CHAPTER FOUR

RESULTS AND DISCUSSIONS

4.1 Introduction

This chapter presents analysis and findings of the study as set out in the research methodology. The analysis was based on financial data collected by use of financial reports of listed commercial banks in Kenya .The objective of this study was on the investigation of corporate governance practices and firm value of listed Commercial Banks in Kenya.

4.2 Descriptive statistics

Table 4.1 presented the results relevant to descriptive statistics for all of the

variables. (Firm value, Ownership Concentration, Board size, Board Independence, audit committee ) employed in this study on commercial banks listed in the NSE from 2011 to 2016, which were explained as follows:

The results on firm value indicate that the average Tobin’s q is 0.24 with values ranging from 0.04 (minimum) to 0.62 (maximum). If Tobin’s q is greater than 1.0, then the market value is greater than the value of the firm’s recorded assets. This suggests that the market value reflects some unmeasured or unrecorded assets of the firm. High Tobin’s q values encourage firms to invest more in capital because they are worth more than the price they paid for them.

On the other hand, if Tobin’s q is less than 1, the market value is less than the recorded value of the assets of the company. This suggests that the market may be undervaluing the firm. When Q is less than parity, the market seems to be saying that the deployed real assets will not earn a sufficient rate of return and that, therefore, the owners of such assets must accept a discount to the replacement value if they desire to sell their assets in the market. Standard variation is a measure used to quantify the amount of variation of a set of data values.

The standard deviation stands at 0.13 which is relatively high compared to the mean indicating that the data points were spread out over a wider range of values. The variance is higher between the firm (0.10) compared to within the same firm over a number of years (0.08)

On average the board has 10 to 11 directors. Similarly, the values are ranging from 7 (minimum) to 13 (maximum) members. Standard deviation is 1.34 which is relatively low when compared to the mean. A low standard deviation indicated that the data points were closely clustered around the mean. It is also evident that the variation is higher between the firms (0.99) than variation within the same firm over the years (0.94)

The results on board independence indicate that the average value is 0.78 with values ranging from 0.40 (minimum) to 0.92 (maximum). This variable has a standard deviation of 0.13. This means that data was highly spread. The discrepancy is higher between the firm (0.11) compared to within the same firm over a number of years (0.07). Higher variance means more dispersion from the mean.

The average value of Ownership Concentration stands at 0.37 with values ranging from 0.12 (minimum) to 0.74 (maximum). Standard deviation in this case is 0.22 which is rather close to the mean. This establishes that the data points were spread out over a range of values. A keen look at the spread shows variation within the firm over the years (0.23) is higher that the variations between the same firm (0.02).

The results in Table 4.1 indicate that the average mean of audit committee stands at 3.77 with values ranging from a minimum of 3 members to a maximum of 5 members. The standard deviation 1s 0.67 and this 1s low demonstrating that the data points were close to the expected value. The variance is lower between the firm (0.37) compared to within the same firm over a number of years (0.57)

Table 4.1 Descriptive statistics of the study variables

Variable Mean Std. Dev. Min Max Observations

Firm Value overall 0.24 0.13 0.04 0.62 N=66

between 0.10 0.07 0.37 N=11

within 0.08 0.05 0.60 T=6

Board Size overall 10.11 1.34 7.00 13.00 N=66

between 0.99 8.50 11.67 N=11

within 0.94 7.44 12.94 T=6

Board Ind. overall 0.78 0.13 0.40 0.92 N=66

between 0.11 0.48 0.90 N=11

within 0.07 0.56 0.92 T=6

Ownership Con. overall 0.37 0.22 0.12 0.74 N=66

between 0.23 0.16 0.74 N=11

within 0.02 0.31 0.46 T=6

Audit Com. overall 3.77 0.67 3.00 5.00 N=66

between 0.37 3.17 4.50 N=11

within 0.57 2.61 5.11 T=6

4.3 Correlation analysis

The study used Karl Pearson correlation which measures the strength between variables and relationships. The coefficient value can range between -1.00 and 1.00. If the coefficient value is in the negative range, then that means the relationship between the variables is negatively correlated, or as one value increases, the other decreases. If the value is in the positive range, then that means the relationship between the variables is positively correlated, or both values increase or decrease together.

The results in Table 4.2 indicate that there is a positive correlation between the firm value and board size (0.6585). This means that if the board size is reduced, the firm value will decrease as well and vice versa. The correlation measurements fall between -1 and +1. If the correlation is closer to +1 the relationship is strong between the variables examined. Guest (2009) articulates that that larger boards are better placed to effectively perform both the agency and resource dependence roles than smaller ones. Kaur (2016) provides a rationale for the adoption of larger boards to yield greater firm value.

There exists a positive correlation between the firm value and board independence at 0.53. Other scholars have found similar results. Raghunandan and Rama (2007) hypothesize non-executive directors are more essential over others in reflecting effective corporate governance. This implies that an increase in board independence is associated with an increase in firm value.

The ownership concentration has shown a negative correlation with the firm value at -0.7285. The negative correlation is strong (closer to -1). This shows that the firm value decreases with an increase in ownership concentration. The results are supported by existing studies that shows a negative relationship between managerial control and Tobin’s Q measurement.

The audit committee has a positive relationship with the firm value at 0.6227. The results show that the increase in audit committee number increases firm value and vice-versa. This can be interpreted to mean that the increase in audit committee increases chances of transparency and accountability in the firms’ thus increasing firm value. Sunday, (2008) contended that large audit committee size gives more skilled individuals serving on the committee leading to enhanced firm reporting .

Table 4.2 Correlation Matrix

Firm Value Board Size Board Ind. Ownership Conc. Audit Comm.

Firm Value 1

Board Size 0.6585 1

Board Ind. 0.5382 0.5088 1

Ownership Conc. -0.7285 0.1086 -0.441 1

Audit Committee 0.6227 -0.3407 -0.0254 0.1414 1

4.4 Exploratory Analysis

4.4.1 Trend plots

In order to gain an insight into the data, trend plots were drawn. The study explored the trend over a six year period; 2011 to 2016. Due to the scaling differenced two graphs were drawn. Figure 4.1 presents trend plot on board size and audit committee (number of audit committee members). The results show a more stable trend, however, firm number nine (HF Group) had an erratic board membership.

Figure 4.2 presents the trend plots for Firm Value, Board Independence and Ownership concentration .The results indicate that there was consistency for the ownership concentration among different commercial banks under consideration except firm number 6 which followed a downward trend and firm 9 which was irregular. It can also be observed that there was an upward trend for all the banks with regards to the board independence except firm number 5 (NIC Bank) and firm number 10 (I;M Bank) which had a downward trend. Firm value varied from one bank to another. As depicted in figure 4.2 firm number 1, 5, 8, 11 had a downward trend depicting a capital increment from the major shareholders as well as the expertise from the rise in number of the non-executive directors. For firm 6, and 10 it was erratic. Firm number 2 and 3 followed an upward trend while others such as firm number 4 (NBK LTD), 7, 9 were relatively stable.

Figure 4.1 Trend plot on board size and audit committee

Figure 4.2: Trend plot on Firm Value, Board Independence and Ownership concentration

4.4.2 Overlay graph

The pictorial presentation in Figure 4.3 shows the overlay of all the listed commercial banks over the period under consideration. From the findings, it can be inferred that all commercial banks had different intercept terms for each entity and again these intercepts were constant over time. This preliminary result affirms that the variation across entities is assumed to be random and uncorrelated with the predictor or independent variable incorporated in the model demonstrating that the appropriate model may be random effects model. The variables assumed to be the same both cross-sectionally and temporally for this type of panel model. The section that follows presents more thorough tests to identify the fitting model.

Figure 4.3 Overlay graph

4.5 Panel Data Diagnostic Tests

To evaluate the correct model for estimation, some diagnostic tests were carried out. The results are presented.

4.5.1 Testing for random effects

The study used Breusch-Pagan Lagrange multiplier (LM) to test whether a panel model or simple OLS can be used. The LM test helps to decide between a random-effects regression and pooled OLS regression. The null hypothesis in the LM test is that variances across entities is zero i.e. no significant difference across units (i.e. no panel effect).

The p-value is less than 0.05 and therefore we reject the null and conclude that random effects are appropriate. This is evidence of significant differences across commercial banks, therefore we cannot run pooled OLS regression.

Table 4.3: Breusch and Pagan Lagrangian multiplier test for random effects

Model Dependent variable ?2-value p-value

1 Firm Value 21.17 0.000

4.5.2 Testing for heteroscedasticity

When heteroscedasticity is present, the standard errors of the estimates are biased and the study should use robust standard errors to correct for the presence of heteroscedasticity (Antonie, Cristescu, ; Cataniciu, 2010; Hoechle, 2007).

The existence of heteroscedasticity problem may bring about overestimation of the model, T statistic becomes smaller and in this way cause the incorrect conclusion. Moreover, the presence of the heteroscedasticity problem will cause the variance to become standard error and indirectly impact the T statistic and F statistic to become incorrect.The study used Modified Wald Test to test for existence for heteroscedasticity. Results in Table 4.4 reveal that the p value is less than 0.05 (p=0.0000). Since p0.05 which results to the acceptance of the null hypothesis. This implies that the random effects model is the appropriate model for analysis.

Table 4.5: Hausman test results

Variable Fixed Random Variable (Diff.) Prob.

Board Size -0.0088 -0.0039 -0.0486 0.0045

Board Independence 0.0672 -0.0414 0.1086 0.0927

Ownership Concentration -0.2771 0.0964 -0.3734 0.5204

Audit Committee -0.0157 0.0152 0.0054 0.0061

Chi square = 6.36, P value =0.1739

4.5.4 Test for autocorrelation

According to Gujarati (2012), autocorrelation may be defined as correlation between

members of series of observations ordered in time or in space. Drukker (2003) argues

that, because autocorrelation in linear panel-data models biases the standard errors and

causes the results to be less efficient, researchers need to identify serial correlation in the

idiosyncratic error term in a panel-data model.

The study used Wooldridge Drukker test for autocorrelation in panel data. The results are presented in table 4.6. It can be observed that p=0.3203 ;0.05. This indicates that there exists no autocorrelation.

Table 4.6 Wooldridge Drukker test results

Model Dependent variable F-value p-value

1 Firm Value 1.093 0.3203

4.5.5 Test for multicollinearity

The study used Variance Inflation Factor (VIF) to test for optimality. As a rule of the thumb, a VIF of 1 indicates no correlation between predictors; a value of between 1 and 5 indicates moderate correlation and a value above 5 indicate that predictor variables are highly correlated (Gujarati, 2012). Table 4.7 indicates a mean VIF of 1.20 which is between 1 and 5 and indeed very close to one. The independent variables are moderately correlated showing nonexistence of multicollinearity.

Table 4.7 Variance Inflation Factor

Variable VIF 1/VIF(Tolerance)

Board Size 1.35 0.7421

Board Independence 1.34 0.7483

Ownership Concentration 1.10 0.9115

Audit Committee 1.02 0.9771

Mean 1.20

4.6 Regression Analysis

Informed by the diagnostic test, the study used the random effects panel model for the analysis. The results are presented in table 4.8.

4.6.1 Relationship between the size of the board and Firm value

The first hypothesis of the study was that there exist no statistically significant relationship between the board size and firm value of listed commercial banks in Kenya. The random effect regression results presented in table 4.8 shows a beta coefficient of 0.034 with a p-value result of 0.000. The p-value ranges between 0-1. A low p-value shows that the relationship between the variables is significant. (Vickers, 2010). In most cases, p-values range from 0.01 to 0.05. If the p-value result shows a low level (below 0.01), the relationship between the variables is significant. A low p-value