Introduction The occurrence and severity of the Indonesian financial crisis in 1997 surprised everyone. Indonesian economic performance has ranked among the best in the world, so that praised by the World Bank (1993) as a part of East Asian miracle. Hill (1998) notes that before the crisis "almost every technical economic indicator looked safe". Economic growth was robust in average of 7.9 percent during 1990 to 1996. The inflation rate associated with the growth path was persistently high, but was still below 10 percent. The average inflation rate associated with the 7.9 percent was 8.3 percent in CPI, and even reached 6.5 percent in 1996. As pointed out by McLeod (1997), inflation was falling, not rising, and the relatively large current account deficit was not caused by unsustainably rapid growth but by high capital inflow, which made sustainable high growth possible, and was itself a response to high returns to investment in Indonesia. In other words, the deficit on current account of the balance of payments looked manageable. The fiscal accounts were in surplus, except a little negative in 1992 and 1993. The structure of savings and investment associated with the growth path was good enough. Official foreign exchange reserves looked adequate and were trending upwards. Even many also believed that Indonesia was in much better position and strategy in responding to the regional currency crisis compared to Thailand. (Feridhanusetyawan, et al, 1998). First, macroeconomic indicators, especially current account deficit was at around 4 percent of GDP while Thailand already reached around 9 percent. Second, Indonesia's Rupiah was not fixed like the Thai Bath, which allowed for some adjustments in responding to the speculative attacks. Third, the slowdown in export growth in 1996 was not as severe as Thailand mainly because real wages grew at much slower rate due to less tight labor market compared to Thailand. In short time after the crisis started, however, it was clear that Indonesia was in much worse condition compared to other Asian countries in crisis. Both foreign and domestic investors have fled, and hundreds of corporations are bankrupt. The banking system has effectively ground to a halt, with very little new lending taking place and dozens of banks insolvent. Domestic demand has plummeted. Thousands of Indonesians have lost their jobs, and millions more face a substantial reduction in their standard of living. (Radelet, 1999) Even after ten years of Indonesian crisis, it is still unclear what really the roots of crisis. The common agreement among scholars is that the existing models so called first generation models (Krugman, 1979; Flood and Garber, 1984) and second generation models (Obstfeld, 1986) failed to explain the Indonesian financial crisis. The failure of the first generation models can be seen from these facts. Government budget was balance or moving into surplus (partly in appropriate fiscal response to higher net private capital flows). Growth in monetary aggregates was fairly high, but cannot be described as runaway monetary expansion. Inflation rate was coming down, nominal GDP growth was largely at levels corresponding to money creation, and Indonesia was at a stage of development where money demand was still growing. The logic of the second-generation crisis model does not apply to the Indonesian financial crisis either. There was no such trade-off between the benefits of a credible exchange rate peg and the costs in terms of higher interest rates, higher unemployment or lower growth of defending the peg before the crisis erupted. Past and expected growth was enviably high, interest rates and sovereign yield spreads were going down, not up, and unemployment was informal (as usual in developing countries). In other words, when the crisis began in mid-1997 Indonesia did not have substantial unemployment nor other apparent incentives to abruptly abandon the pegged exchange rate regimes generally followed in the region in order to pursue a more expansionary policy as suggested by second-generation crisis models.
Research Objective While many studies are still difficult to find common agreement about the causes of the Indonesian financial crisis, this paper argues that the weak Indonesian banking sector as result of improper banking deregulation contributed to the crisis. In this respect, this paper provides an empirical test of the view that the banking was fragile, and to determine the factor contributing to banking fragility by using individual bank balance sheet data.
II.1. Framework Banks continue to dominate the financial systems of most developing and transition countries, as well as Indonesia. As illustration, the study of the World Bank (1997) shows that the ratio of banking sector assets to all financial institutions' assets of Indonesia in 1994 was 0.9. The importance of banking system in Indonesia can also be considered from the ratio of deposit money bank assets to GDP at 0.65 in the same year. In another hand, banking is by nature a high exposure business. As financial intermediaries, it is the business of banks to take on risks passed on them. Banks are typically exposed to a host of different types of risks. Aside from credit risks (default risk) that arise from intermediating between depositors and borrowers, banks also face interest rate risk, currency risk, and market risk. (Bacha, 1998) Based on the importance and the nature of banking system as explained above, it is important to maintain bank soundness. For achieving sound banking system, since the early 1980s, the Indonesian government introduced banking deregulation. It seems such deregulation encouraged operating environment of the Indonesian financial system. However, discussions --even prior to the crisis-- had pointed to the underdevelopment of governance in the Indonesian banking sector, resulting weak banking sector. Some analyses (including Frankel, 1998; Krugman, 1998; Noland, 1998; Yellen, 1998) as summarized by Cabalu (1999), note this stylist features: connected lending (i.e. lending to related parties); excessive concentration of lending to particular borrowers or areas; excessively high loan to valuation ratios; inadequate covenants to restrict the activities of borrowers; lending based on asset values, rather than capacity to service from income; failure to recognize and provide for deterioration in loan quality; lending to firms or individuals as a result of government directive, rather than on a commercial basis; lack of transparency and inadequate prudential regulations; strong expectations of government bailouts should banks get into difficulties. For maintaining sound banking system, Lindgren, et al. (1996) present a framework for sound banking that comprises a supportive operating environment, internal governance, external discipline provided by market forces, and external governance provided by regulation and supervision at the domestic and international levels. (Figure 1) No single of one of these elements is sufficient by itself; each is subject to failure, or may be underdeveloped in a given economy.
Figure 1: Framework of banking sector soundness Supportive operating environment Internal governance Banking sector soundness External governance
Market discipline Regulation and supervision Source: Lindgren, et al. (1996) In analyzing the existing weakness on the Indonesian banking sector, which contributed to the crisis, this paper employs that framework with some modifications. This paper argues that banking deregulation has imposed by the Indonesian government since the early 1980s significantly encouraged the operating environment of the Indonesian financial system. We can note the greater access to banking sector due to increasing the number of banks and branches, increasing bank lending, developing the size of financial sector, maturity structure that beneficial for corporate sector, and better performance and efficiency of banking sector. This good result, however, was undermined by underdevelopment of governance in the Indonesian banking sector. Refer to the framework presented by Lindgren et al. (1996); Indonesia did not yet successfully solve the existing governance failure, both internal and external governance failures. These failures have caused unsoundness in the Indonesian banking sector, indicated by accumulation of non-performing assets in the banking sector. As further will be analyzed, such a banking weakness in the arena of international integration contributed in maintaining dynamics of sustained mispricings and continuous short-term capital inflows. The weak banking sector caused higher interest rates in the domestic markets so that interest rate arbitration does not take place. When interest rate arbitration does not take place, where i(d) > i(f) + e + rp , while large firms could, to some extent, take advantage of low foreign interest rates by borrowing abroad directly rather than via the intermediation of the domestic banking sector, it tended to be the smaller and perhaps less efficient firms that increased their demand for loans from the private banking sector. If this is so, then as far as the domestic banks are concerned, there is an adverse selection problem in terms of the nature of their assets, which further weakens their balance sheets. The proportion of non-performing loans (NPLs) burdening the banks may therefore be expected to rise over time. This in turn makes the supply curve of bank lending shift left (i.e. de facto increase in bank cost structure), leading to a reduction of domestic credit and a rise in the domestic equilibrium interest rate. In fact, if there is asymmetric information such that foreign investors are not aware of the increased NPLs and if consequently country/currency risk premia remain unchanged, we have the paradoxical result that increased domestic financial fragility could induce additional capital inflows in the short to medium term. Figure 2: Conceptual framework and flow of analysis
Banking sector deregulation Better operating environment Governance Problem failure
Weak banking sector
Financial vulnerability Figure 3: Effects of banking deregulation number of banks and branches bank lending size of the financial sector Banking deregulation maturity structure performance: profitability efficiency: ex ante spreads and ex post spreads Figure 4: The governance failures in the Indonesian banking sector Internal governance failure Governance failure Market discipline failure External governance failures Regulatory and supervisory failure
Figure 5: Weak banking sector and build up financial vulnerability
International market integration
Weak banking sector
Capital inflows Interest rate differential External borrowing accumulation, especially in corporate sector
Accumulation of NPLs
Build up financial vulnerability A reasonable place to start is with a traditional model of the effects of financial liberalization against the background of financial repression, with an added assumption that financial intermediation occurs largely through the banking sector, with households placing their savings with the banks and the banks lending to firms for purposes of investment. Figure 6 illustrates this, where the authorities fix an interest rate at level (ic) below the equilibrium one. As a consequence there is excess demand and credit rationing. Financial liberalization allows a (higher) equilibrium rate (ie) to be established. This causes both the supply of loanable funds to increase, as saving is encouraged, and the demand for them to fall, as fewer investment projects appear profitable. Since at the equilibrium interest rate the market is cleared, there is no ration credit by other means. However, the model so far relates to a closed economy and needs to be modified to allow for a liberalized capital account and access to international capital. International financial liberalization may also include trade in financial services. Here the key questions are: how does the domestic rate of interest compare with interest rates abroad, what is expected to happen to the exchange rate, is currency depreciation anticipated, and does the country carry a risk premium because of concerns about default and the small probability of a large negative exogenous shock. Making allowance for these factors, foreign capital will flow in for as long as the domestic rate of interest exceeds the foreign rate. In terms of Figure 7 the supply curve of loans will shift to the right, with Z2 - Z1 reflecting capital inflows. However, of course, this increase in the supply of credit will tend to push down the domestic rate of interest until it is equal to the foreign rate (if), after allowing for expected currency depreciation and risk premia. Thus in equilibrium in a country with both a liberalized domestic financial system and international financial liberalization, the following equality will tend to hold: i(d) = i(f) + e + rp where i(d) is the domestic interest rate, i(f) is the foreign interest rate, e is the expected exchange rate depreciation, and rp is the risk premium. But will it automatically follow that, with e and rp both equal to zero, i(d) = i(f)? The rapid convergence of interest rates may be prevented if the domestic banking sector is relatively inefficient and if capital inflows are intermediated only via the domestic banking sector. In these circumstances i(d) may continue to exceed i(f). However, with fuller international financial liberalization and increasing foreign competition, which may involve the establishment of foreign subsidiaries in the domestic banking sector, the costs of the domestic banks may be expected to fall. This will shift the supply curve of bank lending down and to the right. When interest rate arbitration does not take place, where i(d) > i(f) + e + rp, while large firms could, to some extent, take advantage of low foreign interest rates by borrowing abroad directly rather than via the intermediation of the domestic banking sector, it tended to be the smaller and perhaps less efficient firms that increased their demand for loans from the private banking sector. If this is so, then as far as the domestic banks are concerned, there is an adverse selection problem in terms of the nature of their assets, which further weakens their balance sheets. The proportion of non-performing loans (NPLs) burdening the banks may therefore expected to rise over time. This in turn makes the supply curve of bank lending shift left (i.e. de facto increase in bank cost structure), leading to a reduction of domestic credit and a rise in the domestic equilibrium interest rate. In fact, if there is asymmetric information such that foreign investors are not aware of the increased NPLs and if consequently country/currency risk premia remain unchanged, we have the paradoxical result that increased domestic financial fragility could induce additional capital inflows in the short to medium term. Figure 6: Loan market equilibrium with financial repression and financial liberalization SS0 DD0 i e ic Zd Zz Ze Zs Figure 7: International financial liberalization SS0 SS1 DD1 DD0 i id if Z2 Z1 Z
II.2. Empirical Model and Econometric Procedures The empirical framework used to determine factors contribution to the banking fragility is a "probit" or "normit" model. The regression model is specified as: (1) where is commonly known as a "latent" variable. It is unobserved, and therefore is replaced by an observed dummy variable, , such that: (2) A bank is said to be solvent if it had a capital adequacy ratio (CAR) equal to, or more than, zero during the crisis. A bank is technically bankrupt if the CAR fell below zero. Because the shocks that occurred during the crisis were so large, a bank that maintained its CAR above zero can be called a "resistant" bank. Therefore, the latent variable can be defined as "the ability to resist the crisis". The vector of explanatory variables, , represent the performance and conduct of a bank during the pre-crisis period. If these pre-crisis indicators can significantly explain the variation in , it can be claimed statistically that is detrimental for bank performance during the crisis. Technically, this requires relatively small error terms , in the sense that the variation in mostly belongs to the variation in . In order to accommodate the above idea, the probit estimation technique is used. In Shazam, the probability of occurrence of the dependent variable P(y=1), is described as: (3) Where F(.) represents the cumulative normal density function. The index is a linear function of , but the probabilities are not; therefore, the coefficient must be interpreted carefully. The estimation is done by maximizing the value of the log likelihood function, which is defined as: (4) The maximization of equation (4) is accomplished by non-linear estimation methods. Because it is a concave function, it has a unique solution and trial and error procedures can start from any value. The estimated coefficients tell the effect of a change in the explanatory variable on the index, rather than on the dependent variable. The effect on the dependent variable can be computed as: (5) Where f(.) is the normal density function. It is clear from equation (5) that the effect on the dependent variable is different for each observation. Alternatively, the elasticity can be used, and is defined as: (6) Since the elasticity is different for every observation, either elasticity at means or weighted aggregate elasticity may be used. The elasticity at means is defined as: (7) The weighted aggregate elasticity is computed as: (8) A test of the null hypothesis that all are zero can be carried out by using the log-likelihood as follows: (9) Where L(0) is defined as: (10) Where N is the number of observations and S is the number of successes observed (=1). Various RÂ² can be computed, and the most appealing one can be selected. The Maddala RÂ² is computed as: (11) The Cragg-Uhler RÂ² is defined as: (12) The McFadden RÂ² is: (13) The Chow RÂ² is: (14)