Information about Asia and the Pacific Asia y el Pacífico
Journal Issue
Share
Article

Indonesia: Selected Issues

Author(s):
International Monetary Fund
Published Date:
August 2006
Share
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details

III. Measuring Sovereign and Banking Sector Risk in Indonesia : An Application of the Contingent Claims Approach 1

A. Introduction

1. The balance-sheet approach is a tool developed by economists to help understand the evolution of risks in an economy-wide setting.2 This approach collects information on the size and structure of assets and liabilities of key sectors of an economy, in order to assess the extent of currency and maturity mismatches, or imbalances in the debt and equity structure (such as an excessive reliance on debt). However, balance sheet data do not provide a full picture of all the risks facing a country, because of the contingent nature of many risks. Accounting balance sheets, particularly at the economy-wide level, are also typically valued at full face value (or book value), and are not adjusted to reflect fluctuations in market prices or changes in the likelihood of default. Valuing assets using marked-to-market prices and incorporating contingent liabilities can provide a more complete picture of the risks inherent in a balance sheet.

2. The contingent-claims approach (CCA) provides a methodology to combine balance sheet information with widely used finance and risk management tools to construct marked-to-market balance sheets that better reflect underlying risk. It can be used to derive a set of risk indicators that can serve as barometers of risk and financial sector vulnerability. The CCA has been widely used in financial markets to derive risk indicators for corporations, and its use has been recently extended to sovereign balance sheets and industry-wide balance sheets.3 To date, the CCA has been applied at the sovereign or industry level for illustrative purposes only, since there are numerous challenges in calibrating the methodology without extensive cross-sectional or historical databases like there are for models of the corporate sector.

3. This paper examines the evolution of sovereign and banking-sector risk in Indonesia using the CCA. The first section outlines the methodology and constructs a set of risk-adjusted balance sheets for the period 2000-2005. The second section of the paper assesses the impact of the recent market turbulence in May and June 2006 on risk indicators. The third section of the paper considers the sensitivity of current balance sheets to potential shocks and changes in the structure of debt.

4. The results show a steady improvement in the health of sovereign and bank balance sheets since mid-2001. The recent market volatility has caused risk indicators for the sovereign to show a modest deterioration, with improvements towards the end of June returning indicators to end-2005 levels. For the banking system, risk indicators for state-owned banks deteriorated modestly as a result of recent volatility, reversing some of the gains of 2005.

B. Constructing Risk-Adjusted Balance Sheets for Indonesia

5. To understand changes in the overall level of risk facing a balance sheet, an estimate of the value of total assets and their volatility is needed, since they are typically not observable directly. Because many of the assets on the balance sheet are not traded, and are observed only at infrequent intervals, it is difficult to derive marked-to-market balance sheets. In contrast, many liabilities are traded, and thus can be valued more readily using methods from finance theory to impute the value and volatility of assets using the liability side of the balance sheet. Merton’s (1974) key insight in option pricing theory was that liabilities are contingent claims on total assets, with each liability having a different priority and maturity structure. The most junior liability on the balance sheet can be valued as an implicit call option on total assets. When the value of assets declines relative to the face value of debt, the value of the junior claims declines. Since the liability structure is observed, and many of the liabilities are traded, market prices of different liabilities can be used to derive information on the evolution of total assets. The framework can be applied to individual firms, or at a more aggregated level for an industry or for the sovereign.

The Sovereign Balance Sheet

6. We can use the CCA to estimate the risks to the combined balance sheet of the Indonesian central government and Bank Indonesia,4 following the process in Gray, Merton, and Bodie (2002), and Gapen, et. al (2005). The main elements on the asset side of the public sector balance sheet include international reserves, the net present value of primary surpluses, and the public sector’s monopoly on the issuance of money. These assets are net of any guarantees the public sector may implicitly or explicitly provide to the private sector. The main elements on the liability side of the public sector balance sheet are domestic currency liabilities (domestic currency debt and base money), and foreign currency debt. Thus, the balance sheet of the public sector can be described in the following highly-stylized manner:

7. Estimating the observed value and volatility of sovereign assets directly is difficult, since only international reserves are directly observable on the asset side of the public sector balance sheet. In contrast, each entry on the liability side of the balance sheet is directly observable on a high-frequency basis for Indonesia.5 The CCA uses observed liabilities together with well-known option pricing techniques to derive implied estimates for sovereign asset value and asset volatility. Figure 1 illustrates the different stages of this process.

Figure 1.Overview of CCA for the Sovereign

Table 1.Stylized Sovereign Balance Sheet
AssetsLiabilities
  • International Reserves
  • Net Fiscal Assets (Discounted Value of Primary Fiscal Surpluses)
  • Value of Monopoly over Issue of Money
  • Other Assets less Guarantees
  • Domestic Currency Debt
  • Base Money
  • Foreign Currency Debt

8. Domestic currency liabilities of the sovereign can be modeled as junior claims,6 whereby holders of these liabilities have a residual claim on sovereign assets above what is necessary to service foreign currency debt. If sovereign assets fall to a level where foreign currency debt payments cannot be made, then default is the result. This level is referred to as the distress barrier (DB), and is equivalent to the default-free value of debt.7 Therefore, the value of domestic currency liabilities can be viewed as a call option on sovereign assets with a strike price equal to the level of the distress barrier. Holders of such liabilities receive the maximum of either sovereign assets minus the distress barrier, or nothing in default. The Black-Scholes option pricing formulae can be used to estimate sovereign asset value and volatility with only a few select variables: the value and volatility of domestic currency liabilities (VL and σL, respectively), the distress barrier (DB), the risk-free interest rate (rf), and time (t).8 Once the implied asset values and volatilities are calculated, a range of risk indicators can be derived, including the distance to distress (the number of standard deviations away from the distress barrier), the probability of default, and the credit spread on sovereign assets.

The banking sector balance sheet

9. The process of estimating total assets and their volatility for the banking system is similar to that for the sovereign. However, instead of focusing on the value of domestic currency liabilities, the market value of equity (i.e., total market capitalization from stock price data) and its volatility, together with the distress barrier, can be used to calculate implied assets and their volatility.

10. For Indonesia, the largest private and public banks are included in the analysis. Two groups of banks are defined: the 3 large majority state-owned banks (Bank Mandiri, BNI, and BRI), and the 9 largest private banks.9 The daily market capitalization based on traded stock prices is used to calculate the volatility of bank equity for all 12 banks. The book value of short- and long-term obligations10 are used to calculate the distress barrier for the bank. The distress barrier, market capitalization, and volatility of market capitalization can be used to calculate the implied asset value and implied asset volatility. This is then used to calculate the distance-to-distress, the probability of default, as well as the expected losses of the individual banks.11 Aggregated figures for all private banks and for the state-owned banks are then derived by summing the respective balance sheets and calculating the risk indicators for the two groups of banks.

Baseline results

11. The first step in implementing the CCA for Indonesia is to calibrate the baseline, which was set to June 30, 2006. The estimated stock of local currency liabilities (reserve money and domestic debt) on that date was converted to U.S. dollars, then combined with the U.S. dollar value of external liabilities to derive a distress barrier (in dollars).12 In turn, this was used to solve for implied assets and their volatility plus a range of risk indicators, including the distance to distress, risk-neutral default probabilities, and sovereign credit spreads over U.S. dollar risk-free assets. The same procedure was applied for the 9 private and 3 public banks, using stock market capitalization data to solve for implied assets.13 Then, using the historical data available for Indonesia, a time series of the various risk indicators for the sovereign and the banks were produced from 2000 through to June 30, 2006.14

12. The results show a steady improvement in sovereign balance sheets since mid-2001, with a modest deterioration in risk indicators during the turbulence of August/September 2005, and more recently since May 15, 2006. Figure 2 shows the estimated default probability when 100 percent of expected losses of the banks are assumed to be guaranteed by the sovereign (solid line) and when expected losses are excluded (dashed line). The figure shows a gradual decline in default probabilities, with the exception of a spike in March 2001. This spike was caused by heightened volatility in financial markets in Indonesia and abroad.15 For the later period for which daily information is available, the sovereign spread and default probability move in line with the downward trend in CDS and EMBIG spreads on Indonesian government debt quoted by the market (Figure 3).16

Figure 2.Sovereign Default Probability, 1 Year Horizon, 2000-06.

Source: Fund staff calculations.

Figure 3.Sovereign Default Probability and 5 Year CDS Spread, July 2005-June 2006.

Source: Fund staff calculations, Bloomberg.

13. Balance sheet indicators for the banking system also show a strong improvement in the underlying health of banks over the past five years (Figure 4). The distribution of default risk by assets shown in Figure 5 confirms the general improvement in banking system indicators, with the riskiest banks (those with the highest default probability) accounting for a smaller percentage of total assets over time. Expected losses for the banking system have declined steadily, with sharp falls since end-2003 (Figure 6). This positive trend reflects rising equity valuations and declining volatilities, as balance-sheet structures have improved and non-performing assets have declined. The measure of expected losses for the 12 largest banks moves quite closely with the overall NPL ratio,17 and tends to lead changes in the NPL ratio by as much as two quarters.18

Figure 4.Banking Sector Default Probability, 1 Year Horizon, 2000-06.

Source: Fund staff calculations.

Figure 5.Banking Sector Distribution of Credit Risk by Default Probability, 2003-2005.

Source: Fund staff calculations.

Notes: Figures based on averages for four quarters.

Figure 6.Banking Sector Expected Losses and NPL Ratio, 2000-2005.

Source: Bank Indonesia, Fund staff calculations

C. May-June 2006 Global Market Turbulence

14. Having calibrated the set of risk-adjusted balance sheets, they can be used to understand the impact of recent market turbulence on sovereign and banking sector risk. The estimated sovereign spreads and cumulative default probabilities from the model can be used as the basic metric for comparing and analyzing sensitivities. Spreads provide a valuable and real-time measure of the cost of new borrowing or refinancing. The cumulative probability of default can also be directly related to various rating categories, providing an intuitive benchmark for comparisons. For the banking sector, expected losses provide a convenient measure of the impact of a change in the market environment.

15. The recent volatility experienced by emerging markets since mid-May 2006 provides a natural experiment to illustrate the impact of increased volatility on a variety of risk indicators.Figure 7 shows the market capitalization and volatilities for banks, while Figure 8 shows the recent jump in volatilities for key Indonesian asset markets (overnight interest rate volatility and the bid-ask spread on the rupiah/dollar rate, along with a developed market indicator—the VIX—often used as a proxy for global risk aversion).19 These figures show the increases in volatilities beginning in mid-May, and the subsequent declines in late June, albeit to modestly more elevated levels.

Figure 7.Banking Sector Capitalization and Volatilities, June 2005-June 2006.

Source: Bloomberg, Fund staff calculations.

Figure 8.Key Volatilities, 2005-2006.

Source: Fund staff calculations, Bloomberg.

16. Risk indicators for the sovereign show a modest increase in default probabilities and credit spreads as a result of recent market volatility, then declines toward the end of June.Figure 9 shows the estimated default probabilities for the sovereign on May 8, 2006—just prior to the increased turbulence in emerging markets—which coincided with the low-point for Indonesian default probabilities. The figure also shows the default probabilities on June 15, and then on June 30. As expected, this figure shows that default probabilities increased by about 3 percentage points since the turbulence began (to June 15), but then settled down to around 1¼percent above the May 8 level. Figure 10 shows the model spreads for Indonesian sovereign debt for the baseline on June 30, 2006, as well as the estimated spreads at their low point on May 8, together with their level on June 15. Again this figure shows how the estimated spreads increased across the maturity spectrum for Indonesian sovereign debt by a cumulative 10 basis points to June 15, but then returned to close to their May 8 level by end-June. The actual increase in 5-year CDS spreads observed from May 8 to June 15 was 61 basis points, while EMBIG spreads rose by 44 basis points. On June 30 CDS spreads remained about 76 points above their May 8 level, while EMBIG spreads remained about 44 basis points above May 8 levels. This suggests that the model may underestimate the extent of deterioration in underlying credit risks from May to June. Another explanation may be that CDS and EMBIG spreads overestimate the extent of deterioration in underlying risk, or that they were too low relative to fundamentals in early May, and the recent increase in spreads simply reflects a market correction to more normal levels.

Figure 9.Sovereign Cumulative Default Probabilities, 1-5 Years

Fund staff calculations, S&P.

Figure 10.Estimated Sovereign Spreads, 1-5 Years.

Source: Fund staff calculations.

17. The global market turbulence experienced since May 15 also caused an up-tick in risk indicators for the banking system. Bank capitalization declined by around 15 percent for private banks and 17 percent for state-owned banks (to June 30), while volatilities increased from 29 percent to 33 percent for private banks (reaching as high as 42 percent) and from 34 percent to 46 percent for state-owned banks (reaching as high as 63 percent).20 The reduction in market capitalization and increase in its volatility (particularly for state-owned banks—Figure 7) decreased implied assets and increased their volatility, leading to a decline in distance-to-distress measures and increases in expected losses. These developments reflect increases in interest rate volatilities, as well as a softening of bank earnings for the first quarter of 2006 in the wake of weaker growth and rising NPL ratios. As shown in Figure 11, the recent market turbulence had a larger impact on state-owned banks, reflecting a larger increase in asset volatility (Figure 7). One possible explanation for this result is that state-owned banks have higher NPL ratios, which may imply greater sensitivity of earnings to future growth prospects and interest rate volatility. As uncertainty rises, the volatility of projected earnings is likely to rise more for state-owned banks. Furthermore, the higher debt levels of state-owned banks would imply that higher interest rate volatility would have a greater impact on future debt-servicing costs relative to private banks.

Figure 11.Banking Sector Expected Losses, 1-5 Years, $mn.

Source: Fund staff calculations.

D. Understanding Sensitivities

18. The risk-adjusted balance sheets produced with the CCA can be used to illustrate the sensitivity of Indonesian balance sheets to changes in key parameters and balance sheet structures.Table 2 summarizes the effects of changing different parameters or balance sheet components on the overall distance to distress (D2D) and default probabilities (DP). The previous section demonstrated the impact of increased volatility on a variety of risk indicators, while this section focuses on the impact of changes in exchange rates and the currency composition of sovereign debt.

Table 2.The Effects of Changing Key Variables in the CCA Framework
Change in Input:Effect in Model:Change in Output:
Primary Surplus ↑ASovereignD2D ↑, DP ↓
Reserves ↑ASovereignD2D ↑, DP ↓
Nominal Ex. Rate ↑ASovereignD2D ↑, DP ↓
Dom. Interest Rate ↓ASovereignD2D ↑, DP ↓
DB (Foreign) ↓DB ↓D2D ↑, DP ↓
Ex. Rate Vol. ↑σAD2D ↓, DP↑
Other Vol. ↑σAD2D ↓, DP↑
Stock Mkt. ↓ACorp ↓ ⇒ ABanks ↓ ⇒ Guarantee ↑ ⇒ ASovereignD2D ↓, DP↑
Source: Gray, Lim, and Malone (2006, forthcoming).
Source: Gray, Lim, and Malone (2006, forthcoming).

19. Exchange rate sensitivity can be considered by revaluing the balance sheets for a distribution of different exchange rates to show how the risk indicators might vary with a change in this parameter. This is equivalent to assessing the “partial derivative” of the risk-adjusted balance sheet with respect to the exchange rate.21 Conceptually, the exchange rate is a key price in marking-to-market the sovereign balance sheet, because of the importance of external assets and liabilities in the overall balance sheet. When the exchange rate depreciates, its volatility rises, and the cost of servicing foreign debt increases, increasing the overall level of risk. The distress barrier (measured in dollars) also decreases, as the stock of domestic debt declines in dollar terms, decreasing the overall level of risk. There is thus a non-linear relationship between the level of the exchange rate and the level of implied assets as these two factors (higher volatility and a lower distress barrier) move in different directions, with sovereign assets declining with higher levels of the exchange rate (i.e., a more depreciated exchange rate).

20. Credit spreads are quite sensitive to variations in the exchange rate.Figure 12 shows the distribution of exchange rates that were used to revalue the baseline balance sheet, centered on Rp. 9,263 per U.S. Dollar, with a 95 percent confidence interval from 8,580 to 10,625. This distribution is centered on the exchange rate level at the baseline on June 30, 2006, and is calibrated according to the observed distribution of exchange rates over the last 5 years. Figure 13 shows the associated distribution of credit spreads. For each value of the exchange rate distribution in Figure 12, there is an associated level of credit spread estimated by the model. Plotting all of these combinations together provides a distribution of credit spreads. The 95 percent confidence interval of credit spreads ranges from 185 to 351 basis points, with a mean of 239. Thus, there is a 5 percent chance that spreads could rise above 351 basis points under this exchange rate distribution, and a 5 percent chance that spreads could fall below 185 basis points.

Figure 12.Distribution of Exchange Rates for Simulations.

Source: Bloomberg, Fund staff calculations.

Figure 13.Distribution of 5 Year Credit Spreads From Exchange Rate Simulations.

Source: Fund staff calculations.

21. The next balance sheet sensitivity to be considered is the sensitivity to changes in the structure of debt. The risk-adjusted balance sheets can be used to show the impact of a debt reduction of $10 billion on default probabilities and sovereign spreads, under two different states of the world. First, a reduction in external (dollar-denominated) debt is considered under the conditions prevailing in the baseline, then with a more depreciated and volatile rupiah. The results of this scenario can be compared to an alternative of repaying an equivalent amount of domestic (rupiah-denominated) sovereign debt under the baseline conditions, and then with a higher level and volatility of the exchange rate.22 The impacts of these two scenarios are shown in Table 3.

Table 3.Impact of $10 billion Sovereign Debt Reduction on Risk Indicators
5 Year5 Year
DefaultCredit
ProbabilitySpreads
Baseline 6/30/0618.6%224.3
Impact of:
Repayment of $10bn External−0.1%0.0
Repayment of $10bn External, Higher Volatility1.6%12.4
Repayment of $10bn, Domestic−0.7%−10.9
Repayment of $10bn, Domestic, Higher Volatility0.6%−6.6
Source: Fund staff calculations.Notes: Higher volatility scenario assumes the level and volatility of forward and spot exchange rates rise by 25 percent over the baseline.
Source: Fund staff calculations.Notes: Higher volatility scenario assumes the level and volatility of forward and spot exchange rates rise by 25 percent over the baseline.

22. Repaying the more costly domestic debt brings a greater benefit to the overall sovereign balance sheet.Table 3 shows that sovereign spreads and default probabilities are largely unchanged when long-term foreign debt is repaid, but a reduction in domestic debt has a larger impact. This is because the average interest rate on foreign liabilities is lower than the interest rate on domestically-issued debt, due to the high proportion of external debt that is on concessional terms. Reducing the more expensive liabilities reduces the level of risk in the sovereign balance sheet, as the remaining creditors are more likely to be repaid. This will be reflected in higher values for the remaining liabilities, and lower spreads. In a less favorable state of the world (where volatilities and the level of the exchange rate are 25 percent higher), domestic debt reduction reduces sovereign spreads and default probabilities relative to the baseline and relative to the repayment of external debt.

E. Summary and Conclusions

23. This paper has developed a set of risk-adjusted balance sheets for Indonesia using the CCA. Using data from the liabilities side of the balance sheets of the sovereign and banking sectors, is has been possible to impute the value and volatility of assets and produce a range of risk indicators. These risk indicators are based on forward-looking, marked-to-market balance sheets, and provide information about the distribution of risks and sensitivities to different shocks.

24. Risk indicators for the sovereign and banking system show a steady improvement in balance sheets since mid-2001, with a modest deterioration during the turbulence of August/September 2005. The risk-adjusted balance sheets were also used to assess the impact of recent market volatility since mid-May 2006 on sovereign and banking sector risk indicators. The results for the sovereign show a modest increase in default probabilities and credit spreads as a result of recent market volatility, followed by declines toward the end of June. For the banking system, risk indicators suggest a more significant, though still relatively modest deterioration in expected losses, particularly for state-owned banks. Finally, the paper examined the sensitivity of sovereign risk by considering the impact of changes in exchange rates and debt levels. The results show that credit spreads are quite sensitive to changes in the exchange rate, and that reducing domestic debt levels has a more beneficial impact on risk indicators than reducing foreign debt levels.

References

    AllenMarkChristophRosenbergChristianKellerB.Sester and N.Roubini2002A Balance Sheet Approach to Financial CrisisIMF Working Paper 02/210(Washington:International Monetary Fund), available on the web at: http://www.imf.org/external/pubs/ft/wp/2002/wp02210.pdf.

    • Search Google Scholar
    • Export Citation

    CrosbiePeterJ. and JeffreyR.Bohn2003Modeling Default RiskDecember(San Francisco:Moody’s KMV), available on the web at: http://www.moodyskmv.com/research/whitepaper/ModelingDefaultRisk.pdf.

    • Search Google Scholar
    • Export Citation

    CurrieElizabeth and AntonioVelandia2002Risk Management of Contingent Liabilities Within a Sovereign Asset-Liability Framework, mimeo(Washington:World Bank Group), available on the web at: http://treasury.worldbank.org/web/pdf/currie_velandia_cl.pdf.

    • Search Google Scholar
    • Export Citation

    GapenMichaelT.DaleF.GrayYingbinXiao andCheng HoonLim2004The Contingent Claims Approach to Corporate Vulnerability Analysis: Estimating Default Risk and Economy-wide Risk TransferIMF Working Paper No. 04/121, July(Washington:International Monetary Fund), available on the web at: http://www.imf.org/external/pubs/ft/wp/2004/wp04121.pdf.

    • Search Google Scholar
    • Export Citation

    GapenMichaelT.DaleF.GrayCheng HoonLim and YingbinXiao2005Measuring and Analyzing Sovereign Risk with Contingent ClaimsIMF Working Paper No. 05/155, August(Washington:International Monetary Fund), available on the web at: http://www.imf.org/external/pubs/ft/wp/2005/wp05155.pdf.

    • Search Google Scholar
    • Export Citation

    GrayDaleF.Cheng HoonLim andSamuelMalone2006“A New Debt Sustainability Framework: Incorporating Balance Sheets and Uncertainty,” mimeo(Washington:International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    GrayDaleF.RobertC.Merton andZviBodie2002A New Framework for Analyzing and Managing Macrofinancial RisksMF Risk Working Paper 1-03, available on the web at: http://www.iafe.org/upload/GrayMertonBodieMFRiskWPAugust2003.pdf.

    • Search Google Scholar
    • Export Citation

    International Monetary Fund2004Debt-Related Vulnerabilities and Financial Crises—An Application of the Balance Sheet Approach to Emerging Market CountriesPolicy Development and Review Department, July(Washington:International Monetary Fund), available on the web at: http://www.imf.org/external/np/pdr/bal/2004/eng/070104.pdf.

    • Search Google Scholar
    • Export Citation

    MathisenJohan andAnthonyPellechio2006Using the Balance Sheet Approach in Surveillance: Framework Data Sources and Data AvailabilityIMF Working Paper No. 06/100, April,(Washington:International Monetary Fund), available on the web at: http://www.imf.org/external/pubs/ft/wp/2006/wp06100.pdf.

    • Search Google Scholar
    • Export Citation

    McQuownJohnAndrew1993A Comment on Market vs. Accounting-Based Measures of Default RiskSeptember(San Francisco:KMV), available on the web at: http://www.moodyskmv.com/research/files/wp/A_Comment_on_Market_vs_Accounting_Based_Meas_ures_of_Default_Risk.pdf.

    • Search Google Scholar
    • Export Citation

    MertonRobertC.1974“On the Pricing of Corporate Debt: The Risk Structure of Interest Rates,”Journal of Financevol. 29 (May)pp. 449−70.

    • Search Google Scholar
    • Export Citation
1Prepared by Matthew T. Jones and Dale Gray, with assistance from Yingbin Xiao.
2See Allen et al. (2002),IMF (2004), and Mathisen and Pellechio (2006) for an overview of the balance sheet approach.
3See McQuown (1993) and Crosbie and Bohn (2003) for details of the application to corporations, and Gray, Merton, and Bodie (2002) and Gapen et al. (2004, 2005) for application to the sovereign and industry-wide balance sheets.
4This paper combines balance sheet information from the central government and the central bank. Data on regional governments are not readily available on a timely basis, and their stock of debt is minimal.
5Domestically-issued debt prices are available daily, debt stocks are available monthly. Base money is available daily, and foreign currency debt is available quarterly.
6The CCA assumes foreign currency debt is senior to local currency debt, i.e., governments in distress situations are more likely to first “dilute” the holders of local currency debt (e.g. through inflation), or restructure part of such debt, before defaulting on foreign currency debt. See Gapen et al. (2005). The methodology can be applied with a different seniority structure (e.g., to permit equal seniority for foreign and domestic debt).
7This analysis assumes foreign currency debt includes both external debt and domestic currency dollar-linked debt. The distress barrier is equal to short-term foreign currency debt and 60 percent of long-term foreign currency debt plus interest due up to time t (one year). Research by KMV provides empirical evidence that the fraction of long-term debt for corporates averages around 0.5 to 0.6, and 0.6 to 0.8 for banks. For banks, a value of 0.8 is used for this study, while for the sovereign a value of 0.6 is used. This adjustment is done because assets can fall below total debt for long periods without default if most of the debt is long term, so an adjustment is made to reduce the weight of long-term debt in the distress barrier.
9Three additional banks have data available for some of the period under consideration, but were excluded because they were not available continuously through the period.
10Short-term liabilities include Giro liabilities, other current liabilities, savings deposits, and interbank deposits. Long-term liabilities include time deposits, CD deposits, promissory notes, other long-term loans, and other liabilities. Quarterly balance sheet data were provided by Bank Indonesia.
11Expected losses can be used as a proxy for the value of the “implicit” government guarantee of the banking system.
12Data on the stock of domestically-issued debt was available monthly through June 15, 2006, while the stock of external debt was only available quarterly through March 31, 2006, so this figure was used as a proxy for June 2006 data. External debt levels for Indonesia are relatively stable, so this assumption is unlikely to make a material difference for the results.
13All calculations for the banks were done in rupiah. In the absence of balance sheet data for banks during 2006, end-2005 figures were used to calibrate the distress barrier (short-term debt plus 0.8 × long-term debt). Since debt levels are fairly stable over time, this assumption is unlikely to have a material affect on the results.
14Quarterly data were used from December 2000 to June 2005, then daily data were used through June 30, 2006.
15The volatility of Indonesian and U.S. interest rates rose, together with a rise in forward exchange rate volatility and the volatility of reserve money. The levels of debt did not change much (the distress barrier declined slightly), but a weaker exchange rate caused a decline in the dollar value of liabilities and implied assets. This, together with greater volatility reduced the distance to distress by 25 percent, causing a rise in default probabilities by over five percentage points.
16The correlation between the 1 year default probability and CDS spreads and EMBIG spreads is 0.66 and 0.67, respectively.
17NPL ratio ≡ (Substandard + Doubtful + Loss Loans + Foreclosed Equities + Foreclosed Real Estate + Restructured Loans Classified as Pass and Special Mention)/(Total Loans + Foreclosed Equities + Foreclosed Real Estate).
18The correlation between contemporaneous values is 0.67, but rises to 0.82 with expected losses leading by one quarter, and 0.73 with expected losses leading by 2 quarters.
19The VIX is a volatility index for the Chicago Board Options Exchange, known by its ticker symbol VIX. It is calculated by taking a weighted average of the implied volatility from eight calls and puts on the S&P 100 index.
20Based on the standard deviation taken over the previous 20 business days of the log difference of daily stock market capitalization, annualized by multiplying by √260.
21This distribution was derived by revaluing the balance sheets for each “draw” from the distribution of exchange rates, and deriving the associated risk indicator (such as sovereign spread). The volatility of the exchange rate was estimated using the historical relationship observed between the level of the exchange rate and its volatility over the past 5 years.
22The foreign debt is assumed to be repaid out of long-term fixed-rate debt, which carries a low average interest rate. The domestic debt is assumed to be repaid out of long-term fixed rate debt, which carries a higher average interest rate.

Other Resources Citing This Publication