International Corporate Governance and Finance: Legal, Cultural and Political Explanations

International Corporate Governance and Finance: Legal, Cultural and Political Explanations

Available online at www.sciencedirect.com The International Journal of Accounting 47 (2012) 1 – 43 International Corporate Governance and Finance: L...

549KB Sizes 11 Downloads 228 Views

Available online at www.sciencedirect.com

The International Journal of Accounting 47 (2012) 1 – 43

International Corporate Governance and Finance: Legal, Cultural and Political Explanations☆ Hamadi Matoussi a,⁎, Maha Khemakhem Jardak b a

Manouba University, Tunisia University of Sfax, Tunisia

b

Received 5 April 2010

Abstract We investigate the disparity and the determinants of investor protection around the world and their impact on financial market development. More specifically, we build on Williamson's (2000) new institutional economic model to explain this disparity using legal, political and cultural variables. In 2006, the World Bank established an index to measure the disparity of investor protection regulations across 81 emerging and developed countries. Our results confirm that combining classifications based on legal systems, cultures and polities is important in explaining the disparity of investor protection and market capitalization around the world. In particular, we show that the classical regression analysis is not well suited to approach this question because it gives mitigated results. However, the mediation analysis is more cohesive with our conceptual model: we confirmed an indirect effect through a path analysis involving investor protection as a mediator of the relationship between legal systems and culture on the one hand and stock market capitalization on the other hand. Nevertheless, the polity factor was shown to have only a direct effect on stock market capitalization. © 2011 University of Illinois. All rights reserved. Keywords: International corporate governance; Investor protection; Culture dimension; Legal system; Political system; Mediation analysis; Financial markets



We are grateful for the helpful comments from Professor Hadi Salehi Esfahani, who discussed our paper during the 15th ERF Annual Conference in Cairo, and from Skander Esseghaier, a member of our research Lab at Manouba University. We also thank the participants at the IIAS 2009 Symposium for their useful comments. ⁎ Corresponding author at: 9 Rue Tabriz, Riadh Ennasr I, 2080 ARIANA, Tunisia. E-mail addresses: [email protected], [email protected] (H. Matoussi), [email protected] (M.K. Jardak). 0020-7063/$ - see front matter © 2011 University of Illinois. All rights reserved. doi:10.1016/j.intacc.2011.12.001

2

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

1. Introduction Corporate governance and investor protection are key drivers of market development. Scholars have used many approaches to explain cross country variation. La Porta, Lopez, Shleifer, and Vishny (1998, 1999, 2000, 2002) develop and test the legal origin. Roe (2003) attributes these differences to polity. Licht (2001) advances the difference in culture. In this paper, we investigate the three explanations using a sample of 81 countries. Compared to previous studies, our investigation has two new characteristics. First, it covers developed and emerging economies: Christian, Muslim and other confessional countries; as well as common law and civil law countries. Second, we approach investor protection better because we use three complementary aspects: transparency of transactions (extent of disclosure index), liability for self-dealing (extent of director liability index) and shareholders' ability to sue officers and directors for misconduct. Conceptually, we build on Williamson (2000) new institutional economic model to explain this disparity using legal, political and cultural variables. Specifically, we reduce the four levels of social analysis in Williamson's model to only three: the first level represents the formal and informal institutions of a country (culture, law and polity), the second level captures its governance structure and the third level captures its economic development. Then, we operationalize our conceptual model via a mediation analysis showing that the level of investor protection mediates the impact of the historical evolution of a country on its financial market development. Our results show that legal systems and culture are robust factors and their impact on financial market development is mediated by the level of investor protection. Our results weakly support the impact of political systems. Investigating international corporate governance is important because the differences in corporate governance and ownership structure among countries have persisted during the past half-century despite convergence in economies and business practices. Companies share the same imperatives: the ability to raise new capital, the efficiency of resource allocation, the growth of firm value, and the availability of information to all decisionmakers…These imperatives should drive countries and firms in advanced economies to adopt the same and the most efficient corporate rules and structures. But a simple look at corporate ownership structure around the world shows that there are significant differences in corporate governance structure and ownership concentration. In the United States and in the United Kingdom, publicly traded corporations have diffuse ownership structure, whereas in other advanced economies and especially in Europe, firms continue to have a controlling shareholder. Concentrated family ownership in some European countries (French, Italian, and Spanish firms), strong and powerful managerial control of American and British firms, bank ownership of large blocks in Japan and mandated labor influence in Germany illustrate some of these differences in the ownership patterns in different countries, despite the global convergence of their economic practices and institutions. Recent managerial misbehavior and corporate scandal (such as accounting manipulations, self-dealing behavior, excessive sale of stocks by managers just before a decline of share price) draw much attention to corporate governance. The Enron scandal and bankruptcy raised serious doubts about investor protection in the US and led to more reform and

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

3

regulation of the financial market. 1 On July 30, 2002, the US adopted the Sarbanes–Oxley act in order to improve corporate responsibility and financial disclosures, and to combat corporate and accounting fraud. 2 Various laws and reports around the world responded to restore confidence and to reinforce investor protection. 3 As traditional research, these reforms tried to enhance corporate governance within the framework of agency theory. However, recent research argues that this theory fails to account for key differences across countries (Aguilera & Jackson, 2003; Fligstein & Choo, 2005). Thus, other theories have emerged. Researchers have the challenge to theorize and to test empirically the cross-national diversity in corporate governance and to identify key factors explaining these differences (Bebchuk & Roe, 1999; La Porta, Lopez, Shleifer, & Vishny, 1997; La Porta et al., 1998, 1999, 2000, 2002; Licht, 2001; Roe, 1994, 2003). LLSV (1998, 1999, 2000, 2002) have raised the question of investor protection around the world. They argue that investors' rights depend on the legal rules of the jurisdictions where securities are issued. LLSV attribute the differences in legal rules across countries to the differences in their legal origins. They theorize and test empirically their predictions and find that common law countries (US, UK, Canada…) have more protective laws than civil law countries (France, Germany, Italy…). However, the Enron scandal and its bankruptcy raised serious doubts about investor protection in the US and consequently about the findings of LLSV (1998, 1999, 2000, 2002). Nevertheless, the quick reaction to Enron scandal and the adoption of the Sarbanes–Oxley act of 2002 could be seen as proof of how the US investor is protected. After Enron's scandal, we need to validate empirically this assertion in order to enhance La Porta's approach and the market autoregulation by the law (Sarbanes–Oxley act of 2002). Roe (2003) challenges the LLSV theory and advances the polity to explain the differences in corporate governance around the world. Political institutions and the political orientations of governments, coalitions, ideologies and interest groups are the critical variables. Gourevitch (2005) argues that politics shape corporate governance in creating corporate law. According to Licht (2001) and Licht, Goldschmidt, and Schwartz (2002, 2005), cultural factors seem to be important in explaining the differences in corporate governance and ownership patterns. Most countries around the world have changed their rules and have been inspired by the US reform to protect their investors better. We still have to resolve one question: do these reforms adapt to different national cultures? Will other nations apply these reforms or ignore them? Lately, La Porta, Lopez, and Shleifer (2008) have challenged the cultural and political perspectives and showed that the data support the legal factor in explaining the economic consequences across countries. However, they separately consider cultural dimensions to explain economic consequences. Also, only one empirical indicator (proportional representation) approached polity. In this paper, we adopt a global approach using a synthetic measure for each dimension. We use cross-country analysis to explain differences in corporate governance and their impacts on capital market. Specifically, we compare legal, cultural See: “Recent corporate accounting scandals & the need for corporate governance reforms,” http://www.ipers. org/pdfs/news/corporategovernancereform.pdf. 2 The full text of the act is available at: http://www.sec.gov/about/laws/sae2002.pdf. 3 For example, French law “loi sur la sécurité financière”. 1

4

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

and political explanations of investor protection regulations. We attempt to see if legal rules, culture and politics have a great impact on the actual level of investor protection across countries and on their market development. In particular, we test via a mediation analysis whether the impact of the three factors on market development is direct or indirect. In order to test this impact, we use a database developed by the World Bank from a survey on 175 countries published in Doing Business 2007. This database can be mobilized for research on corporate governance, because the World Bank has been involved in determining the key characteristics of good laws and has made a major contribution by building a valuable index of investor protection. We investigate to what extent legal, cultural and political variables explain the disparity of investor protection and financial market development. The results show that legal origin and culture are key factors. They do have both direct and indirect effects on market capitalization. However, polity has a direct-only effect on market capitalization. The remainder of this paper is organized as follows: Section 2 develops the theoretical framework underlying our investigation, Section 3 presents the data and research design, Section 4 provides the empirical analysis, and Section 5 contains some concluding remarks. 2. The theoretical framework: legal, political and cultural explanations of corporate governance patterns Healy and Palepu (2001, 407) postulate that “A critical challenge for any economy is the optimal allocation of savings to investment opportunities.” They suggest two problems that complicate this allocation: asymmetric information and agency problems. These problems are more critical for Financial Markets' development, where we need the confidence of small investors. Several solutions were suggested: contracting, disclosure, corporate governance, information intermediaries, regulation and corporate control contests. We think all these solutions are linked in one way or another to a key factor: investor protection. However, investor protection is not a regulation or a law, but a “value” requested by individuals who go into Financial Markets. Three “theories,” or approaches, try to explain how to improve or assess investor protection. The object of this section is twofold: (1) develop the spirit of each approach and its empirical investigation, (2) propose a conceptual model which underlies our empirical investigation. 2.1. A survey of previous literature The literature broadly addresses three major issues to explain corporate governance patterns: the legal factor of LLSV (1997, 1998, 1999, 2000, 2002), the political factor of Roe (2003) and Gourevitch (2003, 2005), and the cultural factor of Licht (2001) and Licht et al. (2002, 2005). Roe (2003) political theory challenges the LLSV's legal theory and provides another explanation, centered on political variables, for the differences between countries. The cultural theory (Licht, 2001) argues that differences between national cultures explain the cross country differences in corporate governance.

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

5

2.1.1. The legal factor There are hundreds of legal systems around the world. But despite this variety, researchers try to group them by legal families. The advantage of this classification is that it saves time and energy in description or prediction. The classification depends on the criteria used. In the past, legal systems have often been grouped by geography, race, language, religion or official ideology. Looking at the historical development and substantive features of the legal systems around the world, we can see that many of them fall into one of two families. In the whole of human history only two peoples seem to have founded secular, comprehensive, enduring, and widespread legal systems: the Romans of the ancient world and the Anglo-Normans of the middle ages. The pedigree of civil law goes back to ancient Rome. The common law world begins in England. The common law system resulted from the victory of private landholders over king and nobility. Laws had been adopted to prevent seizure of land by the sovereign. Judges formed common laws to resolve specific disputes. After that, common law spread to British colonies including the United States, Canada, Australia, and India. The civil or Romano Germanic law system is the oldest, the most influential and the most widely distributed around the world. It originated in Roman law, uses statutes and comprehensive codes as its principal means of ordering legal material and it relies greatly on legal scholars to formulate its rules. Scholars have identified three civil law traditions: French, German and Scandinavian. In France, Napoleon created the French civil law system because he did not want judges to have the discretion to restore feudal privileges after the French revolution. The French commercial code was written in 1807 and was brought by army to Belgium, Netherlands, Italy, part of Poland, Saharan Africa, Indochina and French Caribbean islands. France extended her legal influence as well in Luxembourg, Portugal and Spain. It is mainly French civil law that the lawmakers of new nations rely on for inspiration. The German civil law system provides for the independence of judges and the protection of individual property rights. It consists of a hybrid system that has proved effective in promoting economic growth. One proof of the effectiveness of the German system is that Japan and Korea have borrowed it and also experienced economic success. The Scandinavian law system is usually viewed as a part of civil law tradition although its law is less derivative of Roman law than it is of French and German traditions. LLSV (1997, 1998, 1999, 2000, 2002) brought forward the legal factor. They argue that laws and their enforcement are central to understanding the patterns of corporate governance around the world. The legal origin of laws is viewed as the primary factor that affects almost all other variables affecting corporate governance and that exhibits the highest degree of exogeneity. LLSV (1999, 2000, 2002) have shown that common law and civil law systems have an impact on investor protection, ownership structure and financial markets. Common law countries (US, Canada, New Zealand, Australia…) have the strongest protection of outside investors (both shareholders and creditors) and their systems lead to ownership dispersion and a strong market valuation, whereas French civil law countries (French and Spanish colonies) have the weakest protection, leading to ownership concentration. German civil law countries (Germanic countries in Europe and a number of countries in East Asia) and Scandinavian countries are in between and have stronger protection of creditors.

6

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

LLSV (1998) examine empirically how laws protecting investors differ across 49 countries and how the quality of their enforcement varies. However, other researchers criticized LLSV's investor protection index and developed a new index with contradictory findings. Lele and Siems (2006) built a new shareholder protection index for two kinds of investors: active and passive shareholders. Their main findings are that shareholder protection has been improving during the last three decades, the protection of minority shareholders is significantly stronger in blockholder countries, and that convergence in shareholder protection has taken place since 1993 and increased since 2001. They conclude that the differences among the four developed countries challenge the conjecture that there is a distinction between the Anglo-Saxon world and continental Europe. Djankov, Hart, McLiesh, and Shleifer (2006), Djankov, Lopez, La Porta, and Shleifer (2008) built a new indicator of investor protection calculated for 72 countries: the antiself-dealing index against expropriation by insiders. This index focuses on self dealing explicitly while the previous indicators neglected this dimension. Their findings can be summarized as follows: (1) common law countries have more developed stock markets than civil law countries, in particular, French civil law countries 4; (2) common law is a good predictor of the anti-self dealing index; (3) neither measure of public enforcement is associated with stock market development. Finally (4), a comparison between the anti-director right index and the anti-self dealing index indicates that when controlling for the anti-self dealing index, the anti-director index loses significance for stock market capitalization to GDP and ownership concentration. These findings allow the authors to conclude that the anti-self dealing index is a more robust predictor of the development of stock markets than the anti-director right index. 2.1.2. The political factor Historical events such as colonization can affect profoundly corporate governance through the transplantation of corporate governance systems and laws. Societies were forced to take the corporate governance system of their conquerors. The transplant of the common law system to the United States, Canada, Australia and the transplant of the French civil law system to Belgium, Netherlands, Italy, part of Poland, Saharan Africa, Indochina, have affected the ownership structures of these countries and the evolution of their financial institutions (Beck, Demirguc-Kunt, & Levine, 2003). In addition, the civil law or the common law classification is difficult to determine for countries whose legal systems have been transferred from common law to civil law or from civil law to common law. Such countries have mixed systems influenced by both the civil and the common law systems (South Africa, Zambia, Namibia, Botswana, Sri Lanka, and Israel). Furthermore, some countries have mixed systems that incorporate civil or common law with religious law, such as Islamic countries. Another example is India's law which is based both on 4

Specifically, the regulation of self dealing (ex-ante and ex-post private control of self dealing) improves the stock market capitalization to GDP, reduces the private benefit of control, and increases the value of initial public offerings in each country relative to GDP. The ex-post control and the index of self dealing have a positive impact on the number of domestic publicly traded firms. However, only the ex-post private control of self dealing has an effect on ownership concentration (reduces ownership concentration). Anti-self dealing is not associated with ownership concentration.

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

7

common law and separate personal law applied to Muslims, Hindus, and Christians. Thus, the classification of countries into common law and civil law systems is beneficial but has some weaknesses. This reality provides Roe (2003) and Gourevitch (2003) with the evidence to argue that the differences in ownership structure and corporate governance models around the world cannot be explained only by legal origins and quality of laws. Germany and Scandinavia have high quality of laws but do not have dispersed ownership. Therefore, another factor is at work. This factor is politics. For example, Germany and Scandinavia have a high quality of law but concentrated ownership because they have strong labor and strong social democratic parties (Roe, 2003). Class struggle (rising from the conflict among managers, owners and workers) is an important determinant of corporate governance. Shareholders, who fear collusion between managers and workers at their expense, try to protect their interests by concentrating their holdings in blocks. Where workers have power over the control of firms and influence over decision-making, as they do in many of the European social democracies, corporate governance systems tend to favor ownership concentration. Where managers and owners have the power and resources to control the firms, corporate governance institutions favor shareholders over stakeholders. Ownership is dispersed, and workers lack formal power on the boards of directors (US system of governance). Gourevitch (2003) extends the channels of political mechanisms that affect corporate governance and ownership to interest group preferences and cross-class coalitions between owners, managers and workers on one hand and to political institutions such as electoral law, federalism, legislative–executive relations and party systems on the other hand. He argues that corporate governance literature neglected to examine the impact of political institutions on shaping outcomes in the way political groups deal with regulation on this issue. Political forces (political institutions, political orientations of governments, coalitions, ideologies and interest groups) not only define the laws but also determine how the laws actually operate. Variation in the content of laws and enforcement might be the product of variation in political systems. Gourevitch notes that where social democracy is strong, strong labor power presses managers to coalesce with it. Owners must consequently seek other means to control managers, and the best alternative is close ownership or ownership concentration. Thus, in social democracies, shareholder rights are weak and shareholder dispersion is low. 2.1.3. The cultural factor North defines culture as “communities of common ideologies and a common set of rules that all believe in” (North, 1987: 421). He qualifies culture as a robustly useful way of life that responds to changes in prices and costs, drawing the economic system along with it. Culture mobilizes individuals and turns them into a community with shared norms and values, where individuals make choices on the basis of their mental models. Licht (2001) went forward by building a novel theory about the role of culture in the development of corporate governance and financial regulation. It seems that changing the laws on the books and the act of simply writing investor rights into law is not enough and does not guarantee improvement of corporate governance. Theorists, practitioners and policy makers share the view that culture impacts corporate governance and can impede change and legal reforms. Ethnicity, customs, beliefs, shared values, and religion appear

8

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

as primordial factors that affect effective systems of corporate governance. For example, the cultural environment in East–central Europe is a potential impediment to change. After the failure of the communist regimes in 1989–1993, a comparative analysis between the Western European countries and the Eastern European countries that had endured communist rule shows that the communist countries have strongly-authorized cultural embeddedness and hierarchy. These values are compatible with low perceived legality. Achieving social change through legal reform thus faces serious obstacles in these countries. Legal factors alone cannot be effective, because other factors such as culture play an important role. Existing cultural values block changes and generate path dependence. Cultural value adaptation and adjustments favor legal reforms and take place slowly in response to changed life circumstances. Licht et al. (2005, 252) argue that “the link between societal aversion to litigation and high scores on harmony and uncertainty avoidance implies that in such high scoring countries implementing a new legal regime may require alternative to the court system.” Thus, in countries where investor protection cannot occur within the court system, active regulation by the state is required. Licht et al. also argue that “cultural emphases on embeddedness and hierarchy prevalent in many developing and transition economies may be conducive to corruption, in parallel to general disregard of the law.” Countries, such as those in Asia, that develop social norms that do not rely on litigation certainly have other mechanisms of governance than those known in the West. Stulz and Williamson (2003) and Licht et al. (2005) have empirically investigated the cultural factor. They show that differences between national cultures can explain crosscountry differences in corporate governance. Stulz and Williamson (2003) explore whether differences in culture represented by religion and language can explain the differences in investor protection around the world. They use two proxies for culture: religion as a key component of the system of beliefs and language, which is the vehicle to communicate beliefs. The results show that English speaking countries and Protestant countries make it easier for shareholders to vote. The authors examine the correlation between culture proxies and the enforcement of rights (rule of law, corruption, risk of expropriation, accounting standards). They find that language is irrelevant except for accounting standards. Protestant countries have better enforcement and higher standards than Catholic countries. Their results show that the Protestant, Catholic and English speaking countries have higher investor protection than other countries; that Protestant countries have a higher corruption index (less corrupt) than Catholic, Muslim and Buddhist countries; that Protestant and Buddhist countries have a higher repudiation risk; that Protestant and Catholic have a higher expropriation index; that Spanish countries have a lower expropriation index; and finally that English-speaking countries have a higher accounting index than Spanish countries. Licht et al. (2005) investigate in what ways the laws on the books reflect countries' national culture. They use the LLSV dataset to operationalize legal rules, and the cultural value dimension framework to conceptualize culture. More specifically, they use the culture value dimensions identified in cross-cultural psychology to characterize cultures of different societies and measure culture by Schwartz and Hofstede value dimensions. They demonstrate through a comparative international analysis that combining classifications based on cultural dimensions and on the legal families can shed some light on the obscure part of the comparative explanation.

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

9

2.2. The conceptual model 2.2.1. Adaptation of Williamson model According to Licht et al. (2005, 232), the link between culture and law is absent in economic theory. They recall the Williamson (2000) framework, which advances a notional model to capture the new institutional economics. Williamson distinguishes four levels of social analysis corresponding to four principal categories of institutions that shape the economic activity of firms (Williamson, 2000, 597). The first level is called “social embeddedness level”. It represents informal institutions or conventions in the social environment such as customs, mores, traditions and social networks. This level is equivalent in our analysis to the “culture” factor. The second level is referred to as “institutional environment” and corresponds to formal institutions such as polity, judiciary and bureaucracy. 5 The third level represents the “institutions of governance” or the governance of contractual relations. In transaction cost economics, the ultimate unit of activity is the transaction (Commons, 1932, 4). Governance is an effort to craft order, thereby mitigating conflict and realizing mutual gains (Williamson, 2000, 599). Finally, the fourth level represents institutions related to resource allocation and employment within firms. To summarize, we can say that whereas traditional economics rely on rationality and maximizing behavior, institutional economics are much more aware of the importance of history, culture, tradition and other so-called “path dependent” factors in shaping economic behavior. By shaping the environment for economic activity, the informal and formal institutions affect the level of income a country can reach. The formal institutions of a country reflect its social institutions and its history. Hence, Williamson adopts in his model a historical perspective to explain how institutions have evolved. In effect, the economic outcomes (and countries' development) are affected by the governance structure of firms and other organizations. But the governance structure relies greatly on the legal rules in place, which are the result of informal institutions. For Licht et al. (2005), Levels 1 and 2 represent the culture and legal factors, respectively. 6 Corporate governance and investor protection are a result of the interaction of the two levels. However, according to our literature review, the two levels are not disconnected from polity. Political institutions have a great impact on market economy by shaping and safeguarding property rights and providing the market with social stability and cohesion (Borner, Bodmer, & Kobler, 2004; Rodrik, 2000). We do believe that economic change is the result of a historical and dynamic interaction between the three factors: culture, law and polity. Indeed, institutions are made up of formal rules, informal norms and the enforcement characteristics of both. Norms provide the essential “legitimacy” to any set of formal rules and change can never be efficient if law and policy impose it. Nevertheless, polities shape economic performance because they define and enforce the economic rules of the game. But political institutions will be stable “The design instruments at Level 2 include the executive, legislative, judicial, and bureaucratic functions of government as well as the distribution of powers across different levels of government” (Williamson, 2000, 598). 6 “The foregoing analysis points to the assumption that underlies our basic hypotheses: in the long run the content of formal legal rules should be compatible with and partly reflect the prevailing cultural orientations in a society. The present study considers legal rules that pertain to reconciling conflicting economic interests through the court system”, Licht et al. (2005, 233). 5

10

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

only if they are supported by organizations with an interest in their perpetuation. Certainly evolving norms of behavior that will support and legitimize new rules is a lengthy process. While economic growth can occur in the short run with autocratic regimes, long-run economic growth entails the development of the rule of law and the protection of civil and political freedoms. 7 This historical perspective shows that corporate governance cannot be the result of only institutional environment as stated in Williamson model, but the result of the dynamic interaction among culture, law, and polity (see Fig. 1). 8 The solid arrows that go from higher to lower level imply that the higher level imposes constraints on the level immediately below. The reverse dashed arrows signal feedback (Williamson, 2000, 596). 2.2.2. Mediation analysis of corporate governance Institutional economics has the great merit of integrating the tools of economic, historical and cultural analyses. Institutions include formal rules and informal conventions that shape the behavior of members of a society as well as expectations about behavior. North (1991) suggests that institutions provide the incentive structure of an economy and that as the structure evolves, it shapes the direction of the economy toward growth, stagnation or decline. LLSV (1997, 1998, 1999, 2000, 2002) show that law is crucial in understanding the patterns of corporate governance around the world and the impact of legal system on investor protection, ownership structure and financial markets. Roe (2003) and Gourevitch (2003) argue that political systems can explain the differences in ownership structure and corporate governance models around the world. Stulz and Williamson (2003) and Licht et al. (2005) document that cross country differences in corporate governance can be explained by differences between national cultures. Djankov et al. (2008) are interested in linking the regulation of self-dealing to stock market development. They give evidence of differences in financial development across legal families. We learn two key elements from the above analysis. The first element, learned from institutional economics, is the importance of informal and formal institutions in economic outcomes (growth and market development). The second key learning from finance literature is the importance of investor protection in financial market development. We do believe that if we can put the two elements together, we will have a better picture of the link between institutional economics and financial markets. The key element in our view is the process through which institutions impact the stock market. We propose that the 7 North (1991, 97) wrote: “Institutions are the humanly devised constraints that structure political, economic and social interaction. They consist of both informal constraints (sanctions, taboos, customs, traditions, and codes of conduct), and formal rules (constitutions, laws, property rights). Throughout history, institutions have been devised by human beings to create order and reduce uncertainty in exchange. Together with the standard constraints of economics they define the choice set and therefore determine transaction and production costs and hence the profitability and feasibility of engaging in economic activity.” 8 The role of polities is discussed as follows in Williamson (2000, 609): “Thus even if we are confident that “polities significantly shape economic performance because they define and enforce the economic rules,” whereupon “an essential part of development policy is the creation of polities that will create and enforce efficient property rights,” there is the further problem that “we know very little about how to create such polities” (North, 1994, p. 366).”

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

11

Level 1: Informal institutions (customs, norms, mores and traditions) -

Theory: Social Theory

Level 2: Formal legal rules (constitution, law, property rights …) Frequency: 10 to 100 years

-

Theory:

Political system

Frequency: 100 to 1000 years

Economics of property rights Positive political theory

Level 3: Governance structure (of firms) Frequency: 1 to 10 years

-

Theory: Transaction cost economics

Level 4: Resource allocation (marginal analysis of economic outcomes) Frequency: Continuous

-

Theory: Neoclassical economics / agency theory

Fig. 1. Adaptation of Williamson' new institutional economic model.

impact of institutions on financial market development is mediated by investor protection. On one hand, investor protection comes from a lengthy process involving the dynamics of legal, culture and polity factors. On the other hand, investor protection is the quality that investors seek in order to go to capital markets. To assert mediation, Baron and Kenny (1986, 1176) establish that an independent variable X affects a distal dependent variable Y through a mediating variable M when we meet the conditions of Fig. 2. The new institutional economics provide us with a theory that explains how formal and informal institutions (X) affect economic development (Y). Legal,

X, the causal variable, is the history of law, culture and polity; which is represented in our case by the first level of our conceptual model. Y, the outcome variable, is the economic outcome or the Market capitalization; which is represented in our case by the bottom level of our conceptual model. M, the mediator variable, is the corporate governance or investor protection index; which is represented in our case by the intermediate level of our conceptual model.

Fig. 2. Mediation analysis of corporate governance.

12

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

culture and polity theories provide us with arguments that support the impact of legal system, cultural values and political system on investor protection. Hence, the paths a and c are supported by theory. However, for path b we do not have a strong theory, only some empirical evidence. In Baron and Kenny (1986), we should have a significant zero-order effect of the independent variable X (often an experimental manipulation) on the dependent variable Y in order to look for mediation. This “X–Y test” has been labeled the “effect to be mediated.” However, Zhao, Lynch, and Chen (forthcoming) challenged this assertion and qualified the path c just as the total effect, but not the effect to be mediated. For these authors, the indirect effect (a × b) is all that matters. If path a × b is significant, then we have some form of mediation. 3. Data and research design The question that arises here is: To what extent can we explain cross-countries variation in corporate governance through informal and formal institutions (captured by culture, law and polity factors)? And how do corporate governance differences impact capital markets? In this analysis, we add new evidence on corporate governance patterns by investigating investor protection across countries. We do not look only at the legal origin; we also integrate cultural and political dimensions. 3.1. Sample selection and variable measurement We use the World Bank investor protection index to measure investor protection, and the Licht et al. (2005) dataset and “religions of the world” web site for the cultural variables, respectively. We take the data on a country's primary religion and politics from the “Database of Political Institutions” by Beck, Clarke, Groff, Keefer, and Walsh (2001) issued by the World Bank in July 2005 and updated by Keefer (2010). The data on control variables and market capitalization are extracted from the World Bank's World Development Indicators database. Finally, we take the control of corruption index and the rule of law index from Kaufmann, Kraay, and Mastruzzi (2005) database. Combining these datasets provides us with a final sample of 81 countries. 3.1.1. Investor protection index Lawyers generally follow a qualitative approach to deal with law or to compare different legal systems. They do not use numbers and quantitative measures of law, because they think that such measures lead to superficial understanding of different legal systems (Siems, 2005). However, in the investor protection issue, LLSV have attempted to quantify the law in their famous study “law and finance.” They use six items to specify investor protection: voting by mail, blocking shares before meeting, cumulative voting, oppressed minority mechanism, pre-emptive rights to new issues, and share capital required to call an extraordinary shareholder meeting. In the last few years many studies have used the LLSV variables on investor protection (Hope, 2003; Kwok & Tadesse, 2006; Licht et al., 2005; Stulz & Williamson, 2003). Only a few studies have established their proper measure of investor protection and their appropriate dataset (Djankov et al., 2008;

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

13

Gourevitch, 2005; Lele & Siems, 2006). 9 Gourevitch (2005) built a new index of shareholder protection called MSP (Minority Shareholder Protection) for 39 countries. Lele and Siems (2006) established a new shareholder protection index for Germany, France, UK, US, and India over three decades 1975–2005. 10 Recently, Djankov et al. (2008) revised the antidirector right index and calculated this index for 72 countries based on the law and regulation prevailing in 2003. Lately, the World Bank has been interested in what makes countries able to generate growth, enforce investment, secure property rights, and provide public order. It developed a database, which it can mobilize for research on corporate governance. The World Bank has been involved in determining the key characteristics of good laws and made a major contribution to this literature by building a valuable index of investor protection. 11 In this research, we adopt the World Bank's measure of investor protection because we believe this measure is built with great rigor in the method of index construction and because the data covers a recent period and a large number of countries, including developed and emerging economies. The investor protection index was constructed from a survey by the “The International Bank of Reconstruction and Development Association” in 2006. 12 The survey was administered through more than 4400 local experts (lawyers, business consultants, accountants, and government officials). This survey is based on a questionnaire including a short case study that asks contributors how their country's regulations would treat the described transaction as of January 15th 2006. The experts interact with the Doing Business team through conference calls, written correspondence and country visits. Doing Business team members visited 65 countries to verify data and expand the number of respondents. Data from the survey are subjected to a test for robustness that leads to the revision of the information collected. The team members can have multiple interactions with local respondents to clarify misinterpretations of questions as they collect data about laws and regulations. If there are modifications to the laws and regulations in 2005 that affect the answers, the respondents must explain how the modifications change their responses. Respondents provide information on a voluntary basis without expectation of monetary compensation. The data, collected for 175 countries, was published in Doing Business 2007. This data is revised continually as the team receives new information from country visits and recruits other respondents. 3.1.1.1. The construction of the Doing Business investor protection index. The data collected from the survey allows the Doing Business team to build the indicator of investor Siems (2006) argues, “it's doubtful whether the findings of La Porta et al. are accurate. Various studies have identified numerous coding errors (e.g. Braendle, 2006; Cools, 2005). But the main problem is that the limited number of variables hardly provides a meaningful picture of the legal protection of shareholders. The choice of variables by La Porta et al. does not only suffer from a US bias but is also a poor proxy of shareholder protection in general, because their variables do not capture the most significant aspects of the law”. 10 They give a quantification of legal rules through 60 variables to compare variation across countries and across time of legal systems. They measure the level of protection of the active shareholder by an aggregation of 32 variables and the level of protection of passive shareholders by an aggregation of 28 variables. 11 This methodology was originally developed in Djankov et al. (2008) and adopted by Doing Business World Bank with minor changes. See http://www.doingbusiness.org/ for details. 12 This association helps to identify the source of obstacles in doing business in different countries and supports the policymakers in designing reforms. 9

14

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

protection. This index measures “the strength of minority shareholder protections against directors' misuse of corporate assets for personal gain. The indicators distinguish 3 dimensions of investor protection: transparency of transactions (extent of disclosure index), liability for self-dealing (extent of director liability index) and shareholders' ability to sue officers and directors for misconduct (ease of shareholder suits index).” 13 The level of disclosure index: This dimension is measured by five variables and ranges from 0 to 10, with higher values indicating greater disclosure. The level of the director liabilities index: This dimension is measured by seven variables and ranges from 0 to 10, with higher values indicating greater liability of directors. The ease of shareholder suits index: This index is composed of six variables and ranges from 0 to 10, with higher values indicating greater power of shareholders to challenge the transaction. The investor protection index is the average of these three indexes and ranges from 0 to 10, with higher values indicating better investor protection. Singapore and Hong Kong have the highest values of investor protection 9.3 and 9, respectively. Tunisia has 3.3. Doing Business methodology has some limitations. First, the data is collected from the most populated city in the country and may not be representative of the legal practices in other parts of the country. Second, the data focuses on the specific business form and may not be representative of other forms of business in that country. Third, the transaction described in the case study refers to a specific set of issues and does not represent the full set of issues the business encounters. 3.1.2. The culture dimension Hofstede (1980) defines culture as the collective programming of the mind, which distinguishes the members of one group or category of people from another. Culture is learned and is the product of our social environment. The cross-cultural psychology framework defines culture within the operational concept of values. It moves from a general statement about culture to values, value priorities and value dimensions. According to Schwartz and Ros (1995), culture is the process of attribution of values that lead to a norm of behavior and self-imposed codes of conduct. Values are the socially-shared abstract ideas about what is good, right, and desirable in society. The ordered set of values forms a system of values priorities. For extracting values priorities, researchers identify the principal societal

13

To construct these indexes, the respondents were asked to describe the minimum legal requirement regarding: 1) who approves the transaction; 2) what needs to be disclosed to the board, shareholders, stock exchange and regulators; 3) what are the duties of the officers, directors and controlling shareholders; 4) how the transaction validity could be challenged; 5) what kind of actions are available if the buyer suffers damages; 6) what needs to be proven under each cause of action; 7) who has standing to sue under each available cause of action; 8) what is the availability of direct and derivative suits, 9) what access exists to information and discovery rights; 10) what are the potential fines and criminal sanctions (Djankov et al., 2006). The respondents based their answers on the laws and regulations applicable under the case facts and provided the text of laws (civil and commercial codes, stock market acts and regulations, criminal code, civil procedure code), statutes, judicial precedent and regulatory opinions used to answer the questionnaire.

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

15

problems. In response to the social problems identified, scholars define “cultural value dimensions” that reflect the ways for a member of a society to deal with these problems. In their investigation of the impact of culture on investor protection differences around the world, Stulz and Williamson (2003) approached culture by religion. Indeed, religion is a key component of the system of beliefs and values a society shares and it shapes the ethical propositions that govern human behavior. 3.1.2.1. The construction of cultural value dimensions. Two works demonstrate the arguments within literature that addresses the cultural dimension measurement: the work of Hofstede (1991) and that of Schwartz and Ros (1995). Hofstede defines the value dimension as an aspect of culture that can be measured relative to other cultures. In Schwartz and Ros (1995), culture is the process of attribution of values which leads to a norm of behavior and self imposed codes of conduct. In order to extract value priorities, researchers identify the principal societal problems. Then, scholars define “cultural value dimensions” that reflect how members of a society can deal with these problems. We adopt the Hofstede measure because previous research employs it most. Indeed, Hofstede's dimension framework characterizing culture is still the most influential and the most used in international management studies. Furthermore, the complete Schwartz database is not yet available to the public (Kwok & Tadesse, 2006). Related to our research, another reason we use this database is the availability of information. 14 Hofstede derives his cultural value dimensions from a questionnaire distributed in 1967–1973 to over 117,000 employees of IBM Corporation in fifty countries. The questions were designed to identify respondents' value orientations. Once the responses were given, Hofstede computed the scores on each question per national sample group. A factor analysis allowed the identification of the sets of questions that are influenced by a shared dimension. The obtained factors define the cultural value dimensions. He then computed the score of each dimension for every nation by combining the questions that loaded on the relevant factor. Hofstede identified four factors and defined them as four cultural value dimensions: Uncertainty avoidance, Power distance, Individualism–collectivism and Masculinity–feminism. Hofstede (2001) added another value dimension: Long term orientation. In our analysis, we retain only the first four dimensions, because the data are limited for the last one. We use the Hofstede value dimensions from his database. All the scores of the four dimensions are provided for each country in our sample. For Arabic countries without specific scores, we give them the same values given to the “Arab world” overall. Furthermore, we use Hofstede's classification of countries in different cultural regions from the Licht et al. (2005) dataset. 3.1.3. The religion dimension as a proxy for culture Religion is a key component of the system of beliefs. Religion helps guide the ethical propositions that govern human behavior. We adopt the line of reasoning initiated by 14

Hofstede database covers more countries, providing more observations for emerging and developed countries.

16

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Weber who claims that the specific content of religious beliefs may profoundly affect economic behavior. We also rely on the work of Stulz and Williamson (2003) who investigate the impact of culture on investor protection differences around the world. We think that the culture value dimension framework is relevant to operationalize the concept of culture but it does not allow us to counter all aspects of culture. Using both religion and the culture value dimensions can provide, we provide a better picture of how cultural variables give insight into the diversity in investor protection around the world. Religion is widely used as a proxy of culture. LLSV (1999) use religion, measured by the percentage of a country that practices a given religion, as a proxy of culture in their study of government quality. Stulz and Williamson (2003) use countries' predominant religions as proxies of their national cultures. The results show religion's great impact on creditor rights and its smaller impact on investor protection. We retain Stulz and Williamson's measure in the ANOVA, and LLSV's measure in the regression and mediation analyses. 3.1.4. The political dimension Pagano and Volpin (2000), Perotti and von Thadden (2004), and Roe (2005) seek to model political forces shaping investor protection. They present voting models in which voting decisions shape investor protection. These models focus on the possibility of coalitions between insiders and stakeholders against outside shareholders (Pagano & Volpin, 2000; Perotti & von Thadden, 2004) or coalition between insiders and outsiders against stakeholders (Roe, 2005). Gourevitch (2005) empirically investigates the impact of political variables on corporate governance. He argues that corporate governance patterns vary with other features of the economy, among them job security, product market competition, education and training systems, financial structures, income inequality. He called these economic features “institutional complementarities”. Measuring institutional complementarities for countries leads research to group countries according to the degree of coordination into two groups: liberal market economies LME and coordinated market economies CME. Gourevitch measures the institutional complementarities by the coordination index. Pagano and Volpin's (2005) model predicts that proportional electoral systems are conductive to weaker investor protection than majoritarian. For our analysis, we use many proxies for the polity dimension: the total Herfindal and the opposition Herfindal Indexes, the Plurality and Electoral Rules systems and the tenure of the executive system. These data were taken from the “Database of Political Institutions.” 3.1.5. The other variables We use three other variables useful to our analysis: the legal origin, the market capitalization over GDP, the rule of law index and the control of corruption index. Legal origin: We adopt the same methodology as LLSV by classifying countries according to their legal origin, which could be common law, French civil law, German civil law, or Scandinavian law. The rule of law index and the control of corruption index: The rule of law index represents the level of perceived legality in the country. The corruption is the use of public or

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

17

private office or power for personal gains. The corruption is antithetical to the rule of law because corruption has an effect on the lack of respect for law. The control of corruption index measures the extent to which corruption is controlled. We gathered the data on these indexes from International Country Risk Guide. Market capitalization: the financial market is considered the most important indicator of the economy. To judge the economic performance of a country, we usually look at its financial market activity. To make a cross-country comparison, most researchers use the ratio of market capitalization over GDP. In our analysis, we measure this ratio for the year 2005. Table 1 gives the code, the definition and the measurement of the variables used in this study.

3.2. Research design We want to explain capital market development and corporate governance patterns by investigating formal and informal institutions across countries. We hypothesize in our conceptual model above that the impact of formal and informal institutions on financial market development is not direct, but is mediated by the level of investor protection. We can represent this process can by a diagram (Fig. 3), which operationalizes our conceptual model. The diagram seems to be appropriate to structural equation modeling, with the solid lines representing the structural model and the dashed lines representing the measurement model. Despite the superiority of the structure equation modeling, we choose the mediation analysis for its simplicity and because “it provides the researcher with a story about a sequence of effects that leads to something” (Kenny, 2008, 354). 15 However, since our key variables are multidimensional with many proxies for each one, we will start with a factor analysis to extract factors from the initial indicators before conducting a mediation analysis. To conduct a mediation analysis, Judd and Kenny (1981) and Baron and Kenny (1986) recommend three steps to ascertain mediation (see Fig. 4). In Baron and Kenny (1986), the “X–Y test” has been labeled the “effect to be mediated”. For them, without an effect to be mediated, there is no need for further investigating mediation. However, in Zhao et al. (forthcoming), we do not need the X–Y effect to assert mediation. The path c represents only the total effect and we can have mediation even if c is not significant. To establish mediation, all we need is a significant indirect effect (a × b). They recommend running the Preacher–Hayes script and generating “Bootstrap Results for Indirect Effects.”

15 “Iacobucci (2008) argues that structural equation (SEM) approaches dominate the “causal steps” approach of Baron and Kenny (1986). We agree that the SEM approach is superior to Baron and Kenny's because it estimates everything simultaneously instead of assuming that equations 1–3 are independent. However, the greater technical complexity of SEM makes it seem unlikely that SEM will supplant Baron and Kenny's approach soon” (Zhao et al., 2010).

18

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Table 1 Variable definition. Transparency of transactions (Extent of Disclosure Index — Word Bank database) Liability for self-dealing (Extent of Director Liability Index — Word Bank database) Shareholders' ability to sue officers and directors for misconduct (Ease of Shareholder Suit Index — Word Bank database) Invprot_Index (WB) Strength of Investor Protection Index (the average of the three indexes — Word Bank database) Invprot_Factor Strength of Investor Protection Index (extracted factor from the three indexes of the World Bank) Rule_Law Rule of Law Index (Word Bank database) Contr_Curr Control of Corruption Index (Word Bank database) Leg_Fam La Porta Legal Family (1:Common law; 2: French civil law; 3: German civil law; 4: Scandinavian civil law) Common_Law 1: if Common Law countries, 0: otherwise French_Law 1: if French Civil Law countries; 0: otherwise German_Law 1: if German Civil Law countries, 0: otherwise Scan_Law 1: if Scandinavian civil law countries; 0: otherwise trade_Open Trade openness (Word Bank database) Hofst_Region Hofstede Region 1: More developed Latin; 2: Less developed Latin; 3: Anglo; 4: Nordic; 5: Asian; 6: Near eastern; 7: Germanic Geog_Region Geographic region (1:Europe; 2: Latin America; 3: Asia; 4: MENA; 5: North America; 6: South pacific; 7: Africa) Hofst_PDI Hofstede Power Distance Index Hofst_INDIV Hofstede Individualism Index Hofst_MASC Hofstede Masculinity Index Hofst_UAI Hofstede Uncertainty Avoidance Index Hofst_LTO Hofstede LT Orientation Index Relig_Code Religion Code (1: Muslim; 2: Protestant, 3: Catholic, 4: Other) Perc_Muslim Percentage of Muslim among population Perc_Christ Percentage of Christian among population Polity_Herfindal Herfindal government index = the sum of the squared seat shares of all parties in the Index government. Polity_Herfindal Herfindal opposition index = the sum of the squared seat shares of opposition parties Opposition Index Plurality In “plurality” systems, legislators are elected using a winner-take-all/first past the post rule. “1” if this system is used, 0 if it isn't. “1” if there is competition for the seats in a one-party state (LIEC is 4), blank if it is unclear whether there is competition for seats in a one-party state (LIEC is 3.5) and “NA” if there is no competition for seats in a one-party state or if legislators are appointed (LIEC is 3 or lower). In our case, we give 1: if plurality; 0: if not Electoral Rules = 1 if Plurality and Government proportional representation which governs the majority = 0 if proportional Stability indicator Tenure of executive system Stock market Value of listed shares to GDP, calculated using the following deflation method: capitalization {(0.5) ∗ [Ft/P_et + Ft − 1/P_et − 1]} / [GDPt/P_at] where F is stock market capitalization, to GDP P_e is end-of period CPI, and P_a is average annual CPI Standard and Poor's Emerging Market Database (and Emerging Stock Markets Fact book). Data on GDP in US dollars is from the electronic version of the World Development Indicators. End-of period CPI (IFS line 64 M..ZF or, if not available, 64Q..ZF) and annual CPI (IFS line 64..ZF) are from the IMF's International Financial Statistics, October 2005

Disc_Index (WB) Liab_Index (WB) Suits_Index (WB)

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

19

Fig. 3. Empirical model of international corporate governance.

Zhao et al. (forthcoming) identify three patterns consistent with mediation and two with nonmediation: 1. Complementary mediation: Mediated effect (a × b) and direct effect (c) both exist and point at the same direction. 2. Competitive mediation: Mediated effect (a × b) and direct effect (c) both exist and point in opposite directions. 3. Indirect-only mediation: Mediated effect (a × b) exists, but no direct effect. 4. Direct-only nonmediation: Direct effect (c) exists, but no indirect effect. 5. No-effect nonmediation: Neither direct effect nor indirect effect exists.

3.2.1. Testing for mediation According to Judd and Kenny (1981) and Baron and Kenny (1986), three conditions must hold to assert mediation: first, the independent variable X should affect the dependent variable Y (c ≠ 0 in step 1); second, the independent variable X should affect the mediator M (a ≠ 0 in step 2); and third, the mediator M should affect the dependent variable when controlling for X (b ≠ 0 in step 3). 16 If X is no longer significant after controlling for M in step 3, we conclude to full mediation. If X is still significant, we conclude to partial mediation. What is important in mediation analysis is the amount of mediation, called the indirect effect. This effect shows how the effect of X on Y goes through M. The indirect effect is defined as the reduction of the effect of the initial variable on the outcome or (c − c′). This 16

See Baron and Kenny (1986, 1176–1177).

20

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Analysis

Visual Depiction c

Step 1

Conduct a simple regression analysis with X predicting Y to test for path c alone : Y = c X + e1 X

Step 2

Conduct a simple regression analysis with X predicting M to test for path a alone : M = a X + e2

Y a

X

M c’

Step 3

Conduct a multiple regression analysis with X and M predicting Y :

Y = c’ X + b M + e3 X

M

Y

Fig. 4. Steps to conduct mediation analysis.

difference in coefficients is theoretically the same as the product of the effect of X on M times the effect of M on Y or (a × b). Regression Eqs. (2) and (3) in Fig. 4 estimate the parameters a and b used to test the indirect effect.

3.2.2. Calculating the indirect effect Researchers usually use the above approach. However, this approach tends to miss some true mediation effect (Type II errors; see MacKinnon, Fairchild, & Fritz, 2007). An alternative, and preferable approach, is to compute the indirect effect and test for its significance. The regression coefficient b for the indirect effect represents the change in Y for every unit change in X that is mediated by M. Two approaches are proposed to estimate the indirect effect: either compute the difference between the two regression coefficients (c − c′, Judd & Kenny, 1981), or multiply the two regression coefficients (a × b, Sobel, 1982, 1986). 17 Once the indirect effect is calculated, it needs to be tested for its significance. Although there are a few approaches to test for this significance, two approaches McKinnon developed using tailor-made statistics, P and Z, appear to give the highest power. 18 ab ffi. Sobel first proposed the test (1982). Sobel's test equation is: z  value ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 2 b sa þa2 sb

However, Baron and Kenny (1986) recommend using the Aroian version of the Sobel test because it does not make the unnecessary assumption that the product of sa and sb is ab vanishingly small. Aroian test equation is: z  value ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi . 19 2 2 2 2 2 b sa þa2 sb þsa sb

Preacher and Hayes (2004) mention that the Sobel test assumes the normality of the distribution of a × b under the null hypothesis, which is not always the case. They recommend the bootstrapping method. 20 Preacher and Hayes (2004, 2008) made a useful contribution by presenting SAS and SPSS syntax for an alternative “bootstrap” test of the indirect effect 17

The two approaches yield identical values for the indirect effect (McKinnon, Warsi, & Dwyer, 1995). Significance tables for the two approaches are available in McKinnon website: (http://www.public.asu.edu/ ~davidpm/ripl/mediate.htm) and need to be conducted by hand. 19 Aroian (1944/1947). 20 “An alternative approach is to bootstrap the sampling distribution of ab and derive a confidence interval with the empirically derived bootstrapped sampling distribution.” Preacher and Hayes (2004, 721) 18

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

21

that is almost always more powerful than Sobel's test. It generates an empirical sampling distribution of (a × b). 4. Empirical results The objective of our empirical investigations focuses on the link between institutions, investor protection and market development. In a first step, we explore 1) how investor protection varies across countries; and 2) how formal and informal institutions impact investor protection and financial markets development. We will achieve this via descriptive, ANOVA and regression analyses. In a second step, we focus on the process by which formal and informal institutions impact the development of financial markets; more specifically, we show that investor protection mediates the path through which institutions impact financial markets development. We will achieve this through mediation analysis as proposed by Baron and Kenny (1986) and Zhao et al. (forthcoming). 4.1. Descriptive statistics and results of the ANOVA analysis Table 2 displays the summary statistics and the mean differences of synthetic and detailed investor protection indexes according to legal family (panel A), Hofstede classification (panel B), region classification (panel C), religion classification (panel D) and plurality (panel E). We can see from Table 2 that there is some disparity between countries according to most of their investor protection indexes. The Kruskal Wallis test is statistically significant for all factors' classification retained, except the plurality factor. When we look at the details, we realize that the legal classification displays the most important differences (especially between the common law countries and the others). 21 The difference prevails for the synthetic investor protection index (measured either by the World Bank as the average of the three indexes or the factor extracted by a principal component analysis). The same 21

To better understand the across group differences, we run a test for pair ways means' difference for each type of classification. We use both the parametric and non parametric analysis. Table 2 bis summarizes the results of these analyses. Table 2 bis: Pair-ways comparison of means difference Classification Index Disclosure Index Director Liability Index Shareholders Suits Index Investor protection Index (WB) Rule of Law Index Control of Corruption Index Investor Protection Index (Factor analysis)

Legal code No difference Significant differences between common law countries and others of all indexes

Hofstede region No difference No difference

Geographic region No difference No difference

No difference

Difference between Difference Between MENA and others Muslim And others No difference

No difference

Significant differences Difference between between Latin and Europe and others Anglo No difference

Religion code No difference No difference

22

Panel A: legal family Test statistica,b Disclosure Index (B)

Director Liability Index (B)

Chi-square 4.998 19.736 df 3 3 Asymp. sig. .172 .000 a. Kruskal Wallis test b. Grouping variable: legal family (Laporta)

Shareholders Suits Index (WB)

Investor Protection Index (B)

Rule of Law Index (B)

Control of Corruption Index (B)

Investor Protection Factor

9.517 3 .023

18.058 3 .000

14.403 3 .002

12.146 3 .007

18.619 3 .000

Descriptives

Disclosure Index (WB)

Director Liability Index (WD)

Shareholders Suits Index (WB)

1 2 3 4 Total 1 2 3 4 Total 1 2 3 4 Total

N

Mean

Std. deviation

Std. error

95% confidence interval for mean Lower bound

Upper bound

24 29 12 4 69 24 29 12 4 69 24 29 12 4 69

7.0417 5.4828 5.1667 6.5000 6.0290 6.5000 4.0345 3.6667 4.7500 4.8696 6.7500 5.1724 6.1667 7.0000 6.0000

2.51049 2.6134 3.45972 .57735 2.74382 2.18692 1.59201 1.72328 .95743 2.16192 2.06945 2.01900 1.85047 .00000 2.05798

.51245 .48530 .99874 .28868 .33032 .44640 .29563 .49747 .47871 .26026 .42242 .37492 .53418 .00000 .24775

5.9816 4.4887 2.9685 5.5813 5.3698 5.5765 3.4289 2.5717 3.2265 4.3502 5.8761 4.4044 4.9909 7.0000 5.5056

8.1018 6.4769 7.3649 7.4187 6.6881 7.4235 4.6400 4.7616 6.2735 5.3889 7.6239 5.9404 7.3424 7.0000 6.4944

Minimum

Maximum

3.00 .00 .00 6.00 .00 2.00 1.00 1.00 4.00 1.00 2.00 .00 4.00 7.00 .00

10.00 10.00 10.00 7.00 10.00 9.00 7.00 6.00 6.00 9.00 10.00 9.00 9.00 7.00 10.00

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Table 2 Summary statistics and mean difference analysis.

Investor Protection Index (WB)

Control of Corruption Index (WD)

Investor Protection Fact

1: Common law

24 29 12 4 69 24 29 12 4 69 24 29 12 4 69 24 29 12 4 69

6.7583 4.8897 4.9917 6.1000 5.6275 .4167 .0276 .6775 1.9450 .3871 .4354 .0914 .6258 2.1900 .4257 .8211859 .3059250 .2067030 .4305569 .1460642

1.71462 1.11270 1.08079 .48990 1.57217 1.09823 .82878 .97490 .07141 1.02851 1.23072 .81490 1.06133 .15513 1.09859 1.02495776 .63855796 .61087309 .28638524 .92839988

.35000 .20662 .31200 .24495 .18927 .22417 .15390 .28143 .03571 .12382 .25122 .15132 .30638 .07757 .13225 .20921863 .11857724 .17634387 .14319262 .11176621

2: French civil law

6.0343 4.4664 4.3050 5.3205 5.2499 −.0471 −.2877 .0581 1.8314 .1400 −.0843 −.2186 −.0485 1.9431 .1617 .3883842 −.5488195 −.5948332 −.0251460 −.0769618

7.4824 5.3129 5.6784 6.8795 6.0052 .8804 .3428 1.2969 2.0586 .6342 .9551 .4014 1.3002 2.4369 .6896 1.2539876 −.0630306 .1814272 .8862597 .3690902

3: German civil law

4.30 2.70 3.00 5.70 2.70 − 1.38 − 1.22 − 1.15 1.84 − 1.38 − 1.22 − 1.00 − 1.32 2.04 − 1.32 −.68035 − 1.66368 − 1.30750 .19320 − 1.66368

9.70 7.00 7.00 6.70 9.70 1.95 1.78 2.02 1.99 2.02 2.24 1.99 2.12 2.39 2.39 2.57549 .93918 .99860 .76964 2.57549

4: Scandinavian civil law

Panel B: Hofstede regions Test statisticsa,b Disclosure Index (WB)

Director Liability Index (WB)

Chi-square 15.119 9143 df 6 6 Asymp. sig. .019 .166 a. Kruskal Wallis test b. Grouping variable: Hofstede region

Shareholders Suits Index (WB)

Investor Protection Index (WB)

Rule of Law Index (WB)

Control of Corruption Index (WB)

Investor Protection Factor

16.573 6 .011

16.301 6 .012

25.249 6 .000

23.998 6 .001

16.147 6 .013

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Rule of Law Index (WB)

1 2 3 4 Total 1 2 3 4 Total 1 2 3 4 Total 1 2 3 4 Total

(continued on next page) 23

24

Table 2 (continued )

Descriptives N

Disclosure Index (WB)

Director Liability Index (WB)

Shareholders Suits Index (WB)

Investor Protection Index (WB)

1.00 2.00 3.00 4.00 5.00 6.00 7.00 Total 1.00 2.00 3.00 4.00 5.00 6.00 7.00 Total 1.00 2.00 3.00 4.00 5.00 6.00 7.00 Total 1.00 2.00

6 8 7 5 11 3 4 44 6 8 7 5 11 3 4 44 6 8 7 5 11 3 4 44 6 8

Mean

6.8333 5.5000 8.7143 6.0000 7.6364 4.6667 3.5000 6.5455 4.0000 4.3750 7.1429 4.6000 5.1818 3.6667 6.0000 5.0909 5.3333 6.1250 8.2857 6.8000 6.7273 3.0000 5.5000 6.3182 5.3833 5.3375

Std. deviation

Std. error

1.94079 2.77746 1.25357 1.22474 2.65604 3.51188 3.10913 2.71477 2.60768 1.30247 2.54484 89443 2.67650 .57735 2.00000 2.28054 1.21106 2.16712 1.11270 .44721 1.84883 2.64575 2.38048 2.07726 .82321 1.36061

.79232 .98198 .47380 .54772 .80083 2.02759 1.55456 .40927 1.06458 .46049 .96186 .40000 .80699 .33333 1.0000 .34380 .49441 .76619 .42056 .20000 .55744 1.52753 1.19024 .31316 .33607 .48105

95% confidence interval for mean Lower bound

Upper bound

4.7966 3.1780 7.5549 4.4793 5.8520 − 4.0573 − 1.4473 5.7201 1.2634 3.2861 4.7893 3.4894 3.3837 2.2324 2.8176 4.3976 4.0624 4.3132 7.2566 6.2447 5.4852 − 3.5724 1.7121 5.6866 4.5194 4.2000

8.8701 7.8220 9.8736 7.5207 9.4207 13.3907 8.4473 7.3708 6.7366 5.4639 9.4964 5.7106 6.9799 5.1009 9.1824 5.7843 6.6043 7.9368 9.3148 7.3553 7.9693 9.5724 9.2879 6.9497 6.2472 6.4750

Minimum

Maximum

5.00 1.00 7.00 4.00 1.00 1.00 .00 .00 1.00 2.00 2.00 4.00 2.00 3.00 5.00 1.00 4.00 2.00 7.00 6.00 3.00 .00 4.00 .00 4.70 2.70

10.00 8.00 10.00 7.00 10.00 8.00 7.00 10.00 7.00 6.00 9.00 6.00 9.00 4.00 9.00 9.00 7.00 9.00 10.00 7.00 9.00 5.00 9.00 10.00 7.00 6.70

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Panel B: Hofstede regions

Control of Corruption Index (WB)

Investor Protection Factor

1: More developed Latin

7 5 11 3 4 44 6 8 7 5 11 3 4 44 6 8 7 5 11 3 4 44 6 8 7 5 11 3 4 44

8.0429 5.8200 6.5000 3.7667 5.0000 5.9818 .5817 −.1613 1.5229 1.9120 .2645 −.0100 1.6025 .7207 .6467 .0400 1.6929 2.1500 .1582 .0033 1.6975 .8032 −.0471151 −.0156199 1.5871732 .2694759 .6486285 − 1.01652 −.1672050 .3515111

2: Less developed Latin

1.18583 .7631 1.85149 1.32791 2.35089 1.79356 .89130 .93719 .60030 .09628 1.03186 .71337 .57169 1.07730 .87053 .90875 .55150 .16140 1.16466 .44004 .63047 1.13869 .49930856 .79412412 .73558588 .43731886 1.08476796 .75795604 1.41028223 1.06716956 3: Anglo

.44820 .33823 .55824 .76667 1.17544 .27039 .36387 .33135 .22689 .04306 .31112 .41187 .28584 .16241 .35539 .32129 .20845 .07218 .35116 .25406 .31523 .17166 .20384187 .28076528 .27802533 .19557494 .32706985 .43760613 .70514112 .16088186 4: Nordic

6.9461 4.8809 5.2562 .4680 1.2592 5.4365 −.3537 −.9448 .9677 1.7925 −.4287 − 1.7821 .6928 .3932 −.2669 −.7197 1.1828 1.9496 −.6242 − 1.0898 .6943 .4570 −.5711073 −.6795243 .9068697 −.2735272 −.0801285 − 2.8993835 − 2.4112788 .0270619 5: Asian

9.1396 6.7591 7.7438 7.0654 8.7408 6.5271 1.5170 .6223 2.0780 2.0315 .9578 1.7621 2.5122 1.0482 1.5602 .7997 2.2029 2.3504 .9406 1.0964 2.7007 1.1494 .4768771 .6482845 2.2674767 .8124790 1.3773855 .8663509 2.0768687 .6759603

5.70 4.70 3.30 3.00 3.00 2.70 −.56 −1.22 .19 1.78 −1.15 −.76 .76 −1.22 −.44 −1.00 .54 1.99 −1.32 −.47 .76 −1.32 −.44266 −1.66368 .12539 −.37485 −1.06141 −1.57907 −1.30750 −1.66368

6: Near eastern

9.70 6.70 9.30 5.30 8.30 9.70 1.47 1.20 1.95 1.99 1.83 .66 2.02 2.02 1.45 1.34 2.24 2.39 2.24 .40 2.12 2.39 .93918 .73573 2.57549 .76964 2.34653 −.15459 1.83790 2.57549 7: Germanic

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Rule of Law Index (WB)

3.00 4.00 5.00 6.00 7.00 Total 1.00 2.00 3.00 4.00 5.00 6.00 7.00 Total 1.00 2.00 3.00 4.00 5.00 6.00 7.00 Total 1.00 2.00 3.00 4.00 5.00 6.00 7.00 Total

(continued on next page) 25

26

Table 2 (continued ) Panel C: Geographic regions 2: Latin America

3: Asia

4: MENA

5: North America

6: South pacific

7: Africa

Test statisticsa,b Disclosure Index (WB)

Director Liability Index (WB)

Chi-square 14.440 7693 df 6 6 Asymp. sig. .025 .261 a. Kruskal Wallis test b. Grouping variable: geographic region

Shareholders Suits Index (WB)

Investor Protection Index (WB)

Rule of Law Index (WB)

Control of Corruption Index (WB)

Investor Protection Factor

24.868 6 .000

19.605 6 .003

29.922 6 .000

29.046 6 .001

21.119 6 .002

Descriptives N

Disclosure Index (WB)

Director Liability Index (WB)

1 2 3 4 5 6 7 Total 1 2 3 4 5 6 7 Total

24 13 13 17 3 3 7 80 24 13 13 17 3 3 7 80

Mean

5.8750 4.5385 7.3846 5.5294 7.6667 8.6667 4.8571 5.9125 4.2083 4.3077 4.6154 4.2353 7.6667 5.3333 5.7143 4.6000

Std. deviation

Std. error

3.06895 2.29548 2.75495 2.52779 .57735 1.15470 2.11570 2.75219 1.64129 1.93152 3.09673 1.92124 2.30940 3.51188 1.49603 2.17930

.62645 .63665 .76409 .61308 .33333 .66667 .79966 .30770 .33503 .53571 .85888 .46597 1.33333 2.02759 .56544 .24365

95% confidence interval for mean Lower bound

Upper bound

4.5791 3.1513 5.7198 4.2297 6.2324 5.7982 2.9004 5.3000 3.5153 3.1405 2.7440 3.2475 1.9298 − 3.3907 4.3307 4.1150

7.1709 5.9256 9.0494 6.8291 9.1009 11.5351 6.8138 6.5250 4.9014 5.4749 6.4867 5.2231 13.4035 14.0573 7.0979 5.0850

Minimum

Maximum

.00 1.00 1.00 .00 7.00 8.00 3.00 .00 1.00 2.00 .00 .00 5.00 2.00 4.00 .00

10.00 8.00 10.00 9.00 8.00 10.00 8.00 10.00 7.00 8.00 9.00 7.00 9.00 9.00 8.00 9.00

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

1: Europe

Shareholders Suits Index (WB)

Rule of Law Index (WB)

Control of Corruption Index (WB)

Investor Protection Factor

24 13 13 17 3 3 7 80 24 13 13 17 3 3 7 80 24 13 13 17 3 3 7 80 24 13 13 17 3 3 7 80 24 13

6.3333 5.5385 6.4615 3.2353 7.3333 6.6667 6.0000 5.5875 5.4708 4.8000 6.1462 4.3176 7.5333 6.9000 5.5286 5.3625 1.0992 −.3392 .1531 −.2012 .9733 .9600 −.6286 .2743 1.1700 −.2215 −.0023 −.1300 1.0233 1.1100 −.6271 .3121 .0617511 −.3266302

1.46456 2.10616 2.02548 1.92124 2.08167 3.51188 1.41421 2.21442 1.31991 1.18322 2.11645 1.34965 1.32791 2.43311 1.24461 1.65655 .81886 .69079 .98047 .67712 1.26342 1.58660 .52920 1.02973 .90391 .68083 1.13820 .64771 1.25429 1.71222 .57456 1.08605 .74458046 .70018815

.29895 .58414 .56177 .46597 1.20185 2.02759 .53452 .24758 .26942 .32817 .58700 .32734 .76667 1.40475 .47042 .18521 .16715 .19159 .27193 .16422 .72944 .91602 .20002 .11513 .18451 .18883 .31568 .15709 .72416 .98855 .21716 .12142 .15198685 .19419725

5.7149 4.2657 5.2376 2.2475 2.1622 − 2.0573 4.6921 5.0947 4.9135 4.0850 4.8672 3.6237 4.2346 .8558 4.3775 4.9939 .7534 −.7567 −.4394 −.5493 − 2.1652 − 2.9813 − 1.1180 .0451 .7883 −.6330 −.6901 −.4630 − 2.0925 − 3.1434 − 1.1585 .0704 −.2526576 −.7497497

6.9518 6.8112 7.6855 4.2231 12.5045 15.3907 7.3079 6.0803 6.0282 5.5150 7.4251 5.0116 10.8320 12.9442 6.6796 5.7311 1.4449 .0782 .7456 .1470 4.1118 4.9013 −.1391 .5034 1.5517 .1899 .6855 .2030 4.1392 5.3634 −.0958 .5538 .3761599 .0964893

4.00 2.00 2.00 .00 5.00 3.00 4.00 .00 3.00 2.70 2.00 .00 6.00 5.30 4.30 .00 −.84 − 1.22 − 1.15 − 1.81 −.48 −.87 − 1.38 − 1.81 −.74 − 1.00 − 1.32 − 1.27 −.41 −.86 − 1.22 − 1.32 − 1.31589 − 1.66368

9.00 9.00 9.00 6.00 9.00 10.00 8.00 10.00 8.30 6.70 9.30 6.30 8.30 9.70 8.00 9.70 2.02 1.20 1.83 .72 1.81 1.95 .19 2.02 2.39 1.34 2.24 1.13 1.92 2.24 .54 2.39 1.73619 .73573

27

(continued on next page) (continued on next page)

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Investor Protection Index (WB)

1 2 3 4 5 6 7 Total 1 2 3 4 5 6 7 Total 1 2 3 4 5 6 7 Total 1 2 3 4 5 6 7 Total 1 2

28

Table 2 (continued ) Panel C: Geographic regions

N

Investor Protection )Factor Table 2 (continued

3 4 5 6 7 Total

13 17 3 3 7 80

Mean

.4298641 −.7006846 1.2980663 .8402587 .1192421 −.0229738

Std. deviation

Std. error

95% confidence interval for mean Lower bound

1.26545011 .79176422 .88406457 1.51049981 .72056485 .98456374

.35097271 .19203103 .51041492 .87208747 .27234791 .11007757

−.3348398 −1.1077722 −.8980718 −2.9120309 −.5471692 −.2420777

Minimum

Maximum

Upper bound 1.1945679 −.2935970 3.4942044 4.5925482 .7856534 .1961302

−2.10448 −3.24057 .27781 −.18011 −.60381 −3.24057

2.34653 .51516 1.83790 2.57549 1.57504 2.57549

Panel D: religion code Test statisticsa.b Disclosure Index (WB)

Director Liability Index (WB)

Chi-square 10.576 2832 df 3 3 Asymp. sig. .014 .418 a. Kruskal Wallis test b. Grouping variable: religion code

Shareholders Suits Index (WB)

Investor Protection Index (WB)

Rule of Law Index (WB)

Control of Corruption Index (WB)

Investor Protection Factor

16.987 3 .001

10.500 3 .015

18.203 3 .000

19.080 3 .000

10.844 3 .013

Descriptives N

Disclosure Index (WB)

1 2 3 4 Total

26 29 14 10 79

Mean

5.6538 5.0000 6.6429 8.1000 5.8987

Std. deviation

Std. error

2.41565 2.91548 2.73460 1.96921 2.76700

.47375 .54139 .73085 .62272 .31131

95% confidence interval for mean Lower bound

Upper bound

4.6781 3.8910 5.0639 6.6913 5.2790

6.6295 6.1090 8.2218 9.5087 6.5185

Minimum

Maximum

.00 .00 1.00 4.00 .00

10.00 10.00 10.00 10.00 10.00

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Descriptives

Director Liability Index (WB)

Investor Protection Index (WB)

Rule of Law Index (WB)

Control of Corruption Index (WB)

Investor Protection Factor

1: Muslim

26 29 14 10 79 26 29 14 10 79 26 29 14 10 79 26 29 14 10 79 26 29 14 10 79 26 29 14 10 79

4.8077 4.3793 5.2857 4000 4.6329 4.0769 6.1724 6.6429 6.2000 5.5696 4.8385 5.1828 6.1929 6.0900 5.3633 −.3600 .4183 1.1129 .4740 .2923 −.3500 .4928 1.2264 .3930 .3328 −.3704348 −.0886588 .4819955 .3660725 −.0227057 2: Protestant

1.95998 1.93490 2.43148 2.94392 2.17313 2.11515 1.89113 1.69193 2.29976 2.22275 1.50866 1.50927 1.62408 2.07603 1.66712 .66891 .98276 1.05888 .92071 1.02353 .65756 .97822 1.17801 1.09338 1.07705 .89754263 .87738157 .95541910 1.27676774 .99085201

.38438 .35930 .64984 .93095 .24450 .41481 .35117 .45219 .72725 .25008 .29587 .28027 .43405 .65650 .18757 .13118 .18249 .28300 .29116 .11516 .12896 .18165 .31484 .34576 .12118 .17602259 .16292567 .25534649 .40374941 .11147956

4.0160 3.6433 3.8818 1.8940 4.1462 3.2226 5.4531 5.6660 4.5549 5.0718 4.2291 4.6087 5.2551 4.6049 4.9899 −.6302 .0445 .5015 −.1846 .0630 −.6156 .1207 .5463 −.3892 .0915 −.7329601 −.4223969 −.0696471 −.5472722 −.2446444 3: Catholic

5.5993 5.1153 6.6896 6.1060 5.1197 4.9312 6.8918 7.6198 7.8451 6.0675 5.4478 5.7569 7.1306 7.5751 5.7367 −.0898 .7921 1.7242 1.1326 .5215 −.0844 .8649 1.9066 1.1752 .5740 −.0079095 .2450793 1.0336380 1.2794171 .1992331

.00 1.00 1.00 .00 .00 .00 2.00 4.00 2.00 .00 .00 2.70 3.00 2.00 .00 −1.81 −1.22 −.62 −.84 −1.8 −1.27 −1.00 −.82 −.76 −1.27 −3.24057 −1.66368 −1.31589 −2.10448 −3.24057

9.00 9.00 9.00 9.00 9.00 7.00 9.00 10.00 9.00 10.00 8.70 8.30 9.70 9.30 9.70 .72 2.02 1.99 1.83 12.02 1.13 2.12 2.39 2.24 2.39 1.88861 1.77848 2.57549 2.34653 2.57549 4: Other

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Shareholders Suits Index (WB)

1 2 3 4 Total 1 2 3 4 Total 1 2 3 4 Total 1 2 3 4 Total 1 2 3 4 Total 1 2 3 4 Total

(continued on next page) 29

30

Panel E: plurality electoral rule Test statisticsa Disclosure Index (WB) Mann–Whitney U 552.500 Wilcoxon W 1587.500 Z −.391 Asymp. sig. (2-tailed) .696 a. Grouping variable: plurality

Director Liability Index (WB)

Shareholders Suits Index (WB)

Investor Protection Index (WB)

Rule of Law Index (WB)

Control of Corruption Index (WB)

Rule of law_non corruption factor

563.500 914.500 −.260 .795

471.000 1506.000 − 1383 .167

514.500 1549.500 −.845 .398

528.500 1563.500 −.674 .500

474.500 1509.500 − 1319 .187

502.000 1537.000 −.991 .322

Descriptives N

Disclosure Index (WB)

.00 1.00 Total

26 45 71

Mean

5.8462 5.6444 5.7183

Std. deviation

Std. error

2.50906 2.90106 2.74738

.49207 .43246 .32605

95% confidence interval for mean Lower bound

Upper bound

4.8327 4.7729 5.0680

6.8596 6.5160 6.3686

Minimum

Maxim

1.00 .00 .00

10.00 10.00 10.00

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Table 2 (continued )

Director Liability Index (WB)

Shareholders Suits Index (WB)

Rule of Law Index (WB)

Control of Corruption Index (WB)

Investor Protection Factor

1: Plurality in electoral rules

26 45 71 26 45 71 26 45 71 26 45 71 26 45 71 26 45 71

4.6154 4.7556 5.7183 6.2308 5.4889 5.7606 5.5692 5.2844 5.3887 .4777 .2844 .3552 .6012 .2669 .3893 .1161909 −.0593515 .0049316

1.69887 2.42295 2.74738 2.04563 2.24238 2.18740 1.45348 1.87616 1.72805 1.09147 .96332 1.00874 1.10882 1.03435 1.06672 .86580731 1.11372943 1.02696264

.33318 .36119 .32605 .40118 .33427 .25960 .28505 .27968 .20508 .21405 .14360 .11972 .21746 .15419 .12660 .16979878 .16602498 .12187804

3.9292 4.0276 5.0680 5.4045 4.8152 5.2428 4.9822 4.7208 4.9797 .0368 −.0050 .1164 .1533 −.0439 .1368 −.2335163 −.3939529 −.2381465

5.3016 5.4835 6.3686 7.0570 6.1626 6.2783 6.1563 5.8481 5.7978 .9185 .5739 .5940 1.0490 .5776 .6418 .4658980 .2752499 .2480097

2.00 .00 .00 2.00 .00 .00 2.70 .00 .00 − 1.22 − 1.10 − 1.22 − 1.00 − 1.18 − 1.18 − 1.66368 − 3.24057 − 3.24057

9.00 9.00 10.00 9.00 10.00 10.00 8.30 9.70 9.70 1.99 2.02 2.02 2.39 2.24 2.39 1.83790 2.57549 2.57549

0: No plurality in electoral rules

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Investor Protection Index (WB)

.00 1.00 Total .00 1.00 Total .00 1.00 Total .00 1.00 Total .00 1.00 Total .00 1.00 Total

31

32

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

conclusion prevails for the rule of law and control of corruption indexes. However, we do not observe this difference for the disclosure index. Let's look now at the summary statistics. Panel A of Table 2 presents the means of the investor protection index and the dimensions for each of the legal families identified in the literature. We see from panel A that common law countries have the highest score for the all indexes, which are largely above the total sample mean. Scandinavian civil law countries come in second position. However, the mean indexes for Germanic civil law countries and French civil law countries are under the sample mean for all the variables. Disclosure and director liability indexes are higher in French civil law than in Germanic civil law countries. Thus, our results confirm that legal origin remains an important determinant of the investor protection (calculated with the new investor protection index of Doing Business). Common law countries provide the best legal protection to their investors. They require higher disclosure, greater liability of directors and greater power of shareholders to challenge the transaction than civil law countries. Germanic and French civil law countries have similar values for the disclosure index; Scandinavian civil law countries have higher values Germany and France but lower values than common law countries. Director liability is regulated most highly in the Scandinavian countries, French countries and Germanic civil law, in ascending order. Further, based on the index of shareholder suits, litigation is easier in common law countries than it is in civil law countries. The culture classification of Hofstede displays some cross-country differences only for the rule of law, the control of corruption and the investor protection factor. We observe this difference between Latin and Anglo countries. The summary statistics show the highest disclosure index for the Anglo countries, followed by developed Latin countries. The director liability and shareholders suits indexes are the highest for the Anglo countries, followed by the Asian countries. The geographic classification shows a significant difference between Europe and others for the rule of law and the control of corruption indexes. There is a significant difference between MENA countries and others for the shareholder suit index. The summary statistics show the highest scores on rule of law and control of corruption indexes for the European countries, followed by North American countries. The director liability and shareholders suits indexes are the highest for the North American countries, followed by south pacific countries. The MENA countries display the lowest scores for almost all indexes. Finally, we observe a difference between Muslim and others countries for Shareholders Suits, Investor protection, Rule of Law and Control of Corruption Indexes. The most important result from summary statistics is that Muslim countries displayed the lowest scores for almost all indexes. These results are interesting and support the legal and culture theory explanation of cross-country differences in investor protection. However, we need to understand what the determinants of these differences are. In order to do so, we use a regression analysis. 4.2. Cross country variation determinants of investor protection and financial market development: results of the regression analysis Institutional economics shed some light on how formal and informal institution's development creates value to the society. Finance literature focuses on the impact of legal

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

33

system, culture and polity on the level of investor protection. According to our main theories (legal, culture and polity), some researchers explained cross-country variations of investor protection by their legal origins, while others focused on culture or polity. Nevertheless, measuring these variables by one proxy will produce poor outcomes because the variations are multidimensional. Moreover, the databases developed use many proxies for each dimension. In order to have a synthetic measure for each dimension, we use a factor analysis to extract one factor or more for each dimension and then run a regression analysis to examine the explanatory power of investor protection and market capitalization by the three theories. Table 3 exhibits the results of the factor analyses. We see from Table 3 that the legal dimension is captured by two factors with 40% and 32% of explained variance. Factor one represents the common law and the French civil law variables. Factor two represents the German civil law variable. Common law and French civil law variables have opposite signs, which means that they have an opposite effect Table 3 Results of factor extraction analysis. Panel A: Legal Factor extraction Component

Common_Law French_Law German_Law Scan_Law

1

2

−.866 .909 −.081 .048

−.431 −.338 .867 .483

Panel B: Cultural Factor extraction Component

Hofst_PDI Hofst_INDIV Hofst_MASC Hofst_UAI Perc_Muslim Perc_Christ

1

2

3

−.291 .166 −.009 .493 −.779 .908

.813 −.823 −.008 .590 .280 −.044

.202 .186 .989 .014 −.011 −.030

Panel C: Polity Factor extraction Component

Total Herfindahl Index Herfindahl_oppos_Index Plurality Elect_rules Tenure_system_executive Extraction method: principal component analysis. Rotation method: varimax with Kaiser normalization. Rotation converged in 3 iterations.

1

2

.551 .278 .873 .910 .007

.379 .684 .102 .077 −.833

34

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

on the factor. We label these factors respectively “Legal Factor 1” and “Legal Factor 2”. The cultural dimension is captured by three factors with 35%, 24%, and 18% of explained variance. Factor one captures the two religion measures. We label it the “Religion Factor”. Factor two represents power distance and uncertainty avoidance indexes with positive sign and individualism index with negative sign. We label it “Hofstede Factor 1”. Factor three represents the masculinity index with positive sign. We label it “Hofstede Factor 2”. The political dimension is captured by two factors with 46% and 20% of explained variance. Factor one represents the Herfindhal index, plurality and electoral rules with positive sign. Factor two represents the Herfindhal opposition index with a positive sign and tenure of executive system with negative sign. We label these factors “Polity Factor 1” and “Polity Factor 2,” respectively. We will retain these factors will for the regression analysis. Table 4 displays the results. We see from that table that almost all indicators of investor protection are negatively and significantly associated with the two legal factors. Since legal factor 1 is negatively associated with common law variable and positively related to French civil law variable, we can conclude that the common law countries offer more protection to their citizens than French law countries. We can draw the same conclusion for the German civil law countries, because Legal Factor 2 is negatively related to investor protection on the one hand and positively related to German civil law variable on the other hand. These results corroborate the previous findings (Djankov et al., 2006, 2008; La Porta et al., 1998, 1999, 2000, 2002) and confirm the relevance of the legal theory of investor protection, particularly if we know that we used a more reliable and representative database. The culture and polity hypotheses are weakly supported by our results. Only the religion factor has a positive impact on investor protection and shareholders suits indexes at 1% and polity factor 1 has a positive impact on market capitalization at 5%. The latter result supports Pagano and Volpin's (2005) model, which predicts that proportional electoral systems are conducive to weaker investor protection than majoritarian. To summarize, our results show that the legal system is the most important and robust factor in explaining investor protection, but it has no impact on financial markets' development. The culture and polity dimensions do not explain the cross-countries variations in Notes to Table 4: Market capitalization = stock market capitalization/GDP; Investor Protection Index (WB): doing business investor protection index from the World Bank database; Investor Protection (Extracted Factor): extracted factor from the three indexes disclosed by the World Bank; Disclosure Index (WB): the level of disclosure index from the World Bank database; Director Liability Index (WB): the level of the director liabilities index from the World Bank database; Director Liability Index (WB): the ease of shareholder suits index from the World Bank database; Legal Factor 1: Common Law (−), French Law (+); Legal Factor 2: German Law (+) Religion Factor: Perc Muslim (−), Perc Christian (+); Hofstede Factor 1: PDI (+), Individualism (−), UAI (+); Hofstede Factor 2: Masculinity (+) Polity Factor 1: Total Herfindal (+), Plurality (+), Electoral rule (+); Polity Factor 2: Opposition Herfindhal (+), Tenure Executive (−) The sign indicates the nature of association between the variable and the factor. ⁎ p b .010. ⁎⁎ p b .05. ⁎⁎⁎ p b .01.

Dependent variable

Estimation Legal method Factor 1

Market capitalization

Stepwise Enter

Investor Protection Index (WB)

Stepwise Enter

Investor Protection Stepwise (Extracted Factor) Enter Disclosure Index (WB)

Stepwise Enter

Director Liability Index (WB)

Stepwise Enter

Shareholders Suits Index(WB)

Stepwise Enter

Legal Factor 2

Religion Factor

Hofstede Factor 1

Hofstede Polity Factor 2 Factor 1

Polity Factor 2











−.053 (−.655) −.804 (−4798) ⁎⁎⁎ −.044 (−.518) −.496 (−5070) ⁎⁎⁎ −.460 (−3911) ⁎⁎⁎ −.044 (−.518) −.279 (−.700) − 1104 (−4814) ⁎⁎⁎ −.983 (−3606) ⁎⁎⁎ − 1093 (−5144) ⁎⁎⁎ −.928 (−3621) ⁎⁎⁎

.003 (.047) −.538 (− 3266) ⁎⁎⁎ −.641 (−.641) −.303 (− 3158) ⁎⁎⁎ −.369 (− 3490) ⁎⁎⁎ − 1099 (− 3321) ⁎⁎⁎ −.959 (− 2676) ⁎⁎⁎ –

−.000 (−.008) –

−.196 (− 3491) ⁎⁎⁎ −.122 (− 1479) –

−.623 (− 2545) ⁎⁎ – −.346 (− 1502)

−.735 −.301 (− 3631) ⁎⁎⁎ (− 1374) – – .081 (.710) – −.212 (−.548) – −.185 (−.699) .762 (3447) ⁎⁎⁎ .676 (2716) ⁎⁎⁎

−.039 (−.680) –

.148 (2483) ⁎⁎ .148 (2045) ⁎⁎ –

−.019 (−.116) –

−.136 (−.708) –

−.176 (− 1386) −.796 (− 2517) ⁎⁎⁎ −.438 (− 1015) –

.003 (.030) –

−.085 (−.762) –

−.273 (−.859) –

.075 (.199) –

−236 (−.803) –

.125 (.574) –

−.384 (− 1.49) –

−.235 (−.849)

.104 (.508)

−.084 (−.348)

Intercept

.609 (10.015) ⁎⁎⁎ −.093 .619 (− 1247) (9506) ⁎⁎⁎ −.466 5496 (− 2763) ⁎⁎⁎ (9343) ⁎⁎⁎ −.300 5452 (− 1466) (32.269) ⁎⁎⁎ −.275 .071 (− 2800) ⁎⁎⁎ (.749) .042 −.173 (.431) (− 1454) – 5.7 (17.804) ⁎⁎⁎ −.387 5772 (−.958) (17.317) ⁎⁎⁎ – 4838 (21.261) ⁎⁎⁎ −.29 4841 (− 1051) (21.267) ⁎⁎⁎ – 5775 (27.566) ⁎⁎⁎ −.212 5762 (−.816) (26.921) ⁎⁎⁎

Adjusted F test R square .179

13.231 ⁎⁎⁎

.213

3162 ⁎⁎⁎

.397

15.283 ⁎⁎⁎

.388

6877 ⁎⁎⁎

.412

16.179 ⁎⁎⁎

.407

7363 ⁎⁎⁎

.155

6954 ⁎⁎⁎

.119

2643 ⁎⁎

.254

23.178 ⁎⁎⁎

.289

4777 ⁎⁎⁎

.322

16.454 ⁎⁎⁎

.311

5196 ⁎⁎⁎

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Table 4 Results of the regression analysis on legal, culture and polity factors.

35

36

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

investor protection. Finally, it seems that plurality and electoral rules have a positive impact on financial market development. However, the classical regression model does not capture all aspects of our conceptual model, in which the impact of corporate governance is captured through a process where investor protection should mediate the relation between institutions and financial market development. This process is the object of the next subsection. 4.3. Cross country variation determinants of investor protection and market capitalization: results of the mediation analysis We hypothesized that the country level of investor protection mediates the relationship between institutions (approached by legal, culture and polity factors) and financial markets' development. Since we have three initial variables, we will run three regression analyses concerning legal, culture and polity variables, respectively. We also use the bootstrapping method of Preacher and Hayes (2004, 2008) to test the significance of the indirect effect (a × b). 4.3.1. Results of the estimation on legal dimension Table 5 presents the results of investor protection as a mediator of the legal factor. We see from panels A and B that all conditions are met to assert mediation. Investor protection seems to mediate the relation between the two extracted factors and market capitalization. The indirect effect is significant for the two factors. Sobel Test confirms the significance of the indirect effect at 95% and by bootstrap at 90% level for legal factor 1 and by bootstrap at 95% level for Legal Factor 2. However, we have a complementary mediation when we use the Legal Factor 1 and an indirect-only mediation when we use Legal Factor 2. Nevertheless, we give more credit to the results of Legal Factor 1 for many reasons. First, the indirect effect is about 15% for Legal Factor 1 while it is only about 6% for Legal Factor 2. Second, Legal Factor 1 has a better representation of legal variables because it captures a significant portion (40%) of the explained variance and is highly correlated with the two most adopted legal systems in the world (60 of 81 countries in our sample belong to common law or French civil law system), while Legal Factor 2 captures only 32% of explained variance and is highly correlated with the weakly adopted German civil law system (only 13 countries in our sample). Third, the total effect of Legal Factor 1 is about 27%, while for Legal Factor 2, it is only about 5%. To summarize, most of the relationship between legal systems and market capitalization is explained through the indirect pathway (approximately 15%) involving investor protection as mediator, while the direct effect is only about 12%. Hence, we conclude that common law countries provide better protection to their investors, which gives confidence and makes financial markets more attractive. 4.3.2. Results of the estimation on culture dimension Table 6 displays the results of investor protection as a mediator of culture factor. The results of panels A and C do not support mediation. However the results of panel B support complementary mediation. Hence, investor protection seems to mediate only the relationship between the Hofstede Factor 1 and market capitalization. The Hofstede Factor

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

37

Table 5 Summary of multiple regression mediation analyses for the legal system. Panel A: Dependent = stock market capitalization Independent = Legal Factor 1 Mediator = Investor Protection (WB) Analysis 1: Dependent variable: market capitalization 1. Legal Factor 1 (c path) −.2668

.0880

−3.0320⁎⁎⁎

Analysis 2: Dependent variable: Investor Protection (WB) 1. Legal Factor 1 (a path) −.9970

.1661

−6.0023⁎⁎⁎

Analysis 3: Dependent variable: market capitalization (F = 7.3815; adj R2 = .164) 1. Legal Factor 1 (c-prime path) −.1236 .1067 2. Investor Protection (WB) (b path) .1436 .0643

− 1.1583 2.2352⁎⁎

Panel B: Dependent = stock market capitalization Independent = Legal Factor 2 Mediator = Investor Protection (WB) Analysis 1: Dependent variable: market capitalization 1. Legal Factor 2 (c path) −.0481

.0900

−.5341

Analysis 2: Dependent variable: Investor Protection (WB) 1. Legal Factor 2 (a path) .3259

.1949

− 1.6725

Analysis 3: Dependent variable: market capitalization (F = 6.5870; adj R2 = .1467) 1. Legal Factor 2 (c-prime path) .0139 .0845 2. Investor Protection (WB) (b path) .1901 .0531 ⁎ p b .10. ⁎⁎ p b .05. ⁎⁎⁎ p b .01.

.1641 3.5827⁎⁎

Testing for legal mediation Input variable (X)

Mediator variable (M)

Indirect effect (a ∗ b)

Sobel

Legal Factor 1

Investor Protection (WB)

−.1432

−2.1256⁎⁎

Legal Factor 2

Investor Protection (WB)

−.0619

− 1.5294

Bootstrap Lower

Type of mediation Upper

95% confidence intervals −.4134 .0121 90% confidence intervals −.3453 −.005 95% confidence intervals −.2252 −.0108

Complementary mediation according to Sobel test and with bootstrap at 90% Indirect-only mediation according to bootstrap at 95%

1 has a direct effect (c) of about 23% (significant at 1% in panel B) and an indirect effect (a × b) through investor protection of about 8% (significant at 5% according to Sobel and by bootstrap method) on market capitalization. Hence, neither religion nor masculinity seems to have an impact on market development, since our results support nonmediation. However, the results support the role of the three other dimensions of Hofstede (power distance, individualism and uncertainty

38

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Table 6 Summary of multiple regression mediation analyses for the culture dimension. Panel A: Dependent = stock market capitalization Independent = Religion Factor Mediator = Investor Protection (WB) Analysis 1: Dependent variable: market capitalization 1. Religion Factor (c path) −.1192

.0965

Analysis 2: Dependent variable: Investor Protection (WB) 1. Religion Factor (a path) −.1211

.2148

Analysis 3: Dependent variable: market capitalization (F = 7288; adj R2 = .1621) 1. Religion Factor (c-prime path) −.0968 .0889 2. Investor Protection (WB) (b path) .1843 .0516

− 1235 −.5635 − 1.0895 3.5735⁎⁎⁎

Panel B: Dependent = stock market capitalization Independent = Hofstede Factor 1 Mediator = Investor Protection (WB) Analysis 1: Dependent variable: stock market capitalization 1. Hofstede Factor 1 (c path) −.2294

.082

−2.7978⁎⁎⁎

Analysis 2: Dependent variable: Investor Protection (WB) 1. Hofstede Factor 1 (a path) −.5489

.1789

−3.0677⁎⁎⁎

Analysis 3: Dependent variable: stock market capitalization (F = 8.39; adj R2 = .1853) 1. Hofstede Factor 1 (c-prime path) −.1446 .0833 2. Investor Protection (WB) (b path) .1545 .0543

− 1735⁎ 2.8435⁎⁎⁎

Panel C: Dependent = stock market capitalization Independent = Hofstede Factor 2 Mediator = Investor Protection (WB) Analysis 1: Dependent variable: stock market capitalization 1. Hofstede Factor 2 (c path) .004

.0841

Analysis 2: Dependent variable: Investor Protection (WB) 1. Hofstede Factor 2 (a path) −.0816

.1854

Analysis 3: Dependent variable: stock market capitalization (F = 6.6087; adj R2 = .1472) 1. Hofstede Factor 2 (c-prime path) .0194 .0772 2. Investor Protection (WB) (b path) .189 .052 ⁎ p b .10. ⁎⁎ p b .05. ⁎⁎⁎ p b .01.

.047 −3.6352⁎⁎⁎

.2509 3.6352⁎⁎⁎

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

39

Table 6 (continued ) Testing for culture mediation Input variable (X)

Mediator variable (M)

Indirect effect (a ∗ b)

Religion Factor

Investor Protection (WB) Investor Protection (WB) Investor Protection (WB)

−.00223

Hofstede Factor 1 Hofstede Factor 2

−.0848 −.0151

Sobel

Bootstrap Lower −.5569

−2086** −.4286

Type of mediation Upper

95% confidence intervals −.1754 .0414 95% confidence intervals −.2938 −.0045 95% confidence intervals −.0955 .047

Nonmediation according to Sobel test and with bootstrap at 95% confidence intervals Complementary mediation according to Sobel and bootstrap at 95% Nonmediation according to bootstrap at 95%

avoidance). Nevertheless, we give more credit to the Hofstede Factor 1 for at least two reasons. First, it captures a significant part (35%) of the explained variance (compared to the 24% and 17% of the two other factors). To summarize, the complementary mediation of Hofstede Factor 1 through investor protection can be interpreted as follows: as Hofstede Factor 1 is negatively associated with the individualism index and positively associated with power distance index and uncertainty avoidance index, we can say that our results support the conjecture of the culture theory. Specifically, the countries with high individualism, low uncertainty avoidance, and low power distance display more protection to their investors and have more developed financial markets. 4.3.3. Results of the estimation on political dimension Table 7 presents the results of investor protection as a mediator of the political factor. We see from panels A and B that the obtained results do not support any mediation and confirm Direct-only nonmediation. The indirect effect is not significant with Sobel or with bootstrapping. Only a direct effect is supported by the results. Since Polity Factor 1 is positively correlated with the Herfindhal Index, plurality and electoral rules, we can reconfirm our conclusion from multiple regressions that countries with proportional electoral systems are conducive to weaker investor protection and less developed financial markets than those with majoritarian political system. The opposition index and the tenure of the executive system do not have an impact on either investor protection or on financial markets. 5. Conclusion The divergence in corporate governance patterns around the world has been well documented. LLSV work has revolutionized the study of corporate governance by investigating the legal investor protection around the world. Their results confirm that laws do matter. Roe (2003), Pagano and Volpin (2000), and Gourevitch (2003) look at the political process that influences the law and regulations shaping corporate governance. The political system that determines laws influences the struggle for power that shapes corporate governance

40

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Table 7 Summary of multiple regression mediation analyses for the polity dimension. Panel A: Dependent = stock market capitalization Independent = Polity Factor 1 Mediator = Investor Protection (WB) Analysis 1: Dependent variable: market capitalization 1. Polity Factor 1 (c path) .1720

.0649

2.6502⁎⁎⁎

Analysis 2: Dependent variable: Investor Protection (WB) 1. Polity Factor 1 (a path) .3820

.2114

1.8068

Analysis 3: Dependent variable: market capitalization (F = 6.3996; adj R2 = .1617) 1. Polity Factor 1 (c-prime path) .1371 .0644 2. Investor Protection (WB) (b path) .0912 .0399

2.1305⁎⁎ 2.288⁎⁎

Panel B: Dependent = stock market capitalization Independent = Polity Factor 2 Mediator = Investor Protection (WB) Analysis 1: Dependent variable: market capitalization 1. Polity Factor 2 (c path) −.1864

.0659

−2.8283⁎⁎⁎

Analysis 2: Dependent variable: Investor Protection (WB) 1. Polity Factor 2 (a path) −.5214

.2113

−2.4673⁎⁎

Analysis 3: Dependent variable: market capitalization (F = 6.5128; adj R2 = .1645) 1. Polity Factor 2 (c-prime path) −.1425 .0675 2. Investor Protection (WB) (b path) .0841 .0409 ⁎ p b .10. ⁎⁎ p b .05. ⁎⁎⁎ p b .01.

−2.1117⁎⁎ 2.0586⁎⁎

Testing for polity mediation Input variable (X)

Mediator variable (M)

Polity Factor 1 Polity Factor 2

Investor Protection (WB) Investor Protection (WB)

Indirect effect (a ∗ b) .0349 −.0439

Sobel

Bootstrap (95%)

1.4359 − yy036

Lower

Upper

−.0056

.1152

−.1146

.0281

Type of mediation Direct-only nonmediation Direct-only nonmediation

inside the firms. People choose the corporate governance system that hurts them the least; they use politics to express their preferences. Licht (2001), Stulz and Williamson (2003), and Licht et al. (2005) argue that the system of beliefs and values that determines the behavior and actions of individuals within a society can explain the differences in investor protection. In this instance, culture is viewed as the primary cause of all differences in law and regulations across countries and the mother of all path dependence, because culture impedes any changes and reforms that confront people's beliefs and preferences. Political and cultural theories in corporate governance have contested the legal origin of LLSV and have opened the debate to investigate the relevance of each prediction.

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

41

Our paper takes a new approach to examining corporate governance. We first build on Williamson's New Institutional Economics model to assess the link between formal and informal institutions (revealed by cultural values, legal and political systems) on one side and corporate governance on the other side. We show that the four levels of social analysis proposed by Williamson can be reduced to three levels, the first level representing the institutions, the second one representing “corporate governance,” and the third one representing “economic outcomes”. We then operationalize our conceptual model through a mediation analysis, where we argue that the impact of institutions (captured by legal, cultural and political factors) on financial markets is not direct, but can be explained through an indirect pathway involving corporate governance (investor protection). We adopt a cross-section analysis on a sample of 81 countries belonging to different contexts. Our results show that the legal origin is a robust factor with both direct and indirect effects on stock market capitalization. A path analysis confirms the indirect effect involving investor protection as a mediator of the relationship between legal origin and stock market capitalization. The culture hypothesis is strongly supported by our investigation. Countries with high individualism, low uncertainty avoidance and low power distance indexes provide better protection to their investors and have more developed financial markets. However, our results do not support religion and masculinity as determinants of cross-country variations in investor protection and market development. Finally, the results on the political component are mitigated. There is only a direct effect of polity factors on stock market capitalization. Our results do not support the indirect effect of polity. To conclude, our study can be helpful to emergent market economies because it is not easy to develop financial markets only through regulation. These economies need to assess strong governance by looking at global development as it relates to cultural values, democracy, transparency, and enforcement of the law. References Aguilera, Ruth V., & Jackson, G. (2003). The cross-national diversity of corporate governance: Dimensions and determinants. Academy of Management Review, 28(3), 447–465. Aroian, Leo A. (1944/1947). The probability function of the product of two normally distributed variables. Annals of Mathematical Statistics, 18, 265–271. Baron, Reuben M., & Kenny, David A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Bebchuk, L., & Roe, M. (1999). A theory of path dependence in corporate ownership and governance. Stanford Law Review, 52(1), 127–170. Beck, T., Clarke, G., Groff, A., Keefer, P., & Walsh, P. (2001). New tools in comparative political economy: The database of political institutions world bank economic. World Bank Economic Review, 15, 165–176. Beck, T., Demirguc-Kunt, A., & Levine, R. (2003). Law, endowments and finance. Journal of Financial Economics, 70, 137–181. Borner, S., Bodmer, F., & Kobler, M. (2004). Institutional efficiency and its determinants: The role of political factors in economic growth. Paris: Organisation for Economic Cooperation and Development, Development Centre. Braendle, Udo C. (2006). Shareholder Protection in the USA and Germany - On the Fallacy of LLSV. German Law Journal, 7(3), 257–340. Commons, John R. (1932–33). The problems of correlating law, economics and ethics. Wisconsin Law Review, 8(1), 3–26.

42

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

Cools, S. (2005). The real difference in corporate law between the United States and continental Europe: Distribution of powers. Delaware Journal of Corporate Law, 30, 697–766. Djankov, Simeon, Hart, Oliver, McLiesh, Caralee, & Shleifer, Andrei (2006). Debt enforcement around the world. Working paper no. 12807. Cambridge, MA: NBER (December). Djankov, S., Lopez, R., La Porta, F., & Shleifer, A. (2008). The law and economics of self dealing. Journal of Financial Economics, 88(3), 430–465. Fligstein, N., & Choo, J. (2005). Law and corporate governance. Annual Review of Law and Social Science, 1, 61–84. Gourevitch, Philip A. (2003). The politics of corporate governance regulation. The Yale Law Journal, 112, 1829–1880. Gourevitch, Philip A. (2005). Political drivers of corporate governance patterns. Working paper. Georgetown University. Healy, Paul M., & Palepu, Krishna G. (2001). Information asymmetry, corporate disclosure, and the capital markets: A review of the empirical disclosure literature. Journal of Accounting and Economics, 31, 405–440. Hofstede, Geert H. (1980). Culture's consequences: International differences in work-related values. Beverly Hills: Sage. Hofstede, Geert H. (1991). Cultures and organizations: Software of the mind: intercultural cooperation and its importance for survival. London: McGraw-Hill. Hofstede, Geert H. (2001). Culture's consequences: Comparing values, behaviors, institutions and organizations across nations. California: Sage Publications. Hope, Ole K. (2003). Firm-level disclosures and the relative roles of culture and legal origin. Journal of International Financial Management & Accounting, 14(3), 218–248. Iacobucci, D. (2008). Mediation analysis. Thousand Oaks, CA: Sage. Judd, Charles M., & Kenny, David A. (1981). Process analysis: Estimating mediation in treatment evaluations. Evaluation Review, 5, 602–619. Kaufmann, D., Kraay, A., & Mastruzzi, M. (2005). Governance matters IV: Governance indicators for 1996–2004. Unpublished working paper. World Bank. Washington. DC. Keefer, Philip (2010). DPI2010. Database of political institutions: Changes and variable definitions. Development Research Group, World Bank. Kenny, David A. (2008). Reflections on mediation. Organizational Research Methods, 11(2), 353–358. Kwok, Chuck C. Y., & Tadesse, S. (2006). National culture and financial systems. Journal of International Business Studies, 37, 227–247. La Porta, R., Lopez, F., & Shleifer, A. (2008). The economic consequences of legal origins. Journal of Economic Literature, 4(2), 285–332. La Porta, R., Lopez, F., Shleifer, A., & Vishny, Robert W. (1997). Legal determinants of external finance. Journal of Finance, 52, 1131–1150. La Porta, R., Lopez, F., Shleifer, A., & Vishny, Robert W. (1998). Law and finance. Journal of Political Economy, 106, 1113–1155. La Porta, R., Lopez, F., Shleifer, A., & Vishny, Robert W. (1999). Corporate ownership around the world. Journal of Finance, 54, 471–517. La Porta, R., Lopez, F., Shleifer, A., & Vishny, Robert W. (2000). Investor protection and corporate governance. Journal of Financial Economics, 58, 3–28. La Porta, R., Lopez, F., Shleifer, A., & Vishny, Robert W. (2002). Investor protection and corporate valuation. Journal of Finance, 3, 1147–1170. Lele, P., & Siems, Mathias M. (2006). Shareholder protection: A leximetric approach. University of Cambridge, CBR Working Paper No 324. Licht, Amir N. (2001). The mother of all path dependencies: Towards a cross-cultural theory of corporate governance systems. Delaware Journal of Corporate Law, 26(1), 147–205. Licht, Amir N., Goldschmidt, C., & Schwartz, Shalom H. (2002). Culture rules: The foundations of the rule of law and other norms of governance. Journal of Comparative Economics, 35(4), 659–688. Licht, Amir N., Goldschmidt, C., & Schwartz, Shalom H. (2005). Culture, law and corporate governance. International Review of Law and Economics, 25, 229–255. MacKinnon, David P., Fairchild, A. J., & Fritz, Matthew S. (2007). Mediation analysis. Annual Review of Psychology, 58, 593–614.

H. Matoussi, M.K. Jardak / The International Journal of Accounting 47 (2012) 1–43

43

MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995). A simulation study of mediated effect measures. Multivariate Behavioral Research, 30, 41–62. North, Douglass C. (1987). Institutions, transaction costs and economic growth. Economic Inquiry, 25(3), 419–428. North, Douglass C. (1991). Institutions. Journal of Economic Perspectives, 5(1), 97–112. North, Douglass C. (1994). Economic performance through time. American Economic Review, 84(3), 359–368. Pagano, M., & Volpin, Paolo F. (2005). The political economy of corporate governance. American Economic Review, American Economic Association, 95(4), 1005–1030 (September). Perotti, Enrico C., & von Thadden, Ernst-Ludwig (2004). The political economy of dominant investors. Tinbergen Institute discussion papers 04-091/2. Tinbergen Institute. Preacher, Kristopher J., & Hayes, Andrew F. (2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36(4), 717–731. Preacher, Kristopher J., & Hayes, Andrew F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879–891. Rodrik, D. (2000). Institutions for high-quality growth: What they are and how to acquire them. Studies in Comparative International Development, 35(3), 3–31. Roe, Mark J. (1994). Strong managers, weak owners: The political roots of American corporate finance. Princeton, N.J.: Princeton University Press. Roe, Mark J. (2003). Political determinants of corporate governance: Political context, corporate impact. Oxford/New York: Oxford Univ. Press. Roe, Mark J. (2005). Corporate governance: Political and legal perspectives. Edward Elgar Publishing. Schwartz, Shalom H., & Ros, M. (1995). Values in the West: A theoretical and empirical challenge to the individualism– collectivism cultural dimension. World Psychology, 1, 91–93. Siems, M. (2005). What does not work in securities law: A critique of La Porta et al.'s methodology. International Company and Commercial Law Review, 7, 300–305. Siems, M. (2006). Shareholder protection across countries — Is the EU on the right track? Cambridge University: Center of Business Research (Working paper). Sobel, Michael E. (1982). Asymptotic confidence intervals for indirect effects in structural equation models. In S. Leinhardt (Ed.), Sociological methodology (pp. 290–312). Washington DC: American Sociological Association. Sobel, Michael E. (1986). Some new results on indirect effects and their standard errors in covariance structure models. In N. Tuma (Ed.), Sociological methodology (pp. 159–186). Washington, DC: American Sociological Association. Stulz, René M., & Williamson, R. (2003). Culture, openness and finance. Journal of Financial Economics, 70(3), 313–349. Williamson, Oliver E. (2000). The new institutional economics: Taking stock, looking ahead. Journal of Economic Literature, XXXVIII, 595–613. Zhao, X., Lynch, Jr., John G., & Chen, Q. (forthcoming). Reconsidering Baron and Kenny: Myths and truths about mediation analysis. (Published online Feb 15, 2010), Journal of Consumer Research.