Modeling Crisis Evolution and Counterfactual Policy Simulations: A Country Case Study

Azis, Iwan J. Working Paper Modeling Crisis Evolution and Counterfactual Policy Simulations: A Country Case Study ADBI Research Paper Series, No. 23 Provided in Cooperation with: Asian Development Bank Institute (ADBI), Tokyo Suggested Citation: Azis, Iwan J. (2001) : Modeling Crisis Evolution and Counterfactual Policy Simulations: A Country Case Study, ADBI Research Paper Series, No. 23, Asian Development Bank Institute (ADBI), Tokyo, https://hdl.handle.net/11540/4128 This Version is available at: https://hdl.handle.net/10419/111115 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/3.0/igo/ ADB INSTITUTE WORKING PAPER 23 Modeling Crisis Evolution  and Counterfactual  Policy Simulations:  A Country Case Study Iwan J. Azis August 2001 ADB INSTITUTE TOKYO ASIAN DEVELOPMENT BANK INSTITUTE ASIAN DEVELOPMENT BANK INSTITUTE A number of studies have already shown that the effectiveness of raising interest rates to avoid a currency crisis becomes very limited during the crisis. While most of these studies do not really explain the precise mechanisms, the model in this study does. Using the case of one country – Indonesia – it is shown through a financial sector general equilibrium model that a high interest rate policy can be ineffective because the interest rates and the exchange rate channels of transmission of the policy can go in opposite directions. The mechanisms depend crucially on two important elements: economic and political risk factors, and the balance sheet position of the corporate and banking sector. ADB Institute Working Paper Series No. 23 August 2001 Modeling Crisis Evolution and Counterfactual Policy Simulations: A Country Case Study Iwan J. Azis II ADB INSTITUTE WORKING PAPER 23 A dditional copies of the paper are available free from the Asian Development Bank Institute, 8th Floor, Kasumigaseki Building, 3-2-5 Kasumigaseki, Chiyoda-ku, Tokyo 100-6008, Japan.Attention: Publications. Also online at www.adbi.org The Working Paper Series primarily disseminates selected work in progress to facilitate an exchange of ideas within the Institute's constituencies and the wider academic and policy communities. An objective of the series is to circulate primary findings promptly, regardless of the degree of finish. The findings, interpretations, and conclusions are the author's own and are not necessarily endorsed by the Asian Development Bank Institute. They should not be attributed to the Asian Development Bank, its Boards, or any of its member countries. They are published under the responsibility of the Dean of the ADB Institute. The Institute does not guarantee the accuracy or reasonableness of the contents herein and accepts no responsibility whatsoever for any consequences of its use. The term "country", as used in the context of the ADB, refers to a member of the ADB and does not imply any view on the part of the Institute as to sovereignty or independent status. Names of countries or economies mentioned in this series are chosen by the authors, in the exercise of their academic freedom, and the Institute is in no way responsible for such usage. C opyright © 2001Asian Development Bank Institute & the author.All rights reserved. P roduced byADBI Publishing. A BOUT THE AUTHOR Prof. Iwan J. Azis of Cornell University and the University of Indonesia is a regular Visiting Scholar at the ADB Institute. On the topic of the Asian Crisis, in early 1998 he spoke before the Joint Economic Committee (JEC) of the U.S. Congress, and was invited to present his views on the Indonesian case at the IMF meeting in Washington D.C. during the fall of 2000. He has published on subjects such as ASEAN economies, spatial development, impac t s of economic reform, conflicts resolution, exchange rate and capital flows, reform sequencing, and financial crisis. He has authored or co-authored several books, and is currently working on another book on “Modeling Policy Analysis.” He received his BA from the University of Indonesia and his MSc and PhD from Cornell University. During 1984-1993 he served as Chairman, Department of Economics, University of Indonesia, and Director of the World Bank-funded Inter-University Center. III PREFACE The ADB Institute aims to explore the most appropriate development paradigms for Asia composed of well-balanced combinations of the roles of markets, institutions, and governments in the post-crisis period. Under this broad research project on development paradigms, the ADB Institute Working Paper Series will contribute to disseminating works-in-progress as a building block of the project and will invite comments and questions. I trust that thisseries will provoke constructive discussions among policymakers aswell as researchers about where Asian economies should go from the last crisis and current recovery. Masaru Yoshitomi Dean ADB Institute I V ABSTRACT It is importantto understand the causes and mechanisms of the EastAsian crisis in order to analyze the policy responses. Articles and books have been written extensively, but in-depth research in this area is still in its infancy. While analysts continue to work on this topic, any new development that may arise should be evaluated with empirical evidence. Understanding the anatomy of the crisis requires a careful analysis that spells out the details of events (or sequence of events) and the corresponding implications on economic indicators in each country. This study is using the case of one country, i.e., Indonesia. It describes how that country’s economic policies evolved and why some of them may have planted theseeds forthe subsequent crisis. The paper also discusses the dynamics and sequence of events that took place during the episode. The mechanisms of the process are explained using a comprehensive financial sector general equilibrium model, in order to help one better understand how various variables and indicators interacted during the crisis. In the benchmark run, the values of all exogenous variables (including policy variables) and exogenous events that precipitated the crisis are set equal to their actual (observed) values, and the model is used to derive the resulting values of the endogenous variables. The results of the simulation closely replicate the changes and trends that actually occurred. To facilitate the discussions on the effectiveness (or ineffectiveness) of the actual policy response to the crisis, an alternative set of policies is explored. In particular, the author experimented with policies of (partial) debt resolution and of keeping the interest rate from surging continually. The simulations reveal that the country’s macroeconomic conditions would have fared better if a prolonged high-interest rate policy had been avoided. To the extent that counterfactual policies are feasible, most macroeconomic variables would have been better when such a counterfactual policy is combined with partial debt resolution. This conclusion suggests that the initial actions to address the problems of mounting private foreign debts should have been undertaken as quickly as possible. A developing country with limited foreign reserves like Indonesia should seriously consider policies that put limits on its foreign currency debt. The likelihood of facing currency crises associated with not only currency collapse but also large recessions is higher when the foreign currency debt is large. Since the main channel toward recession is deteriorating balance sheets in the corporate and banking sector, irrespective of the country’s exchange rate regime, a currency crisis may still occur under such circumstances. With respect to the high interest rate policy, a number of studies (including at ADB Institute: Ohno et al. (1999:ADBI Working Paper No.6)) have shown that the effectiveness of raising interest rates to avoid a currency crisis becomes very limited during the crisis. Empirical tests using Indonesian data tend to confirm such a conclusion. While most of these studies do not really explicate the mechanisms, the model in this study does. It is shown that a tight monetary policy can be V ineffective because the interest rates and the exchange rate channels of transmission of the policy can go in opposite directions. In fact, some multi-country studies reveal that higher interest rates are associated with real appreciation only in countries that do not suffer from a banking crisis. This condition contrasts almost completely with the Indonesian case, where almost the entire financial sector went into deep trouble following the 1997 shock. Yet, a fairly persistent high interest rate policy was cogently enforced. Clearly, there were serious inconsistencies and flaws in the policy analysis and the policy design. The counterfactual policies explored in this study produce a more favorable trajectory of recovery for Indonesia compared to that under the actual case. These policies are far more essential than the drastic fundamental changes in micro-economic and institutional structure that the IMF prescribed during the early stage of the crisis. Miscalculated policy responses may have blockaded the efficacy of current and future policies. V I TABLE OF CONTENTS About the Author II Preface III Abstract IV Table of Contents VI 1. Introduction 1 2. The Backdrop 2 3. How the Crisis Evolved 5 4. The Model 14 4.1. Financial Sector 14 4.2. Output and Factor Markets 18 5. Model Mechanisms 19 6. Benchmark Simulation 22 7. Counterfactual Policy Simulations 26 8. Closing Remarks 32 Figures and Tables (in body of text) Figure 1. Indonesia’s Real Effective Exchange 5 Figure 2. Movements of the Rupiah, July 1997−Oct 1998 6 Figure 3. Share of Commercial Banks’Time Deposits 8 Figure 4. Foreign Exchange Time Deposits 8 Figure 5. Import Share in Total Intermediate Input, Various Sectors, 1995 vs. 1998 9 Figure 6. Import Share in Total Intermediate Input, Various Sectors, 1998 vs. 1999 10 Figure 7. Indonesia’s Private Foreign Debts 13 Figure 8. Circular Casuality, Multiple Equilibria, and Policy Choices 20 Figure 9. Impacts of Capital Outflows on Financial and Real Sectors 22 Figure 10. Risk Premium 24 Figure 11. Trends of Selected Variables During The Crisis 26 Figure 12. Real GDP: Benchmark & Counterfactuals 29 Figure 13. Exchange Rates: Benchmark & Counterfactuals 30 V II Figure 14. Net Capital Flows: Benchmark & Counterfactuals 30 Figure 15. Prices: Benchmark & Counterfactuals 31 Figure 16. Risk Factor: Benchmark & Counterfactuals 31 Table 1. Summary of Estimated Parameters for Domestic Investment 18 Table 2. Results of Benchmark and Counterfactural Simulations 28 References 34 7 banks had already regained their domination in terms of deposit share. Foreign banks also benefited from a large deposit increase (Figure 3).8 The panic and fears over the fluctuating value of the rupiah also resulted in an increasing share of foreign currency denominated deposits. The asset substitution from rupiah to foreign currencies occurred on a large scale, as indicated in Figure 4. This suggests that many domestic private banks suffered from a major reduction in funds, while state banks and foreign banks found themselves with excess liquidity.9 When the central bank started to act as the ‘lender of last resort’ by injecting liquidity funds known as Bantuan Likuiditas Bank Indonesia or BLBI to a number of private banks (by the end of the year the amount had reached 7 percent of GDP), most of these funds were used by the recipient banks to buy foreign currency. With expectations of further devaluation, the demand for dollars rose, causing many banks to seek rupiah liquidity to change in the forex market. Consequently, interbank rates skyrocketed. It was in such a situation that the injections of BLBI were made. At the end of the day, this liquidity support pushed the rupiah value to further south. In retrospect, the policy response in the financial sector, which ranged from bank closures to liquidity injections, failed to meet the intended purpose, i.e., restoring market confidence. Bank runs and panic worsened the already distressed banking sector, and the impact on the exchange rate was devastating. The Fund’s request for drastic and fundamental changes in the country’s microeconomic and institutional structure created perceptions that the situation was much worse than originally thought. It therefore eroded market confidence further. In the words of ex-IMF staff member Morris Goldstein: “…both the scope and the depth of the Fund’s conditions were excessive…They clearly strayed outside their area of expertise…If a nation is so plagued with problems that it needs to make 140 changes before it can borrow, then maybe the Fund should not lend.” (New York Times, October 21, 2000). Indeed, the IMF had been acting a little like a heart surgeon who, in the middle of an operation, decided to do some work on the lungs and kidneys, too. 8With the expectation that state banks would not go bust, but would be bailed out by the government, most people decided to switch from private national banks to state banks. Some also put their money in foreign banks. 9The market segmentation is evident from the following gap: among small and medium sized banks, the average interbank rates for overnight funds increased from 35 to 57 percent in November 1997, while the rates among the prime banks decreased from 30 to 18 percent (Enoch, Baldwin, Frecaut and Kovanen, 2001). 8 Non-economic factors reinforced the diverging movement of the currency. Uncorroborated political rumors sharpened the fluctuations of the rupiah and the stock market index, especially in the early stage of the crisis.10 Many hoped that the March 1998 presidential election would mark the beginning of the rupiah’s stabilization. The IMF’s early prescription of curtailing the budget to achieve a surplus implied that many spending items, including subsidies, had to be cut. Expectations grew that the 10 When Suharto cancelled his trip to the ASEAN Summit in Kuala Lumpur and to his wife’s resting place in late 1997, the rupiah trembled. In the same month, when rumors spread that Suharto had suffered a stroke (and some even said that he had died), the stock market index plunged, and the rupiah fell to 5,200.00 to the dollar. This rate was considered highly inconsistent with the economic fundamentals even during the time. Figure 3. Share of Commercial Banks' Time Deposits 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1993 1994 1995 1996 Jan.1997 Feb.1997 Mar.1997 Apr.1997 May.1997 Jun.1997 Jul. 1997 Aug. 1997 Sep.1997 Oct.1997 Nov.1997 Dec.1997 Jan.1998 Feb.1998 Mar.1998 Apr.1998 May.1998 Jun.1998 Jul.1998 Aug.1998 Sep.1998 Oct.1998 private national banks state banks foreign & joint banks panic began Figure 4. Foreign Exchange Time Deposits 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 1993 1994 1995 1996 Jan.1997 Feb.1997 Mar.1997 Apr.1997 May.1997 Jun.1997 Jul. 1997 Aug. 1997 Sep.1997 Oct.1997 Nov.1997 Dec.1997 Jan.1998 Feb.1998 Mar.1998 Apr.1998 May.1998 Jun.1998 Jul.1998 Aug.1998 Sep.1998 Oct.1998 percentage of total deposits State Private National Foreign & Joint Commercial 9 prices of rice, utilities, and fuel would increase substantially should the government adhere to the IMF scheme. In the first week of January 1998, Suharto made the surprise announcement that the projected growth of revenue and expenditure for the new government budget would be higher than originally thought. This created public suspicion that the government was simply refusing to acknowledge the real depth of the crisis. The rupiah was quickly dragged down to the 10,000 level.11 Following rumors that the controversial minister Habibie might take the vice presidential post along with the perception that the IMF package so far had failed to restore market confidence, the rupiah subsequently hit its lowest level, 17,000 per US dollar.12 Note: 1 = food crops, 2 = nonfood crops, 3 = livestock, 4 = forestry, 5 = fishery, 6 = oil, LNG, and coal mining, 7 = other mining, 8 = food processing, 9 = textile, 10 = wood processing, 11 = paper, 12 = chemicals, 13 = electricity and water supply, 14 = construction, 15 = trade and storage, 16 = restaurants, 17 = hotels, 18 = land transportation, 19 = air transportation, 20 = finance and insurance, 21 = real estate, 22 = public administration, 23 = social services. The quickly depreciating rupiah caused the price of imported inputs and capital goods to soar. Two possible outcomes became possible: (1) producers could spend more for imported inputs because of rising import prices, or otherwise would be forced to reduce production; (2) producers could substitute imports with domestic products, avoiding a production cut. A careful look at the input-output tables (1995, 1998 and 11 Under strong pressure, however, the government subsequently announced the delay of 15 big infrastructure projects. With President Clinton promising continuing support to Indonesia, Suharto and the IMF’s Michael Camdessus signed a historical agreement on January 15, 1998. 12 Only after the central bank intervened and a revision of the government budget was made the following day, the rupiah rose to 12,000 per US dollar. Realizing how strenuous it was to prop up the rupiah’s value, the government finally announced a temporary freeze on debt servicing in late January. 10 1999) and the production data suggests that the former was clearly the case for Indonesia. Figure 5 plots the Input-Output sectoral share of imports in 1995 versus the corresponding share in 1998. It can be seen that all pairs fall above the 45-degree line, indicating that the import share of total input costs rose between 1995 and 1998 for all sectors. Thus, a depreciating currency and low elasticity of substitution caused firms’ expenditure on intermediate inputs to be increasingly devoted to imported materials.13 When the sectoral share of imports in 1999 is plotted against the corresponding share in 1998 (in Figure 6), only 5 sectorsi.e., nonfood crops (2), oil, LNG and coal (6), food processing (8), paper (11) and electricity/water supply (13)fall below the 45degree line. The other 18 sectors are above the line. Obviously, the elasticity of substitution in most sectors remains low, suggesting that when import prices rise, production tends to decline.14 13 For example, 13 out of 23 sectors experienced at least a 100 percent increase in their import costs: forestry, other mining, food processing, wood processing, paper, electricity, construction, trade & storage, hotel, air transportation, banking, real estate, and public administration. 14 While the correlation coefficient between input coefficients of imported commodities declined, it remained relatively high for the 1998-1999 comparison, i.e., 0.92 for the 1995-1998 comparison, and 0.74 for the 1998-1999 pair. 11 Rising prices, and especially those of essential products, provided fertile ground for further social unrest. Vendors and traders, especially those of ethnic-Chinese descent, were accused of trying to extract a fat profit out of the chaotic situation.15 In early February 1998, a proposal for implementing a currency board system (CBS) was brought to Suharto’s attention.16 Critized by most economistsbut backed by a few members of the business sector (allegedly those who had huge amounts of foreign debts)the CBS concept never really got off the ground. Confusion reigned over the government’s position on the system.17 On March 7, when Suharto was reappointed to his seventh term with Habibie as his vice-president and daughter Tutut as a minister of social affairs, the rupiah slipped and the Jakarta stock index fell further. Many people who had previously anticipated that the March election would mark the beginning of the rupiah’s stabilization, were proven wrong. Political uncertainties continued to plague the nation, hampering its recovery process. In mid-May 1998, the government found itself forced to adhere to the IMF proposal to remove gasoline subsidies. While the original suggestion was to do it in stages, for unknown reasons Suharto impulsively decided to remove the entire fuel and energy subsidies all at once.18 In retrospect, this may well have been the trigger that led to his downfall. As expected, the abrupt and significant price increase caused huge public protests. The government backed down, deciding to postpone the fuel price increase. But the situation worsened, and the student protest gained momentum. On May 13, a peaceful march turned into a terrifying event, later known as the “May riots,” causing much capital to flee.19 Some estimate that the amount of capital outflows reached as much as $28 billion, of which $8 billion was domestically owned.20 The rupiah skidded to 17,500 per dollar on May 19, its lowest level ever. Suharto's 15 It is interesting to observe that riots occurred only in places where the distribution network for essential goods was disrupted. 16 Steve Hanke of the University of Maryland proposed the idea. CBS requires an independent board that handles foreign exchange transactions at a pre-announced rate. Hanke believed that such a board would solve the confidence problem, and thus release pressures on the rupiah and interest rates. But most economists were against the idea, partly because the country’s foreign reserves needed to back up every single rupiah spent was quite small. Furthermore, under CBS, the central bank could no longer extend credits to commercial banks (it could, instead, use only reserve requirements as a policy instrument). 17 Confusion reached its peak when the rupiah strengthened to reach 8,750 per US dollar on February 19. Two days later, probably under the conviction that widespread anticipation for the currency board had caused the strengthening of the rupiah, Suharto instructed the finance ministry and the central bank to make necessary preparations for the inception of CBS. 18 Some believe that at the time Suharto wanted to “test the water.” 19 It was reported that during the riots that there were systematic incidents of molestation and sexual assaults against ethnic-Chinese. Although history has documented periodic resentments toward ethnicChinese during difficult times, such a record of sexual assaults was essentially unprecedented. This phenomenon led to the spreading of wild rumors that a certain powerful figure backed by a strong institution was behind the acts. The number of people who died during the riots was reported to have reached more than 1,000. As for the rape victims, the number is more debatable and difficult to trace. The government and some non-governmental organizations (NGOs) conducted a series of investigations, but a consensus on the exact number of victims has never been reached. 20 During 1998, the capital account remained in deficit (by roughly $3.7 billion) due to capital outflows, despite official capital inflows of US$7.5 billion. FDI also recorded a minus figure of US$1.3 billion, indicating a deterioration in market confidence and investment climate in general. 12 resignation and Habibie’s succession to the presidency failed to restore confidence. Only after the IMF disbursed some of the promised funds did the rupiah slowly begin to regain strength, reaching 13,000 per dollar in mid-July 1998. In retrospect, the policy response at the early stage of the crisis (e.g., bank closures) caused panic, while the austerity program failed to restore market confidence. The high interest rate policy worsened the already weak condition of the banking system, and the tight budget sparked social unrest. Government’s flip-flopping with the program did not help the situation either, and toying with the idea of fixing the exchange rate through a currency board system simply prolonged the uncertainty. One thing, however, is certain: much of the country’s financial sector and many of its large corporate businesses either collapsed or came close to collapse. From this perspective, the explanation of the crisis surely must be financial in nature (Chang & Velasco, 1998). In this context, I do believe that the mismatch between private foreign debt maturity (largely short term) and the country’s capacity to repay (measured by the size of foreign reserves) holds the key to the story (Azis, 1999).21 Yet, it is interesting to note that most analyses of Indonesia’s macroeconomic development up to mid-1997 failed to touch upon this issue. The following data, which is now well-known, only became widely quoted after the crisis burst. Indonesia’s private foreign borrowing increased dramatically during the 1990s. By mid-1997, the amount reached more than US$50 billion, most of which fell under the category of corporate (non-bank) borrowing, with Japanese banks having the largest exposures (Figure 7). More seriously, the proportion of short-term debts (STD) was considerably larger than the long-term borrowings. At the onset of the crisis, STD already made up 170 percent of the foreign reserves.22 This is obviously a strong case of international illiquidity.23 21 In the aftermath of the Asian crisis, the ratio of short-term external debt to reserves has become the IMF’s basic indicator of reserve adequacy. In Stanley Fischer’s words: “…the ratio of short-term external debt to reserves is the single best crisis indicator.” (see Stanley Fischer, “Asia and the IMF,” June, 2001) 22 While the crisis episodes of Latin American countries were often characterized by inappropriate government policies (i.e.weak macroeconomic “fundamentals”), the Asian crisis countries were inferior in terms of international illiquidity (e.g., in mid-1997, the STD/forex ratios were recorded at 170%, 206%, and 145% in Indonesia, Korea and Thailand, respectively, compared to 120% in Mexico and lessthan 100% in Brazil, Peru, Columbia and Chile). 23 The increasing trend of private foreign debts was actually a regional phenomenon. Thanks to widespread optimism about East Asia’s future growth and the celebrated label of the “East Asian Miracle,” many private investors—local and foreign alike—were poised to expand their activities throughout the region. This was the second wave of foreign capital flows into ASEAN, coming mostly from the U.S, Europe and Japan (the first wave occurred during the second half of 1980s, when Japanese FDI in the region surged, following the endaka phenomenon). The relatively high domestic interest rates failed to dampen investment growth, since foreign loans could be obtained easily at a relatively low rate. Furthermore, stable pegged exchange rates were perceived as a guarantee for earning stability. The label “miracle” swayed lenders and the international financial community, leading them to lend recklessly. The fast-growing number of banks and multi-finance corporations, following the 1988 deregulation, also produced considerable effects. Many big companies set up new banks primarily to serve their own oftenrisky projects. Despite regulatory measures formally imposed by the monetary authorities (e.g., legal lending limits, capital adequacy ratios), weak enforcement and compliance discouraged the development of a healthy financial sector. 13 Banks’ foreign borrowings were also in an upward trend, albeit relatively small. But given the quasi-fixed exchange rate system and the full convertibility of the capital account, domestic currency deposits should have also been included in the asset/liability position of the financial system, as additional obligations in international currency. A depositor could withdraw rupiah from a bank to convert it into dollars at the announced parity. In this situation, unless there are sufficient foreign reserves to honor such a demand, a financial system can still suffer from an international illiquidity problem if it holds excessive domestic liabilities. Indonesia’s ratio of M2 to foreign reserves before the onset of the crisis was indeed the highest among the Asian crisis countries (6.3). Yet, it was still smaller than Mexico’s prior to the 1994/95 crisis (9.1). The above suggests that in a liberalized capital account system like in Indonesia, bank deposits, currency mismatches, and short-term foreign debts are important indicators to watch.24 Yet, the issue of short-term private foreign debts did not seem to take center stage in the early policy designs supported by the IMF. The early rescue program dealt with neither debt relief nor moratoriums on existing debts. There were no discussions about “bailing in” foreign creditors. This issue alone has sparked criticism and become a compelling argument for the need to re-design IMF rescue programs (Sachs and Woo, 2000). An interesting question to ask is therefore: in addition to avoiding a higher interest rate policy, would the resulting outcomes have been more favorable had a (partial) debt resolution also been given priority? This will be one of the policy scenarios conducted in the counterfactual exercise in Section 7. To run counterfactual scenarios, a fairly comprehensive model has been constructed. Prior to conducting the counterfactual experiments, the model wil be used to generate trends in some macroeconomic variables based on the actual sequence of events during the crisis. Let me first describe in the next two sections the specifications and mechanisms of the model. 24 In classifying the sources of Indonesia’s vulnerabilities, Summers (2000) assigned a value of “1” (meaning “very serious”) for short-term foreign indebtedness, along with the problem of general governance and banking weaknesses. Figure 7. Indonesia's Private Foreign Debts 0 5 10 15 20 25 30 35 40 45 50 Banks Non Banks Japan USA Others Short term Long term US$billion End of 1995 End of 1996 June '97 Lending Banks Borrowers 14 4. The Model The model used is economy-wide, with endogenous prices in nature. It has a fairly detailed financial sector, designed to capture the mechanisms of the financial shock. In total, there are 893 equations in the model. The following section discusses only the key components of the financial block of the model, and the mechanisms (channels) of influence through which a financial shock such as the one that occurred in the summer 1997, including political instability, affects the system (a complete list of equations is available upon request at inf[email protected]). Some of the parameters and coefficients are calibrated, while others are estimated econometrically. 4.1. Financial Sector In the first stage, gross private capital inflows (PFCAPIN) are specified as a function of interest rate differentials and country risks (labeled RISK), the latter being influenced by the debt service ratio (debt service to exports): )( 10 RISKRFLOANRLOANdegreePFCAPIN −−×+= σσ (1) ∑ ∑ × ×+= pp inl inl pweE DEBSERV RISK 10 αα (2) where RLOAN and RFLOAN are domestic and foreign interest rates, respectively, PFCAPIN and DEBSERV are the gross private capital flows and the debt service, respectively; degree indicates the intensity of capital openness, with its size calibrated from the Social Accounting Matrix (SAM), pwe is the world price of exports, and Eis export volume. The subscripts inl and pare for borrowing institutions and the production sector, respectively. The risk factor, determined by the country’s debtservice ratio (equation 2), affects the gross capital inflows as specified in equation 1. Hence, even when the interest rate RLOAN increases, capital inflows may not increase if the RISK factor also moves upward (equation 1). In standard general equilibrium models, the interest rates act as an equilibrating factor in securing the saving-investment balance. However, during crisis, the interest rates should be treated as policy variables (exogenous) as they were influenced by IMF conditionality requirements and manipulated by the monetary authorities; hence they were exogenously determined. There are three interest rates in the model, i.e., the deposit rate (labeled RT), loan rate (RLOAN), and the central bank certificate or SBI rate (RSBI). All three have usually moved in the same direction, with only a few exceptional cases, e.g., when banks need to adjust the interest rate differential (the gap between RT and RLOAN). Of the three, RSBI is the one that the monetary authority can directly control; hence, this rate is treated exogenously in the model. When RSBI is raised, banks’ portfolios are shifted towards more SBI, as indicated by an increased proportion of SBI in banks’ funds (bb1). 2 1 1brow RLOAN RSBI browlbbl       + + ×= (3) 15 Subsequently, this reduces the amount of loanable funds BANKF, causing the interest rate RLOAN to increase. The specification of BANKF is derived from the commercial bank’s balance sheet: }) 0)(1()( )(){1( CBTRANBANKRESBORROW BORROWrrbankEXRTDGOVTFIEXRTDI TFHEXRTDHCURRENCYdcurWEALBANKbblBANKF combank combank ppp ihh ihhihh +−+ −++×++ ×++×+−= ∑ ∑ (4) where brow1, brow2 and dcur are constant, WEALBANK, CURRENCY, BANKRES, CBTRAN, BORROW and EXR are the bank’s wealth, the amount of currency, bank’s reserves, central bank’s transfers to commercial banks (e.g., BLBI discussed earlier), bank’s foreign borrowing (0BORROW indicates its value at the initial condition), and the nominal exchange rate, respectively. TDH, TFH and TDI, TFI are household time deposits in domestic and foreign currency, and institutional time deposits in domestic and foreign currency, respectively. The subscript ihh refers to household category. While the interest rates are set as policy variables, in practically all crisis countries with the exception of Malaysia, the exchange rate has been allowed to float since August 1997. In this sense, the exchange rate plays an important role in the determination of the saving-investment balance. The phenomenon of capital outflows, particularly as undertaken by foreign investors, was widespread during the early part of the crisis. This is modeled through a shrinking equity asset EQROW in the foreign sector’s balance sheet. In turn, this can spark outflows of other types of assets, and will eventually raise the total outflows, PFCAPOUT (expressed in US$). Next, the exchange rate determination and the role of non-economic factors need to be specified. Since a standard testable uncovered interest parity (UIP) model requires a rational expectation assumption, the corresponding risk premiums (lumped together with expectational errors, ξ )would have a rather loose economic interpretation. The usual assumption that ξ is orthogonal to the interest rate differential (and hence the slope parameter close to unity) is nothing more than a statistical conjecture.25 Hence, alternative interpretations can be suggested, providing scope for introducing other risk factors. The selection of risk factors depends on the prevailing country’s situation. When political factors play a major role, for example, a proxy for political instability, labeled POLRISK, may enter the equation. A simple example of this can be seen in equation 5: POLRISK EXR EXPEXR RFLOANRLOAN +      −+= 1(5) 25 It is not surprising that a clear consensus has not been reached by most empirical tests using UIP models (see for example, Froot, 1989, MacDonald & Taylor, 1992, and Meredith & Chin, 1998). On the other hand, many studies reject the proposition that exchange rate movements are best characterized as a random walk, (Meese & Rogoff , 1983). 16 The expected exchange rate is modeled through: 321 02 2 00 0 δδδ       ×       ×       ×= CBFRM CBFRM RISK RISK PFCAPOUT PFCAPOUT EXREXPEXR (6) where M2CBFR is the ratio of broad money M2 to the central bank’s foreign reserves, and where all variables ending with “0” indicate values at the initial (pre-crisis) period. As the expected exchange rate (EXPEXR) increases, the following alternatives must occur, individually or simultaneously, in order to be consistent with the uncovered interest parity (UIP) equation 5: (i) the interest rate RLOAN must increase, and (ii) the actual exchange rate EXR depreciate. The same alternatives apply to the case where the political instability, POLRISK, worsens. The worsening exchange rate expectation (EXPEXR) can be set off by increased capital outflows, risk factors, and the ratio of M2 to foreign reserves, as specified in equation 6. The money supply is modeled through a money multiplier and high powered money (reserve money), the size of which is determined by the difference between the central bank’s loans CBLNTOT plus transfers CBTRAN plus foreign reserves CBFR (equivalent to NDA plus NFA) and the central bank’s wealth WEALCB plus noninterest bearing government deposits DDGOV and the central bank’s certificate SBI. Hence, multRMMS =2(7) where )()( SBIDDGOVWEALCBCBFREXRCBTRANCBLNTOTRM ++−×++= (8) The money multiplier, mult, fluctuates rather sharply during a crisis, because household behavior varies considerably. Therefore, money multipliers are allowed to vary freely, influenced among other factors by government policy such as reserve requirements (see Harberger, 2000 for a discussion of flexible multipliers during the Asian crisis). The saving-investment closure departs drastically from neo-classical specifications. Private domestic investment in a sector p, i.e., DOMPINVpis determined through an independent function. It has been observed empirically that over an extensive period of time, Indonesia’s sectoral domestic investment was correlated with value added (output accelerator), interest rates and inflation rate (see Throbecke, 1992). In the current model, I have modified the specification by replacing the inflation rate with nominal exchange rate (equation 9) for reasons to be discussed below. Foreign investment FORINV, which is part of net private capital inflows,f 1(1-f2) PFCAP, along with DOMPINVpand exogenous government investment GOVINVp,constitute total investment TOTINVEST, ppp ppp EXRRLOANVADOMPINV 321 )()1( λλλ λ += (9) (10) where VApis the value-added of sector p. EXRPFCAPffGOVINVDOMPINVTOTINVEST ppp ×−++= ∑))1(()( 21 23 As early pressure on the exchange rate emerged following the Thai’s baht depreciation in July 1997, the Indonesian government responded by widening the exchange rate band to 12 percent (Stage 1: July 1997). At the same time, driven by the jitteriness of foreign investors, capital began to leave the country. These outflows, reflected in the model through exogenous EQROW and PFCAPOUT, continued in the following month (August), despite the fact that the interest rate on the Central Bank’s certificate (Sertifikat Bank Indonesia or SBI) was raised. Unable to defend the exchange rate further, in the subsequent stage (Stage 2: August 1997) the government finally decided to float the rupiah. In the model simulation, these two events are captured sequentially.28 In Stage 3 (September 1997), the Central Bank tried to intervene in the forex market by releasing some of its foreign reserves, and the interest rate on SBI was slightly reduced. However, outflows of foreign assets (EQROW) continued, causing net flows to decline. This prompted the government to finally invite in the IMF (Stage 4: October/November 1997). With limited understanding of what caused the crisis at the time, the IMF offered standard prescriptions, i.e., keeping interest rates high (raising them even further from their already high level), closing 16 banks, tightening government outlays, and imposing extensive structural reforms in areas unrelated to financial matters. The closing of banks was done despite the fact that the country had virtually no deposit insurance system.29 The resulting outcome was obvious: a bank run and financial panic. When capital outflows and the rupiah’s depreciation persisted (partly because of the failure to deal with mounting corporate debts), the economic environment quickly turned worse. The financial sector went into a downward spiral, and the entire economy fell into a deep recession. The stock market plunged, and the rupiah continued “to go south.” Pandemonium set in when on January 8 and 9, 1998, people went on a buying spree to hoard foodstuffs, and the rupiah began to experience a severe fall.30 In a standard interest parity model, the country’s risk premium should have surged during the time. This is indeed the case, as can be seen from Figure 10 (note the sharp jump in January 1998). 28 Meanwhile, during the same period, Indonesia also suffered from crop failures due to the fickle global weather (the El Nino phenomenon) and massive haze problems from forest fires. These factors, although unrelated to the financial crisis, affected the country’s food production. Consequently, some adjustments in the food sector productivity parameter have been made. 29 In the end, the government decided to protect small depositors, i.e., with Rp. 10 million deposits, roughly $3,000 at the prevailing exchange rate. These covered 90 percent of total depositors, but only 25 percent of total deposits. 30 The IMF appeared out of touch with these chronological events. In a private conversation with IMF economists in Jakarta in March 2000, I was told that there was no food hoarding and rioting in January 1998 that could have caused the prices of some basic goods, including rice, to soar. This is obviously incorrect. There was hoarding and food rioting, not only in Jakarta but also in many other cities, causing the inflation rate to rise by 13 percent between December 1997 and January 1998. Since the IMF remained convinced that the resulting inflation was a demand phenomenon, the proposed solution continued to be aggregate demand management, i.e., high interest rates. 24 However, given the prevailing interest rates, the recorded jump in the risk premium underestimated the actual size of the exchange rate collapse. The primary reason is that the worsening socio-political conditions began to play a compelling role during the time, and this cannot be fully captured by a standard risk premium index. Indeed, this was a period of great uncertainty over post-Suharto leadership, and riots erupted in a number of regional towns throughout the country following the increase of prices of basic commodities. The failure of the standard UIP models is quite well known. Ex-post deviations from UIP are often attributed to the existence of foreign exchange risk premiums and systematic forecasting errors. Some argue that the risk adverse behavior of agents explains the presence of a risk premium (e.g., Fama, 1984), while the systematic forecast errors may arise because of the existence of irrational traders (e.g., Froot and Thaler, 1990) or, as argued by Lewis (1995), simply due to the presence of expectational errors caused by infrequent shocks/uncertainties. 31 Some observers (McKibbin and Stoeckel, 1999) resolved the problem by adjusting the risk premium exogenously at a rate sufficient to generate the actual degree of exchange rate collapse. I adopt a similar approach, except that for the reasons stated above, the standard risk premium in the UIP equation includes—and in some cases is even dominated by— political risks, labeled POLRISK,in equation 5. For the benchmark simulation, the value of this parameter is adjusted in Stage 5 (January 1998). In the model specification, the collapsed exchange rate causes corporate balance sheets to deteriorate with large negative net-worths (related to unpaid foreign debts). Consequently, domestic investment is dampened, prolonging the recession (equation 9). 31 In general, countries characterized by stable (unstable) monetary policy experience less (more) forward premium bias when the bias is substantially caused by forecast errors arising from changes in monetary regime. Figure 16. Risk Factor: Benchmark and Counterfactuals 0.8 0.85 0.9 0.95 1 1.05 1.1 1.15 Stage 4 Stage 5 Stage 6 Stage 7 Stage 8 Index Benchmark (IMF) Less tight Less tight &debt 25 The significance of the observed declines in output also makes it clear that the problems in the financial and corporate sectors have adverse impacts on productivity. Hence, beginning in Stage 5, downward adjustments in some of the sectoral productivity parameters are made. As demonstrated in Figure 9, the deep recession damaged investors’ confidence further, causing even more capital to leave the country (increased EQROW). Furthermore, political factors (POLRISK) began to play a determining role once again. In May 1998, the Suharto government was in serious trouble, and major riots took place in Jakarta and other large cities, involving looting and burning. The distribution channels of some basic goods were seriously affected, as many food outlets were burnt and damaged. As shown in Figure 10, the country’s risk premium began to creep up again; hence, the value of POLRISK needs to be re-adjusted in Stage 6. Under the Habibie government, uncertainties continued, causing market confidence to remain low. Yet, the political situation became somewhat better than in the preceding stages (POLRISK is adjusted correspondingly).32 This is detected byand captured throughthe unrelenting outflows of capital, despite efforts by the IMF and the government to continue adopting a strategy of monetary tightening. This episode is applied in Stage 7 (December 1998). Only in Stage 8 (March 1999) did the situation begin to get better and the political situation improve somewhat. Signs of recovery emerged, supported further by improved weather conditions that helped the production of many agricultural activities to pick up. By adjusting the relevant exogenous variables in line with the above changes and sequence of events, a set of sequential simulations (from Stage 1 to Stage 8) is conducted. Figure 11 displays the trend of major variables. Note that with the exception of the SBI interest rate (RSBI), all variables shown in the Figure are derived endogenously within the model. Overall, the generated trajectories of these variables are close to the actual trends. Notice also that some dramatic changes occurred in Stages 5 and 6, when the political variable POLRISK began to show its forceful impact on the system (the January and May riots, and the downfall of Suharto). Despite the continued high interest rates, the expected capital inflows did not come in, while outflows continued to rise. This caused a decline in net capital flows, and a collapse in the exchange rate. Real GDP dropped continuously, and supply shock-related inflation surged, reaching over 70 percent. Note also from Figure 11 that by the end of the simulation period (Stage 8) the value of GDP was lower than the pre-crisis level. 32 Actually, a period of deteriorating socio-political conditions occurred between Stage 6 (May 1998) and Stage 7 (December 1998), when there was a series of student demonstrations (in October 1998) demanding Habibie’s resignation and an end to the direct involvement of the armed forces in Indonesian politics. Furthermore, the stern IMF program announced in September, which involved abolishing government subsidies on basic food commodities including rice backfired, sparking widespread protests and damaging the government’s credibility. In terms of the risk premium, Figure 10 shows that there was a spike around October 1998, before it improved towards the end of the year. But since Stage 7 is only associated with the December 1998 period, the October incidents cannot be directly captured in the model. 26 Figure 11. Trends of Selected Variables During The Crisis 0 0.5 1 1.5 2 2.5 3 Benchmark Jul-97 Aug-97 Sep-97 Nov-97 Jan-98 May-98 Dec-98 Mar-99 Index SBI Rate Net Flows Real GDP Exhange Rate Price Index Widened ER Band Floated ER Jiterrines Raised interest BI intervened IMF entry Pol uncertainty Agric Dropped Pol uncertainty Declining Banks May chaos Output dropped Interest surged EXR collapsed Forcing high interest rate Premature recovery Pol improvement Stage 1 Stage 2 Stage 3 Stage 4 Stage 5 Stage 6 Stage 7 Stage 8 7. Counterfactual Policy Simulations In this section, I concentrate on two alternative policies, the results of which will be compared with the benchmark simulation described above. Since the major actual policy response to the crisis was largely influenced by the IMF, I will label the benchmark simulation “Benchmark (IMF).” The two sets of counterfactuals are: (1) a scenario of keeping the interest rates from continually rising, implying that the interest rates are lower than under the actual (benchmark) case; this alternative scenario is labeled “Less tight;” and (2) a scenario identical to number 1, but combined with the restructuring of some foreign debts (labeled “Less tight & less debt”). Obviously, these scenarios contrast with the actual or “Benchmark (IMF)” case.33 As stated in Section 4, there are three interest rates in the model, i.e., the deposit rate, loan rate, and the central bank certificate rate (RSBI). All three have usually moved in the same direction, with only few exceptions. Of the three, RSBI is the one that the monetary authority can directly control; hence, this rate is treated exogenously in the model (set lower than in the benchmark, see Azis, 2001 and Azis et al, 2001). Taking a lesson from the fact that the early interest rate increase (in mid-1997) failed to revive the economy and the exchange rate, the government attempted to lower the interest rates. Indeed, the actual interest rate on RSBI came down slightly in early 33 Another potentially interesting counterfactual scenario would be one in which there is no panic (e.g., no bank closures). This could be done by altering the risk premium in the UIP equation (equation 5) after splitting the risk premium into two types: politically-related and non politically-related risks. Assuming that the non-politically inspired risk is caused partly by panic related to the bank run (the closure of 16 banks), one could adjust (lower) the exogenously determined non politically-related risk premium in order to test a case without panic. 27 1998 (Stage 5), but was subsequently raised again, presumably under recommendations by the IMF (see the actual trend of RSBI, labeled “SBI Rate” in Figure 11). In the counterfactual experiment, it is assumed that the decrease in the interest rates in Stage 5 is larger than in the benchmark, i.e., the assumed rate is lower by roughly 6 basis points, and there is no swing or upward movement of the rate since then (it is kept constant). The scenario of partial debt resolution is conducted by lowering the amount of debt service in Stage 4 and Stage 5 by approximately 10 percent. This implies that the repayment of matured debts is either reduced or postponed. Since most lending banks are non-consolidated, unlike in the case of the 1980s debt crisis in Latin America, and the qualifications/quality of borrowers (mostly in the corporate sector) are so diverse, there were difficulties in arranging a debt resolution plan. Hence, a scenario of only 10 percent debt rescheduling is reasonably justified. Since the IMF made its entry in November 1997roughly equivalent to Stage 4, the starting point of the relevant adjustments to the exogenous variables is in Stage 4. To conduct a proper comparison, the exogenous changes in each stage, with the exception of the interest rates and the level of foreign debt, are kept the same as in the benchmark simulation described in the preceding section.34 In Stages 4 to 7, the influence of the political risks variable (POLRISK)under the two counterfactual scenarios is set smaller, i.e., 17 to 36 percent lower than in the benchmark simulation. This approach is adopted because the political and socioeconomic repercussions of a more reasonable level of interest rates would have been less severe.35 The second counterfactual experiment, labeled “Less tight & less debt,” involves a combination of lower interest rates with a partial resolution of foreign debts. This is done by lowering the value of the variable DEBSERV in Stages 4 and 5, which causes RISK to decline, and consequently PFCAPIN (capital inflows) to increase (see equations 1 and 2). Let me now discuss the results of the counterfactual experiments. Since the different shocks for the experiments are applied starting at Stage 4 (the IMF entry in November 1997), Table 2 and the set of figures in the following discussions show only the trends from Stage 4 to Stage 8. Under the two alternative scenarios, the impacts on output (real GDP) and prices are more favorable than in the “Benchmark (IMF)” scenario. While higher interest rates produce an output-curtailment effect, the corresponding depreciation of the exchange rate appears to be worse. The latter is formed through the following mechanism. As domestic investment (DOMPINV) drops, output is adversely affected; this includes both the domestic output Dand the production for exports E. Consequently, GDP declines and the RISK factor increases (again, see equation 2). In turn, this reduces private 34 When the interest rates are set lower, fewer bankruptcies would be expected. In turn, the resulting fall in construction activities, being the most bank-sensitive sector, would also be less severe. In the two counterfactuals, I accommodate such changes in the relevant stages by setting the (declining) productivity parameters for the sector slightly higher than in the benchmark scenario. 35 When the interest rates surge, the probability of bankruptcies becomes higher. Furthermore, the unfavorable impact on output drags domestic investment further down, causing more severe repercussions on the economy. At some stage, with such a development the political environment can be jeopardized. Hence, under the two counterfactuals, the POLRISK exogenous variable should be set lower than in the “Benchmark (IMF)” scenario. 28 capital inflows, PFCAPIN.AhigherRISK also pushes the expected exchange rate EXPEXR upward, resulting in a severe depreciation of the rupiah. Hence, despite the standard mechanism of the interest parity equation, the intended impact of high interest rates is offset by a rise in EXPEXR through the above channel, causing greater depreciation of the exchange rate.36 Table 2. Results of Benchmark and Counterfactual Simulations Stage 4 Stage 5 Stage 6 Stage 7 Stage 8 Nov-97 Jan-98 May-98 Dec-98 Mar-99 Net flows Benchmark (IMF) 1.0000 1.0737 0.9762 1.0723 1.0966 Less tight 1.0000 1.2621 1.1198 1.1707 1.2058 Less tight & less debt 1.0000 1.1270 0.9556 1.0185 1.0646 Real GDP Benchmark (IMF) 1.0000 0.9408 0.8582 0.8767 0.9481 Less tight 1.0000 0.9869 0.9016 0.9567 1.0380 Less tight & less debt 1.0000 0.9875 0.9022 0.9576 1.0393 Exchange Rate Benchmark (IMF) 1.0000 1.5401 1.9304 1.6775 1.6646 Less tight 1.0000 1.4995 1.7096 1.6248 1.6139 Less tight & less debt 1.0000 1.4231 1.6209 1.5411 1.5313 Price Index Benchmark (IMF) 1.0000 1.5796 1.9805 1.7410 1.6813 Less tight 1.0000 1.5627 1.7780 1.6627 1.6119 Less tight & less debt 1.0000 1.4890 1.6894 1.5817 1.5347 Risk Benchmark (IMF) 1.0000 1.0679 1.0967 1.1096 1.0514 Less tight 1.0000 1.0599 1.0884 1.0496 0.9915 Less tight & less debt 1.0000 0.9066 0.9294 0.8969 0.8478 Source: Model simulations 36 A number of studies have shown that the effectiveness of raising interest rates in order to strengthen the exchange rate disappears during the crisis. Those using the case of Asian countries before and after the crisis also point to a similar conclusion (Ohno, et al, 1999, and Goldfajn & Baig, 1998). Gould and Kamin (1999) show that the exchange rates in the region are not affected by changes in interest rates, but rather are influenced by the credit spread and stock prices. The empirical test using Indonesian data shows that interest rate policy does not help strengthen the exchange rate during crisis. When “the news factor” is taken into account, it appears that only good news has a favorable effect on the rupiah (at a 5% level). The effect of bad news does not seem to be significant (Turongpun, 2001). This is consistent with the findings of Goldfajn & Baig (1998). The main limitation of these models, however, is that they do not really explicate the mechanisms that can explain why higher interest rates can cause a weaker exchange rate. From this perspective, the current model helps to unravel such mechanisms. 29 The above mechanism is repeated and reinforced by the presence of the BernankeGertler-Krugman effects of the exchange rate on investment. As shown in Figure 12 and Table 2, choosing between a high and non-high interest rate policy can result in a difference of real GDP by as much as 5.7% for the entire Stage 4-Stage 8 period, but in Stage 8 alone the gap can be as high as 9.6%. Meanwhile, the exchange rate under the non-high interest rate scenario can be stronger by between 3.9 to 5.5% (Figure 13). By keeping POLRISK identical in the “Less tight” and “Less tight & less debt” scenarios, the exchange rate appears to be stronger in the latter. In terms of net capital flows, there is a fluctuating trend, but “Less tight” is always superior to the other two (Figure 14 and Table 1). The reason why it produces greater inflows than in “Less tight & less debt” is that with debt rescheduling the new inflows appear to be smaller than with no debt rescheduling (any debt resolution tends to deter further inflows), although the outflows are larger in the latter. 30 Figure 13. Exchange Rates: Benchmark & Counterfactuals 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 Stage 4 Stage 5 Stage 6 Stage 7 Stage 8 Index Benchmark (IMF) Less tight Less tight & debt Figure 14. Net Capital Flows: Benchmark & Counterfactuals 0.9 0.95 1 1.05 1.1 1.15 1.2 1.25 1.3 Stage 4 Stage 5 Stage 6 Stage 7 Stage 8 Index Benchmark (IMF) Less tight Less tight & debt 31 Figure 15. Prices: Benchmark & Counterfactuals 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 Stage 4 Stage 5 Stage 6 Stage 7 Stage 8 Index Benchmark (IMF) Less tight Less tight & debt The numerical effects of the scenarios on prices show that up to Stage 7 the “Benchmark (IMF)” produces the highest price index. The gap is largest in Stage 6, though prices tend to converge in the remaining stages. In fact, in Stage 8 the price level under the “Benchmark (IMF)” is slightly lower than under the “Less tight” scenario (Figure 15). However, for the entire period, the poverty line price level is still highest under the “Benchmark (IMF)” experiment. With additional pressures from the RISK factor, the actual (nominal) exchange rate can actually go into a free fall. As stated earlier, the declines in domestic investment and GDP, which prevent the exchange rate-stimulated exports from expanding, are likely to increase the RISK factor. The higher RISK pushes up the expected exchange rate, causing the exchange rate to collapse. From the counterfactual experiments, it appears that lower interest rates and partial debt resolution would have generated a lower risk to the country, as shown in Figure 16 and Table 1. Figure 16. Risk Factor: Benchmark and Counterfactuals 0.8 0.85 0.9 0.95 1 1.05 1.1 1.15 Stage 4 Stage 5 Stage 6 Stage 7 Stage 8 Index Benchmark (IMF) Less tight Less tight & debt 32 It is therefore evident from the above model simulations that the macroeconomic indicators would have fared better if a prolonged high-interest rate policy could have been avoided. Furthermore, with the sole exception of net capital flows, the results generated from the combination of non-high interest rates and partial debt resolution appear to have been most preferable. An intriguing policy question emerges: why was the interest rate raised again despite the fact that the early interest rate surge clearly failed to revive the exchange rate and the economy, and why did not the government put a higher priority on debt resolution? Given the pressure of the IMF letter of intent (LOI), practically no attention—or perhaps none at all—could be directed towards programs other than what had been written into the LOI. The IMF felt that resolving debt problems would be difficult, particularly because most lenders were non-syndicated banks, and the number of borrowers (mostly in the corporate sector) was very large, with a diversity of quality and of intention to repay the debts.37 It is also possible that the IMF thought a debt resolution might become an easy way-out or a quick-fix for the Indonesian government and the private sector, who subsequently might not feel obliged to meet the IMF conditionalities. 8. Closing Remarks I have elaborated the evolution of the crisis in Indonesia by first summarizing the policy environment and the economic condition before the crisis. I have also discussed the dynamics and sequence of events that took place during the episode. The mechanisms of the process are explained using a model. Such mechanisms help one to better understand how various variables and indicators interacted during the crisis. Some of the parameters and coefficients in the model are statistically estimated based on quarterly data for the crisis period, and others are calibrated on the basis of the social accounting matrix (SAM) and the flow-of-funds data. In the benchmark run, the values of all exogenous variables (including policy variables) and exogenous events that precipitated the crisis are set equal to their actual (observed) values, and the model is used to derive the resulting values of the endogenous variables. The latter are, in turn, compared with the actual values of these variables subsequent to the crisis. In general, the results of the simulation closely replicate the changes and trends that actually occurred. To facilitate discussions on the effectiveness (or ineffectiveness) of the actual policy response to the crisis, an alternative set of policies is explored. In particular, I have experimented with policies of (partial) debt resolution and of keeping the interest rate from surging continually. The simulations reveal that the country’s macroeconomic conditions would have fared better if a prolonged high-interest rate policy had been avoided. More importantly, most macroeconomic variables seem to be best when generated from a combination of non-high interest rates and partial debt resolution. This conclusion suggests that initial actions to address the problems of mounting private foreign debts should have been undertaken as quickly as possible. 37 In a seminar held on March 1998, Stanley Fischer (World Bank) mentioned this in response to my question about this issue. In Spring 2001, during his visit to Cornell University, Michael Camdesus (IMF) repeated the same point when I raised the same question.