informality: evidence from Georgia.” International Tax and Public Finance, 23 (5): 834-853. Brunetti, Aymo and Beatrice Weder. 2003. “A free press is bad news for corruption.” Journal of Public Economics, Elsevier, 87 (7-8): 1801-1824. Chemin, Matthieu. 2009. “The impact of the judiciary on entrepreneurship: Evaluation of Pakistan’s “Access to Justice Programme.” Journal of Public Economics, 93 (1-2): 114- 125. Ciccone, Antonio and Elias Papaioannou. 2007. “Red Tape and Delayed Entry.” Journal of the European Economic Association, 5 (2–3): 444–58. Clark, Ximena, David Dollar and Alejandro Micco. 2004. “Port Efficiency, Maritime Transport Costs and Bilateral Trade.” Journal of Development Economics, 75 (2): 417-50. Curtis, Chadwick. 2016. "Economic Reforms and the Evolution of China's Total Factor Productivity." Review of Economic Dynamics, Elsevier for the Society for Economic Dynamics, 21: 225-245. 31    Darnihamedani, Pourya, Joern Hendrich Block, Jolanda Hessels and Aram Simonyan. 2018. “Taxes, start-up costs, and innovative entrepreneurship.” Small Business Economics, 51 (2): 355–369. Demenet, Axel, Mireille Razafindrakoto and François Roubaud. 2016. “Do Informal Businesses Gain From Registration and How? Panel Data Evidence from Vietnam.” World Development, Elsevier, 84(C): 326-341. Djankov, Simeon, Caralee McLiesh, Tatiana Nenova, and Andrei Shleifer. 2003. “Who Owns the Media?” Journal of Law and Economics, 46: 341-382. Djankov, Simeon, Rafael La Porta, Florencio López-de-Silanes and Andrei Shleifer. 2002. “The Regulation of Entry.” Quarterly Journal of Economics, 117 (1): 1–37. Djankov, Simeon, Dorina Georgieva and Rita Ramalho. 2018. “Business regulations and poverty.” Economics Letters, Volume 165: 82-87. Doblas-Madrid, Antonio and Minetti, Raoul. 2013. “Sharing information in the credit market: Contract-level evidence from U.S. firms.” Journal of Financial Economics, 109 (1): 198- 223. Egorov, Gregory, Sergei Guriev, Konstantin Sonin. 2009. “Why resource-poor dictators allow freer media: A theory and evidence from panel data.” American Political Science Review, 103 (4): 645-668. Francken, Nathalie, Bart Minten and Johan F.M. Swinnen. 2009. “Media, Monitoring, and Capture of Public Funds: Evidence from Madagascar.” World Development, Elsevier, 37 (1): 242- 255. García-Sánchez, Isabel-María, Beatriz Cuadrado-Ballesteros and José-Valeriano Frías-Aceituno. 2016. “Does media freedom improve government effectiveness? A comparative cross- country analysis.” European Journal of Law and Economics, Springer, 42 (3): 515-537. Jerbashian Vahagn and Anna Kochanova, 2016. “The impact of doing business regulations on investments in ICT.” Empirical Economics, Springer, 50(3): 991-1008. Klapper, Leora F. and Love, Inessa. 2016. “The Impact of Business Environment Reforms on New Registrations of Limited Liability Companies.” The World Bank Economic Review, 30 (2): 332–353. Klapper, Leora, Luc Laeven and Raghuram Rajan. 2006. “Entry Regulation as a Barrier to Entrepreneurship.” Journal of Financial Economics 82 (3): 591–629. Leeson, Peter and Christopher Coyne. 2007. “The reformers’ dilemma: media, policy ownership, and reform.” European Journal of Law and Economics, Springer, 23 (3): 237-250. Leeson, Peter. 2008. “Media Freedom, Political Knowledge, and Participation.” Journal of Economic Perspectives, 22 (2): 155-169. 32    Maican, Florin G. and Matilda Orth. 2018. “Entry regulations, welfare, and determinants of market structure.” International Economic Review, 59: 727-756. McCombs, Maxwell E. and Donald L. Shaw. 1972. “The Agenda-Setting Function of Mass Media.” The Public Opinion Quarterly, 36 (2): 176-187. Norris, Pippa. 2009. “Public Sentinel: News Media and Governance Reform.” The World Bank: Washington, DC, 2009. Nunn, Nathan. 2007. “Relationship-specificity, incomplete contracts, and the pattern of trade.” The Quarterly Journal of Economics. 122 (2): 569-600. Olper, Alessandro and Johan Swinnen. 2013. “Mass Media and Public Policy: Global Evidence from Agricultural Policies.” The World Bank Economic Review, 27(3): 413–436. Paunov, Caroline, 2016. “Corruption's asymmetric impacts on firm innovation.” Journal of Development Economics, Elsevier, 118 (C): 216-231. Poel, Kevin, Wim Marneffe, Samantha Bielen, Bas van Aarle, Lode Vereeck. 2014. “Administrative Simplification and Economic Growth: A Cross Country Empirical Study.” Journal of Business Administration Research, 3: 45-58. Rocha, Rudi, Gabriel Ulyssea and Laísa Rachter. 2018. "Do lower taxes reduce informality? Evidence from Brazil." Journal of Development Economics, Elsevier, vol. 134(C): 28-49. Stromberg, David. 2001. “Mass media and public policy.” European Economic Review, Elsevier, 45(4-6): 652-663. Ulyssea, Gabriel. 2010. “Regulation of Entry, Labor Market Institutions and the Informal Sector.” Journal of Development Economics, 91 (1): 87-99. Ulyssea, Gabriel. 2018. “Firms, Informality, and Development: Theory and Evidence from Brazil.” American Economic Review, 108 (8): 2015-47. Voltmer, Katrin. 2009. “The Media, Government Accountability, and Citizen Engagement, Public Sentinel: News Media and Governance Reform.” The World Bank: Washington, DC, 2009. 33