Political connections, board of directors and firm performance: The case of Tanzania
Abstract
In this article I use firm-level dataset to offer the first systematic assessment of the effects of political connections on firm performance in Tanzania. My sample consists of all listed firms in Dar-es-saalam Stock Exchange (DSE). I define politically connected firms as follows. A board member who satisfies any of the following two criteria is considered to be a politically connected director: (1) a current or former government official; and (2) a current or former member of a political party in the country. My results indicate that political connections in the board of directors hurt firm performance measured by changes in return on assets and return on sales. This finding is robust in various alternative regression estimated. This suggests that firms with politically-connected directors exhibit significantly worse firm performance than non-connected counterparts.Overall, the results indicate that the appointment of politically-connected directors does not enhance firm efficacy butrather fulfil political goals of politicians.
Key words: Political connections, board of directors, firm performance
1Introduction
The board of directors is a critical corporate governance mechanism, however our knowledge of what makes boards effective is quite limited. Previous research has much focused on how board structure or ownership structure affects firm performance. To date, relatively little research has investigated the impact of politically connected boards on firm performance. This research is motivated by two main issues. First, the literature that links politically connected boards and firm performance is very scarce. Second, to my knowledge there is no systematic study that has analysed the relationship between the two issues in the Tanzanian context.The primary objective of this research is to investigate the nature and extent of politically connected board members and their effect on the firm performance.
The main findings and contributions of this paper are as follows. First, I find that political connections and firm performance are negatively correlated.Both measures of political connections exhibits significant negative relationship with firm performance measures namely changes in return on assets and return on sales. This finding is robust in various alternative regression estimated. This suggests that firms with politically-connected directors exhibit significantly worse financial performance than non-connected counterparts.As far as I know, this paper is the first systematic study that analyse the effect of politically-connected directors on corporate performance. This study extends the limited literature that examines the link between politically-connected directors and firm performance.
Secondly, I find that to some extend blockholders ownership affects firm performance adversely. This is consistent with “Type ІІ agency problem” which argues that controlling shareholders expropriate minority shareholders through tunneling. That means blockholders collude with firm managers to expropriate company assets at their own benefit. On the other hand, this is contrary to the notion that blockholders are monitors of managers and helps to improve firm performance.This finding extends the previous literature that investigates the link between corporate governance attributes and firm performance.
The rest of the paper is organised as follows. Section 2 discusses the literature review on politically-connected directors and firm performance. Section 3presents definition of variables and econometric specification used in the regression analysis. Section 4 discusses descriptive statistics of the variables used in the paper. In Section 5, we present and discuss main findings of the paper. Section 6 discusses robustness results. Finally, in Section 7 we present conclusion and future research opportunities.
2Relevant literature
The literature that links political connections and corporate governance issues is limited. The perception that political control over firms’ decision making is detrimental to firm performance is logical and, to some extent, the corporate governance literature has demonstrated evidence on this link. Most theoretical arguments rely on the assumption that politicians use firms to pursue political and social objectives such as reducing income and regional inequality, correcting market failures and providing excessive employment (Boycko et al., 1996, Shleifer and Vishny, 1994, Shleifer and Vishny, 1998). All these factors are rationally detrimental to the firm’s financial performance. In fact, it is mostly assumed in the literature that, in the absence of political control, shareholders and managers have an incentive to maximize profits.
On the other hand, Brada (1996) viewed politicians as individuals who have incentive to prevent controlling shareholders and managers from engaging in behaviour that reduces the amount of corporate’s resources. In this context, the net effect of political control on firm performance depends on the balance between political costs and incentive problems of managers and shareholders.
Chang and Wong (2004) examined the impact of political control on Chinese listed corporations. These authors documented two major findings. First, they observed that party political control restrains the largest shareholders from expropriation but that the existing level of party control is unsatisfactory to control the largest shareholders. Second, their findings revealed that decision-making power of local party committees relative to managers is negatively related to firm performance. Another study by Leuz and Oberholzer-Gee (2006) examined firms with political connections in relation to global financing and corporate transparency. Their main findings revealed that politically connected firms are less likely to have publicly traded foreign securities. In this paper, they find that politically related firms influence long-run performance and alter firms’ financing strategies.
The study by Faccio et al. (2006) have analysed the probability of government bailouts of 450 politically connected firms from 35 countries during 1997–2002. Their major finding showed that politically connected firms are significantly more likely to be bailed out than similar non-connected firms. Besides, their findings revealed that politically connected firms are disproportionately more likely to be bailed-out when the International Monetary Fund or the World Bank provides financial assistance to the firm’s home government. In addition, among bailed-out firms, those that are politically connected exhibit significantly worse financial performance than their non-connected counterparts at the time of and following the bailout. The main implication of these findings is that politically connected firms are associated with poor financial performance.
Fan et al. (2007) investigated politically connected CEOs with post-IPO performance in newly partially privatised Chinese firms. In this paper, CEO’s political connection is defined as serving as a current or former government bureaucrat. Their finding revealed two interesting findings. Firstly, they find that firms with politically connected CEOs underperform those without politically connected CEOs by almost 18% based on three-year post-IPO stock returns and have poorer three-year post-IPO earnings growth, sales growth, and change in returns on sales. Secondly, firms led by politically connected CEOs do not consider professional backgrounds in directors’ recruitment. These firms are more likely to appoint incompetent officials to the board of directors rather than directors with relevant qualifications.
The study by Boubakri et al. (2008) explored the extent of political connections in newly privatised firms in 27 developing, and 14 developed countries over the period 1980 to 2002. First, they observed that 87 firms have a politician or a former politician on their board of directors. These politically connected firms are characterised by high leverage and operate in regulated industries. Second, the authors revealed that the probability of seeing political connections in these firms is positively linked to government residual ownership, and negatively linked to foreign ownership. Finally, they found that politically connected firms have poor accounting performance compared to non-connected firms.
Another paper by Frye and Iwasaki (2011) examined whether the government sends directors to weak or strong firms. In this study the performance of the firms is based on the frequency of their dividend payments. The comparison of the frequency of dividend payments by firms with and without government directors reveals that the frequency of dividend payments by firms with government directors is higher than that by firms without them. Furthermore, the number of dividend payments is positively related with the total number of state representatives on a board. This finding is inconsistent with the managerial discipline ideal type which suggests that state directors are sent to poorly performing firms, but is consistent with the rent-extraction and collusion ideal types.
Recent research by Menozzi et al. (2012) revealed that politically connected directors exert a positive and significant effect on employment, while they impact negatively on firm performance. This negative effect on corporate performance is consistent to many previous studies such asFaccio et al. (2006),Fan et al. (2007) and Boubakri et al. (2008). Overall, the extant literature on this topic has provided insufficient systematic evidence that politically connected boards worsens corporate performance. For this reason, further organised empirical analysis is important to enhance our understanding on the topic and corroborate findings from the previous studies.
3Methodology
3.1Data
The data used in this paper is collected at firm-level for all companies listed in the Dar essalaam Stock Exchange (DSE). Despite being listed at the DSE notall companies deposit annual reports in the African Financial Database or even their websites. For this reason, I have used two main techniques to hand collect these data. First, I have collected data from annual reports for each company from the time being listed to the financial year ended in 2012. Second, I have constructed questionnaire that asks information required for those companies that do not deposit their annual reports in the African Financial Database and websites.
3.2Dependant variable
Corporate performance is the dependent variable used in this paper. However, the concept of performance is a contentious issue in finance largely due to its multidimensional meanings. Generally, performance measures can be categorised as being either accounting-based or stock market-based metrics. The most commonly used accounting-based performance proxies are return on assets (ROA) and return on equity (ROE) or return on investment (ROI). There are various previous studies that have employed these measures, e.g. (Gorton and Rosen, 1995, Mehran, 1995, Ang et al., 2002). On the other hand, the frequently used stock market-based performance measures are market value of equity to book value of equity (MBVR) and Tobin’s Q, e.g. (McConnell and Servaes, 1990, Morck et al., 1988, Zhou, 2001).
However, it should be noted that the choice of performance measure entirely depends on the objective of the analysis. The focus of this research is to investigate the impact of politically-connected board members on corporate performance. In this paper, I employed changes in return on assets (ROA) andchanges in return on sales (ROS). Consistent to Megginson et al. (1994), I considered changes in return on sales as an appropriate measure because it is less sensitive to inflation and accounting conventions. In addition, manyprevious studies used these measures of firm performance, e.g. (Fan et al., 2007, Boubakri et al., 2008). I then define return on assets (ROA) as the ratio of the company’s earnings before interest and taxation divided by total assets, whereas return on sales (ROS) isdefined as the ratio of the company’s earnings before interest and taxation divided by sales.
3.3Independent variables
3.3.1Political connections
This is the key independent variable of this paper. The criterion for a political connected board member is as follows. A board member who satisfies any of the following two criteria is defined as a politically connected director: (1) a current or former government official; and (2) a current or former member of a political party in the country. This definition closely follows the recent study byChen et al. (2011). This information is collected from the directors’ biography in the annual report of each firm. I then measure political connection in two ways. First, the variable is computed as the number of politicians who are board members at the end of the year. Second, the variable is computed as the percentage of politically-connected members on the board of directors at the end of the year.
3.3.2Control variables
I controlled for several variables. I have included various corporate governance variables since they have great impact on firm performance. Directors’ remuneration is defined as the logarithm of total emoluments paid to all directors at the end of the year. The extant literature suggests that firm performance is highly influenced by directors’ remuneration(Ozkan, 2011, Brick et al., 2006).Blockholders ownership is defined as the percentage of shares owned by larger organisations or individuals with substantial shareholding amounting to or exceeding three percent of the firms’ share capital. The ownership is disclosed if it amounts to three percent or more of the shares outstanding. I aggregate all such three percent (or more) holding for a specific firm. Goyal and Park (2002) andDenis et al. (1997) used similar technique in defining ownership structure variables.
The proportion of non-executive directors is another variable that is included in the regression specifications. I use this approach as non-executive directors are regarded as corporate monitors and most corporate governance codes recommends boards should include non-executives to enhance shareholders wealth e.g.(Greenbury, 1995, Cadbury, 1992). Additionally, the inclusion of this variable is common in practice, e.g.(Conyon and Peck, 1998, Doucouliagos et al., 2007).Following the standard literature on firm performance, I have included firm size measured by logarithms of total assets.
3.4Econometric specification
Initially, I performed Hausman specification test to identify more efficient models between fixed and random effects regressions. I then estimated fixed effects regression following the results of the test performed.
4Descriptive statistics
The mean of the return on assets is 0.19, with a standarddeviation of 0.39. The range of return on assets is from -0.05 to 3.19.The mean of the return on sales is 0.22, with a standard deviation of 0.35. The highest return on sales is 2.88 while the lowest is -0.06. The data analysed reveals that, on average about 25% of board members are politicians. A firm with larger number of politicians accounts to about 57% of the board members. The data also showed that two firms had no politicians on their boards at certain time during the sample period.
The mean for blockholders is 69.08% with standard deviation of 21.23. The highest blockholder is 100% while the lowest is 34.80%. The average proportional of non-executive directors is 85.13%, and its standard deviation is also 9.36. The range of proportional of non-executive directors is from 50% to 100%. The mean of logarithm of directors’ remuneration is 4.22 and also has a standard deviation of 1.38. The value ranges from 1.10 to 6.71.For firm size, measured by logarithm of total assets, the mean reported is 11.24 and the range is from a lowest 8.03 to the highest 14.81. The standard deviation for firm size is 1.86.
Table І: Descriptive statistics
This table reports descriptive statistics of the variables used in this paper. These variables are defined as follows. Return on assets (ROA) is defined as the ratio of the company’s earnings before interest and taxation divided by total assets. ROS is defined as the ratio of the company’s earnings before interest and taxation divided by sales. No of politician is the number of politicians who are board members at the end of the year. Detailed explanation of identifying politician is included in the main text of this paper. Politician (%) is the proportion of politician available in the board at the end of the year. Blockholders ownership is defined as the percentage of shares owned by larger organisations or individuals with substantial shareholding amounting to or exceeding three percent of the firms’ share capital. I aggregate all such three percent (or more) holding for a specific firm. Non-executive directors (%) is the proportion of non-executive directors available in the board at the end of the year. Log (directors’ remuneration) is the logarithm of the total emoluments paid to all directors at the end of the year.Firm size is defined as common logarithm of company's total assets at the end of the year.
Variables / Observations / Mean / Std / Minimum / MaximumROA / 82 / 0.19 / 0.39 / -0.05 / 3.19
ROS / 66 / 0.22 / 0.35 / -0.06 / 2.88
No of politicians / 83 / 2.13 / 1.24 / 0.00 / 4.00
Politician (%) / 80 / 24.97 / 12.60 / 0.00 / 57.14
Blockholder (%) / 68 / 69.08 / 21.23 / 34.80 / 100.00
Non-executive directors (%) / 80 / 85.13 / 9.36 / 50.00 / 100.00
Log (Directors’ remuneration) / 75 / 4.22 / 1.38 / 1.10 / 6.71
Firm size / 82 / 11.24 / 1.86 / 8.03 / 14.81
Table ІІ reports the Pearson correlation matrix for the variables. Among the independent variables, no correlation exceeds 0.42, which is far below the typical threshold of 0.7 that is used to identify the presence of multicollinearity(Lind et al., 2005). To further ensure that multicollinearity is not a problem, I calculate variance inflation factors (VIF) for each independent variable. The VIFs, which are reported in the last two rows of Table ІІ, never exceed 9.34 which is lower than the typical threshold of 10(Belsley et al., 1980).
Table ІІ: Correlation matrix
This table reports correlation matrix of all variables used in the regression analysis. These variables are defined as follows.Return on assets (ROA) is defined as the ratio of the company’s earnings before interest and taxation divided by total assets. ROS is defined as the ratio of the company’s earnings before interest and taxation divided by sales. No of politician is the number of politicians who are board members at the end of the year. Detailed explanation of identifying politician is included in the main text of this paper. Politician (%) is the proportion of politician available in the board at the end of the year. Blockholders ownership is defined as the percentage of shares owned by larger organisations or individuals with substantial shareholding amounting to or exceeding three percent of the firms’ share capital. I aggregate all such three percent (or more) holding for a specific firm. Non-executive directors (%) is the proportion of non-executive directors available in the board at the end of the year. Log (directors’ remuneration) is the logarithm of the total emoluments paid to all directors at the end of the year Firm size is defined as common logarithm of company's total assets at the end of the year.The last two rows shows the Variance Inflation Factor (VIF) as a key measure of multicollinearity problem. In theestimation, none of the explanatory variables in the regression equations revealed VIF coefficient above 9.34. This figure is below the conventional rule of thumb of 10. This suggests multicollinearity is not a problem in the whole analysis. Bolded figures indicate significant correlations between variables.