Islamic Bank Efficiency in Indonesia: Stochastic Frontier Analysis

Abstract

This research is conducted to measure the efficiency level of Islamic banking in Indonesia and also to analyze the factors that can affect its efficiency level. This research used a purposive sampling technique to determine the sample size that will be used, with criteria that the bank has been operating since 2010 and consistently published its financial reports during the research period from 2011 until 2019; therefore, the total sample obtained was 11 samples. Analysis for efficiency level is done by using linear programming Stochastic Frontier Analysis (SFA), with test tool in the form of Frontier 4.1 and Eviews9 to find out what factors that affect efficiency. Efficiency test is done by involving input and output, while influence test used bank-specific variables comprising bank size, bank financial ratio, and macro-economy variable. Research result shows that there are only two banks that are almost close to being fully efficient firms, but the result still does not indicate that Islamic bank works efficiently. Results of the influence test show that factors affecting Islamic banking efficiency in Indonesia are bank size, Capital Adequacy Ratio (CAR), Non-Performing Finance (NPF), and Financing to Deposit Ratio (FDR), while other factors are not influential over the study period.

1. Introduction

Islamic banking defines a bank as an entity that collects funds from the community in the form of deposits and distributes them back to the community in the form of financing or other types to improve the community’s living standard. UU No. 21 the Year 2008 about Islamic Banking, states that Islamic Bank is a bank that manages its business activity based on Islamic principles or Islamic law arranged in Majelis Ulama Indonesia guidance such as justice and balance principle (‘adl wa tawazun), benefit (maslahah), universalism (alamiyah), however, it does not contain gharar (uncertainty, hazard, chance or risk), maysir (speculation or gambling), riba (usury, or unjust, exploitative gains), zalim and haram (act that is forbidden) object. The fulfillment of Islamic principles become important because essentially Islamic banks must offer products appropriate with Islamic principles.

Islamic bank in Indonesia was established in 1992 through the establishment of Bank Muamalat Indonesia (BMI). Since its first emergence, several policies have been implemented and several changes have taken place to increase the performance of Islamic banks. This development is supported by the creation of more Islamic banks. Since then the number of Islamic banking outlets has increased, comprising 14 fully Shariah-compliant banks, 24 Islamic banking units, and 158 Shariah-compliant rural banks. This development is also marked by the success of Islamic banks in Indonesia. Based on the 2019 Islamic Finance Development Indicator (IFDI), Indonesia occupied the first rank in Islamic Banking and Finance (IBF).

One of the main indicators of Islamic financial economy development is the presence of the Islamic banking industry in society. Based on this development, research must be done related to Islamic bank performance to understand how efficient Islamic bank manages input and output from the cost side, and also what factors affect Islamic banking efficiency. This research then becomes important because research about Islamic bank efficiency is rare; therefore, research must be conducted to ensure Islamic banks’ superior performance and ability to compete with conventional banks.

The main objective of this research is to measure the performance of Islamic banks in Indonesia by examining cost efficiency using the Stochastic Frontier Analysis (SFA) parametric test to prove the influence of bank financial ratios and macroeconomy on cost efficiency during the 2011-2019 period. Succinctly, research questions are:

1. How is the performance of Islamic banks in Indonesia during the research period?

2. Do bank financial ratios and macroeconomy affect Islamic bank efficiency in Indonesia during the research period?

Novelty contribution in this research is that this is the first research that measures Islamic bank cost efficiency in Indonesia by using the SFA method since the Islamic banking industry started growing and developing. Second, to the best of the researcher’s knowledge, this research is different from the previous research in terms of input and output usage, and also with regard to the variables that are used to observe factors that affect Islamic bank cost efficiency. Third, there is not many research that examines the influence of bank size, bank financial ratios, and macroeconomy on Islamic bank cost efficiency in Indonesia during the research period.

The results of this research hopefully can give a conceptual contribution and add knowledge related to the Islamic banking industry. This research is strengthened by the intermediation approach which according to Berger and Young (1997) felt more appropriate in evaluating banking performance because of its business’ main function as collector and distributor of society’s fund, which reflects in the variables used.

2. Literature Review

Efficiency is used to measure the performance of a company. A company can be said to be efficient if it can get maximum output from input given or minimize input used in producing output. Efficiency is a comparison between the output and input used. Efficiency is one of the parameters used in measuring performance theoretically which is the performance that underlines all performance in an organization. Banking efficiency is measured by calculating the ratio between banking output and input.

Farrell (1957) stated that efficiency consists of technical efficiency and allocative efficiency. Technical efficiency reflects the ability of a company in managing the amount of input available to produce a number of outputs. While allocative efficiency reflects the ability of a company in optimizing input usage based on price structure and production technology, both measurements are called economy efficiency. A company can be called efficient if it can minimize production costs with technology and valid market price.

According to Mlima and Hjalmarsson (2002) in measuring efficiency, financial ratio calculation with non-parametric and parametric approaches can be used. Data Envelopment Analysis (DEA) is a non-parametric approach and is a linear programming technique. It is a non-parametric technique used in the estimation of production functions and has been used extensively to estimate measures of technical efficiency. The Stochastic Frontier Approach (SFA) is parametric. Distribution Free Approach (DFA) uses the average cost function residual of the presumed using panel data to calculate cost frontier efficiency. Free Disposal Hull (FDH) model is a non-parametric method to measure the efficiency of production units or decision-making units. Thick Frontier Approach (TFA) is a parametric approach. In this study, a Stochastic Frontier Approach (SFA) as a parametric approach will be used.

Stochastic Frontier Approach (SFA) was first put forward by Meeusen and van Den Broeck (1977) to measure bank efficiency. According to Coelli et al. (2003), SFA has more advantages compared to other methods. First, it involves disturbance terms that represent disturbances, measurement errors, and exogenous surprises that are out of control; environment variable that is easier to be treated. They suggested that there are advantages and disadvantages associated with the use of DEA and SFA and that neither approach strictly dominates the other. Rather, there may be context-specific issues or subjective factors that may influence the choice of one technique in a particular study.

Kumbhakar and Lovell (2000) stated that measuring cost efficiency in the SFA method can be done by using an outputoriented approach for technical efficiency measurement and an input-oriented approach to measure cost efficiency. Technical efficiency can be measured by using the production frontier and cost efficiency is measured by using the cost frontier. Ansari (2007) used the distribution-free approach (DFA) to estimate levels of cost efficiency of individual banks operating in Pakistan. These levels of efficiency are also analyzed under CAMELS indicators to provide micro insights into their financial standings to justify their prevailing positions.

Mohamad et al. (2008) did research to observe the efficiency of 80 banks in 21 Islamic and conventional countries, from the cost and benefit side by using the Stochastic Frontier Approach (SFA). Besides, they also assessed efficiency from the size side, age, and bank territory. The findings suggest that there are no significant differences between the overall efficiency results of conventional versus Islamic banks. However, there is substantial room for improvement in cost minimization and profit maximization in both banking systems.

Belanès et al. (2015) investigated the influence of the subprime crisis on the efficiency of Islamic banks in the GCC region using data envelopment analysis for the period spanning from 2005 to 2011. They focused on three aspects of efficiency, namely overall technical efficiency, pure technical efficiency, and scale efficiency. Empirical findings highlight a slight decline in Islamic bank efficiency further to the subprime crisis just like their conventional peers all over the world. Eyceyurt Batir et al., (2017) examined the technical, allocative, and cost-efficiency of conventional and participation banks in Turkey with the data envelopment analysis (DEA) method. In the wake of finding technical, allocative, and cost efficiency results by DEA intermediation approach, Tobit regression analysis is used to determine the factors influencing the efficiency. The results of DEA indicate that average participation of bank efficiency is higher than the average conventional bank efficiency each year. Regarding Tobit regression analysis, while expenses and loan quality has a significantly negative relationship with the efficiency of conventional banks, they have a significantly positive relationship with the efficiency of participation banks.

Kumar et al. (2020) researched from 2005 until 2017 on private banks in India by using Data Envelopment Analysis (DEA) method. The findings showed that private bank efficiency level in India operates in level more than 0,9, and the rest for 3 years on 0,6 until 0,7. Some researchers have conducted studies to observe Indonesian banking efficiency. Margono et al. (2010) studied technology advancement and bank productivity in Indonesia during 1993-2000. The research results showed that the average banking efficiency level before and after the crisis was 79,67% and 53,40%. Besar (2011) research result shows that the average cost efficiency is about 40%-50% and government bank tends to be more efficient compared to foreign ownership. Anwar (2016) did a comparison of conventional and Islamic bank efficiency in Indonesia from 2002-2010 by using Data Envelopment Analysis (DEA) with a data panel of 116 banks. Results showed that Islamic banks are more efficient in the small business funding model (SBF).

Zuhroh et al. (2015) aimed at estimating the cost efficiency of Islamic Banks –using stochastic cost frontier approach, determining cost inefficiency sources, and analyzing the influence of managerial competency and structural variable on cost efficiency. The samples of observation are three full Islamic banking systems and 19 conventional banks listed on the Indonesia Stock Exchange (IDX) purposefully from 2004.03 – 2010.4. The result showed that Islamic Banks are superior in the achievement of technical efficiency, but the average cost efficiency is much lower than conventional banks. Yulita & Rizal (2016) used DEA (Data Envelopment Analysis) to measure Islamic bank efficiency levels between Malaysia and Indonesia during 2011-2014. The findings showed that Islamic banks in Indonesia are more efficient than Islamic banks in Malaysia; however, there is no significant difference. Another research by Banna et al. (2017) studied the banking industry in Bangladesh during 2000-2013 by using Data Envelopment Analysis (DEA). Findings showed that the banking sector in Bangladesh showed the highest efficiency level in 2001 and the lowest in 2010. Besides, this research showed that bank size, capital adequacy ratio, equity return average, and interest rate showed significant influence on bank efficiency in Bangladesh.

Anwar (2019) examined commercial bank cost efficiency in Indonesia during 2002-2010 by using Battese-Coelli 1992 (BC92) method with TOBIT regression. The result showed that banking efficiency tends to decrease during the research period. Besides, the research stated that Indonesia’s banking cost efficiency is affected by bank size, profitability, adequate financial capital, loan to deposit, and credit risk management, and macroeconomy. Agustina et al. (2019) measured and analyzed the technical efficiency of Indonesian Islamic rural banks by using balanced panel data of Indonesian Islamic rural banks from quartile I-2011 to quartile IV-2016. The sample includes 58 Islamic rural banks with a total of 1392 observations. By using stochastic frontier analysis, the result shows that the average technical efficiency of Indonesian Islamic rural banks reached 86% and there is still 14 percent that can be optimized. Puteh et al. (2018) researched the Islamic banking industry in Indonesia during 2012-2016 with a comparison formula - business charge with business revenue (BOPO). The research result showed Islamic bank in Indonesia is not efficient yet.

Muttaqin et al. (2020) used three stages to find out the conditions of the efficiency level of Islamic banking. Frontier and Stochastic Frontier Approaches are used to calculate the efficiency level and then averaged. Last, the determinants of efficiency were conducted by the Tobit Model. The study found the average efficiency level is 83.51% and is classified as less efficient. The Tobit model showed that all of the variables don’t have significant effect on efficiency level of Islamic banking, except the ROA. Le (2020) focused on researching the factors affecting the retail banking efficiency of Vietcombank branches in the Mekong-Delta region. By collecting data from financial statements from 15 branches of VCB in the Mekong-Delta Region between 2015 and 2018, the paper applies DEA estimation to measure the effectiveness of retail banking activities and uses the Tobit regression model to identify factors affecting retail banking efficiency. The results demonstrated that the retail banking efficiency of branches averaged 52.5% during the period. The results also showed that bank scale-related factors, capital adequacy, credit quality, time-specific, and region have impact significantly upon retail banking efficiency.

3. Research Methods and Materials

This research uses secondary data in the form of panel data or longitudinal data that is a group of individual data comprising Islamic general bank data in Indonesia. This research used a purposive sampling technique to determine the sample size that will be used, with criteria that the bank has been operating since 2010 and consistently published its financial reports during the research period from 2011 until 2019. Data is collected through secondary data collection, obtained from the financial reports published on Bank Indonesia, Financial Service Authority, and Islamic bank website.

Efficiency is a comparison between the output and input used. A corporation can be called efficient if it can get maximum output from input given or minimize input that is used in producing output. In this research, the Stochastic Frontier Approach (SFA) method, a parametric approach in measuring efficiency level.

In estimating cost efficiency, this research uses the translog function for the total cost with several output variables and input price. Fund price, labor price, and financial capital price are input variables. Output variables are the total loan, securities investment, and other revenue. Data analysis used is descriptive analysis to describe researched variables which is efficient Descriptive analysis is used by using Frontier 4.1. measuring tool while for regression test we use Eviews 9. The regression equation is determined as follows:

\(\begin{aligned} \mathrm{CE}_{\mathrm{it}} &=\alpha_{0}+\beta_{1} \mathrm{LNSIZEit}+\beta_{2} \mathrm{ROAit}+\beta_{3} \mathrm{NOMit} \\ &+\beta_{4} \mathrm{NPF}_{\mathrm{it}}+\ln \beta_{5} \mathrm{FDR}_{\mathrm{it}}+\mathrm{n} \beta_{6} \mathrm{CAR}_{\mathrm{it}}+\beta_{7} \mathrm{NFLt} \\ &+\beta_{8} \mathrm{GDP}_{\mathrm{t}}+\beta_{9} \mathrm{RATE}_{\mathrm{t}}+\varepsilon_{\mathrm{it}} \end{aligned}\)

CE_it variable shows the estimation of bank cost efficiency–i in period-t. LNSIZE_it is a natural logarithm from the main capital. The main capital usage is done considering the minimum limit of main capital rules based on Bank Indonesia Regulation No 14/26/PBI/2012 date 27 December 2012 about Business Activity and Office Network. ROA_it ratio is a proxy for bank profitability. CAR_it ratio is the ratio that reflects bank solvency risk. NOM_it represents the susceptibility ratio. For liquidity, we use LDR_it ratio. NPF_it ratio shows bank management ability in managing the funding problem that is given by the bank. FDR_it is the ratio between the funding amount given by the bank with third party fund received by the bank. INFL_t is the annual inflation level that reflects the increase of a whole percentage in the consumer price index for all goods and services in Indonesia. GDP_t is the gross domestic product in real price growth level. RATE_t is the American dollar exchange rate in Rupiah Indonesia (IDR). ε_it shows an error.

4. Results and Discussion

4.1. Descriptive Analysis

Table 1 below explains input and output data during 2010- 2016. This research refers to bank function as intermediation institution.

Table 1: Efficiency Input and Output Variables (in million IDR)

The bank’s annual total cost during the research period is Rp. 1,040 billion with a minimum of Rp. 16 billion and a maximum of Rp. 8,335 billion. The total cost difference of banks in Indonesia reflects the difference in the operation scale of big banks and small banks in Indonesia. The price of the fund (PF) is calculated with the total sharing profit cost divided by the total deposit. It can be seen that the average fund price is about 225% of the total fund accumulated. Labor price (PL) is calculated with employee cost divided by the total asset, and capital price (PC) is calculated with the non-interest cost divided by the fixed assets. On average, the labor price of banks in Indonesia is about 2,01% of total assets. To calculate the capital price, the non-interest cost component is calculated from total cost minus interest cost and personnel cost. Output variable, that is, the total loan given to related party shows an average value of Rp.6,892.299 million. This indicates that the total loan given by the bank from the total asset owned is Rp.6,893,299 million. This value indicates that Islamic banks can attract society’s trust as a non-riba funding institution. Besides, that, the average investment also has a high value of Rp.2,167 billion with the lowest amount of investment is Rp.675 million. Meanwhile, banking also needs to increase another amount of revenue which at the moment is about 185 billion. Maximum and minimum data is obtained from one of the Islamic bank data which is closed at the end period of research.

Based on Table 2, banking profitability can be seen based on Return on Asset (ROA) ratio. ROA value can be used to see how far investment is able to give profit appropriate to a certain asset level. ROA in Table 2 shows a low value, even though a minimum ROA (healthy ROA) according to rules is 1.5 which means the condition at the moment shows that ROA of Islamic banks is below the minimum limit. This indicates that the enhancement of Islamic banks’ profit is lower compare to asset enhancement. If seen from Table 2. The ROA ratio tends to be better in 2011, 2012, 2013, and 2019. This becomes a challenge that needs more concern because the presence of Islamic banks in the middle of the rapid growth of conventional banks indicates Islamic banks need to show optimum performance. Capital Adequacy Ratio (CAR) value shows that bank tends to be more able in funding provision, which means banks are able to show its ability in providing fund which is then used to overcome loss as well as risk possibility in the future.

Table 2: Islamic Banking Financial Ratio​​​​​​​

Net Operating Margin (NOM) ratio shows the ability of productive assets in generating profits which is obtained from the difference between fund distribution revenue minus profit sharing with the operational cost divided by the average productive assets. As seen from Table 2, in several years, the ratio shows a very low value, that is <1%. This indicates that the ability of Islamic banks in making profits is still low and banks are not able to manage capital for operational activities. Non-Performing Financing (NPF) ratio of Islamic banks indicate that bank is in healthy condition by having a maximum ratio of 5%. If the NPF ratio is higher, then the funding risk that must be borne by the bank will also be higher, implying that impaired credit is also higher. Therefore, banks must have a large reserve fund to cover that risk. It is seen that the NPF ratio of Islamic banks is still ‘healthy’. Financing to Deposit Ratio (FDR) ratio is used to measure the bank’s ability to fulfill all short-term obligations at the due date. Islamic banks have good liquidity if they’re able to fulfill all funding needs to an external party, as well as, able to return third-party funds when asked. Therefore, the higher the FDR ratio, the lower the liquidity ability of the Islamic bank.

4.2. Indonesia Islamic Banking Efficiency

SFA approach is applied to panel data from 11 Islamic General Banks in Indonesia in 2011-2019. The total number of observations to calculate the efficiency value of all banking units is 99. Frontier efficiency analysis is the best practice analysis from all data set used. Its efficiency value is a comparison of all banks (business unit) and their observation year. This approach gives the simplest output and can be compared between business units directly, and can track the efficiency level of a bank; however, this explicitly assumes that technology is constant from time to time. SFA efficiency score between 0 and 1 (or 0% and 100%), where 1 (100%) represents fully efficient firm or fully technical efficient firm, and less than 0% represents inefficient firm or technically inefficient firm.

Conclusion of descriptive statistic estimation result of general bank efficiency level in Indonesia can be seen in Table 3:

Table 3: Cost Efficiency Estimates​​​​​​​

Table 3 shows during the research sample, the efficiency result tends to be near 0 (zero) which means that value represents an inefficient firm or the firm is technically inefficient. However, bank samples 3, 6, and 10 show a higher value compared to other banks. This indicates that the bank almost represents a fully efficient firm or value almost near 1 (one).

4.3. Determining Factor of Islamic Bank Cost Efficiency in Indonesia

This part will explain the determining factor that can affect Islamic banking cost efficiency in Indonesia during the 2011-2019 period. In this research, cost efficiency will act as a dependent variable and financial ratio and macroeconomy are independent variables.

Table 4: Model Election Test Result

Information: significant α=5%​​​​​​​

The Chow test is a statistical and econometric test of whether the coefficients in two linear regressions on different data sets are equal. The result shows the Chow test is insignificant (p-value 0,6706 is bigger than 5%). Therefore, H0 is accepted and H1 is rejected. The model follows the Common Effect Model (CEM). The Hausman Test (also called the Hausman specification test) detects endogenous regressors (predictor variables) in a regression model. In panel data analysis, the Hausman test can help you to choose between a fixedeffects model or a random-effects model. For the Hausman test, the test result is insignificant (p-value 0,7297 is bigger than 5%). Therefore, H0 is accepted and H1 is rejected. The Lagrange Multiplier (LM) test is a general principle for testing hypotheses about parameters in a likelihood framework. The hypothesis under test is expressed as one or more constraints on the values of parameters. The Lagrange Multiplier test result is insignificant (p-value 0,2535 is bigger than 5%) and follows a common effect. Therefore, it can be concluded that the research model used is the Common Effect Model (CEM).

Furthermore, the correlation analysis is performed, as shown in Table 5, to observe the relationship between independent variables. This aims to confirm that all explanatory variables have a low correlation to each other, in other words, this is an effort to avoid multicollinearity problems.

Table 5: Correlation Matrix of Independent Variables

Information: LNSize is natural logarithm from bank size used is main capital; ROA is Return on Asset; NOM is Net Operating Margin; NPF is Non-Performing Financing; FDR is Financing to Deposit Ratio; CAR is Capital Adequacy Ratio; INFL is annual inflation value; GDP is Gross Domestic Product; Rate is rupiah exchange value toward American dollar.​​​​​​​

Research result shows that bank size (LnSize) has a positive influence on Islamic bank cost efficiency with a significance level of 0.05 (α=5%). This indicates that the presence of capital is a factor that can increase bank cost efficiency, and policy by the government about the minimum limit of main capital enforcement felt appropriate. Another factor that affects cost efficiency is Capital Adequacy Ratio (CAR) value which is significant at α=10%. Even though the significant level is quite high, this ratio is more appropriate in reflecting solvency risk. This finding also depicts that bank capital adequacy becomes very important to support operational activities, therefore, costs can be reduced. The high level of Islamic bank obedience in fulfilling capital adequacy ratio has a positive influence in increasing bank efficiency.

The negative relation between Non-Performing Financing (NPF) and cost efficiency with a significance level at α=1% shows an inversely proportional relation. It is assumed that the bank does not limit expenditure for credit analysis, therefore the bank is not efficient in terms of its cost, or other words, a higher cost ratio in Islamic banks will disrupt bank operational activities. The Financing to Deposit Ratio (FDR) has a positive influence with a significance level at α=1%, which means the higher funding distribution, the better the Islamic banks’ performance, and hence, they will become more efficient.

Other variables such as ROA, NOM, inflation, GDP, and Rate assessed do not have a significant influence on the enhancement of Islamic bank cost efficiency in Indonesia during the research period. However, it was found that the factors that affect cost efficiency during the research period of 2011-2019 are LnSize, CAR, NPF, and FDR.

Table 6: Common Effect Model (CEM) Regression

Information: LNSize is natural logarithm from bank size used is main capital; ROA is Return on Asset; NOM is Net Operating Margin; NPF is Non-Performing Financing; FDR is Financing to Deposit Ratio; CAR is Capital Adequacy Ratio; INFL is inflation annual value; GDP is Gross Domestic Product; Rate is rupiah exchange value towards American dollar

*significant α=1%; ** significant α=5%; ***significant α=10%

5. Conclusion

Research about Islamic bank cost efficiency in Indonesia has not been done much. This research states that from 11 research sample banks, only 2 banks had an efficiency value that was almost close to being a fully efficient firm, while 9 others still must improve their input or output variables (operational activities). Meanwhile, a regression analysis is done for one of two purposes: In order to predict the value of the dependent variable for individuals for whom some information concerning the explanatory variables is available, or in order to estimate the effect of some explanatory variable on the dependent variable. The regression test is done to observe the variable/factor that affects bank cost efficiency. From the research results, it was seen that bank size, CAR, NPF, and FDR are the most influential factors on Islamic bank cost efficiency in Indonesia. This research result gives several implications to policy-makers and regulatory authorities to continually supervise Islamic bank activities especially in maintaining the bank’s main capital within the determined limit. Besides, the government also needs to maintain macroeconomic stability so that Islamic banks are maintained. Monitoring and surveillance of the level of banking efficiency can be carried out with a focus on the determinants of the influencing factors (positive and negative) on Islamic bank efficiency. The level of bank efficiency can be boosted by increased competition in the industry so that further research can add competition variables. Moreover, this research also gives implications for Islamic bank agents to be able to manage bank operational activity more efficiently, increase capital size, manage profitability, rentability, and bank solvency.