Corruption and its diverse effect on credit risk: global evidence

Future Business Journal

Table 8 Robustness test (alternative corruption measurement)

Variables	Full sample		High corruption		Low corruption
Variables	Coefficient	t-stat	Coefficient	t-stat	Coefficient	t-stat
Control of Corruption	− 0.012	− 3.551 (***)	0.007	0.770	− 0.015	0.004 (***)
Capitalization	0.494	30.992 (***)	0.562	25.596 (***)	0.409	0.023 (***)
Credit Disclosure Index	0.076	2.281 (**)	0.072	1.056	0.118	0.039 (***)
GDP Growth	− 0.065	− 4.541 (***)	− 0.038	− 2.282	− 0.151	0.031 (***)
Inflation	− 0.003	− 0.742	0.000	0.075 (**)	− 0.041	0.017 (**)
Public Debt	− 0.001	− 0.428	− 0.006	− 0.884	0.001	0.003
Remittance	0.010	0.854	− 0.002	− 0.155	0.011	0.021
Trade Openness	0.004	2.486 (**)	− 0.001	− 0.334	0.007	0.002 (***)
Unemployment	0.062	4.688 (***)	− 0.016	− 0.806	0.137	0.018 (***)
Constant	0.027	0.099	0.791	2.208	− 0.297	0.532
Adjusted r-square	0.290		0.328		0.281
F-value	146.548 (***)		85.438 (***)		72.229 (***)
Observations	3200		1764		1437

We have performed panel least square regression based on the model: \({\text{NPL}}_{it} = \alpha_{i} + \beta_{1} {\text{Corruption}}_{it} + \mathop \sum \nolimits_{i = 1}^{i} \beta_{2} {\text{Controls}}_{it} + \varepsilon_{it}\). However, the measure of corruption is changed in this regression analysis. We have used control of corruption scores from the World Governance Indicator, published by World Bank in this model. The control of corruption scores ranges from approximately − 2.5 to 2.5. A negative score indicates higher utilization of public power for private gain and vice versa. Hausman test score indicates that the fixed effect model is appropriate for the study. Therefore, we report fixed effect regression scores in Table 8.
Asterisk ** and *** represent significance level at 5 and 1% respectively.