This Selected Issues paper assesses Indonesia’s trade integration relative to underlying country characteristics. The paper analyzes Indonesia’s vulnerabilities, especially compared with the eve of the crisis in 1997. Various indicators suggest that the underlying fundamentals are significantly stronger. The paper examines key features of the financial safety net (FSN) in view of international standards and concludes that the current system is capable of timely addressing bank problems. It looks at determinants of, and constraints to, credit growth in recent years.

Abstract

This Selected Issues paper assesses Indonesia’s trade integration relative to underlying country characteristics. The paper analyzes Indonesia’s vulnerabilities, especially compared with the eve of the crisis in 1997. Various indicators suggest that the underlying fundamentals are significantly stronger. The paper examines key features of the financial safety net (FSN) in view of international standards and concludes that the current system is capable of timely addressing bank problems. It looks at determinants of, and constraints to, credit growth in recent years.

V. Post Crisis Credit Expansion in Indonesia46

A. Introduction

82. Credit growth in Indonesia has not been as strong as might be expected given the strength of its recovery since the crisis. This is not surprising; countries that have experienced a banking crisis do not experience contemporaneous loan growth with the return to positive GDP growth.47 Nevertheless, the level of credit to GDP usually returns to pre-crisis levels after a few years. In the case of Indonesia, the level of private sector lending to GDP continues to be significantly below pre-crisis levels and is very low relative to other countries in the region, despite lending having increased by 80 percent during the past three years.

83. The authorities have expressed concern that the banking system is not adequately performing its financial intermediation role and contributing to economic growth. In particular, they have identified the need for funding of infrastructure projects and small-and medium- sized businesses. There is also concern that unless the banking sector ramps up the rate of growth of lending the authorities will not achieve their economic growth targets. Bank Indonesia (BI) has for the past several years announced loan growth targets for the banking sector. During 2004 and 2005, lending grew at an annual rate of about 20 percent. However, loan growth slowed in 2006 (to a rate lower than the BI target rate) in response to an increase in interest rates and in energy and food price increases.

84. The authorities have undertaken several policy initiatives to stimulate lending. Beginning in late 2006, in an effort to spur additional lending, BI announced a series of relaxations in prudential regulations, including a reduction of capital requirements by lowering risk weights, as well as a relaxation of provisioning and loan classification rules, especially for borrowers who had previously defaulted on loans. In addition, the government has established growth performance benchmarks for state banks and some elements of government have put pressure on them to make infrastructure loans and to fund the construction of toll roads.

85. The low credit flows reflect both supply and demand factors. The actions taken by the authorities are an attempt to influence the supply of credit. However, most independent analysts believe that bank lending in Indonesia is demand constrained rather than rationed. This view is supported by the empirical analysis presented in this paper. However, with the exception of limited corporate bond issuance, there is an absence of long-term lending in Indonesia, and likely unmet demand for such credit. In the absence of data, this is, however, difficult to measure. This paper examines the structural and economic factors that influence the supply and demand for credit. Two sets of empirical analysis are presented to examine the importance of these factors and to test whether lending is supply or demand constrained in Indonesia. It also examines the implications of dramatically increasing loan growth and presents some policy alternatives to stimulate the extension of long-term financing.

B. Credit Growth in Indonesia

Aggregate Trends

86. Private sector credit in Indonesia, particularly to businesses, has recovered slowly in the aftermath of the Asian crisis. Nevertheless,nominal growth in private sector credit averaged 20 percent over the period 2000-2005 and then decelerated to 15 percent in 2006. In real terms, credit growth averaged 13 percent and 0.5 percent respectively. Withinthese broad trends, lending to businesses has grown slowly at a nominal rate of 13 percent compared with 33 percent for consumer lending.

87. Growth in private sector credit has decoupled from the trends in GDP and deposits as a result of the crisis. Prior to the crisis, the credit cycle tracked the economic cycle very closely and also trended with deposit growth. Subsequently, however, these trends diverged as banks replaced their troubled loans with illiquid recapitalization bonds and in more recent times have allocated excess liquidity to SBIs (certificates issued by Bank Indonesia (BI)). However, the flow of deposits into the banking sector continued unabated while credit to the private sector declined significantly in the aftermath of the crisis and has recovered slowly thereafter.48 In part, this resulted from the initial illiquidity of the recapitalization bonds.49

88. Beginning in August 2005, the growth in private sector credit decelerated. The slow down came on the heels of an acceleration in inflation and the subsequent hike in interest rates, which caused domestic demand to slow. The decline was broad based across sectors, borrowers and banks, although there were some differences in the magnitude. Consumer lending appeared to respond more to the rise in interest rates than did business lending, and the decline in borrowing was more pronounced for the agriculture and manufacturing sectors. Nevertheless, the various categories of banks reduced loan growth by similar magnitudes.

89. Intermediation ratios have remained low and have shown little recovery since thecrisis. The credit to GDP ratio increased from 20 percent in 2000 to 26 percent in 2005 before declining slightly in 2006. The intermediation ratios also remain well below some of Indonesia’s regional peers. Malaysia, Thailand, and Korea exhibit credit/GDP ratios that are above 90 percent. The loan-to-deposit ratios for Indonesia are also lower than those of the other countries, possibly reflecting the relatively more inflationary environment and the higher real lending rates that borrowers face (Figure V.1).

Figure V.1.
Figure V.1.

Indonesia: Credit Relative to Other Asian Countires, 1990 - 2006

(In percent).

Citation: IMF Staff Country Reports 2007, 273; 10.5089/9781451818413.002.A005

Policy Response

90. In 2004 the authorities began to use moral suasion to encourage loan growth and followed with relaxation of prudential standards. BI announced growth projections that it thought were consistent with the government’s growth targets, though these were in no way obligatory. In 2005, BI introduced a tightening of certain prudential regulations relating to loan classification and provisioning that were consistent with international standards. However, it introduced a structure of reserve requirement penalties that progressively penalized banks with low loan-to-deposit ratios. These surcharges ranged from 1 percent for banks with loan-to-deposit ratios between 75-89 percent to a 5 percent penalty for banks with ratios below 40 percent. Beginning in late 2006 and early 2007. BI began to relax somewhat its prudential standards relating to capital, provisioning and loan classification (see Box V). State banks have been put under pressure by some elements of government to grow lending and to finance infrastructure projects, despite a lack of clarity as to the real maturity of the loans and the available collateral.

Recent changes in Prudential Regulations

To facilitate bank lending to the private sector, BI has over the last 15 months progressively relaxed a number of the prudential regulations. The initial measures, the changes that were introduced and the potential impact of these changes are discussed below.

Initial measures

In July 2005, BI issued regulations to strengthen asset classification including: (i) introducing a uniform loan classification standard that requires that if a borrower misses a payment with one bank, other creditors must reclassify the loans at the same lower level; (ii) setting out the definition of, and the limits on, large exposures to both related (10 percent) and non-related (25 percent) parties; (iii) requiring banks to consider off balance sheet items in terms of asset classification; (iv) discontinuing implicit forbearance granted for restructured loans and setting out clear and specific criteria for classifying restructured assets; and (v) requiring BI classification of credit to prevail if there is a difference of opinion between the bank and the supervisors;

Prudential measures were also introduced to curb reckless lending to consumers. In November 2005, a ruling was made that requires banks to ensure that cardholders pay a minimum monthly payment of ten percent of outstanding debt. Effective December 28, 2005, and in reaction to the growing consumer credit card debt and default, BI issued a ruling requiring a minimum salary of three times the local minimum wage to qualify for a credit card and for banks which issue the cards to limit the line of credit to twice a card holders monthly salary.

The Changes

In January 2006, BI modified the uniform loan classification standards so as to apply only to the 50 borrowers whose loans are the largest exposures of a bank, or to loans of Rp 25 billion (US$2.7 million) or greater. BI also indicated plans to lower this ceiling to Rp 10 billion (US$1.1 million) in the six months, and Rp 5 billion (US$530,000) in 12 months, but stated that it may delay implementation of the standards for loans of Rp 500 million (US$53,000) or greater until mid-2007. Concurrently, the credit risk weight for residential mortgage loans was reduced from 50 percent to 40 percent, while the credit risk weight for small business loans was reduced from 100 percent to 85 percent. In October 2006, BI announced easing of related lending rules

In April 2007, BI relaxed regulations restricting lending to defaulting borrowers. Amendments were made to the asset classification and provisioning rules, including: (i) relaxing the criteria used to identify problem loans so as to allow qualifying banks to classify loans of a certain threshold using backward looking criteria that rely solely on past-due or delinquency status; (ii) liberalizing the uniform classification rules (UCLL) so as to provide circumstances under which the rule does not apply. Under the revised ULCC, a bank is now allowed to extend loans to a defaulting borrower if the funds are to be used to fund different projects and there is a clear separation of cash flows between projects; (iii) expanding the use of eligible collateral for purposes of determining loan loss provisioning to include machinery and warehouse; and (iv) relaxing the criteria used to classify a placement made to a rural bank, as part of the government’s credit linkage program.

Assessment

The measures announced in 2005 had the potential to encourage prudent lending by banks. Technically, the regulations had potential to increase loans that are classified as non-performing and therefore also provisioned by banks. Moreover, they put pressure on borrowers who wished to increase borrowing to settle their outstanding nonperforming obligations. These changes were consistent with international best practices in that they required banks to adequately assess and provision for credit risk.

The reversals on the other hand entail a number of general and specific risks. The policy reversals could undermine BI policy credibility. The backward looking criteria could delay recognition of NPLs and the associated provisions. It is also not clear as to how the relaxation of backward looking criteria will lead to increased lending since it deals primarily with the process used to determine when a loan becomes a problem rather than address loan origination and approval standards. Similarly, the relaxation of the risk weighting for mortgages presents challenges given the problems in Indonesia’s legal-foreclosure framework and the lack of reliable historical loss data to support the lower risk weights. Finally, the expansion of the use of eligible collateral to include higher risk and relatively illiquid forms of collateral will require maintenance of reliable collateral appraisal programs and, in the absence of such review systems, entails an increase in the riskiness of associated lending without necessary provisions. However, it should be noted that those private banks are unlikely to deviate from international practice and thus the changes will have little impact on their lending. The impact on state owned banks is less clear.

Sources: Bank of Indonesia and IMF staff.

91. Staff analysis illustrates, however, that if banks were to reduce liquid assets and increase loans to 75 percent of assets, the system would be more vulnerable to an economic downturn. Indeed, an economic downturn that resulted in one third of the loans in each classification category being downgraded would result in 7 of the 15 largest banks’ capital falling below the 8 percent minimum capital requirement and an additional 4 banks’ capital adequacy ratio (CAR) falling to below 10 percent.

92. official attempts to increase credit, as a means to stimulate growth, can have adverse consequences. In response to official support the Korean credit card companies, some of which were subsidiaries of banks, dramatically increased the number of credit cards outstanding to over 100 million (approximately 4 cards for every Korean adult).50 This was accomplished through a significant relaxation in underwriting standards and consumers were encouraged to use the cards to increase consumption. The end result was that by November 2003, 34.2 percent of credit card receivables had become nonperforming. As a result, in 2004 the Korean Development Bank had to acquire credit card companies from commercial firms, and parent banks had to merge their credit card subsidiaries into the parent so that loans could be restructured at a loss.

C. Factors Impacting Credit Growth

93. For Indonesia, both demand and supply factors appear to have affected credit growth.

  • on the supply side, banks are liquid and the lending capacity has been increasing. However, a number of factors constrain banks ability or willingness to lend. Prominent among the factors are the credit risk of the corporates, the unfavorable legal and judiciary framework for enforcing creditor rights, and weaknesses in the infrastructure for assessing credit risk. The strengthening of prudential regulations after the crisis, while strengthening the resiliency of the system, may have also dampened lending relative to its pre-crisis levels.

  • on the demand side, economic activity has been on the upswing. However, a number of factors appear to be dampening demand, including high lending rates, high inflation rates, increasing unemployment, and the trend by corporates to deleverage.

94. This section empirically analyzes the relative importance of the various factors on the supply and demand for credit in Indonesia. The analysis applies a standard supply and demand model for credit and draws on the work of Louis Catao (1997). The model and empirical specifications can be found in Appendix I.

95. The supply equations performed well both in terms of economic theory and statistics. All variables have the expected sign. Specifically, increases in the supply of bank credit to the private sector in Indonesia can be explained by the increase in bank liquidity as reflected in lending capacity and the positive spread on loans. The negative sign on the NPL variable suggests that credit risk has been a major influence on banks lending decisions and that banks have curtailed credit supply in the face of high nonperforming loans (NPLs).

96. The demand for credit is a bit more difficult to model.Most of the variables have the expected sign. Demand is clearly related to macroeconomic conditions and inversely to interest rates charged on loans. The excess debt variable yielded a positive sign contrary to expectations. However, since private sector indebtedness has been declining, this result indicates that borrowers reduced their demand for credit as their indebtedness declined (effectively, they deleveraged).

A01ufig25

Indonesia: Reserve Requirements and Excess Reserves

Citation: IMF Staff Country Reports 2007, 273; 10.5089/9781451818413.002.A005

Supply factors

97. Indonesian banks have substantial lending capacity. Although, after 2004, statutory reserve requirements were increased rapidly and progressively, banks’ excess reserves are substantial and have been increasing.51 The loan to deposit ratio is still low at 65 percent and lower than ratios reported by the other countries in the region. The restructuring has also led to improved capitalization (20 percent CAR at end-2006) while NPLs have trended down. Both the short and long term specifications find that banks’ lending capacity is a significant factor in determining the supply of credit.

98. until recently banks were relatively constrained in their ability to divest recap bonds. The original bonds issued by the government were fixed rate and illiquid. Several years ago, a significant portion were converted to floating rate, for which there still is only a limited market. However, during the past two years foreign portfolio investors have become willing to acquire the fixed rate bonds.

99. The balance sheet structure of the banks could limit banks ability to finance long term assets. Similar to many countries in the region, Indonesian banks obtain most of their funding from short-term deposits. As of end December 2006, more than 90 percent of bank deposits were less than one month in maturity. Prudent asset-liability management would therefore call for banks to offer short-term, floating-rate loans. In addition, on the asset side, recapitalization bonds and other public securities still account for significant shares (18 percent) and these reduce the bank credit available for private investment while at the same time resulting in seemingly high capital adequacy ratios.52

100. Corporate profitability and leverage has improved but not enough, thus credit risk remains high. Data compiled by Moody’s KMV indicates that Indonesian corporate groups have a higher default probability than corporate groups in Korea, Thailand, and Malaysia. Indonesian corporates have restructured their financial statements through agreements with creditors more slowly than those in neighboring countries with the result that nonperforming loans to these firms are still high. The empirical analysis found that NPLs are a negative factor influencing the supply of credit in Indonesia. Moreover, reflecting concerns about credit risk, foreign investors have preferred to invest in public sector securities rather than corporate bonds. Uncertainty about the performance of issuers has also been a major factor in the sluggish growth of the corporate bond market while signals of higher default risk have been amplified by the failure of some corporations to meet their bond obligations.

101. The credit risk is accentuated by weaknesses in the legal and judicial systems. In the aftermath of the Asian crisis, the authorities introduced a number of reforms in the legal and judicial framework, including establishing commercial courts. However, much remains to be done to develop an effective bankruptcy regime and a judiciary capable of encouraging investor confidence. In recent studies by the World Bank, Indonesia ranked 5 on the scale of 1-10 for the “index of effective regulation on secured lending through collateral and bankruptcy laws.” On procedures to enforce contracts, the World Bank found that enforcement of contacts, including loan agreements is relatively more difficult in Indonesia than in most countries. Legal limitations to the seizure of collateral property, together with the relatively high cost and usually lengthy judicial process, reduce banks willingness to lend, especially to corporates who have the financial resources to fight the banks in court.

102. Banks’ ability to expand lending is also impaired by weaknesses in information infrastructure to facilitate assessment of borrower creditworthiness. In June 2006, BI launched a credit information bureau for banks and financial institutions to use for managing lending risk. However, the debtor information system, while helpful, does not include information on a prospective borrower’s standing with nonbank institutions. With coverage of only 0.1 percent of the population, the registry does not yet facilitate an adequate assessment of borrower risk. Under these circumstances banks face difficulties screening out sound from risky borrowers and this makes lending to small- and medium-sized firms and consumers much more risky and difficult to justify.

103. The strengthening of prudential regulations could also have induced greater selectivity in lending by banks and reduced lending. After the crisis, Indonesia made greater strides to strengthen banking supervision, align the regulatory framework with international practices and improve risk management capabilities in the banking sector, especially in state-owned banks. In addition, the banks that were taken over by the government and sold to foreigners have resulted in some of the largest private banks introducing international risk management practices. Consequently, banks are more likely to exercise greater prudence in lending than was the case prior to the crisis.

104. until recently, the state banks faced additional hurdles in resolving their NPLs and this overhang served as an additional deterrent to lending. Indonesia views the assets of the state banks as state assets with the result that state banks were prohibited from writing off loans, selling assets at a discount, or offering debt forgiveness as part of a loan restructure. These limitations have been recently relaxed and guidance is being drafted to permit the sale of assets at a discount, as well as restructurings that include debt forgiveness.

105. Finally, government paper may have “crowded out” of private sector credit to some extent. In the aftermath of the crisis, banks’ commercial NPLs were transferred to the government at book value and replaced with recap bonds. The net result was a marked reduction of bank exposure to the private sector in favor of government paper. The share of government paper (as a percent of total assets), though trending down, has continued to remain high and there continues to be a strong appetite for BI paper among the banks. At the state banks, the demand for BI paper has been compounded by a rapid increase in volatile deposits resulting from fiscal decentralization.

D. Demand Factors

106. The sustained growth in GDP augurs well for long term demand for credit. Since 2000, real GDP has been growing at an average rate of 5 percent, driven largely by domestic demand except for 2006 when it was export led. Credit demand would be expected to increase with this increased level of economic activity since banks are the primary source of commercial credit.53 However, growth has also not been sufficiently broad based to generate sustained credit growth. Much of the lending to consumers has been for motor bikes while the property sector, which in many countries has been the engine of consumer lending, has stagnated, due to structural weaknesses in the housing industry and housing finance.

107. Borrowers have also faced relatively high real lending rates. During most of the post crisis period real lending rates have been higher than in neighboring countries. Real lending rates have remained high and banks issuing credit cards charge rates annualized at about 26-42 percent. Moreover, in August 2005 BI hiked interest rates in a bid to curb inflationary pressures. The combination of higher interest rates and the increase in production costs resulting from higher oil prices led many firms to postpone investment and reduced demand temporarily. During this same period defaults on credit cards and motor bikes also peaked.54 Consequently, investment continued to decelerate through the end of 2005 after rebounding in the preceding year and the growth in lending to both consumers and businesses declined sharply. Demand for credit from large corporates may also have been constrained by the deleveraging that has occurred at many of these “groups,” as owners have repatriated funds and invested them in their businesses, in lieu of borrowing from banks.

uA01fig26

Indonesia: Real Lending Rates relative to other Asian Countries, 2001-2006

Citation: IMF Staff Country Reports 2007, 273; 10.5089/9781451818413.002.A005

E. Is Credit Supply or Demand Constrained?

108. In analyzing the appropriateness of policy measures to stimulate credit growth it is important to understand whether credit has been supply or demand constrained. An econometric model was constructed to analyze private sector lending in Indonesia to answer this question. The model uses many of the same determinants of the supply and demand for credit discussed above. The theoretical underpinning for the model derives from the work of Stiglitz and Weiss (1981). The underlying premise is that at a prevalent interest rate, credit can still be supply constrained. Due to adverse selection, banks might choose to apply quantitative rationing rather than increase lending rates due to the risk that higher lending rates will attract only unworthy borrowers that are not expected to repay their loans.

Disequilibrium model

109. A disequilibrium model is used to investigate whether low real credit observed is supply or demand constrained. Following a number of existing studies of credit rationing,55 a disequilibrium framework is used to investigate the behavior of real credit in Indonesia during the period 1990-2006, as well as during the post crisis period. At any given time, observed real credit could be due to low demand, low supply, or both. The disequilibrium framework allows for the identification of the underlying constraints on credit in a switching regression framework by imposing ex-ante different restrictions on supply and demand functions.

110. At a given time, real credit supply does not have to equal credit demand, if lending interest rates do not adjust sufficiently and or credit rationing occurs, or if there are directed credits. The actual level of real credit will then be determined as follows:56

Ct=min(Cts,Ctd)(1)

The choice of variables used to determine credit supply and demand are guided by Ghosh and Ghosh (1999) who investigated real credit in Indonesia, Korea, and Thailand in the late 1990s, Canales-Krilijenko and Gelos (2006) who studied real credit in Uruguay after the 2002 crisis, and Barajas and Steiner (2002) who attempted to explain credit stagnation in Latin America. The analysis in this case, however, is constrained by the limited availability of potential explanatory variables with sufficiently long time series.

111. The supply of credit is modeled as a function of the spread between real lending rate and banks’ cost of funds (rl -rd), banks’ lending capacity (l), and an indicator of borrowers’ ability to repay (GDP, x).57

Cts=β0s+β1s(rtlendingrtdeposit)+β2s*lt+β3s*xt+ɛtd(2)

112. The demand of credit is assumed to depend on real lending rate (r), current output determining the need for working capital (y), output gap (ygap), an indicator of future economic activity (stock market index, s), and inflationt).58

Ctd=β0dβ1d*rt+β2d*yt+β3d*ytgap+β4d*stp+β5d*πt+ɛtd(3)

113. The probability that at any given time real actual credit is supply constrained is determined as follows:

θt=Prob(CtD>CtS)=Φ(CtdCtsσs2+σd2)(4)

Where σs and σd are estimated standard errors of the credit demand and credit supply equations and Ф[•] the cumulative Normal distribution function. Under this setting, one can derive the density function h(Ct) and the associated log likelihood function to be maximized subject to the parameter values:

Σi=0Tlogh(Ci)(5)

Maximization of the log likelihood function allows for estimation of credit demand and supply equations and the probability that the observed credit is supply or demand constrained. OLS is used to estimate the starting values for maximum likelihood estimation. The goodness of fit can be gauged by how well the minimum of the estimated credit supply and demand tracks the actual credit.59 As in previous studies the equations are estimated in term levels, although observed real credit is not stationary. The results are valid as long as the determinants of credit supply and credit demand are cointegrated, which is indeed the case.

114. The results suggest that in Indonesia, credit is demand constrained (see Figure V.1 below and Table V.1 in the Appendix II). During the period immediately following the crisis, credit demand substantially exceeded credit supply, with the probability value of unity. However, following the crisis period both the supply and demand for credit have increased but the results suggest that it is credit demand that is constraining actual credit growth. These results are confirmed both by specifications covering the period from 1992 through 2006 and the post-crisis period beginning with the third quarter of 2000. Figure V.1 is based on the specification (1) in Table V.1, Appendix II; however, different specifications show a very similar picture.

115. For the post-crisis period the empirical analysis finds that most of the key variables are significant and have the correct sign. In the demand equation the real lending rate and the stock market index, are significant with the expected signs.60 In the supply equation lending capacity is significant with the correct sign in most of the specifications.61 In all specifications, measures of output were significant with the expected sign. In all specifications that exclude the crisis period the interest rate spread is significant with the expected positive sign. NPLs are significant with the expected sign for the time period for which data is available.

Figure V.4.
Figure V.4.

Indonesia: Empirical Estimation of Credit Demand and Supply 1/

Citation: IMF Staff Country Reports 2007, 273; 10.5089/9781451818413.002.A005

1/ Figure is based on specification (1) in the Table of Appendix II.Notes: Excess credit demand is defined as the difference between credit demand and supply in percent of credit supply, (Cd-Cs)*100/Cs, and presented on the left axis. Probability is one when credit is supply constrained and zero when credit is demand constrained and is presented on the right axis.

116. Discussions with analysts and bankers confirm the finding that credit is currently demand constrained. Analysts in Singapore and bankers in Jakarta who uniformly indicated that there was a lack of demand for credit from credit worthy borrowers. In fact, several banks noted that they had significant undrawn commitments to the commercial sector. These commitments give the borrowers the right to borrow at pre-specified terms and the fact that they have not been drawn down the loans indicates either weak demand for credit by these borrowers or that they have access to alternative sources of cheaper funds.

Conclusions from empirical work

  • During the past few years lending has been demand constrained or close to equilibrium. A limitation of this empirical work is that it does not capture the demand for long-term credit. Aside from limited corporate bond financing, there has been very little long-term lending in Indonesia and thus the data used in the analysis presented above does not reflect the demand or supply of such credit.

  • The credit risk associated with lending to the large corporates continues to be problematic. Many of these firms, or their affiliates, continue to have NPLs in the banking sector left over from the crisis and this has had a statistically significant adverse effect on the supply of credit (in both models analyzed). There is a perception that with a weak judicial framework that these borrowers will be able to block enforcement of a loan agreement should they wish not to repay a loan. Moreover, with BI’s stated intention to move toward Basel II, banks recognize that they will have to hold additional capital against these loans.

F. Policy Conclusions

117. Inasmuch as the credit has been demand constrained, policies aimed at increasing the supply of short-term lending will not be effective in stimulating loan growth. Moreover, policies aimed at relaxing prudential norms away from those considered to be consistent with international standards risk sending the wrong signal to banks and supervisors.

118. Pressuring banks to make longer-term loans with no guaranteed short-term exit will increase the liquidity risk in the system. Indonesian banks have an average maturity on their liabilities of about one month. For these institutions to make long-term loans (even with floating rates), in the absence of a secondary market, increases liquidity risk in the system.

119. Given the relative underdevelopment of sources of long-term credit, there is potential for stimulating increases in such lending. Anecdotal evidence suggests that there is likely unmet demand for long term credit to finance infrastructure, housing, and other long term projects. For example, in Jakarta buyers of condominiums must be able to pay the entire cost of the apartment over three years, which clearly constrains demand to upper income buyers.

120. The authorities need to move forward with measures to stimulate the creation of a market for asset backed securities. The draft matrix of policies to enhance the financial sector contains reference to measures that need to be taken to develop a mortgage backed securities market. This could be expanded to cover other types of assets. The government could also explore expanding the role of Ascrindo, a state owned loan guarantee company, to include mortgages for lower-middle income borrowers. This, in conjunction with a secondary market would help develop long term mortgage lending in Indonesia.

APPENDIX I: Determinants of Long- and Short-Run Supply and Demand for

Credit

The long run model

121. The supply of bank credit is specified as a log-linear function of the lending capacity of the banking system (LC) and of the lending interest rates (i) as follows:

S=α0LCα1iα2ɛs(6)

Where ε is an error term.

The long-run demand for credit is a positive function of GDP and negatively related to the interest rate,

D=β0GDPβ1iβ2ɛd(7)

And in the long run, supply and demand for credit converges.

S=D=ActualCredit(8)

The short run model

122. The short run specification of the model is derived by applying the log operator and taking the first difference of equations (6) to (8), and adding to these equations any extra variable which may have a short run impact on credit. The analysis in Section III highlighted the factors that impact on credit supply to include the credit risk of the corporates, the banks balance sheet structure, crowding out by the government, the legal and judiciary framework, weaknesses in information infrastructure, and bank regulations. Unfortunately, there are no available indices to capture trends in many of these variables and the developments do not lend themselves to being proxied by dummies.

123. Thus, the empirical specification used for the short run version of the supply equation (5) is:

ΔSt=α10+α11ΔLCt+α12Δispreadtα13Δ(NPLratio)tα15ɛs1+Ut(9)

Where St is private sector credit, LC is lending capacity of banks defined as deposits and foreign liabilities, ispread is the interest rate spread computed as the difference between lending and deposit rates, NPL ratio denotes the credit risk, £ is the residual of equation (6) or “error correction term” and U is a normally distributed residual term, Ut ~N(0,σu2).

By postulating that current changes in credit supply respond to deviations between the actual level of credit and its long run supply, the error correction term ensures consistency between the short and the long run results of the model.62

124. Similarly, the short run credit demand is specified as:

ΔDt=β01+β11E(ΔGDP)tβ21Δitβ31excessdebttβ41ɛt+Zt(10)

Where E (GDP) is the expected output which is proxied by a production index, i is the lending rate, excess debt is measured by the deviation of private sector debt from its long run trend and £ is the estimated residual of equation (7), i.e., an error correction.63

125. The equations were estimated using a Vector Error Correction Model (VECM) and Least Squares estimation (OLS). The sample period is 2001M1-2006M12. The empirical investigation began with an analysis of the time series properties of the variables. The augmented Dickey-Fuller (ADF) test and the Phillip Perron (PP) tests were used to determine the order of integration of the data compiled for each variables (see Appendix for the results). Following on this procedure, we applied the Johansen (1995) cointegration methodology. The VECM has the distinct advantage of ensuring consistency between the long run and the short run equations through the error correction term.

126. The results are presented in the tables below. Appendix Table 1 presents the results for the short-run supply and demand equations and Appendix Table 2 presents the results for the long-run specifications

Table 1.

Indonesia: Determinants of Private Sector Credit

Results of the Ordinary Least Square (OLS) estmation

article image
1/ All variables are in logarithms, with the exception of interest rate and interest spread.

denote 5 and 1 percent significance, respectively.

The results of the long-run model confirm the existence of a significant relationship between the variables. The results are consistent with the theoretical model and the estimated coefficients have the expected signs and are statistically significant at 5 percent or very close. The model also proved to be robust to a number of specification tests for autocorrelation and unit root of the residuals.

Appendix Table 1:.

Results of Different Specifications of the Disequilibrium Model for Real Credit

article image
Notes: All variables are deflated by the WPI; all variables are in logs expect for real interest rates. t-statistics in parenthesis;

represent significance at 10, 5, and 1 percent respectively.

References

  • Juda Agung (2002) “Credit Crunch in Indonesia in the Aftermath of the Crisis: Facts, Causes and Policy Implications” Bank Indonesia, Jakarta.

    • Search Google Scholar
    • Export Citation
  • Kusmiarso Bambang (2006) Housing and Mortgages Markets in the SEACEN Countries, The South East Asian Central Banks (SEACEN), Kuala Lumpur, Malaysia.

    • Search Google Scholar
    • Export Citation
  • Barajas, Adolfo, and Roberto Steiner, 2002, “Credit Stagnation in Latin America,” IMF Working Paper 02/53 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Calvo, Guillermo A., Alejandro Izquierdo, and Ernesto Talvi, 2006, “Phoenix Miracles in Emerging Markets: Recovering Without Credit from Systemic Financial Crises” (Unpublished manuscript; Washington: Inter-American Development Bank).

    • Search Google Scholar
    • Export Citation
  • Canales-Kriljenko, Jorge Iván, and Gastón Gelos, 2006, “Post-Crisis Credit: Facts, Lessons, and Prospects,” IMF Country Report No. 06/427 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Catao, Luis (1997) “Bank Credit in Argentina in the Aftermath of the Mexican Crisis: Supply or Demand Constrained?” International Monetary Fund (WP/97/32).

    • Search Google Scholar
    • Export Citation
  • Ghosh, Swati R., and Atish Ghosh, 1999, “East Asia in the Aftermath: Was There a Credit Crunch?” IMF Working Paper 99/38 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Kiefer, Nicholas, 1980, “A Note on Regime Classification in Disequilibrium Models,” Review of Economic Studies, Vol. 17, pp. 637-639.

    • Search Google Scholar
    • Export Citation
  • Laffont, Jean-Jacques, and Rene Garcia, 1977, “Disequilibrium Econometrics for Business Loans,” Econometrica, Vol. 45.

  • Maddala, G., and Forrest Nelson, 1974, “Maximum Likelihood Methods for Models of Markets in Disequilibrium,” Econometrica, Vol. 42, No. 6, pp. 1013-1030.

    • Search Google Scholar
    • Export Citation
  • Pazarbasioglu, Ceyla, 1996, “A Credit Crunch? A Case Study of Finland in the Aftermath of the Banking Crisis,” IMF Working Paper 96/135 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Sealey, C.W., 1979, “Credit Rationing in the Commercial Loan Market. Estimates of a Structural Model under Conditions of Disequilibrium,” Journal of Finance, Vol. 34, pp. 689-702.

    • Search Google Scholar
    • Export Citation
  • Stiglitz, Joseph, and Andrew Weiss, 1981, “Credit Rationing in Markets with Imperfect Information,” American Economic Review, Vol. 71, pp. 393-410.

    • Search Google Scholar
    • Export Citation
  • Tornell, Aaron, and Frank Westermann, 2002, “The Credit Channel in Middle Income Countries,” NBER Working Paper No. 9355, (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • World Bank (2006) “Unlocking Indonesia’s Domestic financial Resources: the role of Non-Bank Financial Institutions,” Washington DC.

    • Search Google Scholar
    • Export Citation
46

Prepared by Inutu Lukonga, Elina Ribakova, and Steven Seelig.

47

In addition to the countries affected by the Asian crisis, Sweden, Russia, Uruguay, and others have seen positive loan growth with a lag after GDP growth becomes positive. The lags have typically been between 18 months and three years.

48

Prepared by Inutu Lukonga, Elina Ribakova, and Steven Seelig.

49

Recently these bonds have become more liquid as international investors have been willing to purchase them.

50

IMF. “Republic of Korea—Staff Report for the 2003 Article IV Consultation,” Washington: February 2,2004.

51

Deposit accounts in foreign exchange are subject to a 3 percent reserve requirement while accounts in rupiah are subject to a daily reserve requirement in the range of 5 percent to 8 percent, depending on the total amount of deposits. Effective September 8, 2005, reserve requirements were raised by an additional 1 percent to 5 percent based on the loan to deposit ratio.

52

Given that government securities are zero risk weighted, the capital adequacy ratios of Indonesian banks are higher than if the banks held loans that are risk weighted.

53

Domestic capital markets are not yet a significant source of financing for long term funds for Indonesian corporates. The recent Supreme Court ruling regarding APP bonds is likely to make this financing even more difficult and calls the legality of all corporate bonds into question Also, loan securitization is yet to develop. Rather than provide long term finance to corporates, pensions and insurance firms currently invest a significant portion of their resources in short-term bank deposits, in essence transforming scarce long term resources into short-term assets.

54

As of August 2005, credit card defaults were Rp 1.08 trillion or 7.2 percent of total credit card debt.

55

Pazarbasioglu (1997), Ghosh and Ghosh (1999), Barajas and Steiner (2002), and Canales-Kriljenko, and Gelos, (2006) among others.

56

Following the method by Maddala (1974).

57

All data are quarterly and cover the period 1992:Q1 to 2006.Q1. Data sources include International Financial Statistics and Biropustat Statistik. All variables are deflated by the Wholesale Price Index (2000=100). Consistent with earlier studies and APD practice, the WPI was used as the deflator because it is more stable. All variables are in logs, apart from the real interest rate and the output gap which is defined as the difference between current output and trend output in percent of trend output. Inflation is defined as the percentage change in the CPI over the previous quarter.

58

A simple measure of output gap is calculated following earlier studies using the Hodrick-Prescott filter to estimate trend. Jakarta stock market index is used as an indicator of future economic activity. Inflation is included to reflect the fact that borrowers benefit from inflation.

59

A measure of the robustness of a disequilibrium model is whether the probability that the underlying hypothesis that credit is supply constrained approaches 1, and that if demand constrained the probability approaches zero.

60

Due to the limited number of observations for the post-crisis specification for credit demand, other variables were excluded to assure sufficient degrees of freedom. In specifications that include the longer period, with and without crisis, the output gap is also significant and with the correct sign.

61

In the specification covering only the post-crisis period lending capacity is insignificant. This is consistent with the results for equation (4) where the lending capacity (more broadly defined) elasticity was very low.

62

Ideally, the equation should have included dummies for regulations, the legal framework for contract enforcement and credit registry, but these were omitted due to insufficient information on how they have progressed.

63

Ideally, the short run demand equation should have included unemployment among the explanatory variables, but data limitations precluded this possibility.

Indonesia: Selected Issues
Author: International Monetary Fund
  • View in gallery

    Indonesia: Credit Relative to Other Asian Countires, 1990 - 2006

    (In percent).

  • View in gallery

    Indonesia: Reserve Requirements and Excess Reserves

  • View in gallery

    Indonesia: Real Lending Rates relative to other Asian Countries, 2001-2006

  • View in gallery

    Indonesia: Empirical Estimation of Credit Demand and Supply 1/