Determinants of capital buffer in Islamic banks: the lesson from Indonesia

Abstract

Our research explores the determinants of capital buffers in Indonesian Islamic banking, which is ranked 10th in the world. The explanatory variables consist of competition, profit loss sharing (PLS) financing, and bank-specific variables such as bank size, stability, total financing, efficiency and financing risk. All Islamic banks, 34 banks, are selected for the purpose of this study, covering from 2014 to 2020 and using quarterly data. The dynamic panel regression is employed and estimated using the bias-corrected least-squares dummy variable (LSDVC) that generates a more robust estimator than the widely used GMM method for the case of the small cross-section unit with the unbalanced panel. The findings evidently show that low competition discourages banks from having a high capital buffer. Stability, bank size and high financing default encourage banks to build up a high capital buffer. By contrast, PLS financing, total financing and operating inefficiency lower capital buffers. Interestingly, being a full-fledged Islamic bank gives more benefits than an Islamic bank window in building up a capital buffer. Our study has important implications for the Indonesian Financial Service Authority as a policymaker. Converting an Islamic bank window to a full-fledged Islamic bank through a spin-off policy must be implemented immediately according to Indonesian Islamic bank law no 23/2008 to encourage strong and sound Islamic banking in Indonesia.

Keywords:

• Islamic bank
• capital buffer
• competition
• bank-specific variable
• Indonesia

JEL:

• G 20
• G21
• G3

REVIEWING EDITOR:

• Len Tiu Wright, De Montfort University Faculty of Business and Law, United Kingdom

SUBJECTS:

• Economics
• Finance
• Business, Management and Accounting

1. Introduction

The soundness and stability of the banking system became very important after the global financial crisis in 2007–2010. Bank capital is an important factor that plays an important role in banking soundness and stability. The global financial crisis implies the importance of capital adequacy for banks. The regulator imposes a minimum capital requirement on the banking industry. If a bank fails to meet the minimum requirements, it will face sanctions set by the regulations of the country where the bank is located (Valencia & Bolaños, 2018).

Indeed, bank regulators have also recognized the importance of capital buffers after the global financial crisis. The Basel Committee issued regulatory standards relating to capital buffers to support a strong and resilient global banking system, especially in addressing liquidity problems. Capital buffer is capital that exceeds the minimum requirement (Jokipii & Milne, 2008; Shim, 2013). Basel III involves two capital buffers, namely the countercyclical buffer and the capital conservation buffer. The countercyclical buffer is aimed to ensure that the movements of the capital buffer occur well throughout the business cycle. It is set to vary by the percentage of risk-weighted assets (RWA). In contrast, the capital conservation buffer is formed outside of periods of the financial crisis (Abbas et al., 2021).

The importance of capital buffers for bank stability has encouraged many studies to explore the determinants of capital buffers in the banking industry. However, most of the studies were conducted for conventional banks. Some factors affect the capital buffer in conventional banks, namely market competition (Mateev et al., 2021), bank stability (Daoud & Kammoun, 2020), bank size (Wang, 2017), operating efficiency (Alsharif, 2021), financing risk (Liu, 2015), bank liquidity (Abbas et al., 2021) and the business cycle (Valencia & Bolaños, 2018). While research on the capital buffer for Islamic banking is still rare. Daher et al. (2015) and (Bougatef & Korbi, 2019) focus on the role of risk exposure in determining capital buffer. By contrast, Maharani and Setiyono (2018), Maatoug et al. (2019) and Atikah et al. (2023) focus on the role of the business cycle in determining capital buffers.

The limited research on capital buffers in Islamic banks opens opportunities to enhance previous studies while looking for knowledge gaps. One important aspect that can influence the capital buffer is the type of Islamic bank financing. Islamic bank financing is grouped into two, namely profit-loss sharing (PLS) financing and non-PLS financing. This first type of financing is the core of Islamic banks which makes Islamic banks different from conventional banks (Ibrahim & Alam, 2018). However, PLS financing is a type of financing that causes high financing risks due to moral hazard and asymmetric information (Azmat et al., 2015; Ibrahim & Alam, 2018). The high risk of financing certainly reduces profits and further reduces the capital buffer of Islamic banks. However, previous research has not addressed the role of PLS financing on capital buffers in Islamic banking.

The study aims to investigate Islamic banks’ capital buffers in Indonesian Islamic banking. Islamic banking in Indonesia is ranked 10th in the world, and PLS financing is higher compared to other countries (Risfandy et al., 2020). Accordingly, it is interesting to investigate the determinant of capital buffers in Islamic banking in Indonesia considering that PLS financing is relatively higher than other countries. Regarding bank capital, the Indonesian Financial Services Authority (IFSA), as a regulator in the Indonesian banking industry, has determined the minimum bank capital as measured by the capital adequacy ratio (CAR), and it accounts for 8%. In order for bank capital to be properly maintained and so that banks can anticipate losses due to excessive credit growth or financing, the Financial Services Authority January 2016 issued regulation No. 11/POJK.03/2016, which regulates additional capital as a buffer called Countercyclical Capital Buffer (OJK, 2016). The regulation of the capital buffer is related to the procyclical behavior of bank lending which tends to increase when the economy is booming (Auer et al., 2022).

Some contributions are expected from our study to the empirical Islamic banking literature. First, empirical studies on capital buffers in Islamic banking are limited. Second, we incorporate PLS financing as the main financing contract of Islamic banks in influencing the capital buffer. PLS financing is high risk due to being subject to moral hazard and asymmetric information (Ibrahim & Alam, 2018). This work, to the best of our knowledge, is the initial work addressing this issue of the role of PLS financing on capital buffer. Third, Islamic banking in Indonesia comprises full-fledged Islamic banks and Islamic bank windows. Full-pledged Islamic banks represent large Islamic banks while Islamic bank windows are small Islamic banks. Hence, the behavior of capital buffers between the two Islamic banks is apparently different. It is very interesting, therefore, to explore the determinants of a bank’s capital buffer between full-pledged Islamic banks and Islamic bank windows.

2. Literature review and hypothesis

2.1. Theoretical background

Bank as a financial intermediary is a highly controlled business sector associated with capital structure. The bank capital consists of compulsory and voluntary capital. Compulsory capital is the obligation to provide the minimum capital set by the policymaker. The voluntary capital is the capital requirement decided by banks to run their business according to the theory of a firm’s capital structure (Lepetit et al., 2015). Indeed, every bank usually holds a capital buffer, which is the bank capital that exceeds the capital requirements set by the policymaker.

Capital buffer is a crucial condition for soundness and stability of banks. There are several plausible reasons why banks hold capital buffers (Huang & Xiong, 2015; Atikah et al., 2023). First, banks have a capital buffer in an effort to meet the minimum capital requirements. Capital buffer protects against high adjustment costs that arise because banks violate the minimum capital threshold (Ayuso et al., 2004). Second, a higher capital buffer indicates that the bank has a strong and sound performance (Huang & Xiong, 2015). Third, a higher capital buffer also indicates that banks can overcome unexpected shock problems and reduce the possibility of bankruptcy, especially as the economy experiences a downturn (Nier & Baumann, 2006).

2.2. Literature review

A strand of literature has analyzed the factors that influence capital buffers. Some factors affecting capital buffer are competition (Mateev et al., 2021), bank stability (Daoud & Kammoun, 2020), bank size (Wang, 2017), efficiency (Abbas et al., 2022), loan loss provision (Liu, 2015) and liquidity (Abbas et al., 2021). High market competition reduces intermediation costs, but capital regulation will increase intermediation costs. Thus, a high-competition market can reduce the capital buffer (Mia, 2022). Study by Islam et al. (2020), market competition as measured by the Lerner index has a positive effect on bank capital in Malaysia, Singapore, Indonesia and Thailand. Mateev et al. (2021) indicate that market competition has an impact on bank capital, both conventional banks and Islamic banks. Adesina and Mwamba (2018) found higher capital buffer regulation associated with higher market power.

Bank stability reflects a healthy banking condition, where the bank avoids bankruptcy problems, which is often measured by a Z-score (Čihák & Hesse, 2008; Pambuko et al., 2018). Daoud and Kammoun (2020) obviously claim that better financial stability would strengthen bank capital. Danarsari et al. (2018) found that increasing the capital buffer would increase bank stability Rubio and Yao (2020) said that the optimal countercyclical buffer in the Basel III regulations resulted in a greater risk of financial stability than the accumulation of additional bank capital, so that the implementation of the capital buffer must be more aggressive in a world with low-interest rates.

Total assets represent bank size (Ibrahim et al., 2017). Banks with large assets indicate large banks, and vice versa. Small assets show small banks. Larger banks are more resilient in the face of negative shocks to their capital. This is due to the fact that large banks are able to diversify their investments, not just in one sector, so they can obtain higher returns (Kartika et al., 2022). Large banks also have high stability, making it easier to obtain equity (Octavianus & Fachrudin, 2022). Banks can arrange RWA to increase the capital buffer so larger Islamic banks can generate a high capital buffer.

The core business of the bank as a financial intermediary is financing. However, banks must be able to maintain their liquidity by providing sufficient funds to meet the withdrawal of funds and fulfill financing commitments (Anisa & Sutrisno, 2020). Islamic bank liquidity is measured by the financing-to-deposit ratio (FDR), which is the ratio of financing provided to third-party funds (Priyadi et al., 2021). The higher financing indicates the greater the bank’s ability to obtain profitability because the main income of Islamic banks is from financing. High profits increase the ability of banks to provide a capital buffer (Nurhikmah & Pangestuty, 2020; Abbas et al., 2021). However, if there are many problems with financing, increasing financing requires additional funds taken from the capital buffer (Bitar et al., 2018).

Islamic banks are required to work efficiently by reducing operational costs. The more efficient a bank is, the higher the opportunity for getting a high profit. Consequently, high profit can be used to build up the capital buffer so that efficiency will boost the capital buffer. However, if the bank operates inefficiently, it will result in a decrease in profits and can even suffer losses. Bank losses must be covered with capital because one of the functions of capital is to buffer losses incurred. As a result, the inefficiency of bank operations reduces the capital buffer (Anisa & Sutrisno, 2020; Abbas et al., 2022).

One of the main income sources in Islamic banks comes from PLS financing, where income depends on customer profits. Therefore, Islamic banks’ financing is a high risk that comes from PLS financing (Widarjono et al., 2022a). In general, the financing default of Islamic banks is higher than conventional banks in Indonesia (Sukmana & Febriyati, 2016), Malaysia (Wasiaturrahma et al., 2020) and Pakistan (Aman et al., 2016). The high financing risk is also due to agency problems. Bank owners want high profits. Consequently, bank management must meet these demands by selecting less selective in providing financing, which results in high financing risk. Bank Indonesia, through regulation No: 14/15/PBI/2012, requires banks to set aside financing losses in the form of a financing loss provision (FLP). FLP is very beneficial for banks because it can be a solution in dealing with risks to become prudential banking. The high risk of financing causes banks to increase their capital buffer to anticipate high financing defaults (Liu, 2015; Wang, 2017).

2.3. Hypothesis development

Capital buffer is the holding of capital above the regulatory threshold suggested by policymakers (Jokipii & Milne, 2008; Shim, 2013). The capital buffer arises because banks are imprudent in providing credit or financing, so they have the potential to experience losses due to many financing payments that are not smooth. The losses incurred must be covered by bank capital and if the bank capital is relatively small, it will have an impact on bank bankruptcy. The Basel Committee for Bank Supervision in December 2010 issued an official regulation aimed at creating a resilient global banking system, namely the regulation on capital buffers, which is devoted to addressing liquidity problems (Atikah). The capital buffer in Basel III regulates a countercyclical buffer and the capital conservation buffer to encounter various risks in the banking system (Abbas et al., 2021).

2.3.1. Market competition and capital buffer

It is undeniable that the competition in the banking industry is very high. Allen et al. (2011) suggest that high competition encourages banks to build up their capital, as this condition lures creditworthy borrowers and prompts a commitment to monitoring. High market competition has a positive effect on bank capital in Malaysia, Singapore, Indonesia, and Thailand (Islam et al., 2020). Adesina and Mwamba (2018) found that high market competition increases the capital buffer. High market competition, however, reduces the capital buffer in the banking case in Bangladesh (Mia, 2022). Mateev et al. (2021) said that market competition increases bank capital in both conventional banks and Islamic banks.

H₁

Market competition negatively affects the capital buffer

2.3.2. PLS financing and capital buffer

Islamic bank financing comprises PLS and non-PLS financing (Čihák & Hesse, 2010; Ibrahim & Alam, 2018). PLS financing consists of joint ventures (Musyarkah) and risk-sharing (Mudharabah), while non-PLS financing consists of a margin scheme (Murabahah), contracts scheme (Istisna and Salam) and leasing scheme (Ijarah). PLS financing faces a high risk due to the problem of moral hazard and asymmetric information, so they prefer to disburse their funds in Murabahah financing (Meslier et al., 2020). Some studies documented that PLS reduces profits (Risfandy, 2018; Sutrisno & Widarjono, 2022). The decline in profits subsequently reduces retained earnings and further reduces the ability to increase the capital buffer.

H₂

PLS financing has a negative impact on the capital buffer

2.3.3. Bank stability and capital buffer

Every bank is required to be able to maintain banking stability in various economic conditions. Bank stability reflects a healthy bank condition and can avoid bankruptcy risk (Čihák & Hesse, 2008; Pambuko et al., 2018). According to the pecking order theory, stable banks prefer internal funds to enhance their capital (Bitar et al., 2018). Therefore, one predicts bank stability boots capital buffer. Daoud and Kammoun (2020) found that better financial stability will strengthen bank capital as well as a capital buffer.

H₃

Bank stability positively influences the capital buffer

2.3.4. Bank size and capital buffer

Bank size is measured by total assets. Large banks are less likely to face negative shocks to their capital (Kartika et al., 2022). This is due to the fact that large banks are able to diversify their investments not only in one area but also in various investment fields, according to portfolio theory. Banks can determine RWA as a measure of the CAR (Rubio & Yao, 2020). The decrease in RWA has an impact on increasing the capital buffer so that the larger bank can boost the capital buffer.

H₄

Bank size positively influences the capital buffer

2.3.5. Financing and capital buffer

Financing is the main activity of Islamic banks. If the quality of financing is good, then higher financing provides opportunities to enhance profitability, thereby increasing the capital buffer (Atici & Gursoy, 2012; Nurhikmah & Pangestuty, 2020). However, if the quality of financing is bad, high financing causes high financing defaults, which will reduce profitability (Kartika et al., 2022). As a result, financing has a negative effect on the capital buffer because increasing credit requires additional funds taken from the capital buffer (Liu, 2015).

H₅

Financing affects the capital buffer

2.3.6. Efficiency and capital buffer

In order to encourage high profits, banks must be able to operate efficiently. Bank efficiency is measured by the cost-to-income ratio (CIR) (Trinugroho et al., 2018). The higher the CIR is, the more inefficient bank operations, and conversely, a low CIR indicates low operating efficiency (Banna et al., 2017). The high CIR will reduce profits and further reduce the capital buffer. Thus, CIR has a negative effect on the capital buffer (Hisan et al., 2020; Nurhikmah & Pangestuty, 2020; Anisa & Sutrisno, 2020).

H₆

Inefficiency is negatively linked to the capital buffer

2.3.7. Financing risk and capital buffer

Islamic bank financing risk is often measured by non-performing financing (NPF) (Muhammad et al., 2020; Priyadi et al., 2021) and FLP (Ozili & Outa, 2017). In this study, financing risk is measured by NPF The NPF is the total financing default divided by the total financing (Widarjono et al., 2020). The high NPF causes an increase in the capital buffer in an effort to anticipate financing defaults (Anisa & Sutrisno, 2020).

H₇

Non-performing financing positively affects the capital buffer

2.3.8. COVID-19 and capital buffer

Economic downturns, such as COVID-19 worsen macroeconomic conditions and further affect banking conditions. COVID-19 caused Indonesia’s economic growth to decline and even negative economic growth occurred in the second quarter of 2020. Low economic growth caused higher financing risk (Alabbad & Schertler, 2022; El-Chaarani, 2023). Several studies documented that capital buffers are negatively related to the business cycle in the economy (Chen & Chen, 2014, Valencia & Bolaños, 2018; Maatoug et al., 2019; Atikah et al., 2023). As a result, banks must provide higher capital buffers to anticipate COVID-19.

H₈

COVID-19 positively affects the capital buffer

3. Data and research method

3.1. Data

As a predominantly Muslim country, Indonesia started with Islamic banking in 1992 through banking law No. 7/1992. The development of Islamic banks is growing rapidly when the government specifically regulates Islamic banking through Islamic Banking Law No. 23/2008 (Widarjono et al., 2022b). As of 2020, Islamic banks are 34, encompassing 12 full-fledged Islamic banks and 22 Islamic bank windows. Our study examines all Islamic banks. The period study is seven years, using quarterly data dan covering from 2014 to 2020. Our data set is an unbalanced panel with 742 observations. All data is retrieved from the IFSA.

3.2. Empirical method

We employ the regression method to empirically investigate the determinant of capital buffer in Indonesian Islamic banks following previous studies such as Daher et al. (2015), Maatoug et al. (2019) and Atikah et al. (2023). Our data set is panel data with an unbalanced panel. We employ dynamic panel data, following the equation: (1) $${\text{Buf}}_{\text{it}}={\delta }_0+{\delta }_1{\text{Buf}}_{\text{it}-1}+{\delta }_2\text{PLS}+{\delta }_3{\text{ALerner}}_{\text{it}}+{\delta }_{4}Z-{\text{score}}_{\text{it}}+{\delta }_5{\text{Lasset}}_{\text{it}}+{\delta }_6{\text{Fin}}_{\text{it}}+{\delta }_7{\text{CIR}}_{\text{it}}+{\delta }_8{\text{NPF}}_{\text{it}}+{\delta }_9{\text{Covid}}_{\text{it}}+e_{\text{it}}$$(1) where Capbuf is capital buffer, ALerner stands for adjusted Lerner index, PLS stands for PLS financing, Z-score is the stability of Islamic bank, the asset is total assets, Fin is the total financing, CIR is the cost-income ratio (CIR), NPF is non-performing financing, Covid is Covid-19 outbreak. Our study expresses total assets in a natural logarithm.

Our dependent variable is the capital buffer. It is the difference between the capital ratio and the minimum requirement set by the regulator (Jokipii & Milne, 2008; Shim, 2013). A capital buffer is formed to consider the instability of the capitalized level of banks under capital regulation.

Our study measures competition using the Lerner Index. We use the Adjusted Lerner index. This adjusted Lerner Index is more robust than the standard Lerner index by relaxing the assumption of profit as well as cost efficiency (Koetter et al., 2012; Tan & Floros, 2018; Risfandy et al., 2022). The Adjusted Lerner index is calculated as follows: (2) $$\begin{array}{l}\mathit{\text{Aspiratio}}{n}_i={\beta }_0+{\beta }_1\mathit{\text{sales role}}\hspace{0.17em}\mathit{\text{focu}}{s}_i+{\beta }_2\mathit{\text{satisfactio}}{n}_i\\ +{\beta }_3\mathit{\text{stres}}{s}_i+{\beta }_4\mathit{\text{performanc}}{e}_i+{\beta }_5\mathit{\text{allocatio}}{n}_i\\ +{\beta }_6\mathit{\text{sales role}}\hspace{0.17em}\mathit{\text{focus correctio}}{n}_i+{\epsilon }_{i2}\end{array}$$(2) where Alerner is adjusted Lerner Index, NI is net income, TC is total income, MC is marginal cost and EA is earning asset.

Our study employs the trans-log cost function with two inputs to derive the marginal cost. The trans-log cost function with two inputs is more applicable to estimate cost function in emerging markets such as Indonesia (Fu et al., 2014; Widarjono et al., 2023) as follows (3) $${\text{TC}}_{\text{it}}={\delta }_0+{\displaystyle \sum }_{k=1}^2{\delta }_1{\text{LM}}_{k,\text{it}}+0.5{\displaystyle \sum }_{k=1}^2{\displaystyle \sum }_{l=1}^2{\sigma }_{\text{kl}}{\text{LM}}_{k,\text{it}}{\text{LM}}_{l,\text{it}}+{\sigma }_1{\text{Lasset}}_{\text{it}}+0.5{\delta }_2{({\text{Lasset}}_{\text{it}})}^2+{\displaystyle \sum }_{k=1}^2{\sigma }_{2k}{\text{LAsset}}_{\text{it}}{\text{LM}}_{k,\text{it}}+{\epsilon }_{\text{it}}$$(3)

Total cost (TC) comprises costs of profit-and-loss sharing and costs of other operational expenses. M1 shows the profit-and-loss sharing expense divided by total customer deposits. M2 indicates the other operating expense divided by total fixed assets. Finally, MC is calculated using the following formula: (4) $${\text{MC}}_{\text{it}}=\left({\sigma }_1+{\sigma }_2{\text{Lasset}}_{\text{it}}+{\displaystyle \sum }_{k=1}^2{\delta }_{2k}{\text{LM}}_{k,\text{it}}\right)\frac{{\text{TC}}_{\text{it}}}{{\text{LAsset}}_{\text{it}}}$$(4)

The lower Lerner index indicates low market power and vice versa. Banks can determine high pricing above their marginal cost as their Lerner index is high. A High Lerner index is linked to low competition in the banking market (Fu et al., 2014).

PLS financing is the ratio of total financing to total assets (Risfandy et al., 2020). The Z-score is commonly recognized in the banking literature as a proxy for bank stability. The Z-score is calculated as the sum of return on asset and CAR divided by the standard deviation of returns (Čihák & Hesse, 2010) as follows: (5) $$Z-\text{score}=\frac{\text{ROA}+\text{CAR}}{\text{SDROA}}$$(5)

Total assets represent bank size. The financing is total financing divided by total assets. CIR indicates operating efficiency. NPF shows financing risk. Covid is the covid-19 outbreak, starting in the second quarter of 2020 and representing the business cycle. Table 1 exhibits the variable definition and predicted sign of the explanatory variable.

Table 1. Variable and expected sign.

Variable	Definition	Predicted sign
Dependent variable
Capital buffer	Difference between the total capital ratio and regulatory minimum requirement
Independent variable
Adjusted Lerner Index	Efficiency-adjusted Lerner index	−
Bank-specific variable
PLS	PLS financing divided by total financing	−
Z-score	Risk-sharing financing divided by the total asset	+
Asset	Total asset	+
Fin	The financing divided by asset	−
CIR	The cost divided by income	−
NPF	The financing defaults divided by the total financing ratio	−
Covid-19	Covid-19 outbreak	+

This study employs panel data. There are two types of panel data regression, encompassing static and dynamic panels. This study uses dynamic panel model estimation, given that the behavior of the capital buffer is persistent over time in financial intermediaries, such as Islamic banks (Daher et al., 2015; Bougatef & Korbi, 2019). The GMM method is commonly employed to estimate the dynamic panel regression, both the difference GMM (Arellano & Bond, 1991) as well as the system GMM (Blundell & Bond, 1998). However, the GMM method produces a consistent estimator if the number of cross-sectional observations is very large (Ibrahim & Law, 2020; Khan et al., 2021). In addition to the GMM method, dynamic panel regression can be estimated using the bias-corrected least-squares dummy variable (LSDVC) method proposed by Bruno (2005a) and Bruno (2005b). The LSDVC method was developed specifically for panel data with a small number of cross-section observations and unbalanced panel data. This study uses the LSDVC method by considering that our data is the unbalanced panel with a small number of cross-section units (34 banks). This method has recently been widely used to estimate Islamic bank performance (Ibrahim et al., 2017; Sunarsih et al., 2022).

4. Results and discussion

Table 2 describes the descriptive statistics of the variables studied. The average capital buffer is 13.04% with a standard deviation of 6.43%. Indonesian Financial Service Authority stipulates that the minimum CAR is 8%. The average value of CAR in this study is 21.26%. IFSA has also categorized a bank’s soundness based on CAR. IFSA has classified banks based on their CAR, namely (1) CAR ≥ 12%; (2) 9% ≤ CAR < 12%; (3) 8% ≤ CAR < 9%; (4) 6% < CAR < 8%; (5) CAR ≤ 6%. Group 1 is a very healthy bank, Group 2 is a healthy bank, Group 3 is a fairly healthy bank, Group 4 is a slightly less healthy bank and Group 5 is an unhealthy bank. Based on the average CAR, Islamic banking in Indonesia performs soundness associated with CAR as well as a capital buffer.

Table 2. Summary statistics.

Variable	Mean	Standard deviation	Minimum	Maximum
Buf	0.1316	0.0660	0.0216	0.8065
Alerner	0.5591	0.5140	−4.6689	1.8998
PLS	0.3819	0.2761	0.0000	1.0000
Z-score	0.3827	0.4239	0.0033	2.8055
Asset	14.0365	20.3620	0.3516	126.9000
Fin	0.7531	0.1658	0.2711	1.2572
CIR	0.8578	0.1513	0.1684	2.1740
NPF	0.0389	0.0414	0.0001	0.4399
COVID	0.1334	0.3403	0.0000	1.0000

The correlation between the independent variables affects the estimator results. A high correlation above 0.85 will produce a less robust estimator. Table 3 displays the partial correlation between the explanatory variables studied. Each partial correlation does not exceed 0.5. This correlation coefficient shows that the model does not contain multicollinearity problems so it produces a robust estimator.

Table 3. Correlation matrix.

	Buf	Alerner	PLS	Zscore	Lasset	Fin	CIR	NPF
Buf	1
Alerner	0.0491	1
PLS	−0.0845	0.0168	1
Zscore	0.3112	−0.1075	0.0813	1
Lasset	−0.2276	0.0414	0.1241	0.1664	1
Fin	−0.0909	−0.2524	−0.0984	−0.0811	0.2736	1
CIR	−0.2250	−0.0102	0.1224	−0.0263	0.2390	0.2679	1
NPF	−0.0698	0.0284	−0.0549	−0.2377	−0.0448	0.1971	0.4350	1
COVID	0.1113	0.0086	0.1086	0.0138	0.0711	−0.1581	−0.0188	−0.0470

4.1. Baseline regression

The first discussion starts with the case of all Islamic banks in Indonesia. The method used is the dynamic regression panel method using the LSDVC method. To guarantee consistent results, this study utilizes all biased correction methods consisting of Arellano-Bond (AB), Blundell-Bond (BB) and Aderson Hisiao (AH) proposed by Bruno (2005b). The computed standard error is calculated using the bootstrap method with 100 replications. The estimation results are shown in Table 4. The lag of the dependent variable, Buf (−1), is positive and significant. These findings remain the same for three biased corrected methods. These results infer that capital buffer is permanent from time to time, implying that dynamic panel data is applicable. These results confirm the existing works of Daher et al. (2015), Ghosh (2017) and Bougatef and Korbi (2019) using the GMM method in estimating dynamic panel regression.

Table 4. Baseline results.

Variable	The LSDVC method
	AH	AB	BB
Buffer (−1)	0.1763***	0.1819***	0.2267***
	(0.0332)	(0.0320)	(0.0342)
Market variable
Alerner	−0.0059*	−0.0058**	−0.0059*
	(0.0039)	(0.0035)	(0.0043)
Bank variable
PLS	−0.0238*	−0.0230**	−0.0241*
	(0.0142)	(0.0127)	(0.0166)
Z-score	0.2135***	0.2120***	0.2124***
	(0.0138)	(0.0125)	(0.0158)
Lasset	0.0179***	0.0173***	0.0164***
	(0.0044)	(0.0039)	(0.0051)
FIN	−0.0210*	−0.0203*	−0.0208
	(0.0144)	(0.0129)	(0.0164)
CIR	−0.1492***	−0.1490***	−0.1473***
	(0.0136)	(0.0122)	(0.0155)
NPF	0.3053***	0.3029***	0.2945***
	(0.0718)	(0.0647)	(0.0825)
External variable
COVID	0.0080**	0.0079**	0.0074**
	(0.0039)	(0.0036)	(0.0044)
No. of observations	742	742	742
No. of Islamic banks	34	34	34

The bootstrap standard errors are presented in parentheses with 100 replications.

***p < 0.01, **p < 0.05 and *p < 0.1.

Our first analysis begins with the variable of market power as measured by the adjusted Lerner Index (Alerner). The high adjusted Lerner index indicates a low competitive market (imperfect competition) and conversely, the low adjusted Lerner index indicates high competition (perfect competition). The coefficient of Alerner is negative and significant for all biased corrected methods. These results show that the more uncompetitive the market is the lower the capital buffer. Conversely, the more competitive the market is, the greater the capital buffer. Banks tend to have a higher capital buffer when operating in a highly competitive market but hold less capital buffer in a low, competitive market because low competition does not require banks to have a high capital buffer in attracting creditworthy customers (Etudaiye-Muhtar & Abdul-Baki, 2021). These results are in line with research from Adesina and Mwamba (2018) and Islam et al. (2020).

Turning to PLS financing, PLS financing as the main financing in Islamic banks negatively affects the capital buffer using the bias-corrected method of AB and AH but is negative and not significant in the case of BB. The high PLS financing reduces the capital buffer. PLS financing is a type of financing that differentiates it from conventional banks. PLS financing is risky financing because of the moral hazard and asymmetric information. Research from Sutrisno and Widarjono (2022) found that PLS financing has a negative effect on Islamic banking profits. The decline in profits will further reduce the capital buffer.

Bank stability, as measured by the Z-score, has a positive effect on the capital buffer, implying that a bank with high stability will increase the capital buffer. These results are consistent with using all bias-corrected models of AB, BB and AH approaches. Every bank management is required to be able to maintain banking stability in various economic conditions. Bank stability reflects a healthy bank condition with a high profit rate and can avoid bankruptcy risk. Bank stability, in turn, can increase capital and capital buffers (Hunjra et al., 2020; Daoud & Kammoun, 2020).

Total assets represent Islamic bank size. Total assets are positively associated with the capital buffer and are consistent with the three bias-corrected methods. Large Islamic banks are banks that are more experienced, more efficient because of economies of scale, are more able to diversify their portfolios, and have an easy channel to capital markets (Bitar et al., 2018; Smaoui et al., 2020). The advantage of large banks is thus able to obtain higher profits compared to small Islamic banks. As a result, large banks can build a higher capital buffer. This study confirms previous research on conventional banking cases in Indonesia (Nurhikmah & Pangestuty 2020), but it is a contrast to the study of Abbas et al. (2023) for banks in emerging economies.

The next independent variable is financing. Financing is negatively related to the capital buffer, suggesting that higher financing reduces the capital buffer. The three bias-corrected methods produce the same finding. Financing is the main activity of Islamic banks to gain profit. Expansive financing, if managed properly, increases profits, but expansive financing reduces profits if it is not properly monitored (Kartika et al., 2022). Islamic banks in Indonesia are less experienced in managing their financing compared to conventional banks. This is indicated by the higher NPF of Islamic banks compared to the non-performing loans (NPL) of conventional banks. For example, the NPF of Islamic banks in the 2015–2020 period was 3.95%, while the NPL of concessional banks was 2.83% (IFSA). This finding confirms the results of a study by Ghosh (2017).

The next bank-specific variable is efficiency as calculated by the CIR. Operational inefficiency negatively affects the capital buffer and is consistent with all bias-corrected methods (AB, BB and AH). The efficiency level of banking in Indonesia is lower compared to conventional banks which is indicated by the more expensive products of Islamic banks compared to conventional bank products (Widarjono & Rafik, 2023). Low efficiency will lead to a decrease in profits and further reduce retained earnings, which cannot be used to build up the high capital buffer (Anisa & Sutrisno, 2020). This study confirms the previous study conducted by Abbas et al. (2022) for Islamic banks in 35 countries.

The next explanatory variable is financing risk as measured by NPF. The NPF coefficient is positive and significant, consistent using each bias-corrected method. These results indicate that a bank with a high NPF also has a high capital buffer. NPF illustrates the low quality of Islamic bank financing portfolios. High financing defaults encourage banks to increase their capital buffer to anticipate impaired financing. Accumulation of capital buffer is to avoid insolvency and to meet regulatory requirements (Shim, 2013). This research is in line with the results of previous studies, such as Saadaoui (2015) and Bougatef and Korbi (2019).

The last explanatory variable is COVID-19. COVID-19 has a positive effect on capital buffers, meaning that the economic downturn has caused banks to increase their capital to anticipate unexpected shocks. This finding is in line with the view that capital buffers are procyclical. Our study supports studies from Chen and Chen (2014), Valencia and Bolaños (2018), Maatoug et al. (2019) and Atikah et al. (2023), but this is contradicted by the findings from Huang and Xiong (2015).

4.2. Further analysis

As for Islamic bank law no 23/2008, there are two types of Islamic banks, consisting of full-fledged Islamic banks as well as the conventional bank that has Islamic business lines (Islamic bank windows). As mentioned before, Islamic bank in Indonesia comprises 22 Islamic bank windows and 12 full-fledged Islamic banks. Both types of banks have different management, including how to manage capital as well as a capital buffer. Accordingly, this article split Islamic banks into two types for further analysis, consisting of Islamic bank windows and full-fledged Islamic banks.

Table 5 shows the findings by type of bank. Our study estimates using all bias-corrected methods to warrant consistent results. We begin with the full-fledged Islamic bank that is presented in the right part of Table 5. The lag of buffer, Buf (−1), is positive and significant using all biased corrected methods, meaning that capital buffer is also persistent for full-fledged Islamic banks. Market power is negatively associated with the capital buffer. Bank stability positively affects capital buffer. Capital buffer is positively associated with bank size. Financing and PLS financing have a negative effect on capital buffers. Inefficiency also reduces capital buffer. Lastly, highly impaired financing reduces capital buffer. These findings are consistent with the case of all Islamic banks.

Table 5. Full-fledged Islamic banks vs. Islamic bank windows.

	Full-fledged Islamic banks			Islamic bank windows
Variable	AH	AB	BB	AH	AB	BB
Buffer (−1)	0.3827***	0.4127***	0.4790***	0.0056	0.0023	0.0507*
	(0.0565)	(0.0402)	(0.0442)	(0.0336)	(0.0295)	(0.0391)
Market variable
Alerner	−0.0091	−0.0088***	−0.0082**	−0.0045	−0.0045	−0.0048
	(0.0084)	(0.0032)	(0.0043)	(0.0055)	(0.0046)	(0.0071)
Bank variable
PLS	−0.0720	−0.0683***	−0.0786***	−0.0563***	−0.0555***	−0.0549***
	(0.0616)	(0.0241)	(0.0336)	(0.0157)	(0.0130)	(0.0205)
Z-score	0.1482***	0.1444***	0.1464***	0.9451***	0.9483***	0.9547***
	(0.0256)	(0.0101)	(0.0135)	(0.0577)	(0.0479)	(0.0745)
Lasset	0.0250*	0.0230***	0.0206**	−0.0047	−0.0046	−0.0055
	(0.0195)	(0.0073)	(0.0103)	(0.0080)	(0.0067)	(0.0107)
FIN	−0.0538*	−0.0519***	−0.0549**	0.0081	0.0076	0.0068
	(0.0408)	(0.0157)	(0.0214)	(0.0195)	(0.0162)	(0.0253)
CIR	−0.0762**	−0.0755***	−0.0733***	−0.1006***	−0.1003***	−0.0958***
	(0.0358)	(0.0138)	(0.0187)	(0.0213)	(0.0177)	(0.0275)
NPF	0.1733	0.1726***	0.1623**	0.2177**	0.2180***	0.2187*
	(0.1598)	(0.0601)	(0.0826)	(0.1018)	(0.0851)	(0.1336)
External variable
COVID	0.0066	0.0064	0.0058	0.0047	0.0048	0.0043
	(0.0140)	(0.0053)	(0.0071)	(0.0050)	(0.0041)	(0.0063)
No. of Islamic banks	12	12	12	22	22	22
No. of observations	338	338	338	404	404	404

The bootstrap standard errors are presented in parentheses with 100 replications.

***p < 0.01, **p < 0.05 and *p < 0.1.

Turning to Islamic bank windows, we estimate using all bias-corrected methods to produce reliable findings. The lag of the capital buffer is positive and significant for the BB method. This finding also indicates that capital buffer is permanent over time for Islamic bank windows. Bank stability positively influences capital buffer. PLS financing has a negative impact on capital buffers. The inefficient operation also deteriorates the capital buffer. Finally, high NPF encourages Islamic bank windows to build up more capital buffers. The rest of the independent variables are not significant.

Some important findings are interesting by splitting both types of Islamic banks. On average, the assets of full-fledged Islamic banks and Islamic bank windows are IDR 22, 5089 trillion and 6948 trillion, respectively. Therefore, the full-fledged Islamic banks represent large banks and Islamic bank windows indicate small banks. Large Islamic banks can gain economies of scale, likely diversify their portfolios and easily access capital markets (Ibrahim et al., 2017; Smaoui et al., 2020). Full-fledged Islamic banks can capitalize on imperfect markets because they do not need a high capital buffer compared to small banks in low-competition situations. Large banks benefit from their size since bank size encourages capital buffers. The unfavorable impact of inefficiency is also more pronounced in Islamic bank windows than in their counterpart full-fledged Islamic banks. Large Islamic banks can reduce the negative impact of a decrease in the capital buffer due to financing defaults compared to small banks. However, the effect of stability on the capital buffer is stronger for small banks than for large banks. In addition, the unfavorable effect of financing and PLS financing on the capital buffer is smaller on Islamic bank windows than on full-fledged Islamic banks. However, overall, being a full-fledged Islamic bank generates many benefits compared to Islamic bank windows associated with a capital buffer.

4.3. Robustness test

Our study conducts a robustness test to warrant our results for baseline results. We employ different estimation methods using static panel regression by considering that our data panel is small in cross-section units. We employ the pooled least squared (PLS) with robust standard error, fixed effect (FE) and random effect (RE). The results of the static panel are presented in Table 6. The best model for static panel regression is a FE based on the F test, LM test and Hausman test. Low competition with higher Adjusted Lerner index, financing, PLS financing and inefficiency are negatively related to capital buffer. Stability, bank size and high financing defaults positively influence capital buffer. These findings remain unchanged and consistent with the result of dynamic panel regression using the LSDVC approach.

Table 6. Robustness test results.

Variable	PLS	FE	RE
Market variable
Alerner	0.0160***	−0.0082***	−0.0069**
	(0.0051)	(0.0028)	(0.0030)
Bank variable
PLS	−0.0147*	−0.0218**	−0.0113
	(0.0109)	(0.0123)	(0.0123)
Z-score	0.0616***	0.2285***	0.1889***
	(0.0060)	(0.0115)	(0.0111)
Lasset	−0.0141***	0.0208***	0.0123***
	(0.0019)	(0.0040)	(0.0037)
FIN	0.0415***	−0.0285**	−0.0305***
	(0.0148)	(0.0123)	(0.0130)
CIR	−0.0863**	−0.1496***	−0.1497***
	(0.0462)	(0.0123)	(0.0130)
NPF	0.1233	0.3160***	0.4110***
	(0.1606)	(0.0573)	(0.0589)
External variable
COVID	0.0286***	0.0081**	0.0119***
	(0.0066)	(0.0036)	(0.0039)
Cons	0.3596***	−0.1312**	0.0215
	(0.0231)	(0.0606)	(0.0569)
R-squared	0.2441	0.5205	0.5112
No. of Islamic banks	34	34	34
No. of observations	742	742	742
Diagnostic test
F-test	61.8400
LM-test	1440.7500
Hausman	191.0300

The standard errors are presented in parentheses. ***p < 0.01, **p < 0.05 and *p < 0.1.

5. Conclusion

This article explores the determinant of capital buffer in the Indonesian Islamic banking Industry. The explanatory variables consist of market variables and some bank-characteristic variables, comprising PLS financing, stability, bank size, total financing, efficiency and financing risk. Our findings confirm that high competition market requires Islamic banks to build up capital buffer. Stability, bank size and high financing risk clearly support banks to accumulate more capital buffers. Total financing, PLS financing and inefficiency discourage high capital buffers. Capital buffer in Islamic banks is procyclical as well. More importantly, full-fledged Islamic banks give many advantages compared to Islamic bank windows in accumulating capital and capital buffers, for which a high capital buffer is a key factor in reducing bank defaults.

This study has some important findings. First, PLS financing as equity-based financing lowers the bank capital; accordingly, Islamic banks prefer to disburse non-PLS financing compared to PLS financing. Second, economic downturns such as COVID-19 require banks to hold more capital. Although holding a higher capital buffer is likely costly, it can overcome the problem of unexpected shocks during economic downturns such as the COVID-19 outbreak. Third, a full-fledged Islamic bank is better than an Islamic bank window in building up capital buffers. Our result has policy implications for Otoritas Jasa Keuangan (OJK) (Indonesian Financial Service Authority) as a policy maker. OJK must immediately transform Islamic bank windows into full-fledged Islamic banks starting in 2023. According to Islamic bank law no 23/2008, all Islamic bank windows must be converted to a full-fledged Islamic bank through a spin-off policy (Trinugroho et al., 2021). This policy requires that every Islamic bank must add additional equity. The study by IFSA found that the minimum capital for full-fledged Islamic banks is approximately IDR 0.8 trillion to 1.2 trillion.

Public interest statement

Islamic Bank, as a financial intermediary, is very vulnerable to shocks from both internal and external factors. The main cause of the global financial crisis in 1998 was due to the problem of capital adequacy. Therefore, capital adequacy, including capital buffer, is a critical factor that plays an important role in Islamic bank stability. This study explores the determinant of capital buffer in Indonesian Islamic banks which are ranked number 10 in the world. Our study employs market conditions and some bank-specific variables in determining capital buffers. Low competition discourages banks from accumulating capital buffers. Large banks, stability and high financing defaults encourage banks to build up capital buffers. However, PLS financing, total financing and inefficiency lower capital buffers. More interestingly, the high capital buffer is more pronounced for full-fledged Islamic banks than Islamic bank windows.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was funded by the Center for Management Studies, Department of Management, Faculty of Business and Economics, Universitas Islam Indonesia, Yogyakarta, Indonesia, under Grant Number: 59/P/HibahKolaborasi/PPMJur.Mj/VI/2023.

Notes on contributors

Sutrisno Sutrisno

Sutrisno, Sutrisno, Dr. is an associated professor at the Department of Management, Faculty of Business and Economics, Universitas Islam Indonesia (UII). His research interests are in Islamic banking and finance. He has published some articles in Asian Economic and Financial Review and Risks.

Agus Widarjono

Agus Widarjono, Ph.D., is a professor at the Department of Economics, Faculty of Business and Economics, Universitas Islam Indonesia (UII). His research interests are in Islamic banking and finance and Islamic Economics. He has published some articles in the Journal of Islamic Marketing, Cogent Economics and Finance, Asian Economic and Financial Review, Risks, Cogent Business and Management, and ISRA International Journal of Islamic Finance.