Table 3 shows the descriptive statistics of the dependent and independent variables: non-performing loan ratio, credit growth, profitability, operating efficiency, capital, and income diversification of 36 commercial banks listed at the Indonesian Stock Exchange over the period of 20082015.

The descriptive statistics in Table 3 show that the average NPL, credit growth, profitability, efficiency, capital ratio, and diversification rates are 3.2 per cent, 24.6 per cent, 0.9 per cent, 84.3 per cent, 12.5 per cent, and 10.6 per cent, respectively. The sample also clearly demonstrates that the commercial banks in Indonesia greatly vary. For instance, despite the fact that the average NPL ratio of 3.2 per cent is lower than the OJKs requirement of 5 per cent, the lowest NPL ratio is 0.1 per cent and the highest reaches 51 per cent. 6 per cent, implying that banks generally increase their loans by almost 25 per cent from last years total credits. In addition, there is a bank issuing loans by 5 times higher compared to last years book, but there is also a bank that even reduces its credits by 93 per cent. This huge variation also appears in the return on average asset (ROAA), operating efficiency (EF), capital (CAP), and income diversification (DIV). The static model applied in this study uses a fixed-effects estimation model. The results of the model are given in Table 4.

This is a preventive action to mitigate potential default risk to finally lower the possibility of having loan defaults in the banks asset

The results in Table 3 confirm that the five bank-specific factors are joint determinants of the NPL ratio indicated by a significant F-test (0.00) and a highly adjusted R 2 ( per cent). Nevertheless, only profitability (ROAA) and credit growth (CG) are found to be significant. Both profitability and credit growth have negative relationships with the NPL. The results suggest that, at the 1 per cent significance level, a 1 per cent increase in profitability lowers the NPL by 5 per cent. Similarly, at the 5 per cent significance level, a 1 per cent increase in credit growth decreases NPL by 2.64 per cent.

Comparing our results with those of earlier studies, the negative relationship between profitability and the NPLs of Indonesian commercial banks diverges from what is proven by Rajan (1994) but is similar to the UT car title loans work of Berger and De Young (1997). In addition, the negative influence of credit growth on NPL contrasts the positive relationship found by Keeton (1999) and Ahmad and Bashir (2013); however, it supports the finding of Boudriga et al. (2010). In terms of operating efficiency, bank capital, and income diversification, our findings show that these factors do not influence NPLs, which conflicts with most of the previous research in various countries. However, an insignificant relationship between income diversification and NPL is persistent with what is found by Hu et al. (2004).

This is sensible because, when banks are highly profitable, they can perform proper management practices, including undertaking day-today operations as well as managing loan portfolios

The negative relationship between profitability and NPL suggests that the more profit a bank gets, the less likely the bank is to have NPLs in the respective year. More specifically, because they can afford a sufficient operating budget, they can perform adequate loan management such as underwriting, monitoring, and controlling.

The negative impact of credit growth on NPL implies that banks issuing more credit tend to be more concentrated on credit activities so that they are more skillful in working with credit scoring, evaluation, and monitoring systems. By being specialized in lending activities, these banks can eventually reduce the number of NPLs in the current year. Another possibility to explain the negative relationship between credit growth and NPLs is that banks apply a liberal credit policy to reach a particular NPL ratio, especially to meet the OJKs requirement of 5 per cent. This liberal policy is carried out by restructuring the former credit terms of the insolvent loans, such as extending the terms of loans or softening the covenants, so that the debtors can remain current. In this case, the restructured loans might be recognized as the new loans, which are solvent. Alternatively, the banks can simply lend new money in order to increase total loans to get a lower NPL ratio. This makes sense because the NPL ratio is calculated by dividing the amount of non-performing loans by the total loans. Accordingly, an increase in credit supply is followed by the lower NPL ratio.