Research Data and Results of the Depositor Model
4.3 Analysis and Discussions of the Estimation Result
4.3.3 Discussion
4.3.3.1 The Association of Deposit Growth with Bank Fundamentals
141 Table 4.9 continued
Arellano–Bond test for
AR(1) -3.403 -2.308 -2.584
0.0007*** 0.021*** 0.0098***
Arellano–Bond test for
AR(2) 1.921 1.304 1.243
0.1548 0.1923 0.214
This table presents the results from the two-step Generalized Method of Moments System estimations.
Coefficients and standard errors (in parentheses) are from the second step. The Sargan and Arellano–Bond tests are from the second. The estimation uses quarterly observations over the period 2001–2011. The dependent variable is DEP_IR (deposit interest rate). The independent variables include: CAR, NPL, ROA, OPEX, NIM, LDR, BANK_SIZE (total asset of banks), and TOT_DG (total deposit growth). Control variables for general macroeconomic conditions include: GDP_GR (the growth rate in GDP); INF_RT (inflation rate); and EXC_RT (the annual average of exchange rate IDR/USD scaled in IDR000). All variables are transformed using the natural logarithmic transformation. Three dummy variables are BLA_GR (1 = Quarter I of 2001 to Quarter IV of 2004 and 0 otherwise), LIST_BANK (1=listed bank and o otherwise and FORG_BANK (1 = foreign bank and 0 otherwise).
* Indicate statistical significance at the 10% level (2-tailed)
** Indicate statistical significance at the 5% level (2-tailed)
*** Indicate statistical significance at the 1% level (2-tailed)
142 Regression results of the estimated relationship between deposit and CAMEL ratios are summarized in Table 4.10.
The changes in deposits were found to be strongly correlated with the movement of each of the six CAMEL ratios used in this study, as can be seen in the summary of regression results46. These results indicate that Indonesian depositors have the ability to deduce information from the published financial data and to exert market discipline accordingly.
These results in general support the hypothesis that, to a certain extent, some unsophisticated markets in developing economies are still able to exert market discipline (Levy-Yeyati et al., 2004a).
Table 4.10 Summary of the Results for CAMEL Ratios Variables Sign Blanket guarantee
Period
Limited guarantee Period TOTAL DEPOSIT
GROWTH
CAR + Significant Significant
NPL - Significant Significant*
ROA + Significant Significant*
OPEX - Significant Significant
NIM + Significant Significant
LDR - Significant* Significant*
BANK SIZE + Significant Significant
DEP_IR + Significant Significant
TIME DEPOSIT GROWTH
CAR + Significant Significant
NPL - Significant Significant*
ROA + Not significant Not significant
OPEX - Significant Significant*
NIM + Significant* Significant
LDR - Significant* Significant*
BANK SIZE + Significant Significant
UNINSURED DEPOSIT GROWTH
CAR + N/A Significant
NPL - N/A Not significant
ROA + N/A Significant*
OPEX - N/A Not significant
NIM + N/A Significant*
LDR - N/A Significant*
BANK SIZE + N/A Significant
* The sign is contrary to the expectation
46 Regression results for ROA and NIM for all periods of observation were insignificant, but if the observation period is divided into the full guarantee and the limited guarantee, the results were significant.
143 In order to identify the differences in behaviour between retail and wholesale investors, this study used time deposits and uninsured deposits as signals to identify the sensitivity of wholesale depositors on bank risk. The growth of time deposits is correlated significantly with 5 out of the 6 CAMEL variables. For uninsured deposits47, the deposit growth was only influenced significantly by four out of the six CAMEL variables. These results indicate a weaker market discipline by the holders of time deposits and uninsured deposits.
Under the assumption that a majority of time deposits and uninsured deposits are owned by large individuals and institutional depositors, these results contradict the premise that sophisticated wholesale investors have a much better capacity to impose discipline on financial institutions. Theoretically, wholesale depositors have a higher capacity to access and utilize information to make decisions (Garten, 1986), and they have significant incentives to monitor banks since their funds are not insured (Flannery, 2001). Therefore, the weaker discipline by wholesale depositors indicates the possibility of moral hazard implications due to a FSN, with wholesale depositors taking advantage of the government guarantee for their own benefit. This view concurs with that of Huang and Ratnovski (2011). As stated in the literature, a safety net has potentially lower risk of default, which enables wholesale depositors to invest money in a particular bank. This offers higher return regardless of its risk.
The exploitation of government guarantees by wholesale depositors could also possibly occur due to a condition in which small banks in Indonesia have a tendency to rely on wholesale funds48. This could provide opportunities for wholesale depositors to dictate to bank management for the provision of high interest rates to ensure depositors retain funds in their banks. This evidence, perhaps, could partially explain the phenomenon of the “deposit interest rate wars” that frequently occur in the Indonesian market, particularly in the midst of financial market turbulence (Bank Indonesia, 2009b).
The indicator of a bank capital adequacy (CAR) is one of the main factors that influences the growth of deposits in Indonesia. This concurs with the work of Kameyama et al. (2006). In the Indonesian market, uninsured deposits, time deposits, and total deposits grow faster in banks with a stronger capital base because these banks are considered less risky. The ability
47The growth of uninsured deposit is observed during the limited guarantee period from Quarter I of 2006 to Quarter IV of 2011.
48Table 4.2 shows that small banks in general have a higher proportion of time and uninsured deposit than large banks. Majority of the Indonesian small banks are categorized as regional development banks, non-foreign exchange commercial banks, and joint venture banks.
144 of banks with higher capital buffers to attract greater deposits is a signal of the presence of market discipline (Barajas & Steiner, 2000).
The results presented in Table 4.10 indicate the presence of mixed results on the correlation between NPL and deposits over the periods of blanket guarantee and limited guarantee.
During the blanket guarantee period, the sign on the correlations is negative, which is consistent with existing literature that maintains that if market discipline exists, an increase in distressed assets would have a negative effect on deposit growth (Wu & Bowe, 2012). An increase in non-performing loans would be interpreted as an increase in bank credit risk and a reduction of profit because the bank has to set aside some provision to cover potential losses.
In the Indonesian case, over the period of blanket guarantee, many banks recorded high NPL ratios that were above the central bank regulation of a maximum of 5% during the period immediately following the 1997 financial crisis. One of the main concerns of both bank management and regulators was how to write down these underperforming and nonperforming assets from bank balance sheets. The finding of the current study suggests that Indonesian depositors are also concerned about credit risk and, therefore, exercise market discipline by punishing banks that have a high NPL ratio (Kameyama et al., 2006).
In contrast, the sign on the relationship between NPL and deposit is positive during the limited guarantee period. One possible explanation relates to the condition of high NPL rates during the period of limited guarantee. In this period, the Indonesian banking sector had recovered from the financial crisis and the NPL ratio was not a major issue since most of the banks had complied with the central bank regulation to maintain their NPL under 5% (Bank Indonesia, 2009b). In this period, an increase in the provision of credits might not be viewed simply as a deterioration of bank credits, but could also be seen as the ability of bank management to obtain more deposits to finance new loans, which in turn would generate more income in the future. If this assumption is accurate, this could partially explain the tendency of higher deposit growth for banks with a higher NPL growth during the limited guarantee period.
Regression results of the relationship between the ROA variable and deposit growth suggests that the Indonesian depositors imposed discipline by investing their funds in more profitable banks, particularly during the full guarantee period. This finding is consistent with the hypothesis that a bank with a stronger earning quality indicator would likely be able to attract more deposits (Levy-Yeyati et al., 2004a). Nevertheless, over the period of limited guarantee, the relationship between ROA and deposit growth was negative, which indicates that the
145 improvement in bank profitability is followed by a decline in the rate of deposit growth49. This relationship, in a way, can be interpreted as an indication of a weaker market discipline.
In this case, banks with low ROA indicators are more active in the market to gather deposits to finance their operations. Alternatively, the negative correlation can be seen as a signal of a different kind of market discipline action if depositors perceive that banks with a higher rate of ROA pose higher risks. If this is the case, market discipline during the limited guarantee period is demonstrated by the withdrawal of funds from banks that are perceived as having higher returns from risky investments - the additional risk does not justify the additional returns.
The empirical results of this study suggest that an increase in NIM would increase the volume of total deposits. NIM serves as an indicator of profitability which is calculated as the difference between interest earned and the interest paid by the banks. This could be increased by either reducing deposit interest rates (interest expense) or by increasing credit interest rates (interest revenue). If the increase in NIM was caused by lowering interest expenses, then it can be assumed that market discipline is effective. Generally, large banks are able to keep their deposit interest rates low because these banks have wider retail deposit bases that provide cheap funds. In contrast, if the improvement in a bank’s NIM is the result of an increase in interest revenue from riskier investments that offer higher rates of returns, then depositors would probably react by decreasing their balances or, in extreme cases, withdraw their money completely. This might be the case for small banks that have limited access to low-cost retail deposits and therefore rely on expensive wholesale funding sources. To sum up, without any information on the actual reason for the increase in the NIM ratio, the impact of this variable on market discipline cannot be accurately determined.
The management efficiency variable as measured by the ratio of OPEX indicates a negative relationship implying that an increase in the value of the OPEX ratio results in a decline in the volume of deposits. This finding is consistent with the view that less efficient banks are expected to have higher expenditure; hence depositors would punish these inefficient banks by withdrawing their deposits. However, there is a possibility that the increases in expenditure are the result of greater investments in improving customer service. According to Martinez-Peria and Schmukler (2001), if the control variable for the quality of service cannot be measured, the effect of this efficiency variable cannot be determined. As this study does
49 A similar situation was found in the Chinese market by Wu and Bowe (2012), who argue that this negative response of investors to an increase in reported bank earnings may represent the depositor perception that such earnings are potentially subject to manipulation by management.
146 not measure the quality of service variable, the impact of this variable on market discipline in the Indonesian market is undetermined.
The liquidity risk indicator, as measured by the LDR, indicates a positive relationship between the growth of deposit and LDR ratio, either during the blanket guarantee or the limited guarantee period. This sign on the relationship can be interpreted as the willingness of depositors to increase the size of their balances enabling the banks to increase their lending.
These results are contrary to expectations as suggested by existing literature, which maintains a reverse correlation between a liquidity variable with the growth of deposits (Barajas &
Steiner, 2000; Demirgüç-Kunt & Huizinga, 2003; Martinez-Peria & Schmukler, 2001). The previous findings suggest that at any given level of total deposits, a more prudent bank will issue fewer loans in order to maintain access to more liquid assets in the event of unexpected declines in the balances (Martinez-Peria & Schmukler, 2001; Wu & Bowe, 2012). However, in the Indonesian market, findings suggest that during the observation period of this study, an increase in LDR might not be seen as an increase in liquidity risk, but rather as an increase in bank capacity to generate revenue from new loans. Before the 1997 financial crisis, the average LDR of Indonesian banks was above 100% (Batunanggar, 2002), whereas the mean and median of LDR for sample banks in this study was 81.97% and 70.50% respectively, as shown in Table 4.4. Moreover, this ratio is comparatively small in comparison to other ASEAN member countries (Bank Indonesia, 2012b)50. If this assumption is valid, then it can be concluded that Indonesian depositors exercised discipline by rewarding banks that have the ability to generate higher levels of revenue through disbursing more loans at any given level of deposits.
Besides the CAMEL indicators, deposit interest rate is one of the main factors affecting the growth of deposits, as can be seen in Table 4.10. If market discipline exists, for a given increase in the interest rate, banks that are perceived as financially healthy may experience a growth in their deposits (Boyd & De Nicolo, 2003). Hadad et al. (2011) confirmed the existence of market discipline by depositors in Indonesia, which was imposed through a market price mechanism, where depositors demanded higher rates of return to compensate for additional risks. However, if market discipline is effective, banks cannot continuously adjust interest rates to attract new deposits, unless this is accompanied by an improvement in their risk profile (Boyd & De Nicolo, 2003). The provision of a FSN in the Indonesian banking
50 To stimulate a higher level of LDR, the central bank issued a regulation on 1 March 2011 that linked the bank reserve requirement. The benchmark for a prudent LDR is 78% to 100%. A bank with a LDR lower or higher than the benchmark will be charged additional reserve requirements.
147 sector might distort the market price mechanism by enabling weak banks to compete for deposits. As documented in Enoch et al. (2001), the central bank introduced a cap in deposit rates in 1998. This was done in order to minimize the moral hazard implications arising from weaker banks attempting to increase their deposit balances through offering unsustainably high interest rates. Such actions could lead to solvency issues. This cap on the interest rate is no longer in place despite the potential of the moral hazard implications of the FSN.
Therefore, the role of a banking supervisor is to ensure that only healthy banks have access to compete for deposits using the market price mechanism, such as competitive interest rates (Boyd & De Nicolo, 2003).