Empirical Results and Analysis

Random Effect Model REM. Out of these 3 models, Chow and Haussman test will be used to choose one best method.

4. Empirical Results and Analysis

Sample used in this research that has met the criteria is 470 years of observation. On the first phase, we will do multicollinearity test between variables by undertaking Pearson correlation analysis as shown in table 1. Table 1.Correlation Analysis Result GENDER GROWTH PAST WORKING EXPERIENCE PEND. PENG. PROFIL FOTO PROFIT TAN. UK. GENDER 1.000 0.032 0.065 0.016 0.024 0.076 0.036 0.158 0.162 GROWTH 0.032 1.000 0.054 0.151 0.034 0.055 0.129 0.003 0.038 PAST_WORKING_ EXPERIENCE 0.065 0.054 1.000 0.232 0.180 0.112 0.573 0.064 0.434 EDUCATION 0.016 0.151 0.232 1.000 0.073 0.051 0.211 0.071 0.129 EXPERIENCE 0.024 0.034 0.180 0.073 1.000 0.017 0.202 0.021 0.111 PROFILE_PHOTO 0.076 0.055 0.112 0.051 0.017 1.000 0.000 0.090 0.180 PROFITABILITY 0.036 0.129 0.573 0.211 0.202 0.000 1.000 0.170 0.372 TANGIBILITY 0.158 0.003 0.064 0.071 0.021 0.090 0.170 1.000 0.013 SIZE 0.162 0.038 0.434 0.129 0.111 0.180 0.372 0.013 1.000 As shown in table 1, it uses all variables with correlation coefficient less than 0.6 by making the correlation value to absolute, except for past working experience and profile photo. The next test is the variance test for inflation factor with the result of all variables are below cut-off 5, such as: Gender with VIF value of 1.074; growth with VIF value of 1.056; past working performance with VIF value of 1.730; education with VIF value of 1.102; experience with VIF value of 1.056; profile photo with VIF value of 1.078; profitability with VIF value of 1.711; tangibility with VIF value of 1.122; and size with VIF value of 1.344. Due to this result, it can be concluded that there is no linkage between independent variables, and no variable need to be removed from the research. After seeing the correlation between independent variables, the result of the data processed must be analyzed through descriptive statistic as shown in table 2. From table 2, it is seen that the average debt to totalasset is less than 20, and the firm debt to equity ratio is reaching the average of 60. Table 2 Descriptive Statistic Means Modus Min. Max. LTD_TA 0.1898 0.0001 0.8056 LTD_TE 0.5935 0.0002 16.9192 PROFILE_PHOTO 2 1 4 EDUCATION 2 1 4 EXPERIENCE 1 1 GENDER 1 1 PAST_WORKING_ EXPERIENCE 0.0824 -0.5392 0.7423 PROFITABILITY 0.0618 -0.7557 0.9660 SIZE 28.0635 21.1949 32.9378 TANGIBILITY 0.57911 0.1223 0.9945 GROWTH 0.1428 -1.4918 2.3491 The financial report shows that managerial overconfident reflected from the profile photo in financial report, therefore most of the executive profile photos are 2, which means there is another party beside the executive that also has his profile photo in the financial report. The other party is identified to be the commissioner. As from the education, most executives are bachelor graduates. While from the experience, most executives have been officiated as chief officers CEO, CFO, COO, CIO, and other equivalents position either at current firm or at previous companies. In terms of gender, most executives are male. The next managerial overconfident factor is related to CEO past working performance. The average of past working performance, which is shown by the ability to generate cash flow from operational activity, is reaching the range of 8 to the total asset. Furthermore, further discussion to the main model will be made. After applying Chow and Hausmann test to 3 models of data panel, such as CE, FEM, and REM, the Chow test generates significant result, where FEM gives more estimation than CE. After that, Haussman test is done to choose either REM or FEM is the best model. The chosen model to be interpreted is FEM. Table 3 Inferential Statistic Result for Model 1 Variable CE FEM-1.1 FEM-1.2. REM Beta t beta t Beta t beta t PROFILE_PHOTO 0.0002 0.02 -0.0200 -2.86 -0.0199 -2.86 -0.0190 -2.78 EDUCATION -0.0122 -1.17 0.0220 1.84 0.0220 1.84 0.0155 1.39 EXPERIENCE -0.0321 -2.34 0.0363 2.49 0.0359 2.49 0.0270 1.96 GENDER -0.0459 -1.46 -0.0091 -0.19 -0.0036 -0.08 PAST_WORKING_ PERFORMANCE -0.1238 -1.77 -0.1668 -3.32 -0.1670 -3.33 -0.1664 -3.37 PROFITABILITY -0.0682 -0.86 0.0866 1.54 0.0871 1.55 0.0803 1.45 SIZE 0.0162 4.30 -0.0129 -1.70 -0.0128 -1.69 -0.0004 -0.07 TANGIBILITY 0.2946 8.68 0.0624 1.51 0.0621 1.50 0.1200 3.20 GROWTH 0.0289 1.41 0.0198 1.44 0.0198 1.44 0.0173 1.28 R Squared 0.1916 0.7573 0.7573 0.0767 Adjusted R Squared 0.1758 0.7251 0.7257 0.0587 F Statistics 12.1190 23.4971 23.9872 4.2505 Information: Significance on α = 5; significance on α = 10 From the F test, we can see that FEM model 1 is significant, which means that altogether the independent variable is affecting the firm financing decision. In table 3, especially model 1 that uses FEM as seen from 5 measurements of managerial overconfidence, 4 proxies are found to be significant, such as profile photo, education, experience, and past working performance. This result implies that managerial overconfidence variables in the executive will determine a firm financing decision. Two of 5 managerial overconfidence approaches, such as profile photo and past working performance appear to be negative significance, while education and experience are positive significance, and gender appears to give no significance effect. Table 3 shows that the result of profile photo is different from the hypotheses, where apparently, the higher the confidence of an executive, the lower the debt used. This result is consistence to the research done by Kiong-Ting et al. 2016, which means that the more confident the executive, the less is the debt used by the firm. Therefore, this result explains that overconfident executive chooses to utilize internal funding for its new projects in expectation to give additional value for the shareholders. Meanwhile, education variable is found to be positive significance, which means that the higher the level of education, the higher the confidence of the executive will likely to raise the usage of debt. The confident executive will have high assurance that the new project invested is a right choice, which is why the executive is certain on its ability to pay-off the debt. Likewise to the working experience, it also shows a positive significance result. It appears that by having a working experience, a CEO will be able to encounter many situations by using existing information and tends to be unbiased, and this will form its confidence. CEO experience at its previous specific position CEO, CFO, COO, CIO, and other equivalent positions either at the current firm or the previous firm, will be useful in overcoming problems that might appear at new projects, so CEO tends to use debt financing. From table 3, we can see that gender has no effect on the firm financing decision. With the education and experience owned by CEO, and by focusing on the past working experience, the firm will be the main factor in shaping CEO confidence when it comes to choose whether to use debt or not. It is also presented in table 3, the second FEM model by removing gender variable that is not significant. By removing insignificant factor like gender, the result of the test showsconsistency on both sides; direction and significance. This research also does robustness test by using long term debt to total equity as dependent variable. The result of this test can be seen on table 4. FEM 2.1 model is a model that uses all independent variables as used in FEM 1 on table 3. The generated result from the test is education, experience, and past working experience are proven to be affecting, while gender is consistently showing insignificant effect to a firm debt usage decision. However, there has been a shift, where profile photo that is previously significant, has now become insignificant. The test is said to be robust because 3 out of 5 managerial overconfidence measurements, have direction and significance that suit the main model. Table4 Inferential Statistic Result for Model 2 Variable FEM 2.1. FEM 2.2. FEM2.3 beta T beta t beta t PROFIL_PHOTO -0.1322 -1.54 -0.1317 -1.53 EDUCATION 0.3311 2.26 0.3313 2.26 0.3387 2.31 EXPERIENCE 0.4542 2.53 0.4474 2.52 0.4622 2.61 GENDER -0.1606 -0.27 PAST_WORKING_ EXPERIENCE -2.6065 -4.23 -2.6101 -4.24 -2.6445 -4.29 PROFITABILITY 2.0150 2.93 2.0235 2.95 2.0790 3.03 SIZE -0.2914 -3.12 -0.2898 -3.11 -0.3129 -3.40 TANGIBILITY -0.7761 -1.53 -0.7817 -1.54 -0.8169 -1.61 GROWTH 0.2773 1.64 0.2762 1.64 0.2835 1.68 R Squared 0.3777 0.3776 0.3741 Adjusted R Squared 0.2951 0.2967 0.2944 F Statistics 4.5704 4.6640 4.6921 Information: Significant on α = 5; significant on α = 10 In table 4, a test is also done to remove insignificant factors and those are gender in model FEM 2.2.; also, gender and profile photo in model FEM 2.3. The result shows consistency in direction and significance of 3 variables, such as education, experience, and past working experience to financing policy in a firm. As for the control variable in model FEM 1 and model 2, the result is that firm size has a negative significant result to financing. This result means that the bigger the firm size, it tends to not use much debt. It is because a big size firm tends to dominate market, both service or product market, so the firm earns higher profit. Firm with a high profit like Unilever Indonesia tends to avoid the option of using debt in its financing policy. For the other control variables, those are tangibility and growth, a consistent result of being insignificant either on the first and second model is found. Meanwhile, for profitability, there has been a significant change, which on the first model, it is proven to be insignificant, yet on the second model, it turns out to be negative significant.

5. Conclusion