Un/desired impact of capital buffers: Evidence from Indonesian bank profitability and risk-taking

Figures & data

Figure 1. Average capital buffer in Indonesia, source: OJK, Indonesia Financial Services Authority.

Table 1. Descriptive statistics of the variables

Variable	Measurements	Mean	Std.Dev.	Min	Max
ROA	Net income/total assets	0.019	0.018	−0.159	0.074
ROE	Net income/total equity	0.121	0.122	−0.890	0.484
NIM	(Interest income – interest expense)/avg. interest-earning assets	0.059	0.025	−0.063	0.194
RWATA	Risk-weighted assets/total assets	0.677	0.142	0.009	1.457
NPL	Impaired loans/total loans	0.013	0.012	0.001	0.099
Z-score	(ROA + equity/assets)/standard dev. ROA	9.004	5.243	0.046	52.771
Buffer	(Tier 1 + Tier 2 over RWA) − a minimum capital requirement of 8%	0.160	0.157	0.016	1.734
Size	The logarithm of total assets	16.529	1.731	11.981	21.075
Loans	Total loans/total earning assets	0.663	0.107	0.207	0.884
Liquidity	Total liquid assets/total assets	0.254	0.040	0.185	0.364
Listed	A dummy that equals 1 if the bank is listed in a stock market, and 0 otherwise	0.380	0.486	0.000	1.000
GDP	Annual GDP growth rate	4.727	2.112	−2.069	6.224
Inflation	Annual inflation growth rate	4.484	1.410	1.921	6.413

Note: The table shows the list of variables and brief descriptions with summary statistics of them.

Table 2. Partial correlation matrix

Variable	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)
ROA	1
ROE	0.857***	1
NIM	0.432***	0.510***	1
RWATA	−0.031**	−0.217***	−0.250***	1
NPL	−0.383***	−0.403***	−0.196***	0.080**	1
Z-score	0.190***	−0.043*	0.041**	0.084***	−0.104***	1
Buffer	−0.140***	−0.298***	0.022*	−0.010*	−0.063**	0.653***	1
Size	0.131***	0.174***	−0.164***	0.130***	−0.054**	−0.203***	−0.295***	1
Loans	0.073**	0.061*	0.137***	0.370***	0.063**	−0.284***	−0.297***	−0.022**	1
Liquidity	0.083***	0.031*	−0.174***	0.135***	−0.049**	−0.033**	−0.125***	0.225***	−0.031**	1
Listed	−0.229***	−0.176***	−0.207***	0.085***	0.193***	−0.166***	−0.131***	0.386***	−0.016	−0.097***	1
GDP	−0.198***	−0.225***	−0.177***	−0.035**	0.052**	0.042**	0.036**	0.241***	0.043**	0.008	0.070**	1
Inflation	0.203***	0.246***	0.124***	0.068**	−0.122***	−0.047**	−0.093***	−0.143***	0.103***	0.051**	−0.055**	−0.382***	1

Note: ***,**,* represent statistical significance at 1%, 5%, and 10%, respectively.

Table 3. Results for the overall sample of the profitability and risk model

	Profitability			Risk
	ROA	ROE	NIM	RWATA	NPL	Z-score
Risk	−0.0322**	−0.2366*	−0.0343**
	(.0149)	(.1312)	(.0157)
Profit				−0.911**	−0.183**	24.731
				(.376)	(0.075)	(27.319)
Buffer	−0.0498***	−0.4404***	−0.0143	−0.161***	−0.024***	13.529***
	(.0159)	(.1484)	(.0137)	(.059)	(.009)	(3.608)
Size	−0.0036**	−0.0228*	−0.0017	−0.009	0.002	−1.395***
	(.0017)	(.0128)	(.0021)	(.015)	(.001)	(.528)
Loans	0.0119	0.0485	0.0867***	0.516***	0.007	2.284
	(.0101)	(.0883)	(.0324)	(.078)	(.009)	(3.175)
Liquidity	−0.0499*	−0.3114*	−0.0618	0.136	−0.006	−1.591
	(.0270)	(.1646)	(.0637)	(.258)	(.015)	(7.406)
Listed	−0.0112	−0.0143	−0.0010	−0.040	0.009**	1.103
	(.0129)	(.0793)	(.0082)	(.062)	(.004)	(1.157)
GDP	−0.0009	−0.0924*	−0.0253***	0.032	−0.006	1.759**
	(.0078)	(.0559)	(.0086)	(.046)	(.004)	(.891)
Inflation	0.0004	−0.0016	−0.0013**	0.003	0.0001	−0.093
	(.0004)	(.0030)	(.0005)	(.002)	(.0002)	(.079)
L1. Profit	0.2849***	0.4287***	0.7258***
	(.0993)	(.0967)	(.1365)
L1. Risk				−0.0701	0.319***	0.078
				(.087)	(.075)	(.117)
Cons	0.1703	3.5039**	0.7735***	−0.463	0.154	−19.688
	(.1967)	(1.4776)	(.2338)	(1.110)	(.114)	(18.960)
Obs.	891			891
AR 2 p-value	0.617	0.327	0.374	0.573	0.411	0.163
Hansen J-test	0.243	0.058	0.060	0.066	0.469	0.092

Note: Table 3 used a two-step system-GMM method to measure the impact of capital buffers on banks’ profitability and risk. Banks’ capital buffer is a ratio of risk-based capital to risk-weighted assets minus 8%, and all other variables are defined in Table 1. Three varying measures are considered for each model specified in Eq. (1)(1) $Profi{t}_{it}={\alpha }_0+{\alpha }_{1}Ris{k}_{it}+{\alpha }_{2}Buffe{r}_{it}+{\alpha }_{3}Profi{t}_{i,t-1}+\alpha {\sum }^{Y_i}+{\epsilon }_{it}$(1) and (2(2) $Ris{k}_{it}={\beta }_0+{\beta }_{1}Profi{t}_{it}+{\beta }_{2}Buffe{r}_{it}+{\beta }_{3}Ris{k}_{i,t-1}+\beta {\sum }^{{\mathrm{X}}_i}+{\mu }_{it}$(2) ). AR (2) is the p-value for the test of second-order autocorrelation, the null hypothesis of the serial correlation test is that the errors exhibit no second-order serial correlation. Hansen’s J-test stands for the p-value of Hansen’s J diagnostic test for instrument validity. The null hypothesis of the Hansen test is that the instruments used are not correlated with residuals (over-identifying restrictions). Robust standard errors are reported in parentheses. ***, **, and * denote statistical significance at 1%, 5% and 10% levels, respectively.

Table 4. Estimation results for subsamples based on bank ownership—a robustness check

	Private’s bank			Government’s bank
	ROA	ROE	NIM	ROA	ROE	NIM
Risk	−0.0487***	−0.2853***	−0.0159	−0.0005	−0.1237	−0.0249
	(.0151)	(.1104)	(.0202)	(.0126)	(.1453)	(.0202)
Buffer	−0.0459***	−0.4056**	−0.0112	−0.1116	−0.6305***	0.0408
	(.0159)	(.1827)	(.0137)	(.0882)	(.1446)	(.0454)
Size	−0.0048***	−0.0211	−0.0009	0.0001	0.0046	−0.0223***
	(.0016)	(.0128)	(.0024)	(.0040)	(.0213)	(.0060)
Loans	0.0191	0.0090	0.0591**	0.0049	0.0337	0.0913*
	(.0117)	(.1387)	(.0252)	(.0144)	(.1481)	(0913)
Liquidity	−0.0318*	−0.2687	−0.0097	−0.2638***	−0.7610	−0.0624
	(.0184)	(.1671)	(.0809)	(.0926)	(.7501)	(.1578)
Listed	−0.0117	−0.0681	−0.0065	0.0012	−0.1005	−0.0120***
	(.0166)	(.0993)	(.0108)	(.0108)	(.1971)	(.0046)
GDP	0.0146*	0.0400*	−0.0128	−0.0246**	−0.2670***	−0.0111
	(.0079)	(.0656)	(.0101)	(.0119)	(.0571)	(.0197)
Inflation	0.0010**	0.0056*	−0.0009**	−0.0005	−0.0037	−0.0015
	(.0004)	(.0033)	(.0004)	(.0004)	(.0031)	(.0010)
L1. Profit	0.2419*	0.3512***	0.6015**	0.1562	0.3453***	0.2240
	(.1332)	(.1154)	(.2601)	(.2456)	(.0902)	(.1444)
Cons	−0.2470	−0.1569	0.3781	0.9637***	8.461***	0.7625*
	(.1999)	(1.6574)	(.2930)	(.3478)	(1.505)	(.4504)
Obs.	561			330
AR 2 p-value	0.418	0.151	0.410	0.674	0.608	0.586
Hansen J-test	0.627	0.268	0.216	0.703	0.735	0.563

Note: Table 4 used a two-step system GMM method to measure the impact of capital buffers on banks’ profitability. Banks’ capital buffer is a ratio of risk-based capital to risk-weighted assets minus 8%, and all other variables are defined in Table 1. Three varying measures are considered for the model specified in Eq. (1)(1) $Profi{t}_{it}={\alpha }_0+{\alpha }_{1}Ris{k}_{it}+{\alpha }_{2}Buffe{r}_{it}+{\alpha }_{3}Profi{t}_{i,t-1}+\alpha {\sum }^{Y_i}+{\epsilon }_{it}$(1) . AR (2) is the p-value for the test of second-order autocorrelation, and the null hypothesis of the serial correlation test is that the errors exhibit no second-order serial correlation. Hansen’s J-test stands for the p-value of Hansen’s J diagnostic test for instrument validity. The null hypothesis of the Hansen test is that the instruments used are not correlated with residuals (over-identifying restrictions). Robust standard errors are reported in parentheses. ***, **, and * denote statistical significance at 1%, 5% and 10% levels, respectively.

Table 5. Estimation results for subsamples based on bank ownership—a robustness check

	Private’s bank			Government’s bank
	RWATA	NPL	Z-score	RWATA	NPL	Z-score
Profit	−1.1064***	−0.2062*	40.282	−0.2080	−0.2096	15.713
	(.3498)	(.1176)	(33.969)	(1.235)	(.1390)	(15.237)
Buffer	−0.1144**	−0.0215*	12.217***	−0.2660	−0.0496*	26.110***
	(.0581)	(.0111)	(4.379)	(.2298)	(.0286)	(2.075)
Size	−0.0193	0.0022	−1.827**	−0.0398	0.0038	−1.576***
	(.0212)	(.0024)	(.8424)	(.0311)	(.0025)	(.198)
Loans	0.5697***	−0.0002	−1.888	0.5554***	−0.0024	8.469***
	(.1131)	(.0119)	(4.984)	(.0764)	(.0126)	(1.042)
Liquidity	0.1050	−0.0101	2.084	0.8259	−0.0695	0.621
	(.3004)	(.0234)	(9.944)	(.5980)	(.0565)	(8.266)
Listed	−0.0242	0.0097	1.562	0.0078	0.0160***	0.515
	(.0973)	(.0062)	(1.625)	(.0374)	(.0030)	(0.625)
GDP	0.0952	−0.0100	2.686	0.0562	−0.0105	2.144***
	(.0670)	(.0078)	(1.747)	(.1093)	(.0116)	(.568)
Inflation	0.0036	−0.0001	−0.108	0.0025	0.0002	−0.025
	(.0031)	(.0005)	(.108)	(.0040)	(.0002)	(.039)
L1. Risk	−0.0388	0.2777***	0.128	−0.0060	0.3981***	0.093
	(.1043)	(.1001)	(.187)	(.1608)	(.1259)	(.0795)
Cons	0.1703	0.2612	−39.451	−1.431	0.3081	−35.136
	(.1967)	(.1910)	(33.917)	(2.993)	(.2936)	(19.736)
Obs.	561			330
AR 2 p-value	0.188	0.640	0.194	0.128	0.366	0.305
Hansen J-test	0.190	0.157	0.583	0.438	0.870	0.807

Note: Table 5 used a two-step system GMM method to measure the impact of capital buffers on banks’ risk. Banks’ capital buffer is a ratio of risk-based capital to risk-weighted assets minus 8%, and all other variables are defined in Table 1. Three varying measures are considered for the model specified in Eq. (2)(2) $Ris{k}_{it}={\beta }_0+{\beta }_{1}Profi{t}_{it}+{\beta }_{2}Buffe{r}_{it}+{\beta }_{3}Ris{k}_{i,t-1}+\beta {\sum }^{{\mathrm{X}}_i}+{\mu }_{it}$(2) . AR (2) is the p-value for the test of second-order autocorrelation, and the null hypothesis of the serial correlation test is that the errors exhibit no second-order serial correlation. Hansen’s J-test stands for the p-value of Hansen’s J diagnostic test for instrument validity. The null hypothesis of the Hansen test is that the instruments used are not correlated with residuals (over-identifying restrictions). Robust standard errors are reported in parentheses. ***, **, and * denote statistical significance at 1%, 5% and 10% levels, respectively.

Table 6. Estimation results for the overall sample using different estimation methods—a robustness check

	Profitability				Risk
	ROA	ROE		NIM	RWATA		NPL	Z-score
Risk	−0.0192***	−0.2602***		−0.0137***
	(.0042)	(.0281)		(.0048)
Profit					−1.226***		−0.0836***	7.371
					(.1826)		(.0204)	(5.320)
Buffer	−0.0014***	−0.0106***		0.0032***	0.0785***		−0.0035***	5.173***
	(.0005)	(.0011)		(.0003)	(.0051)		(.0002)	(.207)
Size	0.0016***	0.0138***		0.0003	0.0093***		−0.0004**	−0.121**
	(.0002)	(.0018)		(.0002)	(.0021)		(.0002)	(.059)
Loans	0.0156***	0.1449***		0.0197***	0.2864***		0.0071**	−2.616***
	(.0045)	(.0287)		(.0047)	(.0297)		(.0029)	(.830)
Liquidity	−0.0127	−0.0452		−0.0226**	0.1718**		0.0044	5.461**
	(.0111)	(.0693)		(.0115)	(.0761)		(.0080)	(2.164)
Listed	−0.0042***	−0.0301***		−0.0029***	−0.0097		−0.0024***	0.242
	(.0009)	(.0060)		(.0009)	(.0070)		(.0007)	(.195)
GDP	−0.0178***	−0.2147***		−0.0227***	0.0643		0.0079*	0.099
	(.0057)	(.0362)		(.0058)	(.0415)		(.0043)	(1.114)
Inflation	0.0003	0.0003		−0.0002	0.0145***		0.0002	−0.015
	(.0004)	(.0026)		(.0004)	(.0029)		(.0003)	(.080)
L1. Profit	0.6756***	0.5696***		0.7716***
	(.0203)	(.0203)		(.0192)
L1. Risk					0.6019***		0.4921***	0.765***
					(.0207)		(.0278)	(.018)
Cons	0.4854***	5.8690***		0.6509***	−2.0689*		−0.2154*	−2.9846
	(.1600)	(1.0037)		(.1636)	(1.1497)		(.1201)	(31.61)
Obs.	891				891
R-Square	0.548	0.583		0.730	0.573		0.415	0.784
Chi2	1559.86	1765.81		2496.01	1553.30		858.14	3790.60
Chi2 P-value	0.0000	0.0000		0.0000	0.0000		0.0000	0.0000
Harvey LM Test	0.582	0.526		0.804	0.384		0.295	0.694
Ovr. System autocorrelation- Harvey LM test= 7.278 P-value= 0.4004			Contemporaneous correlation – Breusch-Pagan LM test= 1049.08 Degrees of freedom= 21.0 P-value= 0.000			Order & rank condition – System is identified

Note: The estimation is performed using three-stage least squares (3SLS) for the models specified in Eq. (1)(1) $Profi{t}_{it}={\alpha }_0+{\alpha }_{1}Ris{k}_{it}+{\alpha }_{2}Buffe{r}_{it}+{\alpha }_{3}Profi{t}_{i,t-1}+\alpha {\sum }^{Y_i}+{\epsilon }_{it}$(1) and (2(2) $Ris{k}_{it}={\beta }_0+{\beta }_{1}Profi{t}_{it}+{\beta }_{2}Buffe{r}_{it}+{\beta }_{3}Ris{k}_{i,t-1}+\beta {\sum }^{{\mathrm{X}}_i}+{\mu }_{it}$(2) ) with three varying measures considered for each model. Harvey LM is the P-value of the test for serial autocorrelation, with the null hypothesis that the errors are uncorrelated over time or across observations. The Breusch-Pagan LM test is used to check for the presence of contemporaneous correlation (cross-sectional dependency), with the null hypothesis being that evidence of cross-sectional dependence exists within the panel data. The order and rank condition help to determine whether the system of equations is identified and whether the estimated parameters are reliable. Robust standard errors are reported in parentheses. ***, **, and * denote statistical significance at 1%, 5% and 10% levels, respectively.

Data availability statement

The data that support the findings of this study are available from the corresponding author, Hendrawan, upon reasonable request.