Credit risk in Islamic and conventional banking

This study investigates differences in the credit profiles of Islamic and conventional banks in a single market (Bahrain) and attempts to identify the factors responsible for those differences. By focusing on one market, the impacts of cultural factors, structural differences in the banking industry and alternate regulatory regimes are mitigated. The study utilizes panel data from the Bankscope database for fifty-six conventional banks and twenty-five Islamic banks between 1987 and 2014.

Using a pooled ordinary least squares (OLS) model, in which credit risk is defined as loan loss reserves divided by gross loans, we find Islamic banks have lower credit risk than their conventional counterparts. Control variables indicated by previous studies are incoporated into the model. These include the ratio of equity capital to assets, loans to earning assets and bank size (log of total assets), all of which are inversely related to credit risk. The robustness of the OLS results is confirmed using a log transformation of the credit risk measure as the dependent variable.

In order to examine further the effect of the control variables on credit risk, interaction dummies were introduced. The regression results show that three variables contribute to lower credit risk for Islamic bamks: equity capital to assets, liquidity (liquid assets to deposits and short-term funding) and cost inefficiency (operating costs to earnings). These results are consistent with previous studies, which have found that Islamic banks are better capitalized, generate more income from loans and are less efficient than conventional banks.

The study then uses a logit model to assess the likelihood of credit risk in Islamic and conventional banks. Classifying credit risk into quartiles, and focussing on the highest risk and lowest risk categories, we find that equity capital to assets, loans to earning assets and size increase the likelihood of banks having low credit risk. The values of these variables are higher for Islamic banks, consistent with the results of our OLS regressions.

In contrast, the results for the liquidity measure are mixed. The OLS regressions indicate a positive relationship between liquidity and risk. Likewise, the logit model suggests that Islamic banks are more likely to have high credit risk as a result of an increase in liquidity. In contrast, the liquidity interaction variable indicates that the liquidity variable contributes to lower credit risk in Islamic banks. This suggests that further analysis is required.