Members e from above model -7.39 is

Members X2= Board meeting frequencyX3= Audit committee meeting frequency X4= CEOs’ compensation Moreover ? is the constant or intercept. ? representscoefficient or slope.  ? 1, ? 2,? 3 and ? 4 are slope or regression coefficients and e is errorterm for the model. 4. 0 Results and DiscussionsAnalysisand interpretation of collected data are discussed in this section of thisresearch paper. This section discusses the inferentialstatistics of the data. 4.

1 Resultsof selected state owned commercial banksFollowing results are getting from ordinary leastsquare (0LS) multiple regression analysis based on investigated state ownedcommercial banks. Table 2: Model summary and ANOVA analysis forinvestigated State owned commercial banks Details ROA ROE Sample size 12 12 Multiple correlation coefficient (R) 0. 822 0. 689 Co-efficient of determination (R2) 0. 864 0. 593 Adjusted R2 0. 580 0. 019 Standard error of the estimate 0.

86960 7. 22236 F value 4. 827 2. 535 Source: Appendix table 3 to 84. 1. 1Influential CG factor analysisModel 1: ROA= -7. 79+ 0. 30(x1) + 0.

187(x2)+ (-1. 19) (x3) + 0. 112(x4) + eFrom above model -7. 39 is the constant value thatindicates if the independent variables has zero impact on ROA then the companieswill face loss about -7. 39 million BDT.

Most influencing variable for ROA isboard meeting frequency then CEO’s compensation and most uninfluential variableis audit meeting frequency. Model 2: ROE= -33. 186+ 1.

826 (x1) + 1. 646(x2)+ (- 7. 82) (x3) + 0.

293(x4) + eFor model-2 the constant value is -33. 186 thatindicates if the independent variables has zero impact on ROE then the companieswill face loss about -33. 186 million BDT . Most influencing variable are boredsize and board meeting frequency and most uninfluential variable is audit committeemeeting frequency on ROE.

4. 1. 2 Co-efficientof correlation (R): Co-efficient of correlation measures the strength ofthe linear relationship between dependent variable and independent variables. Multiple co-efficient of correlation for ROA is 82. 2% which means dependentvariable and independent variables are very strongly positively correlated. Onthe other hand, co-efficient of correlation for ROE is 68. 9% which just exceedsthe moderate level but positively correlated.

4. 1. 3 Co-efficientof determination (R2): Co-efficient of determination measures the percentageor proportion of total variation in dependent variable explained by theindependent variables. R2 is 86. 4% which is very high for ROA.

Itmeans independent variables can explain perfectly 86. 4 % variation of ROA. The co-efficientof determination is 59. 3% for ROE. 4. 1. 4 AdjustedR2Adjusted r2 measure whether the model isfit or not that is adjusted for the number of explanatory variables in themodel. If more useful explanatory variable will add in the model, the moreadjusted r2 will increase.

For ROA adjusted r2 is 58%. Onthe other hand, for ROE it is 1. 9%.

There may be other variables which affectmore on the banks financial performance. 4. 1. 5 Standarderror of estimate: Astandard error is the standarddeviation of the samplingdistribution of a statistic.

In this analysis standard errors are 0. 86960 and 7. 22236respectively for ROA and ROE. 4. 1.

6 F-test: Here level of significance is 5%. Degrees of freedomfor the Numerator and denominator are respectively 4 and 7.  Conditionrule: If f-calculated> f-tab then the overall model issignificantIf f-calculated < f-tab then the overall mode isinsignificantTherefore f table value found is 4. 12 from fdistribution table. For ROA f-calculated value is greater than the f-tabvalue.

So the null hypothesis is rejected and alternate hypothesis is acceptedthat means there has impact of corporate governance on ROA and the model issignificant. On the other hand for ROE, f-tab value is greater than the f-calculatedvalue so we cannot reject the null hypothesis. Also it is clear that theoverall model is insignificant and there has no impact of corporate governanceon ROE.

4. 2 Resultsof selected private commercial banksFollowing results are getting from the ordinary leastsquare (OLS) multiple regression analysis based on selected private commercialbanks. Table 3: Model summary and ANOVA analysis for selectedprivate commercial banks. Details ROA   Sample size 12 12 Multiple correlation coefficient (R) 0. 792 0. 692 Co-efficient of determination (R2) 0. 609 0. 568 Adjusted R2 0.

302 0. 223 Standard error of the estimate . 38539 3. 6342 F value 1.

376 1. 494 Source: Appendix table 9 to 144. 2. 1Influential CG factor analysisModel 1: ROA: 1. 475+ 0.

063(x1) + 0. 082(x2)+ -7. 56(x3) + -0. 043(x4) +eFrom above model-1 the constant value is 1. 475 thatindicates if the independent variables has zero impact on ROA then the companieswill generate about 1.

475 million BDT. Most influencing variable for ROA isboard meeting frequency then board size and most uninfluential variable isAudit committee meeting frequency. Model 2: ROE: 10. 4534+ 0. 682(x1) + 0. 069(x2)+ -6.

30 (x3) + -0. 147(x4) +eFrom above model-2 the constant value is 10. 4534 thatindicates if the independent variables has zero impact on ROE then the companieswill generate about BDT 10. 4534 million. Also in this case most influencingvariable for ROE is board size then board meeting frequency and mostuninfluential variable is Audit committee meeting frequency.

4. 2. 2 Co-efficientof correlation: For selected private banks multiple correlation ofcoefficient is 79. 2% and 69. 9% respectively for ROA and ROE which are definedas perfectly positively correlated between independent variables and dependentvariable. 4. 2. 3 Co-efficientof determination (R2): Co-efficient of determination is approximately 60% forROA which means independent variables can express almost 60% variation in ROA.

On the other hand for ROE is 56. 8%. 4. 2. 4 AdjustedR2: Adjusted R2 is 30. 2% and 22.

3% respectivelyfor ROA and ROE. There must be other variables which have more influence onselected private banks’ financial performance than those are taken. 4. 1.

5Standard error of estimate: Astandard error is the standarddeviation of the samplingdistribution of a statistic. In the analysis standard errors are 0. 38539 and3.

6342respectively for ROA and ROE. 4. 2.

5 F test: Here level of significance is 5%. Degrees of freedomfor the Numerator and denominator are respectively 4 and 7.  As the calculated fval