Government ownership and corporate governance: Evidence from the EU

Journal of Banking & Finance 36 (2012) 2917–2934

Contents lists available at SciVerse ScienceDirect

Journal of Banking & Finance journal homepage: www.elsevier.com/locate/jbf

Government ownership and corporate governance: Evidence from the EU Ginka Borisova a, Paul Brockman b,⇑, Jesus M. Salas b, Andrey Zagorchev c a

Department of Finance, College of Business, Iowa State University, Ames, IA 50011, USA Perella Department of Finance, College of Business and Economics, Lehigh University, Bethlehem, PA 18015, USA c Division of Business, Concord University, Athens, WV 24712, USA b

a r t i c l e

i n f o

Article history: Available online 8 February 2012 JEL classification: G32 G34 Keywords: Corporate governance Government ownership Legal origin Golden shares

a b s t r a c t The ongoing global financial crisis has led to the largest increase in state intervention since the Great Depression. Direct government ownership in publicly-traded corporations has increased dramatically since 2008. How will this increase in public ownership affect the governance of these erstwhile private companies? We examine the impact of government ownership on corporate governance using a sample of firms from the European Union, a region that is relatively familiar with active government participation. Our main finding is that government ownership is associated with lower governance quality. We further show that while government intervention is negatively related to governance quality in civil law countries, it is positively related to governance quality in common law countries. Finally, we find that the preferential voting rights of golden shares are especially damaging to governance quality. Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction The debate over government involvement in private enterprises has been reopened because of the worldwide intervention by governments that followed the recent financial crisis. The last few years have brought a significant shift from free market capitalism towards state ownership and control in countries around the world, particularly in areas of corporate governance. In March 2009, the Obama administration used the threat of refusing bailout money to force out General Motors CEO Rick Wagoner and most of the firm’s board members. The three top executives of Caisse d’Epargne, France’s third-largest retail bank, resigned in October 2008 after the bank suffered a loss of €600 million from trading derivatives. President Nicolas Sarkozy and the French government urged the bank’s executives to ‘‘take the consequences’’ of the huge losses and welcomed the resignations of the bank’s leadership.1 Against this backdrop, we examine the impact of government influence on corporate governance practices at firms with government ownership. It is possible that increasing levels of government ownership will lead to greater monitoring in particular, and improved governance in general, because of governments’ monopoly on the use of coercive power. Governments wield bigger sticks and carrots

⇑ Corresponding author. Tel.: +1 610 758 2914; fax: +1 610 758 6429. E-mail addresses: [email protected] (G. Borisova), [email protected] (P. Brock man), [email protected] (J.M. Salas), [email protected] (A. Zagorchev). 1 Sources: Wall Street Journal (March 30, 2009), Economist (November 29, 2008), and Bloomberg (October 20, 2008). 0378-4266/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.jbankfin.2012.01.008

than any potential private-sector partner. For example, state shareholders were ready to oust board members opposed to the appointment of the government’s choice for the new chairman of Thales, a company partially owned by the French state, in May 2009.2 On the other hand, it is not at all clear that government interests are aligned with shareholder interests. Governments are much more likely to use their influence to maximize employment, for example, even if the hiring of additional workers is not warranted based on a firm-level, cost-benefit analysis (Megginson, 2005). State owners could be reluctant to allow corporate governance improvements that would interfere with their non-profit-maximizing goals. In this study, we draw on a sample of 373 companies from 14 European Union (EU) countries during the period 2003–2008 to test the impact of government ownership on corporate governance. We utilize RiskMetrics Corporate Governance Quotient (CGQ), which incorporates the most widely used corporate governance proxies. This measure is popular in studies with non-US samples because international corporate governance data are difficult to collect and because comparing non-standardized governance measures across countries can be problematic.3 Since CGQ is a rating based on multiple corporate governance measures, we also examine specific subcomponents that have received significant

2

Source: The Financial Times (May 19, 2009). For a sample of recent studies that compare corporate governance across countries using CGQ components, see Doidge et al. (2007), Aggarwal et al. (2009), Chhaochharia and Laeven (2009), Bruno and Claessens (2010), and Aggarwal et al. (2011). 3

2918

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

attention in the literature (e.g., Bruno and Claessens, 2010): board independence, the presence of board committees, board entrenchment, committee independence, board transparency, and CEO power. By considering these elements, we address the central question of the analysis: how do the corporate governance practices of firms with state ownership compare to those of firms not held by the state? We measure state presence as the percentage of government ownership in our sample of publicly-traded corporations. Because the state frequently maintains partial ownership after privatizations, we use a pool of privatized firms as a starting point for the construction of our sample. Next, we examine the influence of legal origin on the relation between government ownership and corporate governance. Previous research (e.g., La Porta et al., 1998, 1999b) suggests that a country’s legal and regulatory framework plays a significant role in capital market development, governance, and the behavior of the state. Finally, we investigate how the presence of additional voting rights through golden shares affects the government ownership-corporate governance relation. This latter part of the study examines whether the type (as opposed to the amount) of government ownership has an incremental effect on corporate governance quality. Our first set of empirical results shows that government ownership is generally detrimental to good corporate governance, consistent with the state’s interests not being completely in line with those of the firm. Isolating national government holdings from total government ownership, we find that the negative effect of state holdings is mostly due to national ownership. We also find some evidence that central governments have a particularly damaging impact on governance by decreasing the number of board committees and augmenting CEO power, implying that governments attempt to maintain control by concentrating power within a firm. Our second set of results confirms that the legal framework of common law (civil law) countries has a positive (negative) incremental impact on the relation between government ownership and corporate governance, with the state furthering the practices fostered by its legal system. Finally, we show that the use of golden shares has an incrementally negative effect on corporate governance. While government ownership is generally associated with weaker corporate governance, the government’s ability to leverage up its voting power through golden shares is especially damaging. Throughout the study, we are mindful of the possible influence of endogeneity on our empirical results. We address these potential concerns in several ways. First, we use lagged government ownership throughout the study. Second, the government ownership we collect is retained ownership in previously state-owned enterprises that are now owned in part by the private sector. After a government sells part of its holdings in a company, it often maintains a large ownership stake. As time passes, the state continues to decrease its holdings through subsequent tranche sales. Therefore, the residual government ownership level in such firms is not a result of active investment policies due to contemporaneous corporate governance quality, which precludes the reading of reverse causality in our results. Third, we use fixed effects in all of our models, and fourth, we use an instrumental variable approach to further control for the association between state ownership and corporate governance quality. All of these econometric methods reveal consistent empirical results. The primary contribution of our study is to expand knowledge of the governance role played by state actors. Denis and McConnell (2003) identify this aspect of corporate governance as an especially underdeveloped area in the literature. Addressing this relative void is all the more important in the context of the 2008 global financial crisis and the concomitant rise of government ownership. Related regional or country-specific studies also argue that government ownership can have significant effects on firm policy. Price et al.

(2011) characterize the government ownership and frequent bailouts prevalent in Mexico until the late-1990s as engendering an apathetic view towards corporate governance. In a study of Argentine banks, Berger et al. (2005) associate the governance fostered by state ownership with inefficiencies and poor performance. Fan et al. (2007) show that Chinese government-owned firms which retain politically-connected CEOs perform worse than those with unconnected CEOs. They also find that politically-connected CEOs are more likely to appoint other bureaucrats to the board of directors. These results suggest that government intervention in corporations is negatively related to corporate governance quality, an idea that Ferri (2009) discusses in the context of the Chinese banking sector. The remainder of the paper is organized as follows. In Section 2, we discuss why government owners are of special interest to academics and practitioners. We describe our data and methodology in Section 3. In Section 4, we present and discuss our empirical results and perform robustness checks. We briefly conclude in Section 5.

2. Why are governments unique owners? 2.1. The state as a conflicted institutional owner Besides its persistent role in the corporate world, there are many characteristics that make the state a unique owner worthy of study. Governments and institutional investors are similar in many ways because of their significant resources and power. However, governments and institutions can often have very different objectives, which can lead to disparate outcomes when it comes to their effect on corporate governance. Governments tend to have much deeper pockets than institutions, and in most developed economies, governments can leverage themselves almost infinitely, using implicit guarantees to secure debt financing for state-controlled firms (Borisova and Megginson, 2011). The relative ease with which these firms secure financing could discourage monitoring, allowing agency problems to develop. Additionally, governments have the ability to create regulations that can positively or negatively affect a company, even to the point of forcing it to shut down. Institutional owners arguably have an informational advantage over the average investor because of their superior research and analytical skills. Governments have a different informational advantage in their ability to demand any and all information about a firm through regulatory or legal means. Perhaps the biggest difference (and likely the most relevant to this study) between institutions and governments is that institutions’ sole objective is to profit from their investments. The state might want profitable investments as well, but their motivation can also include the reduction of unemployment, increase in tax collection, and overall stability of the financial system. Government owners could work to prevent governance improvements that would curtail the realization of these alternative goals. In short, governments and institutions share some similarities, but governments have much greater access to resources than institutions and might own firms for reasons other than the maximization of wealth. Because institutional investors are influential and prevalent in financial markets, there is a vast literature exploring how they affect corporate governance quality. For example, Gillan and Starks (2000) find that financial institutions are more successful in instituting shareholder proposals. When financial institutions face too much resistance, they just ‘‘vote with their feet’’ and sell out (Parrino et al., 2003), damaging companies’ reputations. However, due to bureaucratic or social policies, government owners may be somewhat less mobile than these institutional investors and therefore more likely to maintain their stake and desire for governance

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

reform within a firm. In this way the state could be similar to a stable institutional owner, which Elyasiani and Jia (2010) associate with ameliorating corporate governance within a firm. However, Ruiz-Mallorquí and Santana-Martín (2011) show that benefits accrue to firms with higher (or dominant) institutional ownership only when monitoring by these owners is not weakened by strong business ties to the firms. State ownership, often motivated by the desire to keep a nationally important firm viable, could be susceptible to having its monitoring functions compromised due to the government’s broad network of connections and varied political objectives. Using CGQ component data, Aggarwal et al. (2011) empirically find that institutional owners engender good governance practices, specifically when these owners are from nations with legal systems that traditionally support shareholders’ rights (i.e., common law nations). A related literature has examined the effect of government pension fund intervention on corporate governance. For example, Del Guercio and Hawkins (1999) find that even though pension funds (many of which are governmentcontrolled) vary in the way they request changes in corporate policy, they are usually successful in promoting change. Public pension funds, however, could also be guided or restricted by suboptimal government directives, as discussed by Angelidis and Tessaromatis (2010). The business and financial press often reports on specific cases, showing that governance policy can be conflicted in firms with state ownership. For instance, the privatized bank Societé Générale (still partially owned by the government-run institution Caisse des Dépôts et Consignations) had several powerful antitakeover provisions in place as of early 2008, despite the fact that CEO and chairman Daniel Bouton authored a commissioned report in 2002 promoting improved corporate governance. Furthermore, Bouton stepped down from his position as CEO following a trading scandal and questions about money laundering in May 2008, temporarily separating the chairman and CEO roles (as often recommended in governance codes to limit CEO power). However, these roles were reunited when CEO Frédéric Oudéa, a former civil servant and adviser to French President Nicolas Sarkozy, replaced Bouton as chairman in May 2009; subsequent institutional shareholder requests to again disassociate these positions in May 2010 were denied by the board of Societé Générale.4

2919

corporate governance policies under common law but shun them under civil law. Supporting this notion, Fligstein and Choo (2005) state that legal tradition is a crucial factor for understanding firm-level corporate governance norms within a country. Investigating the governance structure of government-linked companies (GLCs) in Singapore, Ang and Ding (2006) find that GLCs practice better corporate governance compared to a control group of non-GLCs. Their study highlights how the government can monitor firms without exercising day-to-day operating control, and it also explains how Singaporean GLCs have started implementing international best practices in their corporate governance structures. In Australia, another common law country, government-owned corporations have adopted several principles which are beneficial for corporate governance (McDonough, 1998). These principles include efficiency in production and allocation of resources, management independence, and accountability for overall firm performance. In the Chinese context, however, political considerations often impede state-owned enterprises from achieving more efficiency, wealth maximization, and stronger managerial incentives (Clarke, 2003). As an example, Tian and Estrin (2007) show that debt financing in Chinese firms fails to achieve corporate governance improvements because of government ownership that encourages soft budget constraints. Previous research indicates that governance changes within a firm under the direct influence of government ownership warrant empirical investigation, particularly in light of media reports on the behavior of state-owned firms. On the one hand, the government’s unique role as a powerful and coercive owner provides opportunities for governance improvements through widespread monitoring and direct enforcement of recommended policies. On the other hand, state owners could also engender poor corporate governance when political decisions and non-profit-maximizing goals trump best practices. To the best of our knowledge, our multi-country study is the first to test the empirical question of whether governments have different corporate governance motivations based on ownership levels and legal origins.

3. Data and methodology 3.1. Data

2.2. The influence of legal systems According to the ‘‘development’’ view of government ownership as described in La Porta et al. (2002), the government can intervene to support firms and markets through its legal powers when necessary. In contrast, the alternative ‘‘political’’ view of government ownership in La Porta et al. (2002) emphasizes politicians’ desire to guide firm investments in order to achieve political goals, such as providing employment or benefits to favored constituents. In general, common law countries appear to emphasize the development view of government participation, while civil law countries seem to embrace the political view. Consistent with this interpretation, La Porta et al. (2008) suggest that common law nations handle social issues using legal means, while civil law nations employ greater direct government control. Therefore, government ownership under common law is less prevalent and often associated with maintaining financial markets and supporting firms in times of crisis, while redirecting company resources via state ownership can be a channel that civil law governments use to further their political agendas. In turn, state owners could encourage stronger 4 Sources: The Financial Times (February 12, 2008 and May 26, 2010) and The Deal (May 6, 2009).

We study companies in the EU for a number of reasons. First, a large number of firms in the EU have significant government participation (Bortolotti and Faccio, 2009). Using this framework, we will be able to generalize our results and conclusions to other environments (e.g., the US) where government participation in the corporate sector is on the rise. Second, there are fewer cross-country differences in this region than in most others, partially due to the integration fostered by the EU. This context will make it easier to form an overall conclusion about the role of government ownership on corporate governance. Third, corporate governance and other relevant data are widely available for the EU area. We first identify all companies in the EU that are covered by both Thomson Worldscope and RiskMetrics (formerly Institutional Shareholder Services (ISS)). In the data collection process, company names from Thomson Worldscope are manually matched with firm names from the CGQ database. We collect the relevant corporate data from Thomson Worldscope and drop companies that have no available financial data. RiskMetrics contains global corporate governance data on a firm level, with monthly updates beginning November 2003. Since the RiskMetrics database is updated on the first day of the month, CGQ scores are matched to the previous month’s accounting data. We match January 2009 CGQ data, for example, to the December 2008 financial data.

2920

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

Overall, RiskMetrics covers 1911 companies from the EU. After dropping firms without corresponding financial data, the sample with CGQ data includes 1327 companies from 14 EU countries from January 2003 to June 2009. To define a group of firms with significant government involvement from this sample, we identify a pool of state-owned enterprises in which the state has subsequently sold at least some of its equity stake to the private sector (i.e., privatized firms). We therefore base our final sample construction on these government-owned companies, identified by the Privatization Barometer (PB) database and by the Thomson ONE Banker M&A Deals Analysis. We then use a propensity score matching procedure to match these government-owned companies with non-government-owned firms using firm size, leverage, sales growth, ROA, capital expenditure, stock value traded as a percentage of GDP, GDP growth, GDP per capita, financial structure, and legal origin. In the first step, this matching procedure uses a probit regression to determine the characteristics that best identify government-owned companies. In the second step, the procedure finds non-government-owned companies that are most similar to government-owned firms based on these characteristics. In the final sample of firms, the largest difference in the propensity scores between government-owned firms and non-government-owned firms is 0.0051. Once we obtain a matched sample, we collect government ownership for both the government-owned firms and their matches using Thomson Ownership data, supplemented with information from the companies’ annual reports and the PB database. Total state ownership consists of holdings by central/national and local governments, with the former contributing the bulk of the government stake in our sample. We therefore differentiate between total and national government ownership in our analysis to determine how much of the relation between corporate governance practices and state ownership is linked specifically to holdings by central governments. On average, governments own 11.8% of our stateowned sample.5 Our final sample consists of 133 governmentowned companies in a sample of 373 companies from 14 EU countries for the period 2003–2008.6 Our firm-level corporate governance quality dependent variable is the RiskMetrics CGQ score that incorporates many of the governance proxies used in the literature (e.g., Doidge et al., 2007). CGQ also has the most widespread coverage in the industry, including ratings on 8000 firms in more than 30 countries. Perhaps the biggest advantage of CGQ scores is that the same entity produces all ratings across different companies and countries. Consequently, CGQ reduces the bias that arises when one tries to compare governance structures across countries. Further, RiskMetrics accounts for country-level differences in regulation, and adjustments are made to the CGQ scores when these regulations affect governance practices (ISS, 2003). The RiskMetrics CGQ raw scores are weighted scores of eight categories,7 and the composite CGQ rating comprises these core categories: (1) board of directors, (2) audit issues, (3) charter and bylaw provisions, (4) anti-takeover provisions, (5) executive and director compensation, (6) progressive practices, (7) 5 Although on average the state is not a majority owner in our sample firms, its presence as a blockholder likely has more significant implications for the firm when compared to other investors. Aggarwal et al. (2011) make a similar point that institutional investors with small stakes in a firm can greatly influence governance practices. Government ownership also represents 0.11% of our non-governmentowned company sample. We find traces of state holdings, primarily by local governments, in a few of these companies. 6 Each firm-year observation of a government-owned firm could be matched to a non-government-owned firm-year for each of the years between 2003 and 2008. Because of this, for each government-owned firm there are six possible nongovernment-owned firm-year matches and consequently more non-governmentowned than government-owned firms in our sample. 7 For more details of the components in the CGQ scores, see Bruno and Claessens (2010).

ownership, and (8) director education. According to the RiskMetrics coding scheme, higher CGQ scores indicate better corporate governance of firms. Following Bruno and Claessens (2010), we also extract six subcomponents of the CGQ score that are related to measures commonly used in the corporate governance literature. The six subcomponents are based on board independence, the number of board committees, board entrenchment, committee independence, board transparency, and CEO power, as described below:8 (a) Board Independence is a dummy variable that takes a value of 1 if the board is controlled by a majority of independent outsiders and zero otherwise.9 Weisbach (1988) finds that boards dominated by outside directors are more likely to remove top management after poor performance than boards dominated by inside directors. Boubakri et al. (2008) discuss how state owners often appoint government officials to the board of directors and find that these politically-connected firms are associated with poor performance even after privatization.10 (b) Board committees are a major focus of the Sarbanes–Oxley Act of 2002 because of their crucial influence on corporate governance. The Board Committee variable takes values ranging from 0 to 4, where higher values represent the existence of more of the following committees: audit, compensation, governance, and nomination. Although certain major stock exchanges (e.g., NYSE) require some of these committees as prerequisites for listing, their presence is still optional for many publicly-traded firms. Uzun et al. (2004) find some evidence linking the absence of an audit committee to fraudulent activity within a firm. (c) Gompers et al. (2003) and Bebchuk et al. (2009) find that firms with more antitakeover provisions have lower returns. Thus, we incorporate Board Entrenchment as a corporate governance component, taking values ranging from 0 to 4, where higher values are associated with fewer anti-takeover provisions. Although hostile takeovers have traditionally been uncommon in Europe (outside of the UK), Martynova and Renneboog (2008) discuss an increase in managementopposed European takeovers starting in the mid-1990s due to wider share ownership and institutional reforms. Jenkinson and Ljungqvist (2001) also point out that hostile takeover activity in Germany, for example, is often overlooked because it takes the more subtle form of control shifts facilitated by the accumulation of ownership blocks. (d) Committee Independence is a measure ranging from 0 to 3, with a point given when each of the nomination, compensation, and audit committees entirely comprises independent members. Klein (2002) and Davidson et al. (1998), for example, find evidence of firm benefits related to the independence of the audit and compensation committees. (e) Board Transparency is a combination of three dummy variables: the first dummy variable equals one if the auditor is ratified at fiscal year end, the second dummy variable equals one if audit expenses are strictly audit fees, and the third dummy variable equals one if the CEO is not involved in related party transactions. The board transparency measure

8 For consistency, all the corporate governance measures increase with quality, so that a higher governance measure means better corporate governance. 9 According to RiskMetrics proxy analyses, board independence for two-tier board structures in German firms, for example, is based exclusively on supervisory board members, since the management board consists only of company executives. 10 RiskMetrics Group (2007) considers government representatives on the board to be non-independent directors.

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

therefore takes values ranging from 0 to 3. Aggarwal et al. (2009) observe firm value losses associated with the lack of auditor ratification at an annual shareholder meeting. (f) CEO Power also ranges from 0 to 3. This measure increases with the separation of CEO and the chairman of the board, board independence, and the presence of a former CEO on the current board. Pi and Timme (1993) find a lower return on assets for banks when the chairman is also the CEO. We also gather data on a variety of firm-level and country-level independent variables. The following firm-level measures can potentially affect corporate governance and are used as control variables. We measure firm size as the log of the market value of total assets in millions of US dollars, and leverage is the ratio of the book value of debt to the book value of assets. Sales growth is the change in current period sales divided by the sales in the previous period, and serves as a proxy for growth opportunities. Return on assets (ROA) measures firm profitability and is the sum of net income and interest expense on debt minus capitalized interest, both of which are after taxes, divided by total assets. According to Gompers et al. (2003), firms with better corporate governance have better performance. Thus, we expect a positive coefficient on ROA throughout our regressions. The quick ratio is equal to the sum of cash, cash equivalents, and receivables divided by current liabilities, and it proxies for financial flexibility and liquidity. We also include capital expenditures, the firm’s funds used to acquire or upgrade its physical assets, scaled by total assets. External finance is capital expenditures minus cash holdings divided by capital expenditures. Several papers suggest the use of dependence on external finance as an alternative measure of growth opportunities (e.g., Durnev and Kim, 2005). Dividend dummy is a binary variable that takes a value of one if a firm pays dividends in a given year, and zero otherwise. Years since initial sale of ownership measures the number of years since the government first began to sell its ownership stake in a firm. Institutional ownership represents ownership by institutional investors as a percentage of shares outstanding. Next, we incorporate several country-level metrics as control variables. Following La Porta et al. (1998), we classify EU countries as civil law or common law systems based on their legal origin. The civil law dummy variable takes a value of one if a country has a civil law system, and zero if a country has a common law system. The stock market capitalization as a percentage of GDP measures the relative size of the stock market in each country. We measure GDP per capita as the log of gross domestic product per capita based on constant 2000 US dollars. GDP growth is the annual percentage growth rate of GDP calculated at market prices using a constant local currency. We collect GDP data from the World Bank’s World Development Indicators (WDI) Database. Appendix A presents definitions of all variables. Table 1 presents summary statistics of the variables used in our study. We report accounting-related summary statistics in Panel A, corporate governance statistics in Panel B, ownership characteristics in Panel C, legal origin in Panel D, economic and country variables in Panel E, sample construction figures in Panel F, and the number of firms per country in Panel G. Panel A shows that the average firm has log of total assets equal to 9.071 million dollars (i.e., a mean of 25 billion dollars in assets), a leverage ratio of 31%, sales growth of 9.48%, and ROA of 6.23%. Roughly 91% of these firms are dividend payers. Panel B shows that the average CGQ score is 32.47, with considerable variation between 12.30 at the 5th percentile and 63.00 at the 95th percentile. Panel C shows that our sample firms have 8.38% government ownership on average, the bulk of which is made up of holdings by national governments (6.72%). Moreover, the government owns golden shares in 18.4% of

2921

the firm-year observations in the sample. Panel D tells us that almost 83% of our sample observations are from firms residing in civil law countries, and the summary statistics in Panel E confirm that the sample consists of wealthy, well-developed countries. Panel F describes the construction of our final sample number of 373 firms. In Panel G, we provide a breakdown of the number of firms per country. In Table 2, we compare firm-level characteristics for government-owned firms and non-government-owned firms. There are no significant differences between these two groups (partly by construction) in terms of sales growth, capital expenditures, payout policy, institutional ownership, and the need for external financing. There are significant differences with respect to assets, leverage, ROA, and the quick ratio. Firms with government ownership are more levered, less profitable, and have lower quick ratios. More importantly, these univariate results suggest that government-owned firms have better corporate governance than their privately-owned counterparts. Government-owned firms score significantly better on their overall CGQ measure, board independence, board entrenchment, committee independence, and board transparency. In contrast, government-owned firms score significantly worse with respect to board committees and CEO power. As we will see in the next section, however, many of these univariate results will change substantially after controlling for firm characteristics in a multivariate setting. In Table 2, we also compare the CGQ score between common law firms and civil law firms. The results indicate that firms operating in civil law countries have significantly lower CGQ scores compared to firms operating in common law countries. 3.2. Methodology We use different methodologies to test our main hypothesis that government ownership is significantly related to corporate governance. First, we use firm random effects to regress corporate governance scores (CGQs) on the government ownership variable, a set of explanatory variables, and controls for year, industry, and country fixed effects. Doidge et al. (2007) discuss the importance of country fixed effects, which they find to capture significantly more of the variation in governance ratings than traditional country-level independent variables. Next, we use logit regressions to test whether the likelihood of having an independent board (board independence) is affected by corporate governance. Finally, we use ordered logit regressions to test whether the existence of board committees (board committee), antitakeover provisions (board entrenchment), the independence of key committees (committee independence), appropriate board transactions (board transparency), or limits to the CEO’s role (CEO power) are related to government ownership. In all models, we use firm-level clustered standard errors to control for time-series dependence. We estimate the following model:

yit ¼ a þ b1 Git þ b2 X it þ ck þ lj þ mt þ eit

ð1Þ

where yit represents the vector of CGQs, a is the constant term, b1 represents a vector of coefficients, Git is a vector of government ownership, b2 is a vector of coefficients, Xit is a matrix of explanatory variables including both firm-level and country-level variables, ck (k = 1, . . . , 25) represents the industry fixed effects, lj (j = 1, . . . ,14) are the country fixed effects, mt (t = 1, . . . ,6) represents the annual fixed effects, and eit is the error term. The primary objective of this study is to empirically determine the sign and magnitude of the b1 coefficient related to government ownership variables. A positive coefficient for b1 suggests that government ownership has beneficial effects and leads to the

2922

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

Table 1 Summary statistics and sample of firms. Variable

N. obs.

Panel A. Accounting statistics Log (assets) Leverage Sales growth ROA Quick ratio Capital expend./Assets External finance Dividend dummy

989 989 989 989 989 989 989 989

9.071 0.306 9.480 6.226 0.913 0.057 1.747 0.914

1.494 0.184 22.085 8.339 0.578 0.053 11.133 0.280

6.806 0.049 9.820 2.580 0.360 0.010 8.242 0.000

11.668 0.587 34.850 16.810 1.780 0.130 0.978 1.000

Panel B. Corporate governance Corporate Governance Quotient (CGQ) score Board independence Board committee Board entrenchment Committee independence Board transparency CEO power

989 806 989 989 989 989 809

32.474 0.452 2.186 1.181 0.410 1.276 1.240

22.101 0.498 1.398 0.535 0.891 0.681 0.883

12.300 0.000 0.000 0.000 0.000 0.000 0.000

63.000 1.000 4.000 2.000 3.000 2.000 3.000

Panel C. Ownership characteristics Total government ownership National government ownership Institutional ownership Years since initial sale of ownership Golden share

989 989 989 989 989

8.384 6.721 31.718 8.197 0.184

17.549 15.994 23.486 7.961 0.388

0.000 0.000 3.340 0.000 0.000

51.120 50.820 74.580 21.000 1.000

Panel D. Legal origin Civil law dummy

989

0.828

0.377

0.000

1.000

Panel E. Economic and country variables Financial structure dummy Stock market cap. Stock value traded (% of GDP) GDP growth Log (GDP per capita)

989 989 989 989 989

0.654 85.243 102.094 2.326 10.020

0.476 35.832 68.349 1.258 0.245

0.000 39.252 16.770 0.076 9.513

1.000 145.864 206.148 4.376 10.314

Panel F. Sample of firms Number of firms covered in the CGQ database by RiskMetrics Minus firms without corresponding financial data in Thomson One Number of firms with financial and CGQ data Minus the firms without a corresponding match Final sample of firms with government ownership data Panel G. Number of firms per country Austria Belgium Denmark Finland France Germany Greece Ireland Italy Netherlands Portugal Spain Sweden United Kingdom

Mean

Std. dev.

5th percentile

95th percentile

1911 (584) 1327 (954) 373

13 11 10 18 62 41 20 4 31 29 10 36 17 71

This table presents summary statistics of firm characteristics, corporate governance, government ownership, and country level variables for 373 EU companies between 2003 and 2008. Log (assets) is the log of the market value of assets in millions of US$. Leverage is the ratio of total debt to total assets. Sales growth is the percentage sales growth computed as the change in current period sales divided by the sales in the previous period. ROA is the return on assets calculated as [net income + interest after taxes/total assets]. Quick ratio equals [(cash + cash equivalents + receivables)/current liabilities]. Cap. expend./Assets refers to the firm’s funds used to acquire or upgrade its physical assets divided by total assets. External finance is calculated for all companies as capital expenditures minus cash holdings divided by capital expenditures. Dividend dummy is a dummy variable that takes a value of one if a firm pays dividends in year t, and zero otherwise. CGQ score is the corporate governance quotient score calculated by RiskMetrics. Board independence takes a value of 1 if the board is controlled by a majority of independent outsiders, and zero otherwise. Board committee takes values ranging from 0 to 4, where higher values represent more committees. Board entrenchment takes values ranging from 0 to 4, where higher values are associated with fewer anti-takeover provisions. Committee independence takes values ranging from 0 to 3, where higher values are associated with more independent outside directors serving on committees. Board transparency takes values ranging from 0 to 3, where higher values are associated with a more transparent board. CEO power takes values ranging from 0 to 3, where higher values indicate less CEO power. Total government ownership measures the residual stake held by the national and local government in the company. National government ownership represents the level of ownership by the national government. Institutional ownership represents the ownership by institutional investors. The number of years since the government first began to sell its ownership stake in a firm is captured in the Years since initial sale of ownership variable. Golden share takes a value of 1 if the government has retained a golden share, and zero otherwise. Civil law dummy takes a value of one if a country has a civil law system, and zero if a country has a common law system. Financial structure dummy takes a value of one if a country has a bank-based system and zero if a country has a market-based system. Stock market cap. (% of GDP), Stock value traded (% of GDP), GDP growth, and log (GDP per capita) are from the World Bank’s WDI Database. Descriptions about the final sample of firms with government ownership data and their distribution across countries are also presented.

2923

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934 Table 2 Differences between firms partially owned by the government and firms not owned by the government and between common law firms and civil law firms. Firm variable

Summary variable

Log (assets)

Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs. Ownership N. obs.

Leverage Sales growth ROA Quick ratio Cap. expend./Assets External finance Dividend dummy Institutional ownership CGQ score Board independence Board committee Board entrenchment Committee independence Board transparency CEO power

mean mean mean mean mean mean mean mean mean mean mean mean mean mean mean mean

Lag of total government ownership and golden share (GS) Zero percent and no GS

Non-zero percent/GS

8.84 556 0.30 556 9.64 556 6.78 556 0.97 556 0.06 556 1.87 556 0.91 556 32.74 556 30.34 556 0.42 460 2.26 556 1.12 556 0.25 556 1.24 556 1.34 463

9.37 433 0.32 433 9.28 433 5.51 433 0.84 433 0.06 433 1.60 433 0.92 433 30.41 433 35.22 433 0.50 346 2.10 433 1.26 433 0.61 433 1.33 433 1.10 346

p-value for diff. of means test

0.000 0.098 0.801 0.017 0.001 0.848 0.705 0.782 0.123 0.001 0.017 0.077 0.000 0.000 0.043 0.000

Firm variable

Summary variable

Common law

Civil law

p-value for difference of means test

CGQ score

Legal origin mean N. obs.

56.28 170

27.53 819

0.000

This table presents tests of differences in means of firm characteristics between firms without government ownership or a golden share and firms with government ownership and/or a golden share. The table also presents test of differences in means of CGQ between common law firms and civil law firms. Log (assets) is the log of the market value of assets in millions of US$. Leverage is the ratio of total debt to total assets. Sales growth is the percentage sales growth computed as the change in current period sales divided by the sales in the previous period. ROA is the return on assets calculated as [net income + interest after taxes/total assets]. Quick ratio equals [(cash + cash equivalents + receivables)/current liabilities]. Cap. expend./Assets refers to the firm’s funds used to acquire or upgrade its physical assets divided by total assets. External finance is calculated for all companies as capital expenditures minus cash holdings divided by capital expenditures. Dividend dummy is a dummy variable that takes a value of one if a firm pays dividends in year t, and zero otherwise. Institutional ownership represents the ownership by institutional investors. CGQ score is the corporate governance quotient score calculated by RiskMetrics. Board independence takes a value of 1 if the board is controlled by a majority of independent outsiders, and zero otherwise. Board committee takes values ranging from 0 to 4, where higher values represent more committees. Board entrenchment takes values ranging from 0 to 4, where higher values are associated with fewer anti-takeover provisions. Committee independence takes values ranging from 0 to 3, where higher values are associated with more independent outside directors serving on committees. Board transparency takes values ranging from 0 to 3, where higher values are associated with a more transparent board. CEO power takes values ranging from 0 to 3, where higher values indicate less CEO power. The p-values report the significance of the differences in means tests.

improvement of corporate governance quality. By including a number of firm-level and country-level control variables that can explain the variation in corporate governance quality, we isolate the possibility that the results may be distorted by firm-level financial measures or other fixed effects. To reduce the likelihood of endogeneity, we use one-year lags of the independent firmlevel financial variables and county-level metrics in the regressions. Although government ownership in our sample is usually retained from the time of privatization, and despite using lagged independent variables and fixed effects throughout the paper, it is still possible that our results are subject to endogeneity concerns. At the end of our empirical analyses, we implement additional robustness tests to address endogeneity issues. Specifically, we employ a two-stage least squares instrumental variable approach in which we use privatization and country factors as instruments for government ownership (and the interaction of government ownership with civil law). The results from instrumental variable regressions confirm our original empirical findings.

4. Empirical results 4.1. Government ownership and corporate governance quality The first question that we address is simply whether government ownership is related to corporate governance. To do this, we regress our governance quality proxies against government ownership. In Table 3, we present results in which the CGQ broad score is the dependent variable. Models 1 and 2 examine the relation between total government ownership and corporate governance without and with country dummy variables, respectively. Models 3 and 4 examine this relation for national ownership only – again, both without and with country dummy variables, respectively. Our main variables of interest are the state ownership variables: total government ownership and national ownership only.11 11 In unreported results, we isolate local government ownership to determine its association with corporate governance practices (e.g., Chen et al., 2009). However, possibly due to the generally low levels of local government ownership in our sample, we find almost no significant relations between the subset of local government-only stakes and corporate governance.

2924

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

Table 3 Multivariate analysis of government ownership and corporate governance. Model 1 Total government ownership

***

0.136 (3.37)

Model 2

Leverage Sales growth ROA Quick ratio Cap. expend./Assets External finance Dividend dummy Years since initial sale of ownership Institutional ownership GDP growth Log (GDP per capita) Stock market cap. Intercept Country dummies Observations R-squared

Model 4

0.105** (2.44) 2.344*** (5.34) 1.983 (0.79) 0.010 (0.63) 0.063 (1.26) 2.831*** (3.04) 12.195 (1.10) 0.084** (2.06) 1.542 (0.86) 0.084 (1.10) 0.055** (2.06) 4.530*** (5.03) 74.031** (2.23) 0.241*** (3.81) 740.077** (2.16) Yes 989 0.63

0.100 (2.67)

3.727*** (7.00) 1.375 (0.45) 0.000 (0.01) 0.012 (0.19) 3.430*** (3.03) 17.901 (1.29) 0.086* (1.83) 3.058 (1.37) 0.128 (1.25) 0.039 (1.17) 1.261** (1.98) 26.528*** (7.14) 0.163*** (6.37) 269.367*** (7.22)

2.268*** (5.18) 2.164 (0.86) 0.008 (0.52) 0.060 (1.21) 2.727*** (2.93) 11.292 (1.01) 0.083** (2.05) 1.486 (0.83) 0.080 (1.05) 0.055** (2.07) 4.520*** (5.01) 72.733** (2.20) 0.242*** (3.81) 727.670** (2.14)

0.161*** (3.80) 3.793*** (7.16) 1.208 (0.40) 0.002 (0.07) 0.014 (0.23) 3.507*** (3.09) 18.784 (1.36) 0.089* (1.88) 3.118 (1.40) 0.123 (1.20) 0.036 (1.07) 1.328** (2.08) 26.136*** (7.04) 0.169*** (6.62) 267.225*** (7.18)

No 989 0.49

Yes 989 0.63

No 989 0.49

National government ownership Log (assets)

Model 3

***

This table presents firm random effects regression results. The dependent variable is the RiskMetrics corporate governance quotient, and the explanatory variables follow. Firm data are from Thomson Worldscope. Total government ownership measures the residual stake held by the national and local government in the company. National government ownership represents the level of ownership by the central government. Log (assets) is the log of the market value of assets in millions of US$. Leverage is the ratio of total debt to total assets. Sales growth is the percentage sales growth computed as the change in current period sales divided by the sales in the previous period. ROA is the return on assets calculated as [net income + interest after taxes/total assets]. Quick ratio equals [(cash + cash equivalents + receivables)/current liabilities]. Cap. expend./Assets refers to the firm’s funds used to acquire or upgrade its physical assets divided by total assets. External finance is calculated for all companies as capital expenditures minus cash holdings divided by capital expenditures. Dividend dummy is a dummy variable that takes a value of one if a firm pays dividends in year t, and zero otherwise. Years since initial sale of ownership indicates the number of years since the government first began to sell its ownership stake in the firm. Institutional ownership represents the ownership by institutional investors. GDP growth, Log (GDP per capita), and Stock market cap. (% of GDP) are from the World Bank’s WDI Database. In addition, we include industry and year dummy variables in all models. The presented models use one year lags of the explanatory variables. The values in parentheses are t-statistics based on standard errors that are clustered by firms. * Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level.

The estimated coefficient for the total government ownership variable is negative and significant in Model 1 (0.136) and Model 2 (0.100).12 This result suggests that government ownership has a detrimental impact on governance, consistent with the state’s political objectives interfering with a firm’s recommended corporate practices. Our findings in Models 3 and 4 show that this detrimental effect is attributable to national ownership. The estimated coefficient for national ownership is negative and significant in Model 3 (0.161) and Model 4 (0.105). As discussed in the previous section, we include a number of control variables in our regressions, including year, industry, and

country dummy variables.13 As expected, larger firms consistently have better governance quality than smaller firms throughout the specifications. Larger firms typically have more analyst coverage than smaller firms, and this greater scrutiny could encourage better governance practices. We also find that firms with good governance tend to be in countries with high GDP growth and a larger stock market capitalization. This latter result supports the idea that firms which are more likely to issue equity are also more concerned with protecting shareholders through rules of governance (Dahya et al., 2008). Higher institutional ownership is not linked with improved corporate governance, echoing the relative ineffectiveness of these

12 We examine the economic significance of our results by following the measure used in Chhaochharia and Laeven (2009). Specifically, we multiply the regression coefficient of 0.136 (0.100) from Model 1 (2) of Table 3 by the standard deviation of our total government ownership variable (17.549) and then divide this product by the standard deviation of our CGQ measure (22.101). This calculation yields a change in CGQ that is roughly 11% (8%) of its standard deviation. For comparison purposes, Chhaochharia and Laeven (2009) report a change that is 7% of the standard deviation of their dependent variable.

13 In unreported tests, we control for two additional country-level variables (i.e., foreign national debt as a percentage of GDP and foreign national debt service as a percentage of exports) in all regressions from Tables 3–9. We collect both variables from the financial risk ratings of the International Country Risk Guide (ICRG). Morey et al. (2009) show the relevance of ICRG’s financial risk measures to corporate governance levels in their sample countries. The empirical results after controlling for these two additional variables are similar to those reported herein.

2925

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934 Table 4 Multivariate analysis of government ownership and disaggregated governance measures. Board independence Model 1 Total government ownership

Total government ownership

0.018*** (2.97)

806

989 Committee independence

989

Model 5

Model 6

Model 7

Model 8

0.007 (1.18)

0.002 (0.14) 0.008 (1.16)

989

Model 9

989

989

0.002 (0.13) 989

989

CEO power Model 10

Model 11

Model 12

**

0.003 (0.63)

National government ownership Observations

Model 4

806 Board entrenchment

Board transparency

Total government ownership

Model 3 0.019*** (3.56)

0.002 (0.23)

National government ownership Observations

Model 2

0.002 (0.22)

National government ownership Observations

Board committee

0.015 (2.29) 0.000 (0.03) 989

809

0.014** (2.02) 809

This table presents logit and ordered logit results of governance and government ownership. The dependent variables are Board independence in the first two models, Board committee in models 3 and 4, Board entrenchment for models 5 and 6, Committee independence for models 7 and 8, Board transparency in models 9 and 10 and CEO power in the last two models. Board independence takes a value of 1 if the board is controlled by a majority of independent outsiders, and zero otherwise. Board committee takes values ranging from 0 to 4, where higher values represent more committees. Board entrenchment takes values ranging from 0 to 4, where higher values are associated with fewer anti-takeover provisions. Committee independence takes values ranging from 0 to 3, where higher values are associated with more independent outside directors serving on committees. Board transparency takes values ranging from 0 to 3, where higher values are associated with a more transparent board. CEO power takes values ranging from 0 to 3, where higher values indicate less CEO power. Total government ownership measures the residual stake held by the national and local government in the company. National government ownership represents the level of ownership by the national government. All models include the control variables used in the previous table, including industry and year dummy variables. The presented models use one year lags of the explanatory variables. The values in parentheses are t-statistics based on standard errors that are clustered by firms.  indicate significance at the 10% levels, respectively. ** Significance at the 5% level. *** Significance at the 1% level.

monitors in a European setting, as discussed by Renneboog (2000) and Brunello et al. (2003). In Table 4, we present results for the six subcomponent measures of corporate governance quality (i.e., board independence, board committee, board entrenchment, committee independence, board transparency, and CEO power). Total government ownership has a significantly negative impact on board committees (Model 3 coefficient = 0.019) and CEO power (Model 11 coefficient = 0.015). Similar to our results in Table 3, we also find that the detrimental impact of government ownership on governance is due to national ownership. For the board committee subcomponent, the estimated coefficient for national ownership is negative and significant (Model 4 coefficient = 0.018), as it is for CEO power (Model 12 coefficient = 0.014). The subcomponent results show that government ownership has an insignificant effect on board independence, board entrenchment, committee independence, and board transparency. Overall, these empirical results suggest that government ownership is harmful to corporate governance, mostly due to national ownership and its relation to fewer board committees and a greater amount of CEO power. These findings suggest that the state encourages a centralized power structure in firms, perhaps to streamline the decision-making process or to facilitate a transfer of control when intervention is required. The regressions up to this point implicitly assume that all governments are alike. We know from the law and finance literature that countries with civil law traditions tend to have weaker shareholder rights than those with common law traditions. We therefore expect that disaggregating these different legal environments can further explain the effect of state owners on

firm-level corporate governance. In the next subsection, we test whether governments in common law countries have different incentives than those in civil law countries when they have ownership stakes in publicly-traded corporations. 4.2. Government incentives, government ownership, and corporate governance In our next set of results, we examine the marginal impact of the firm’s legal environment (i.e., legal origin) on the relation between government ownership and corporate governance. Based on previous research (e.g., La Porta et al., 1998, 1999b; Aggarwal et al., 2011), we expect that civil law country governments will have a more negative impact on this relation than common law country governments. Extant literature highlights the heritage of civil law systems as a way for the state to assert economic control, in contrast to the protection traditionally afforded individuals and businesses under common law. Further, press reports highlight how the support of motivated governments can help foster improved corporate governance practices. For instance, despite complaints from chairmen at some of the top UK firms, the British government incorporated recommendations (such as having more independent directors) from a 2003 report on board policies by Sir Derek Higgs into a new code of governance rules.14 Our research design captures the marginal impact of civil law countries by adding an interaction term between government ownership and a civil law dummy variable. Because legal origin 14

Source: The Financial Times (January 31, 2003 and October 27, 2003).

2926

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

Table 5 Multivariate analysis of incentives, government ownership and corporate governance. Model 1 Total government ownership

*

0.073 (1.95)

Total government ownership  Civil law dummy

Model 2

Model 3

Model 4

0.092** (2.27)

⁄

0.411 (1.72) 0.485⁄⁄ (2.01)

3.973*** (8.64) 2.389 (0.79) 0.006 (0.31) 0.036 (0.61) 3.571*** (3.49) 13.912 (1.01) 0.074 (1.61) 1.759 (0.87) 0.002 (0.03) 0.036 (1.17) 0.652 (1.06) 20.391*** (5.67) 0.042* (1.65) 23.536*** (9.57) 177.834*** (4.76)

3.988*** (8.64) 2.396 (0.79) 0.006 (0.32) 0.036 (0.62) 3.582*** (3.50) 13.829 (1.00) 0.075 (1.62) 1.808 (0.89) 0.003 (0.04) 0.036 (1.15) 0.671 (1.09) 20.413*** (5.68) 0.042* (1.65) 23.333*** (9.36) 178.409*** (4.77)

4.010*** (8.76) 2.290 (0.76) 0.007 (0.36) 0.037 (0.64) 3.633*** (3.55) 14.600 (1.06) 0.076* (1.66) 1.800 (0.89) 0.000 (0.00) 0.039 (1.24) 0.705 (1.14) 20.210*** (5.65) 0.046* (1.82) 23.374*** (9.51) 177.181*** (4.77)

0.252⁄ (1.70) 0.346⁄⁄ (2.24) 4.020*** (8.75) 2.294 (0.76) 0.007 (0.36) 0.037 (0.64) 3.642*** (3.55) 14.550 (1.06) 0.077* (1.67) 1.833 (0.90) 0.001 (0.02) 0.038 (1.22) 0.717 (1.16) 20.222*** (5.65) 0.047* (1.82) 23.250*** (9.35) 177.552*** (4.77)

989 0.56

989 0.56

989 0.56

989 0.56

National government ownership National government ownership  Civil law dummy Log (assets) Leverage Sales growth ROA Quick ratio Cap. expend./Assets External finance Dividend dummy Years since initial sale of ownership Institutional ownership GDP growth Log (GDP per capita) Stock market cap. Civil law dummy Intercept Observations R-squared

This table presents firm random effects regression results. The dependent variable is the RiskMetrics corporate governance quotient, and the explanatory variables follow. Firm data are from Thomson Worldscope. Total government ownership measures the stake held by the national and local government in the company. National government ownership represents the level of ownership by the central government. Log (assets) is the log of the market value of assets in millions of US$. Leverage is the ratio of total debt to total assets. Sales growth is the percentage sales growth computed as the change in current period sales divided by the sales in the previous period. ROA is the return on assets calculated as [net income + interest after taxes/total assets]. Quick ratio equals [(cash + cash equivalents + receivables)/current liabilities]. Cap. expend./Assets refers to the firm’s funds used to acquire or upgrade its physical assets divided by total assets. External finance is calculated for all companies as capital expenditures minus cash holdings divided by capital expenditures. Dividend dummy is a dummy variable that takes a value of one if a firm pays dividends in year t, and zero otherwise. Years since initial sale of ownership indicates the number of years since the government first began to sell its ownership stake in the firm. Institutional ownership represents the ownership by institutional investors. GDP growth, Log (GDP per capita), and Stock market cap. (% of GDP) are from the World Bank’s WDI Database. Civil law dummy takes a value of one if a country has a civil law system, and zero if a country has a common law system. This variable is interacted with government ownership to compare effects across legal systems. In addition, we include industry and year dummy variables in all models. The presented models use one year lags of the explanatory variables. The values in parentheses are t-statistics based on standard errors that are clustered by firms. * Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level.

is a linear function of country dummies, we omit country dummy variables in these regressions. Each regression has two government ownership coefficients: one on government ownership and one on the interaction between government ownership and the civil law dummy. The effect of government ownership on corporate governance in common law countries is simply the coefficient on the government ownership variable. For civil law countries, the marginal effect of government ownership is the sum of the coefficients on both the government ownership variable and the interaction term. Table 5 presents our main results in which we use the broad CGQ corporate governance score to measure corporate governance quality. The coefficient on the newly added civil law dummy variable is persistently negative and significant, consistent with

evidence in La Porta et al. (1999b) that this legal system traditionally favors the government at the expense of other market participants. We also see in Model 2 that the impact of common law on the relation between total government ownership and corporate governance is positive (0.411) and significant. In sharp contrast, the impact of civil law on the relation between government ownership and corporate governance is negative (0.411– 0.485  (1) = 0.074) and significant.15 Model 4 shows that this same pattern of significance is present when only considering national ownership. That is, the impact of common law on the relation 15 We bold both the common law coefficient and the interaction term coefficient when the sum of the coefficients of government ownership and its interaction with civil law is significant using a chi-square test.

2927

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934 Table 6 Multivariate analysis of incentives, government ownership and disaggregated governance measures. Board independence Model 1 Total government ownership Total government ownership  Civil law dummy

Model 2

0.218 (0.67) 0.225 (0.69) 30.819** (2.58) 30.823** (2.11)

National government ownership  Civil law dummy 806

806

Board entrenchment Model 5 Total government ownership Total government ownership  Civil law dummy

Model 6

989

Model 9

989

Model 7

989

Model 8

0.023 (0.86) 0.059⁄ (1.82) 989

Model 10

989

Model 11

Model 12

0.423 (1.50) 0.435 (1.55) 0.091*** (4.43) 0.091*** (4.28)

National government ownership  Civil law dummy

989

CEO power

0.071** (2.48) 0.072** (2.46)

National government ownership

Observations

989

0.173⁄⁄⁄ (6.14) 0.152⁄⁄⁄ (5.37)

Board transparency

Total government ownership ⁄ Civil law dummy

0.007 (0.37) 0.039⁄ (1.82)

0.030 (1.08) 0.061⁄ (1.89)

National government ownership  Civil law dummy

Total government ownership

Model 4

Committee independence

0.171⁄⁄⁄ (6.24) 0.153⁄⁄⁄ (5.54)

National government ownership

Observations

Model 3 0.010 (0.48) 0.040⁄ (1.83)

National government ownership

Observations

Board committee

989

5.288 (0.70) 5.299 (0.70) 809

809

This table presents logit and ordered logit results of governance and government ownership for common vs. civil law countries. The dependent variables are Board independence in the first two models, Board committee in models 3 and 4, Board entrenchment for models 5 and 6, Committee independence for models 7 and 8, Board transparency in models 9 and 10 and CEO power in the last two models. Board independence takes a value of 1 if the board is controlled by a majority of independent outsiders, and zero otherwise. Board committee takes values ranging from 0 to 4, where higher values represent more committees. Board entrenchment takes values ranging from 0 to 4, where higher values are associated with fewer anti-takeover provisions. Committee independence takes values ranging from 0 to 3, where higher values are associated with more independent outside directors serving on committees. Board transparency takes values ranging from 0 to 3, where higher values are associated with a more transparent board. CEO power takes values ranging from 0 to 3, where higher values indicate less CEO power. Total government ownership measures the residual stake held by the national and local government in the company. National government ownership represents the level of ownership by the national government. All models include the control variables used in the previous table, including industry and year dummy variables. The presented models use one year lags of the explanatory variables. The values in parentheses are t-statistics based on standard errors that are clustered by firms. The sum of the bold coefficients is significant at the 5% level. * Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level.

between central government ownership and corporate governance is positive (0.252) and significant, while the impact of civil law on this relation is negative (0.252–0.346  (1) = 0.094) and significant.16 Next, we examine our six subcomponents of CGQ to further identify the ways in which government ownership affects corporate governance, and Table 6 shows the results of these measures of corporate governance quality. The Board Independence results (Models 1 and 2) show that the impact of common law is positive and significant for national ownership. For Board Committee (Models 3 and 4), we find that the impact of civil law on the government ownershipcorporate governance relation is negative and significant for total

16 In additional robustness tests, we include squared government ownership variables in all Table 5 regressions to check for nonlinearities. The results show no significant quadratic effects when considering the full sample of government-owned firms or the civil law subsample. We find some evidence of quadratic effects for national ownership in common law subsamples. The negative and significant coefficients on these squared terms suggest that common law governance benefits increase at a decreasing rate.

government ownership and national ownership. Neither of the common law coefficients is significant. The Board Entrenchment results (Models 5 and 6) show that the impact of common law is positive and significant for total government ownership, as well as for national ownership alone. Civil law government ownership is linked to a worse entrenchment score than common law government ownership, although the overall effect of state ownership is positive when considering both coefficients (e.g., 0.171– 0.153  (1) = 0.018, for total government ownership). The Committee Independence results (Models 7 and 8) show that the impact of civil law is negative and significant for total government ownership and national ownership. Neither of the common law coefficients is significant. The Board Transparency regressions (Models 9 and 10) indicate that common law effects are positive and significant for total government ownership and national ownership. Considering the sum of the coefficients of government ownership and its interaction with civil law, only total government ownership for civil law firms yields a significant, negative association with CEO Power (Models 11 and 12).

2928

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

Overall, the results in Tables 5 and 6 are consistent with the notion that the legal framework of common law countries has a positive impact on the relation between government ownership and corporate governance. In a legal environment more conducive to good corporate governance and effective monitoring, state owners could enforce regulations and ensure recommended practices are followed at the firm level. Our results also show that the legal framework of civil law countries typically has a negative impact on the relation between government ownership and corporate governance. Despite evidence (e.g., Dahya et al., 2008) that independent directors increase corporate value, particularly in legal systems offering weaker shareholder protection, state owners in civil law nations seem to discourage outsider participation on board committees. More generally, our results show that government owners extend the corporate governance quality traditionally linked to their respective legal systems. Similar to findings in Aggarwal et al. (2009), state owners operating in less-protected legal environments tend to disregard good governance practices, instead of being catalysts for regulation reform.

4.3. Golden shares and corporate governance Another method for governments to retain control of corporations is through the creation and use of golden shares (Bortolotti and Faccio, 2009). The ownership of a golden share within a firm grants the government special controlling powers in that firm, analogous to additional voting rights relative to normal shares of stock.17 We now investigate how these state control mechanisms affect corporate governance quality using a binary variable that takes a value of one if the government has retained a golden share in a firm, and zero otherwise. Our overall expectation is that golden shares will be detrimental to corporate governance – all else equal. We examine the impact of golden shares on the composite CGQ score in Table 7 and then present the effect of golden shares on six subcomponents of CGQ in Table 8. The results in Table 7 confirm that the presence of a golden share reduces the quality of corporate governance in line with its function as a control mechanism of the state, allowing it to override various corporate decisions. We keep the other government ownership variables in these regressions to distinguish the golden share effect from the previously tabulated government ownership effects. The regression results show consistently negative and significant coefficients for the golden share variable across all models (i.e., Models 1–4) in Table 7. We find similar negative and significant results for the subcomponents of board independence, board committee, and board transparency in Table 8. Since golden shares often grant the state the right to appoint board members, government-related directors could be selected to maintain control of the firm, resulting in the associated decrease in board independence. These results further reflect that golden shares often have nationally-centered objectives, such as keeping the general direction of a firm in domestic hands, against the suggestion of Oxelheim and Randøy (2003) that electing foreign independent board members can serve as a way of importing more demanding governance practices. The inverse relation between the number of board committees and a golden share could also result from the perception that the government is already sufficiently monitoring the firm. Additional board committees could (perhaps falsely) seem extraneous with the state nominally supervising matters of board nomination and governance via a golden share. Self-serving government behavior could be behind the significant negative relation between golden shares and board 17 These powers can pertain to veto rights regarding acquisitions, board member selection, and imposing ownership thresholds, among others.

transparency measures, as the state seeks control of auditing decisions and related-party transactions (Guedhami et al., 2009). In contrast, we find little to no impact of golden shares on the subcomponents of board entrenchment, committee independence, and CEO power in Table 8. Overall, our findings in this section suggest that the use of golden shares has an incrementally negative effect on corporate governance. While government ownership is generally associated with weaker corporate governance, the government’s ability to leverage up its voting power through golden shares is particularly detrimental. 4.4. Robustness checks Beyond the previously-discussed characteristics of our data that help alleviate reverse causality concerns, we further address potential issues related to endogeneity by employing a two-stage least squares instrumental variable approach. In Table 9, we instrument government ownership and its interaction with civil law using variables relating to privatization and government behavior. Most crucially, we employ a privatization dummy that is a binary variable equaling one if a firm has been government-owned, and zero otherwise. Since this variable is strongly tied to the firm’s level of state ownership, but not in and of itself reflective of current corporate governance, it serves as an appropriate instrument. We also include a political stability index measuring the likelihood that the government could be destabilized or overthrown by unconstitutional or violent means. Higher values correspond to greater political stability. As discussed in Boubakri et al. (2005), for instance, political stability can influence government ownership decisions, yet should not reflect ongoing corporate governance practices of a firm, making it a suitable exogenous instrument. Additionally, we use a corruption index, which refers to the perception of corruption in the government, and higher values of the index indicate less corruption. Self-serving behavior amongst government officials could encourage greater ownership retention in state-controlled firms. The political stability and corruption indexes are from the World Bank’s Worldwide Governance Indicators. Because we interact government ownership with legal origin in much of our analysis, our instrumental variable analysis will have two first-stage regressions, one for government ownership and one for the interaction of government ownership and the civil law dummy variable. The results in Table 9 confirm our earlier findings that the legal framework associated with common (civil) law countries has a generally positive (negative) impact on corporate governance. The sum of the coefficients for government ownership and its interaction with civil law is significant at the 10% level for Model 1 (total government ownership) and at the 5% level for Model 2 (only national government ownership). Although our analysis employs country fixed effects in previous models, as well as a composite corporate governance measure that accounts for differences in national governance regulation, we formulate additional tests to further control for country-level factors. Following Chhaochharia and Laeven (2009), we calculate an adjusted corporate governance score as the minimum CGQ for all firms in a given country and year subtracted from a specific firm’s CGQ in the corresponding year. This residual score captures the governance quality implemented by a firm above and beyond what is either legally mandated or traditionally practiced by all other firms from the same nation for a specific year. The regression models in Table 10 use the adjusted CGQ score as the dependent variable and present results similar to those in Table 5. Model 2, for example, again shows that common-law firms with state ownership exhibit better average governance practices, while government ownership in civil law countries is linked to a

2929

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934 Table 7 Multivariate analysis of golden share, incentives, ownership, and corporate governance. Model 1 Golden share Total government ownership

**

3.905 (2.01) 0.056 (1.44)

Total government ownership  Civil law dummy

Model 2 **

4.098 (2.10) 0.618** (2.43) 0.676*** (2.65)

Model 3 **

3.813 (1.99)

3.947** (2.05)

0.076* (1.80)

4.072*** (8.78) 2.441 (0.80) 0.005 (0.24) 0.038 (0.63) 3.585*** (3.54) 15.155 (1.10) 0.079* (1.74) 1.613 (0.80) 0.037 (0.41) 0.036 (1.18) 0.660 (1.07) 19.626*** (5.40) 0.034 (1.33) 23.996*** (9.62) 169.448*** (4.50)

4.097*** (8.79) 2.452 (0.81) 0.005 (0.26) 0.039 (0.64) 3.601*** (3.55) 15.098 (1.09) 0.080* (1.76) 1.676 (0.83) 0.040 (0.44) 0.035 (1.15) 0.686 (1.11) 19.621*** (5.40) 0.034 (1.32) 23.734*** (9.42) 169.856*** (4.51)

4.098*** (8.89) 2.367 (0.78) 0.005 (0.29) 0.039 (0.66) 3.639*** (3.59) 15.727 (1.14) 0.081* (1.78) 1.654 (0.82) 0.035 (0.39) 0.038 (1.25) 0.707 (1.14) 19.521*** (5.41) 0.038 (1.46) 23.818*** (9.56) 169.310*** (4.52)

0.441⁄⁄⁄ (2.76) 0.518⁄⁄⁄ (3.18) 4.116*** (8.89) 2.377 (0.78) 0.005 (0.28) 0.040 (0.66) 3.652*** (3.59) 15.696 (1.14) 0.082* (1.80) 1.700 (0.84) 0.038 (0.41) 0.038 (1.23) 0.721 (1.16) 19.516*** (5.41) 0.038 (1.46) 23.650*** (9.42) 169.583*** (4.52)

989 0.56

989 0.56

989 0.56

989 0.56

National government ownership National government ownership  Civil law dummy Log (assets) Leverage Sales growth ROA Quick ratio Cap. expend./Assets External finance Dividend dummy Years since initial sale of ownership Institutional ownership GDP growth Log (GDP per capita) Stock market cap. Civil law dummy Intercept Observations R-squared

Model 4

This table presents firm random effects regression results for analysis of government ownership, golden shares, and corporate governance. The dependent variable is the RiskMetrics corporate governance quotient, and the explanatory variables follow. Firm data are from Thomson Worldscope. Golden share takes a value of 1 if the government has retained a golden share, and zero otherwise. Total government ownership measures the residual stake held by the national and local government in the company. National government ownership represents the level of ownership by the central government. Log (assets) is the log of the market value of assets in millions of US$. Leverage is the ratio of total debt to total assets. Sales growth is the percentage sales growth computed as the change in current period sales divided by the sales in the previous period. ROA is the return on assets calculated as [net income + interest after taxes/total assets]. Quick ratio equals [(cash + cash equivalents + receivables)/current liabilities]. Cap. expend./Assets refers to the firm’s funds used to acquire or upgrade its physical assets divided by total assets. External finance is calculated for all companies as capital expenditures minus cash holdings divided by capital expenditures. Dividend dummy is a dummy variable that takes a value of one if a firm pays dividends in year t, and zero otherwise. Years since initial sale of ownership indicates the number of years since the government first began to sell its ownership stake in the firm. Institutional ownership represents the ownership by institutional investors. GDP growth, Log (GDP per capita), and Stock market cap. (% of GDP) are from the World Bank’s WDI Database. Civil law dummy takes a value of one if a country has a civil law system, and zero if a country has a common law system. In addition, we include industry and year dummy variables in all models. The presented models use one year lags of the explanatory variables. The values in parentheses are t-statistics based on standard errors that are clustered by firms. * Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level.

lower quality of corporate governance. Due to the construction of the dependent variable, the positive and significant coefficient on the civil law dummy indicates a greater disparity in corporate governance in these nations and more room for firms to improve, including those still owned and influenced by the government.18 The results in Table 10 confirm that our main findings are robust to alternative CGQ specifications.

18 The average minimum annual country CGQ score for civil (common) law nations is 3.7 (39.4), with a standard deviation of 18.5 (6.8).

5. Conclusions General Motors, AIG, Fannie Mae, and Citibank, among many others, appear to have survived the recent financial crisis after significant doses of government intervention and equity injections. It is too early to tell exactly how these interventions will affect the future governance of these erstwhile private companies. It is possible, however, to examine the impact of government ownership on corporate governance for EU-based companies in the recent past. The main purpose of this study is to analyze these experiences in order to better understand and forecast the likely consequences of current interventions.

2930

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

Table 8 Multivariate analysis of incentives, golden share, government ownership and disaggregated governance measures. Board independence

Golden share Total government ownership Total government ownership  Civil law dummy

Model 1

Model 2

Model 3

Model 4

0.564* (1.75) 0.264 (0.60) 0.267 (0.61)

0.618* (1.90)

0.708** (2.49) 0.043⁄ (1.79) 0.070⁄⁄⁄ (2.86)

0.736*** (2.62)

32.338** (2.20) 32.338** (2.20)

National government ownership National government ownership ⁄ Civil law dummy Observations

Golden share Total government ownership Total government ownership  Civil law dummy

806

989 Committee independence

989

Model 5

Model 6

Model 7

Model 8

0.232 (0.63) 0.160*** (5.37) 0.143*** (4.78)

0.240 (0.66)

0.384 (1.02) 0.016 (0.51) 0.050 (1.46)

0.392 (1.04)

0.162*** (5.27) 0.142*** (4.63)

National government ownership  Civil law dummy 989

989

Board transparency

Golden share Total government ownership Total government ownership  Civil law dummy

0.010 (0.30) 0.048 (1.35) 989

989

CEO power

Model 9

Model 10

Model 11

Model 12

0.764*** (3.57) 0.110*** (3.95) 0.106*** (3.84)

0.764*** (3.61)

0.016 (0.05) 0.424 (1.48) 0.436 (1.53)

0.047 (0.14)

0.126*** (5.76) 0.122*** (5.54)

National government ownership National government ownership ⁄ Civil law dummy Observations

0.041⁄ (1.81) 0.070⁄⁄⁄ (2.96)

806 Board entrenchment

National government ownership

Observations

Board committee

989

989

5.354 (0.71) 5.365 (0.72) 809

809

This table presents logit and ordered logit results of government ownership, golden shares, and corporate governance for civil vs. common law countries. The dependent variables are Board independence in the first two models, Board committee in models 3 and 4, Board entrenchment for models 5 and 6, Committee independence for models 7 and 8, Board transparency in models 9 and 10 and CEO power in the last two models. Board independence takes a value of 1 if the board is controlled by a majority of independent outsiders, and zero otherwise. Board committee takes values ranging from 0 to 4, where higher values represent more committees. Board entrenchment takes values ranging from 0 to 4, where higher values are associated with fewer anti-takeover provisions. Committee independence takes values ranging from 0 to 3, where higher values are associated with more independent outside directors serving on committees. Board transparency takes values ranging from 0 to 3, where higher values are associated with a more transparent board. CEO power takes values ranging from 0 to 3, where higher values indicate less CEO power. Golden share takes a value of 1 if the government has retained a golden share, and zero otherwise. All models include the control variables used in the previous table, including industry and year dummy variables. The presented models use one year lags of the explanatory variables. The values in parentheses are t-statistics based on standard errors that are clustered by firms. The sum of the bold coefficients is significant at the 5% level. * Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level.

Our empirical results show that government ownership is generally harmful to the corporate governance of the firm, suggesting that firm value maximization is not always the goal of state owners. This finding intensifies as the government’s power increases, and it is specifically linked to central government ownership. Our results also suggest that national government ownership reduces the number of board committees (particularly under civil law) while increasing the power of CEOs, signaling the state’s intention to consolidate power within a firm and facilitate the transfer of control. Perhaps more significantly, we show that while government intervention is detrimental to governance quality in civil law countries, it is beneficial to governance quality in common law countries; that is, state owners are an extension of their respective legal systems and traditional levels of protection for

shareholders. Lastly, our results indicate that governments’ access to preferential voting rights through golden shares is especially damaging to the quality of corporate governance. Government ownership, while ostensibly similar to institutional ownership, seems to yield little comparable governance benefits, particularly in its more prevalent form under civil law systems. The lower quality of corporate governance linked with state-owned firms could be associated with a general notion that these corporations require (or prefer) less monitoring due to the government’s supervising presence. However, privatization studies such as Boubakri et al. (2005) and D’Souza et al. (2007) contradict the idea of the state as an efficiently regulating owner by relating corporate governance and firm value improvements to government divestiture. Beyond ownership changes, future governance

2931

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934 Table 9 Instrumental variable analysis of government ownership and corporate governance. Model 1

First stage of Model 1 First 1

Total government ownership Total government ownership  Civil law dummy

First 2

Cap. expend./Assets External finance Dividend dummy Years since initial sale of ownership Institutional ownership GDP growth Log (GDP per capita) Stock market cap. Civil law dummy Intercept Observations R-squared

989 0.35

4.247*** (4.62) 5.193 (0.79) 0.016 (0.34) 0.025 (0.18) 4.267** (2.11) 9.019 (0.43) 0.141 (1.28) 6.677 (1.51) 0.081 (0.51) 0.103* (1.66) 0.855 (0.69) 24.801*** (4.63) 0.060 (1.32) 7.875 (0.95) 250.607*** (3.75)

4.510*** (5.11) 4.043 (0.67) 0.007 (0.17) 0.042 (0.32) 5.106** (2.52) 15.276 (0.75) 0.155 (1.46) 6.176 (1.50) 0.090 (0.60) 0.092 (1.55) 1.128 (0.91) 22.681*** (4.79) 0.096** (2.09) 11.383* (1.71) 234.247*** (3.88)

9.783*** (5.74) 15.721*** (7.28) 11.270*** (7.60) 0.574 (1.51) 5.793** (2.12) 0.019 (0.92) 0.029 (0.49) 2.268*** (2.65) 23.268*** (2.62) 0.116*** (2.61) 2.053 (1.20) 0.399*** (4.03) 0.150*** (6.72) 1.705*** (3.46) 7.740*** (2.75) 0.137*** (6.17) 6.828*** (3.65) 96.876*** (3.10)

989 0.17

989 0.37

989 0.37

989 0.19

989 0.35

Corruption index

Quick ratio

9.551*** (5.61) 15.771*** (7.31) 11.326*** (7.65) 0.591 (1.56) 5.803** (2.13) 0.018 (0.91) 0.029 (0.49) 2.301*** (2.70) 23.178*** (2.61) 0.118*** (2.64) 2.124 (1.25) 0.386*** (3.90) 0.149*** (6.67) 1.717*** (3.48) 7.812*** (2.78) 0.137*** (6.21) 7.093*** (3.79) 98.222*** (3.14)

11.614*** (6.31) 14.204*** (6.09) 10.839*** (6.77) 0.209 (0.51) 9.020*** (3.06) 0.006 (0.27) -0.011 (0.17) 0.916 (0.99) 10.964 (1.14) 0.108** (2.24) 2.011 (1.09) 0.419*** (3.92) 0.155*** (6.44) 1.111** (2.09) 13.181*** (4.33) 0.075*** (3.12) 7.517*** (3.72) 135.950*** (4.03)

Political stability index

ROA

First 2

11.850*** (6.44) 14.137*** (6.06) 10.777*** (6.73) 0.227 (0.55) 9.011*** (3.05) 0.006 (0.28) 0.011 (0.18) 0.889 (0.96) 11.045 (1.15) 0.106** (2.22) 1.930 (1.05) 0.431*** (4.04) 0.156*** (6.48) 1.097** (2.06) 13.107*** (4.31) 0.074*** (3.10) 7.180*** (3.55) 134.578*** (3.98)

Privatization dummy

Sales growth

First 1

40.128⁄⁄ (2.25) 40.594⁄⁄ (2.26)

National government ownership  Civil law dummy

Leverage

First stage of Model 2

42.943⁄⁄ (2.26) 43.365⁄⁄ (2.26)

National government ownership

Log (assets)

Model 2

This table presents two-stage least squares instrumental variables (IV) regression results. The dependent variable is the RiskMetrics corporate governance quotient, and the explanatory variables follow. Firm data are from Thomson Worldscope. Total government ownership measures the residual stake held by the national and local government in the company. National government ownership represents the level of ownership by the national government. All control variables are defined in Appendix A. The privatization and country variables listed below are used as instruments for government ownership and its interaction with the civil law dummy in the two IV regression models. The privatization dummy is a binary variable that equals one if a firm has been privatized, and zero otherwise. The Political stability index measures the likelihood that the government could be destabilized or overthrown by unconstitutional or violent means, and higher values correspond to greater political stability. The Corruption index measures the perception of corruption in the government, and higher values signify less corruption. The political stability and corruption indexes are from the World Bank’s Worldwide Governance Indicators. The dependent variables in the first stage models are the government ownership variable for First 1 and the government ownership variable interacted with the civil law dummy for First 2. Year and industry dummies are included in all models. The presented models use one year lags of the explanatory variables. The values in parentheses are t-statistics. * Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level.

improvements could emerge from new regulation and better legal protection for shareholders (e.g., La Porta et al., 1999a; Price et al., 2011), with remaining government owners potentially enforcing these reforms from within. Indeed, our results for common law nations provide some hope for current scenarios created by the US bailouts, for instance, by suggesting that the state can be a positive influence on a firm’s corporate governance practices when operating within the appropriate legal framework. Overall, these and related results contribute to our understanding of the

governance role played by state actors. This previously underdeveloped topic is likely to grow in importance as a consequence of the 2008 global financial crisis and the associated increase in government interventions. Acknowledgments We would like to thank the editor (Ike Mathur), an anonymous referee, Shin-Yi Chou, Rafel Crespí, James Dearden, Abe

2932

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

Table 10 Multivariate analysis of government ownership and corporate governance above yearly national minimum standards.

Total government ownership

Model 1

Model 2

*

⁄⁄⁄

0.060 (1.77)

Total government ownership  Civil law dummy

Model 3

Model 4

0.059 (1.56)

0.365 (2.72) 0.426⁄⁄⁄ (3.06)

2.942*** (6.70) 2.759 (1.01) 0.003 (0.16) 0.052 (1.06) 2.105** (2.49) 4.126 (0.43) 0.036 (1.61) 2.179 (1.18) 0.050 (0.65) 0.036 (1.35) 1.483*** (2.82) 2.580 (0.90) 0.027 (1.18) 6.563*** (3.52) 19.532 (0.66)

2.953*** (6.71) 2.770 (1.02) 0.003 (0.15) 0.052 (1.07) 2.112** (2.50) 4.054 (0.42) 0.036 (1.63) 2.226 (1.20) 0.051 (0.66) 0.036 (1.33) 1.473*** (2.79) 2.558 (0.89) 0.027 (1.18) 6.724*** (3.56) 19.044 (0.65)

2.978*** (6.81) 2.635 (0.96) 0.002 (0.11) 0.054 (1.09) 2.140** (2.52) 4.490 (0.47) 0.036 (1.61) 2.193 (1.19) 0.047 (0.60) 0.036 (1.32) 1.455*** (2.74) 2.815 (0.98) 0.030 (1.28) 6.536*** (3.50) 20.845 (0.71)

0.300*** (2.66) 0.360*** (2.97) 2.988*** (6.81) 2.643 (0.97) 0.002 (0.11) 0.054 (1.10) 2.148** (2.53) 4.439 (0.46) 0.036 (1.63) 2.229 (1.21) 0.048 (0.62) 0.035 (1.30) 1.446*** (2.71) 2.801 (0.97) 0.030 (1.28) 6.652*** (3.53) 20.469 (0.69)

989 0.30

989 0.30

989 0.30

989 0.30

National government ownership National government ownership  Civil law dummy Log (assets) Leverage Sales growth ROA Quick ratio Cap. expend./Assets External finance Dividend dummy Years since initial sale of ownership Institutional ownership GDP growth Log (GDP per capita) Stock market cap. Civil law dummy Intercept Observations R-squared

This table presents firm random effects regression results. The dependent variable is the minimum RiskMetrics corporate governance quotient (CGQ) for all firms in a given country and year subtracted from the firm-level CGQ score. The explanatory variables follow. Firm data are from Thomson Worldscope. Total government ownership measures the stake held by the national and local government in the company. National government ownership represents the level of ownership by the central government. Log (assets) is the log of the market value of assets in millions of US$. Leverage is the ratio of total debt to total assets. Sales growth is the percentage sales growth computed as the change in current period sales divided by the sales in the previous period. ROA is the return on assets calculated as [net income + interest after taxes/total assets]. Quick ratio equals [(cash + cash equivalents + receivables)/current liabilities]. Cap. expend./Assets refers to the firm’s funds used to acquire or upgrade its physical assets divided by total assets. External finance is calculated for all companies as capital expenditures minus cash holdings divided by capital expenditures. Dividend dummy is a dummy variable that takes a value of one if a firm pays dividends in year t, and zero otherwise. Years since initial sale of ownership indicates the number of years since the government first began to sell its ownership stake in the firm. Institutional ownership represents the ownership by institutional investors. GDP growth, Log (GDP per capita), and Stock market cap. (% of GDP) are from the World Bank’s WDI Database. Civil law dummy takes a value of one if a country has a civil law system, and zero if a country has a common law system. This variable is interacted with government ownership to compare effects across legal systems. Year and industry dummies are included in all models. The presented models use one year lags of the explanatory variables. The values in parentheses are t-statistics based on standard errors that are clustered by firms. The sum of the bold coefficients is significant at the 5% level. * Significance at the 10% level. ** Significance at the 5% level. *** Significance at the 1% level.

de Jong, Dean Diavatopoulos, Nadia Doytch, Shelly Howton, Art King, Chris Koehn, Mi Luo, Chad Meyerhoefer, David Nawrocki, Mike Pagano, David Shaffer, Stephen Snyder, Tina Yang, Lyubomir Zagorchev, participants of the International Corporate Finance and Governance conference at the University of Twente, the 1st Finance and Corporate Governance conference at La Trobe University, the Asian Finance Association meeting in

Macao, China, the Financial Management Association meeting in Denver, Colorado, and seminar participants at Lehigh University, University of New Haven, and Villanova University for their comments. Ginka Borisova would like to dedicate this work to her father, Ognyan. Mario Stamboliev and Luke Wanninger provided excellent research assistance. All remaining errors are our own.

2933

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

Appendix A. Variable definitions

Variable definitions (continued) Variable

Variable

Definition

Board committee

Board committee takes values ranging from 0 to 4, where higher values represent the existence of more of the following committees: audit, compensation, governance, and nomination Board entrenchment takes values ranging from 0 to 4, where one point is given if a company has (i) no poison pills, (ii) an annuallyelected board, (iii) a majority vote requirement for mergers, and (iv) and a majority vote requirement for charter/bylaw amendments Board independence takes a value of 1 if the board is controlled by a majority of independent outsiders, and zero otherwise Board transparency takes values ranging from 0 to 3, where higher values are based on (i) the auditor’s ratification at fiscal year end, (ii) auditor expenses being strictly related to auditing fees, and (iii) the CEO not being involved in any related party transactions Cap. expend./Assets refers to the firm’s funds used to acquire or upgrade its physical assets (capital expenditures) divided by total assets CEO power takes values ranging from 0 to 3, where higher values indicate (i) the separation of the CEO and the chairman, (ii) board independence, and (iii) the presence of a former CEO on the current board Civil law dummy takes a value of one if a country has a civil law system, and zero if a country has a common law system Committee independence takes values ranging from 0 to 3, with a point given for each of the nomination, compensation, and audit committees that is entirely composed of independent members Corporate Governance Quotient (CGQ) score is calculated by RiskMetrics. The CGQ raw scores are weighted scores of eight categories, and the composite CGQ rating comprises these eight core categories. Higher CGQ scores indicate better corporate governance of firms Corruption index refers to the perception of corruption in the

Board entrenchment

Board independence

Board transparency

Cap. expend./Assets

CEO power

Civil law dummy

Committee independence

Corporate Governance Quotient (CGQ) score

Corruption index

Dividend dummy

External finance

Financial structure dummy

GDP growth

Golden share

Institutional ownership

Leverage

Log (assets)

Log (GDP per capita)

Total government ownership

National government ownership

Political stability index

Privatization dummy

Quick ratio

ROA

Definition government, and higher values of the index indicate less corruption Dividend dummy is a binary variable that takes a value of one if a firm pays dividends in a given year, and zero otherwise External finance is calculated for all companies as capital expenditures minus cash holdings divided by capital expenditures Financial structure dummy takes a value of one if a country has a bankbased system, and zero if a country has a market-based system GDP growth is the annual percentage growth rate of GDP calculated at market prices using a constant local currency Golden share takes a value of 1 if the government has retained a golden share, and zero otherwise Institutional ownership represents the ownership by institutional investors as a percentage of shares outstanding Leverage is the ratio of the book value of debt to the book value of assets Log (assets) is the log of the market value of total assets in millions of US dollars, representing firm size Log (GDP per capita) measures the log of gross domestic product per capita based on constant 2000 US dollars Total government ownership measures the residual stake held by the national and local governments in the company as a percentage of shares outstanding National government ownership represents the level of ownership by the national government as a percentage of shares outstanding Political stability index measures the likelihood that the government could be destabilized or overthrown by unconstitutional or violent means, and higher values correspond to greater political stability Privatization dummy is a dummy variable that equals one if a firm has been government-owned, and zero otherwise Quick ratio is equal to the sum of cash, cash equivalents, and receivables divided by current liabilities Return on assets (ROA) is calculated as the sum of net income and (continued on next page)

2934

G. Borisova et al. / Journal of Banking & Finance 36 (2012) 2917–2934

Variable definitions (continued) Variable

Sales growth

Stock market cap.

Stock value traded (% of GDP)

Years since initial sale of ownership

Definition interest expense on debt minus capitalized interest, both of which are after taxes, divided by total assets Sales growth is the percentage sales growth computed as the change in current period sales divided by the sales in the previous period Stock market capitalization as a percentage of GDP measures the relative size of the stock market Stocks total value traded as a percentage of GDP captures the country’s relative stock market turnover The number of years since the government first began to sell its ownership stake in a firm

References Aggarwal, R., Erel, I., Ferreira, M., Matos, P., 2011. Does governance travel around the world? Evidence from institutional investors. Journal of Financial Economics 100, 154–181. Aggarwal, R., Erel, I., Stulz, R., Williamson, R., 2009. Differences in governance practices between US and foreign firms: Measurement, causes, and consequences. Review of Financial Studies 22, 3131–3169. Ang, J., Ding, D., 2006. Government ownership and the performance of governmentlinked companies: The case of Singapore. Journal of Multinational Financial Management 16, 64–88. Angelidis, T., Tessaromatis, N., 2010. The efficiency of Greek public pension fund portfolios. Journal of Banking and Finance 34, 2158–2167. Bebchuk, L., Cohen, A., Ferrell, A., 2009. What matters in corporate governance? Review of Financial Studies 22, 783–827. Berger, A., Clarke, G., Cull, R., Klapper, L., Udell, G., 2005. Corporate governance and bank performance: a joint analysis of the static, selection, and dynamic effects of domestic, foreign, and state ownership. Journal of Banking and Finance 29, 2179–2221. Borisova, G., Megginson, W., 2011. Does government ownership affect the cost of debt? Evidence from privatization. Review of Financial Studies 24, 2693–2737. Bortolotti, B., Faccio, M., 2009. Government control of privatized firms. Review of Financial Studies 22, 2907–2939. Boubakri, N., Cosset, J., Guedhami, O., 2005. Postprivatization corporate governance: the role of ownership structure and investor protection. Journal of Financial Economics 76, 369–399. Boubakri, N., Cosset, J., Saffar, W., 2008. Political connections of newly privatized firms. Journal of Corporate Finance 14, 654–673. Brunello, G., Graziano, C., Parigi, B., 2003. CEO turnover in insider-dominated boards: the Italian case. Journal of Banking and Finance 27, 1027–1051. Bruno, V., Claessens, S., 2010. Corporate governance and regulation: can there be too much of a good thing? Journal of Financial Intermediation 19, 461–482. Chen, G., Firth, M., Xu, L., 2009. Does the type of ownership control matter? Evidence from China’s listed companies. Journal of Banking and Finance 33, 171–181. Chhaochharia, V., Laeven, L., 2009. Corporate governance norms and practices. Journal of Financial Intermediation 18, 405–431. Clarke, D., 2003. Corporate governance in China: an overview. China Economic Review 14, 494–507. Dahya, J., Dimitrov, O., McConnell, J., 2008. Dominant shareholders, corporate boards, and corporate value: a cross-country analysis. Journal of Financial Economics 87, 73–100. Davidson III, W., Pilger, T., Szakmary, A., 1998. Golden parachutes, board and committee composition, and shareholder wealth. Financial Review 33, 17–32.

Del Guercio, D., Hawkins, J., 1999. The motivation and impact of pension fund activism. Journal of Financial Economics 52, 293–340. Denis, D., McConnell, J., 2003. International corporate governance. Journal of Financial and Quantitative Analysis 38, 1–36. Doidge, C., Karolyi, A., Stulz, R., 2007. Why do countries matter so much for corporate governance? Journal of Financial Economics 86, 1–39. D’Souza, J., Megginson, W., Nash, R., 2007. The effects of changes in corporate governance and restructurings on operating performance: evidence from privatizations. Global Finance Journal 18, 157–184. Durnev, A., Kim, E., 2005. To steal or not to steal: firm attributes, legal environment, and valuation. Journal of Finance 15, 1461–1493. Elyasiani, E., Jia, J., 2010. Distribution of institutional ownership and corporate firm performance. Journal of Banking and Finance 34, 606–620. Fan, J., Wong, T., Zhang, T., 2007. Politically-connected CEOs, corporate governance and post-IPO performance of China’s partially privatized firms. Journal of Financial Economics 84, 330–357. Ferri, G., 2009. Are new tigers supplanting old mammoths in China’s banking system? Evidence from a sample of city commercial banks. Journal of Banking and Finance 33, 131–140. Fligstein, N., Choo, J., 2005. Law and corporate governance. Annual Review of Law and Social Science 1, 61–84. Gillan, S., Starks, L., 2000. Corporate governance proposals and shareholder activism: the role of institutional investors. Journal of Financial Economics 57, 275–305. Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. Quarterly Journal of Economics 118, 107–155. Guedhami, O., Pittman, J., Saffar, W., 2009. Auditor choice in privatized firms: empirical evidence on the role of state and foreign owners. Journal of Accounting and Economics 48, 151–171. Institutional Shareholder Services, 2003. ISS Corporate Governance. Best Practices user Guide and Glossary. Rockville, MD. Jenkinson, T., Ljungqvist, A., 2001. The role of hostile stakes in German corporate governance. Journal of Corporate Finance 7, 397–446. Klein, A., 2002. Audit committee, board of director characteristics, and earnings management. Journal of Accounting and Economics 33, 375–400. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 1999a. Corporate ownership around the world. Journal of Finance 54, 471–517. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2002. Government ownership of banks. Journal of Finance 57, 265–301. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2008. The economic consequences of legal origins. Journal of Economic Literature 46, 285–332. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1998. Law and finance. Journal of Political Economy 106, 1113–1155. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1999b. The quality of government. Journal of Law, Economics, and Organization 15, 222–279. Martynova, M., Renneboog, L., 2008. A century of corporate takeovers: what have we learned and where do we stand? Journal of Banking and Finance 32, 2148– 2177. McDonough, D., 1998. Corporate governance and government owned corporations in Queensland. Bond Law Review 10, 272–312. Megginson, W., 2005. The Financial Economics of Privatization. Oxford University Press, New York. Morey, M., Gottesman, A., Baker, E., Godridge, B., 2009. Does better corporate governance result in higher valuations in emerging markets? Another examination using a new data set. Journal of Banking and Finance 33, 254–262. Oxelheim, L., Randøy, T., 2003. The impact of foreign board membership on firm value. Journal of Banking and Finance 27, 2369–2392. Parrino, R., Sias, R., Starks, L., 2003. Voting with their feet: institutional ownership changes around forced CEO turnover. Journal of Financial Economics 68, 3–46. Pi, L., Timme, S., 1993. Corporate control and bank efficiency. Journal of Banking and Finance 17, 515–530. Price, R., Román, F., Rountree, B., 2011. The impact of governance reform on performance and transparency. Journal of Financial Economics 99, 76–96. Renneboog, L., 2000. Ownership, managerial control and the governance of companies listed on the Brussels stock exchange. Journal of Banking and Finance 24, 1959–1995. RiskMetrics Group, 2007. ISS Governance Services: International Corporate Governance Policy 2008 Updates. New York. Ruiz-Mallorquí, M., Santana-Martín, D., 2011. Dominant institutional owners and firm value. Journal of Banking and Finance 35, 118–129. Tian, L., Estrin, S., 2007. Debt financing, soft budget constraints, and government ownership. Economics of Transition 15, 461–481. Uzun, H., Szewczyk, S., Varma, R., 2004. Board composition and corporate fraud. Financial Analysts Journal 60, 33–43. Weisbach, M., 1988. Outside directors and CEO turnover. Journal of Financial Economics 20, 431–460.