Board structure and dividend smoothing: A case of Pakistani listed firms

The purpose of this study is to empirically examine the effects of ownership and board structure on dividend smoothing for listed non-financial firms for the period 2005-2015 on the Pakistan Stock Exchange (PSX) in the light of agency and information asymmetry theory. Two measures, speed of adjustment and relative volatility are used to estimate dividend smoothing. Tobit regression is used for analyzing the effects of board structure on dividend smoothing measured via speed of adjustment while OLS is used for relative volatility. Additionally, the study applies the technique of principal component analysis (PCA) for board index. The results reveal that firms with large boards have a high level of dividend smoothing in Pakistan. They less frequently increase, cut, initiate and omit dividends. The substitution role of board independence is depicted by the negative association of board independence with dividend smoothing. Furthermore, independent audit committees and dividend smoothing are also found to be negatively associated.


Introduction
Many scholars have found empirical evidence of an increase in prices in response to dividend initiation and negative reaction to dividend omission and cuts (Lintner 1956). This situation is so serious that managers go for external financing and forgo economically attractive projects to avoid dividend cuts (Brav et al 2005;Gordon 1963). The market puts a premium on companies with relatively Z. Ali et al Rashid 2014).
The relationship between dividends and corporate governance is quite hazy in literature. On one hand, companies with weak corporate governance follow a high dividend policy that substitutes for corporate governance. In such companies, management uses dividends to establish trust for future equity raising (Hu and Kumar 2004). On the other hand, according to the outcome hypothesis companies with strong corporate governance where investor rights are strongly protected are empowered to demand higher dividends (Sawicki 2009). Jiraporn et al (2011) also, found that companies with strong corporate governance offer higher dividends.
The literature on corporate governance depicts its linkage with company value, financial performance, takeovers, and reliability of financial information, CEO compensation, capital structure, dividends and dividend smoothing. Gompers et al (2003) constructed a corporate governance index using 24 corporate governance provisions by taking 1500 US companies and found that company value and corporate governance are associated with each other. Bebchuk et al (2008) studied 1800 companies for the period of 1990-2003. They incorporated six corporate governance provisions and found that high entrenchment index is associated with the negative performance of companies. Brown and Caylor (2006) considered 51 corporate governance firm-specific provisions by using 1,868 U.S. firms and found a positive association between corporate governance and firm value. But contrary to these studies Cheung et al (2008) developed a corporate governance index for Chinese companies and found no association between corporate governance and firm value.
More recently Javakhadze et al (2014) by investigating non-financial companies from 24 countries over the period of 1999 to 2011, found that dividend smoothing and corporate governance follow the substitution hypothesis. They found that companies with strong corporate governance exhibit less dividend smoothing. Similarly Leary and Michaely (2011), concluded that companies with the weak corporate governance exhibit more dividend smoothing. The authors found that firms with weak growth potentials and corporate governance, having greater proportions holding by institutions go for higher dividend smoothing.
Keeping in view the above gap in literature, the current study addresses board characteristics i.e. board size, board independence, audit committee independence and CEO/chairperson duality, and investigates its impact on dividend smoothing. Fama and Jensen (1983) argued that the board of directors of a firm contributes significantly to reduce the agency problem as the central focus of corporate governance is to monitor companies' executive management. However, boards can effectively monitor management and can take corrective actions only if they have enough representation of independent directors (Jensen 1986;.

Board size and dividend smoothing
Most of the previous literature on corporate governance has addressed the 68 Business Review: (2019) 14(2):65-91 Board structure and dividend smoothing... board of directors' size, its independence, its composition and board activities.
While concentrating on a sample of 7000 U.S firms over a period of 1990 to 2004Linck et al (2008 found the size of the board, the proportion of independent directors and board leadership to be significant determinants of the board of directors' structure. They further argued that during 1990s large firms have reduced their board size while small firms have maintained its boards flat. Among other studies on corporate board of directors (Boone et al 2007) related the board size to company's life cycle and CEO's links with outside directors. They found that companies increase their board size and appoint more outside directors as they move towards maturity. Further, they also found that board composition is the outcome of the CEO's negotiation with outside directors. Similarly, Vafeas (2000) found the inverse relationship of board meetings frequency and firm's value by addressing 307 U.S. firms for the period 1990-1994. Yermack (1996) concentrated on 452 large U.S firms for the period 1984-1991 and found that small boards are more effective than larger ones. Hermalin and Weisbach (1991) related board structure and board incentives with firm performance for 142 NYSE firms during 1971-1983 but failed to find a significant relationship between the proportion of outside directors and firm performance. On the other hand, Krivogorsky (2006) related board structure and firm performance for 87 European firms and found that the higher the proportion of independent directors, the higher is the firm's performance.
In accordance with the agency cost theory, the size of the firm's board of directors can influence dividend smoothing in two ways. Large boards have more diversity and more external linkages, therefore, they have the ability to monitor executive management effectively (Fiegener et al 2000). In contrast Ghosh and Sirmans (2006) argued that managers exploit this weak monitoring of large boards and boards populated by older directors and pay a lower level of dividends. Therefore, firms with larger boards face a lower level of agency problem according to the agency theory; and exhibit lower levels of dividend smoothing.
On the other side of the coin, Jensen (1993) and Yermack (1996) argued that it is not necessary, that larger boards may be more efficient monitors. They argued that large boards have the issue of coordination between board members. They stated that small boards with enough proportion of independent directors are more efficient and can monitor executive management effectively. Therefore, from this perspective firms with large boards suffer more from the agency problem and smooth their dividends more.
Most of the firms in Pakistan are controlled by families, therefore, they don't appoint directors who limit their business control. Therefore, in large it is expected that boards of directors in Pakistan have weak monitoring role and substitute this by large and smooth dividends which are in accordance with the substitution hypothesis (La Porta et al 2000). Based on the above, the following hypothesis is proposed: H1: There is a positive relationship between the size of the board and dividend smoothing. Shareholders elect the board of directors for the management and control of the firm. Later on the board of directors recruits officers and delegates authority for routine decision making control to the management. However, they retain the authority of hiring and firing and compensating management including the CEO. Naturally, the management tries their best to attain majority of the positions on the board and elect the CEO as the chairman to safeguard their personal interests. Boards having duality and majorly made up of executive directors are usually not good monitors and are not be able to challenge management decisions (Jensen 1993). Non-executive or outside director means, director of the company who is not assigned managerial or administrative responsibilities (SECP Corporate Governance rules, 2013). Corporate governance rules (2013) issued by SECP, had asked all companies operating in Pakistan to have 40% independent directors on their boards by 2015 and a majority of independent directors by 2017. However, only electing independent/ non-executive directors on the board doesn't protect minority shareholders because not all outside directors challenge management rather they safeguard self-interest. Therefore, SECP has addressed characteristics of independent directors in its 2013 rules as some outside directors' monitor but others don't.
Independent boards improve the decision-making process and exhibit good monitoring and supervision. According to the agency theory, independent directors are good monitors and can safeguard shareholder's interest better (Jensen and Meckling 1976;Armstrong et al 2014;Sharma 2011). Independent directors can give independent judgement, free of management influence (Fama and Jensen 1983;Linck et al 2008). These boards can protect minority shareholders from expropriation by the major shareholders and hence reduce both type I and type II agency problem (Raheja 2005;Sharma 2011;Armstrong et al 2014). Stakeholder theory depicts that independent directors outperform insiders in guarding the interests of the stakeholders because they are more experienced and have more ethical and legal obligations (Rodriguez-Dominguez et al 2009). They are more conservative than executive directors and reduce agency cost mounting from information asymmetry between shareholders and management. However this is not always true, sometimes executive directors reveal partial information to boards and to outside directors in order to prevent full information access to shareholders (Yermack 1996;Armstrong et al 2014).
Previous literature depicts that the presence of more executive directors on the board affects disclosure (Cheng and Courtenay 2006) and performance (Cho and Kim 2007) negatively. Ruiz-Barbadillo et al (2009) found that the presence of a higher proportion of executive directors enhances management dominance over the board and ultimately these boards are less likely to support dividend payouts (Deshmukh et al 2013;Banerjee et al 2013). Raheja (2005) found that the optimal number of outside/non-executive directors are dependent on the private benefits of insiders, and found both to be positively linked. Companies having more executive directors on their boards are less likely to use dividend as a monitoring device for agency cost reduction because they can monitor man-Board structure and dividend smoothing... agement by themselves effectively (Maury and Pajuste 2002;Krishna Prasanna 2014). Independent directors on the board ask for large dividends in order to counter the use of free cash flows for self-interest by management (Ghosh and Sirmans 2006). In contrast Cheung et al (2005) for Hong Kong and Su et al (2014) for China found a weak association of dividend payout with boards independence while How et al (2008) found that board independence lowers dividend payouts in Hong Kong. This study precedes with the notion that independent directors are effective monitors and can challenge management decisions. Therefore, we follow the substitution hypothesis that the requirement of smooth dividends will be reduced as long as a company elects more independent nonexecutive directors on its board. Based on the above the following hypothesis is proposed: H2: There is negative relationship between board independence and dividend smoothing.

Audit committee and dividend smoothing
The audit committee is a standing committee of non-executive directors of the board (AICPA). It plays a vital role in corporate governance. According to the Corporate Governance Rules (2013) of the Public-Sector companies issued by the Securities and Exchange Commission of Pakistan, audit committee should be chaired by a non-executive director and neither chairman of the board nor CEO should be its member. The audit committee has the responsibility to monitor the internal audit department, review audit plans, annual and audit reports of the company. They have to look after the appointment of the external auditors, review accounting control and protect company's assets (Setia-Atmaja 2009). Another important responsibility of the audit committee is to make recommendations to the board of directors about the related party transactions during the period (SECP Corporate Governance Rules 2013).
The audit committee of the company is in the best position to identify any overruling by the management and react immediately (Lutz 2015). The Commission further states that company's internal audit is strengthened by its audit committee. The primary purpose of the audit committee is to safeguard the interests of the shareholders and this purpose can be achieved only if the committee is controlled by the external members. McMullen (1996) have linked reliable information with the presence of the audit committee. But Beasley (1996) has presented quite opposite results i.e. the presence of audit committee is not likely to do anything with the reliability of the financial information. Merely the presence of the audit committee cannot stop management from overruling but the audit committee's independence is crucial to curtail the management and safeguard the interests of the shareholders. Audit committee's independence is positively linked with private benefits to the management (Setia-Atmaja 2009). That is the audit committee's role is increasing as chances of expropriation increase. Therefore, the presence of an independent audit committee is likely to reduce information asymmetry and the agency problem. Hence, the preference for a high level of smoothed dividends https://ir.iba.edu.pk/businessreview/vol14/iss2/5 DOI: 10.54784/1990-6587.1018 Z. Ali et al will reduce. Keeping in view the above literature, the following hypothesis is proposed for the impact of the audit committee's independence on dividend smoothing. H3: There is negative relationship between audit committee's independence and dividend smoothing.

Sample
Sample of the study comprises of all listed non-financial firms on the Pakistan Stock Exchange (PSX) for the period 2005-15, however, data on certain variables like Earnings per share (EPS), Dividend per share (DPS) and the total dividend was included for the period ranging 1999-2015. Data is filtered according to table 1. EPS, DPS and data for control variables has been acquired from the balance sheet analysis of the State Bank of Pakistan (SBP). Stock prices information for return volatility has been collected from Khistocks.com and business recorder website. Data on firm's measure of risk (beta) has been collected from opendoors.pk while board structure related data were hand collected from the annual reports of the companies for period of 2005-2015.

Variable measures
Following Leary and Michaely (2011), this study uses two measures of dividend smoothing i.e. the speed of adjustment (SOA) and relative volatility (Rel Vol).

Speed of adjustment (SOA)
The study follows (Lintner 1956) for finding dividend smoothing measures of the listed Pakistani non-financial firms given by: For measuring the level of dividend smoothing across firms, this study follows (Fama and Babiak (1968)) which is a modified form of (Lintner 1956 (Leary and Michaely 2011;Jeong 2011Jeong , 2013 have used this modified (Lintner 1956) model given by: Speed of adjustment (SOA) given by c i , theoretically ranges between 0 and 1. As SOA approaches 1, the level of dividend smoothing lowers and as SOA approaches zero, it reflects a higher level of dividend smoothing. The level of dividend smoothing (SOA) is estimated as β 2 from this equation (Lintner 1956). We follow a two step procedure to measure SOA, introduced by Leary and Michaely (2011) to improve the precision of the estimates. This two step procedure reduces small sample biases (Leary and Michaely 2011). Fist payout ratio is calculated by dividing common dividends by the net income. Then target payout ratio (TPR) is estimated as firm's median payout ratio for the entire sample period. By using the above target payout ratio we calculate the deviation from the target dividend for each firm-year observation through this equation.
Here speed of adjustment (SOA) represented by β 1 is estimated over the rolling window of six years for each firmyear observation. As an alternative measure of dividend smoothing to SOA, we use another proxy called relative volatility.

Relative Volatility (Rel Vol)
Leary and Michaely (2011) introduced another measure of dividend smoothingthe relative volatility of dividends to the corresponding earnings. This study follows this as the second measure of dividend smoothing. First, scaled earnings are calculated by multiplying earnings per share with the firm's target payout ratio in order to remove the scaled effect of the dividend level. Without these two, firms with same earnings volatility and same change in the dividend, the one with the larger dividend exhibits higher dividend volatility. Following Leary and Michaely (2011) quadratic time trend is fitted to both dividend per share and scaled earnings.
Controlling for the linear trend, reports the same level of dividend smoothing for two firms with one targeting specific DPS and the other targeting specific change in DPS while the square time trend inclusion reports the same level of dividend smoothing for a firm targeting same percentage change in dividends. Equation (4) and (5) . Both of these variables (speed of adjustment and relative volatility) are treated as alternative proxies of dividend smoothing. The higher the value of these variables, the lower is the level of dividend smoothing.

Independent variables
Board structure-related variables used are board's size, board independence, audit committee size, audit committees' independence and audit quality of firms in Pakistan. A binary variable represents whether the firm is audited by the No of non-executive directors in audit committee in proportion to the total no of directors in audit committee.
Board structure and dividend smoothing...

Board structure and dividend smoothing
The model depicting the impact of board structure on dividend smoothing is given as Then a board index is developed through principal component analysis and its impact is estimated on dividend smoothing.
3.3.2 Asymmetric dividend smoothing Lintner (1956) assumes that firms adjust their dividends at the same rate (SOA) irrespective of whether the firm's last dividend is below the target payout or above. We followed (Leary and Michaely 2011) to relax this assumption. That is; managers are reluctant to announce cuts in dividends than announcing increases (Brav et al 2005). We classified the deviations into two classes in the following pattern.
Then the above regression is regressed to investigate the asymmetry of dividend smoothing that whether β 1 and β 2 are different. The null hypothesis for the above model β 1 = 0 , β 2 = 0 is to be investigated and both the coefficients need to be positive in order to show convergence towards the target payout ratio. Consequently it is expected that if the cost of announcing an increase in dividend is lower than dividend cuts, then managers are more reluctant to cuts rather than towards announcing increases. Hence they are expected to adjust quickly if they are below the target dividend than when they are adjusting downwards. Ultimately 0 < β 2 < β 1 is expected. Similarly, this study investigates whether the board structure variables impact differently adjustment from below and adjustment from above through the

Frequency of dividend changing events in Pakistan
Following Chemmanur et al (2010) we have classified various dividend changing events in Pakistan that are dividend increases, dividend cuts, continuations, dividend initiations, dividend omissions and other events. Chi-square test of independence is used for detecting association between dividend changing events and corporate governance related variables. https://ir.iba.edu.pk/businessreview/vol14/iss2/5 DOI: 10. 54784/1990-6587.1018 Published by iRepository, December 2020 Board structure and dividend smoothing... is associated with greater chances of payouts. Therefore it is consistent with the outcome hypothesis of corporate governance i.e. good corporate governance results in high and frequent dividends.
The table also depicts that few dividend paying firms have CEO duality than non-paying firms. That is assigning the role of CEO and chairperson to one person reduces the chances of dividend payments, again consistent with the outcome hypothesis of corporate governance. It is also reported in the results that 56.6% of the dividend-paying firms are audited by the big four auditors in Pakistan which is again consistent with the outcome hypothesis. Table 5 compares firm characteristics of high dividend smoothing firms (reflected by low SOA and low relative volatility) with low dividend smoothing firms (having high SOA and high relative volatility). We have estimated the mean of firms' characteristics for each quintile, presented in the five columns.

Smoothing and non-smoothing firms
Column 6 in both panels shows the difference in means of the first and last quintile. The last column shows significance with the help of two-way t-test that means the first and last quintile is significantly different. Both panels of table 5 depict that higher dividend smoothing is associated with large boards. Pakistani corporate structure is dominated by families and in such scenario large boards are perceived to be weak monitors. Therefore an alternative monitoring mechanism is adopted through dividend smoothing which is consistent with hypothesis (1) of this study. Similarly, panel B depicts that, firms with low board independence smooth more as an alternative monitoring mechanism which supports the hypothesis (2). Table 5 depicts that higher dividend smoothing is associated with large audit committees. However, the difference between higher smoothers and lower smoothers with respect to audit committee's independence is insignificant, which may be because of the fact that corporate governance rules (2013)  https://ir.iba.edu.pk/businessreview/vol14/iss2/5 DOI: 10. 54784/1990-6587.1018 Published by iRepository, December 2020 the audit committee. The results for the auditing by any of the four largest auditing firms show that firms audited by them are associated with lower level of dividend smoothing as they are associated with lower level of agency conflict and low-level information asymmetry.

Regression results: Board structure and dividend smoothing
Both columns of the Our result is consistent with the agency view of Jensen (1993), Yermack (1996) and Ghosh and Sirmans (2006), all of these authors associated large boards with weak monitoring because of the coordination issue. Boards have two main roles to perform, resource dependence role and monitoring role. Here this result is consistent with the monitoring role of the board because firms with large boards smooth more which may be because of the weak coordination and weak monitoring of the large boards. Table 6 depicts that firms with more independent boards have a tendency of less smoothing. It means that firms with lower proportion of non-executive directors; faces a high level of agency conflict and greater information asymmetry exists between outsiders and insiders, alternatively such firms use smooth dividends as an alternative monitoring mechanism. This result affirms hypothesis (H2) of the study and is consistent with the agency and information asymmetry theory of dividend smoothing. It is concluded that firm's managers with independent boards opt for less dividend smoothing. This result is consistent with (Javakhadze et al 2014). Firms with more independent audit committees engage in less dividend smoothing. As hypothesized Pakistan's corporate structure is characterized by concentrated family ownership, therefore, chances of expropriation are high in absence of independent audit committees (Raheja 2005). So this study finds evidence of substitution role of dividend smoothing for audit committees independence. These results are consistent with both agency and information asymmetry theory.

Board index
This study uses principal component analysis (PCA) for developing board index (Board index). Following (Tarchouna et al 2017), we have also chosen the first factor which represents the largest variation of board size, board independence, audit size, audit independence, CEO-duality and audit by big four auditing firms of Pakistan during the sample period. PCA combines these six variables into a linear combination. Table 7 depicts weights of all the mentioned six variables in the board index for Pakistani firms during the sample period based on principal components analysis. Following Tarchouna et al (2017) two concerns were addressed before reporting the index. Kaiser-Meyer-Olkin (KMO) test is used to ensure that correlation between variables is higher than the correlation between errors. KMO statistic was 0.626 well above 0.50. Bartlett's test for sphericity (P-value < 0.001) is used to ensure that the variables are factorable.

Board index and dividend smoothing
In  Standard errors in parentheses * * * p < 0.01, * * p < 0.05, * p < 0.1 https://ir.iba.edu.pk/businessreview/vol14/iss2/5 DOI: 10.54784/1990DOI: 10.54784/ -6587.1018 Board structure and dividend smoothing... auditing firms or not. A firm scoring high on board index represents strong corporate governance practices through mentioned variables than the one with a low score. Negative association of board index and dividend smoothing is witnessed, which is in line with the substitution hypothesis. This result is parallel to Leary and Michaely (2011) and Javakhadze et al (2014). As mentioned in the literature review, firms with large boards and audit committees dominated by the executive directors, with entrenched CEOs and not audited by reputed audited firms, have low monitoring ability and greater information asymmetry. Therefore, these results can be explained on the basis of both agency and information asymmetry theory. The results of table 8 are consistent with the substitution hypothesis of La Porta et al (2000). While investigating the asymmetry in dividend smoothing when the firm is above its target level dividend and when it is below the sample was divided into sub samples. Firms' change in dividend from last year is regressed with two variables Dev P and Dev N , one, when firm's target dividend is higher than last year dividend (adjusting from below) and second when firm's target dividend is lower than last year dividend (adjusting from above). The speed of adjustment from below should be higher than speed of adjust- Standard errors in parentheses * * * p < 0.01, * * p < 0.05, * p < 0.1 ment from above (La Porta et al 2000). In line with La Porta et al (2000) firms in Pakistan adjust their dividend at 39.6% (against 34.4% from above) for any increase in the earnings when they are adjusting through increases. The higher speed of adjustment from below, than speed of adjustment from above depicts that Pakistani firms are more reluctant to cut dividends than announcing dividend increases. Then we regress the following regression by adding interaction terms of the board variables with positive deviations and negative deviations.
Here we observe the impact of these variables on the speed of adjustment from above and below. Table 10 depicts that firms with large boards quickly adjust their dividends towards the target level when the managers are adjusting their dividends from below. It means that board size has a positive influence on speed of adjustment from below. And interestingly firms with large boards try to reduce the speed of adjustment from above. It means that board size doesn't alter the asymmetric behavior of dividend smoothing when firms are adjusting dividends from below or from above rather intensify it. Our result is parallel to (Leary and Michaely 2011). Audit size effect is more pronounced when firms are adjusting from below. This result is consistent with (Leary and Michaely 2011). Consistent with H2, firms with more proportionate non-executive directors adjust quickly to target dividend when they are adjusting from below. It is is consistent with agency and information theory of dividend smoothing. Board independence plays the https://ir.iba.edu.pk/businessreview/vol14/iss2/5 DOI: 10.54784/1990-6587.1018 Standard errors in parentheses * * * p < 0.01, * * p < 0.05, * p < 0.1 substitution role for dividend smoothing only when the firm is adjusting its dividend from below. Firms adjusting from above, exhibit different role of board independence. Firms having more independent audit committees, adjust their dividends quickly from below. This result affirms (H3) of the study and is line with the agency and information asymmetry theory of dividend smoothing, it supports the substitution role of audit committee independence with dividend smoothing Business Review: (2019) 14(2):65-91 83 https://ir.iba.edu.pk/businessreview/vol14/iss2/5 DOI: 10.54784/1990DOI: 10.54784/ -6587.1018 Z. Ali et al (Javakhadze et al 2014). While looking at the interaction terms of the audit by big four firms with both positive and negative deviations, positive but statistically insignificant coefficients are observed.

Board structure and dividend changing events
In this section, the authors have reported how frequently Pakistani firms adjust their dividends. Table 11 depicts the frequency of dividend changing events by classifying them on the basis of board structure characteristics. The changes in dividends are increases, decreases, omissions and initiations of dividends. Table  11 reports frequency of dividend cuts (a decrease of more than 10% from last year dividend), dividend increases (increase of more than 10% from last year dividend) and continuations (any dividend change lower than 10% increase and 10% decrease). We have divided the sample into two sub samples based on the board size, such that observations having board size greater than the mean value of board size were termed as firms with large boards while those below mean value were termed as firms with small boards. The first section of panel A of table 11 depicts that firms with large boards have 28.84% dividend increases as compared to 38.71% by the firms with small boards. Furthermore, they have 28.13% cuts against 28.49% by firms with small boards. As it is shown in the table, firms with large boards have continuations of 43.03% against 32.8% by firms with small boards. The null hypothesis of independence of dividend increases and decreases from board size is rejected at 1% significance level. It is therefore concluded that firms with large boards smooth their dividends more than those with small boards. Similarly firms with large audit committees are frequently changing its dividends via increases and decreases. Table 11 confirms the associations of dividend changes with board independence at the significance level of 1% and affirms H2. Firms having more independent audit committees exhibit dividend continuations of 40.2% against 39.1% continuations by firms with low audit independence. Firms audited by big four auditing firms have fewer dividend continuations then those not audited by big four. It is consistent with agency and information asymmetry theory of dividend smoothing. At last, we divided the sample based on the board index developed via principal components analysis. Firms having a score above the sample mean on board index were firms practicing good corporate governance via board related variables and others were termed as firms with low board governance. It can be noticed that firms having a high score of board index were having more dividend increases and cuts than other groups. It suggests that firms with low corporate governance use smooth dividends in substitution. This result is parallel to Leary andMichaely (2011) andJavakhadze et al (2014). In this section, we have divided dividend events into three categories, dividend initiations, dividend omissions and others, where initiations are the event when a firm is moving from zero dividends to a positive dividend and omission is the event when a firm is moving from positive dividend to zero dividend.
It was found that initiations events are more than omissions. Moreover, firms with large boards opt for few dividend initiations and omissions then firms with 84 Business Review: (2019) 14(2):65-91 https://ir.iba.edu.pk/businessreview/vol14/iss2/5 DOI: 10. 54784/1990-6587.1018 Board structure and dividend smoothing...  Table 12 depicts that firms with independent boards experience more initiations and omissions. It depicts that firms with more independent audit committees have a higher number of initiations and omissions than firms with less independent audit committees. We find a significant association of auditing by big four firms with dividend initiations and omissions. While looking at the association of board index with dividend initiations and omission, it is observed that firms having a high score in board index opt for fewer initiations and omissions. These results affirm the substitution role of dividend smoothing.

Conclusion
This study empirically examined the impact of board structure on dividend smoothing in Pakistan. It found answers to the research questions that whether dividend smoothing is the outcome or substitutes for corporate governance. First, the role of board size in dividend smoothing was explored and it was found that firms with large boards are more likely to smooth their dividends in Pakistan. Firms with large boards are associated with more dividend increases and cuts as compared to those with small boards. Similarly, firms with large boards have more initiations and omissions. Firms with large boards quickly adjust their dividends towards the target level when the managers are adjusting their dividends from below. It suggested that large boards have coordination issues and are not efficient monitors. These results are not consistent with the resource dependency theory which is of the notion that a firm's board is a resource and firms with large boards have more linkages with outside and can assist management efficiently in making decisions (Carpenter and Westphal 2001;Hassan 2014) but the results are consistent with monitoring role of boards under the agency view of Jensen (1993), Yermack (1996) and Ghosh and Sirmans (2006). However, when the firm is adjusting from above, then large boards reduce SOA, but it might be due to the asymmetric behavior of dividend smoothing. The study found negative association of board independence with dividend smoothing as reflected by multiple regressions via both proxies (SOA and relative volatility). However, the substitution role of dividend smoothing for board independence was only observed when a firm is adjusting their dividends from below. Firms adjusting from above, observe the different role of board independence but it could be because of the asymmetric behavior of dividend smoothing. The non-parametric analysis revealed that firms with high board independence have exhibited higher frequency of dividend increases, cuts, initiations and omissions than firms with low board independence. It means that firms having a lower proportion of non-executive directors; face a high level of agency conflict and greater information asymmetry exists between outsiders and insiders. Alternatively such firms use smooth dividends as an alternative monitoring mechanism. Therefore, it is concluded that weak monitoring because of the low board independence is substituted with smooth dividends. These results are consistent with the monitoring role of boards. So the substitution effect of dividend smoothing for corporate governance is prevailing in a weak shareholder environment like Pakistan (Leary and Michaely 2011;Javakhadze et al 2014).

Z. Ali et al
Independent audit committees and dividend smoothing are found to be negatively associated. On the contrary firms having few non-executive directors on the boards opt for a higher level of dividend smoothing. The non-parametric analysis revealed that firms with independent committees have exhibited higher frequency of dividend increases, cuts, initiations and omissions than firms with low board independence. As hypothesized Pakistan's corporate structure is characterized by concentrated family ownership, therefore, chances of expropriation are high in the absence of independent audit committees (Raheja 2005; Setia-Atmaja et al 2009). So we find evidence of the substitution role of dividend smoothing for audit committee's independence. The asymmetric analysis reveals that firms having more independent audit committees adjust their dividends quickly from below. While when the firm is adjusting from above then audit committee's independence slows the adjustment of dividends towards the target dividend. So again the asymmetric behavior of dividend smoothing dominates weak monitoring because of the lack of audit committees.
It suggests that firms with large audit size smooth less dividends which is consistent with the agency and information asymmetry theory. Firms with large audit committees face lower agency conflict and are exposed to a lower level of information asymmetry, hence need less dividend smoothing as alternative mechanism according to the substitution hypothesis (La Porta et al 2000). It shows that audit size effect is more pronounced when firms are adjusting from below. This result is consistent with Leary and Michaely (2011). As market reaction to cuts is more severe than for the increase.
This study depicts via multiple regression analysis that firms audited by big four auditing firms of Pakistan exhibit lower levels of dividend smoothing. Similarly, non-parametric analysis depicted that firms audited by big four firms' exhibit a high number of dividend increases, cuts, initiations and omissions than those not audited by the big four. These results are in line with both the agency and information asymmetry theory. Firms audited by the big four face lower information asymmetry as there are higher chances of detecting frauds by the big four auditing firms.
Board index was developed via PCA by incorporating board size, board independence, audit committee size, audit committee's independence, CEO duality and audit quality. The negative association of board index and dividend smoothing is witnessed, which is in line with the substitution hypothesis. Similarly, firms having a high score of board index were having more dividend increases and cut events than other groups. That is, in a weak corporate governance environment, dividend smoothing is an alternative monitoring mechanism. Firms having high agency conflict and higher information asymmetry between insiders and outsiders because of the weak monitoring power in Pakistan, use more smoothed dividends as alternative monitoring (signaling mechanism) in the light of agency (information asymmetry) theory. Finally, the study found that firms with large boards, low board independence, and low independent audit committees smooth more in Pakistan.
In future this research area could be further investigated by examining the role of the board structure on dividend smoothing in the context of better monitoring and in the context of resource dependency roles proxies of boards. Also https://ir.iba.edu.pk/businessreview/vol14/iss2/5 DOI: 10. 54784/1990-6587.1018