Manajemen | Fakultas Ekonomi Universitas Maritim Raja Ali Haji 0007491042000205196

(1)

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=cbie20

Download by: [Universitas Maritim Raja Ali Haji] Date: 19 January 2016, At: 19:53

ISSN: 0007-4918 (Print) 1472-7234 (Online) Journal homepage: http://www.tandfonline.com/loi/cbie20

Indonesia's banking crisis: a new perspective on

$50 billion of losses

Olivier Frécaut

To cite this article: Olivier Frécaut (2004) Indonesia's banking crisis: a new perspective on $50 billion of losses , Bulletin of Indonesian Economic Studies, 40:1, 37-57

To link to this article: http://dx.doi.org/10.1080/0007491042000205196

Published online: 12 Jul 2010.

Submit your article to this journal

Article views: 75

View related articles


(2)

ISSN 0007-4918 print/ISSN 1472-7234 online/04/010037-21 © 2004 Indonesia Project ANU DOI: 10.1080/0007491042000205196

INDONESIA’S BANKING CRISIS:

A NEW PERSPECTIVE ON $50 BILLION OF LOSSES

Olivier Frécaut

International Monetary Fund, Washington DC*

The Asian crisis devastated the Indonesian banking sector and led to astronomical losses, almost entirely paid for by the government, i.e. by the general public. The paper provides a new perspective on the crisis, stressing that bank losses are not the same as losses to the economy: most of the ‘losses’ of the banks are actually transfers to borrowers and depositors. It should be possible to recover part of the amounts concerned through taxation of the major beneficiaries. The paper con-trasts the conventional approach, embedded in business accounting, used to man-age the banking crisis, with an alternative approach that relies on national accounting concepts. It shows how the latter can provide a new perspective, eluci-dating the massive transfers of wealth that took place during the crisis. This sug-gests possible improvements in bank resolution strategies, through the identification and quantification of the main transfers of wealth, followed by their taxation.

INTRODUCTION

The Asian crisis that began in 1997 devastated the Indonesian economy, and in particular its banks. The banking sector posted huge losses, and bank insolven-cies mushroomed, with repeated waves of bank closures. A number of banks, often among the largest, avoided financial collapse only through massive central bank emergency liquidity support. The government eventually had to provide a blanket guarantee of bank liabilities to avert the threat of a system-wide financial meltdown. This and the rehabilitation of the surviving banks ended up imposing enormous costs on the general public by virtue of a dramatic increase in public debt.

This systemic banking crisis, with its major macroeconomic implications, was diagnosed, monitored and managed on the basis of conventional concepts and instruments, i.e. established prudential norms embedded in business accounting. Macroeconomic policy analysis and decision making are not based on business accounting and prudential norms, however; they rely on a different set of con-cepts and definitions, contained in the System of National Accounts (SNA). Could the coherent and comprehensive analytical framework provided by the SNA shed additional light on what happened to the Indonesian banking system? Could the SNA provide a better understanding of events in Indonesia, and point to operationally useful conclusions about diagnosing, containing and preventing systemic banking crises in the future? These are the questions addressed in this paper.


(3)

The next section describes the way the crisis was actually managed, by means of a classic prudential approach. The third section identifies the features that give the SNA the potential to improve on the classic approach. On this basis, the fourth section provides a new perspective on the events of the banking crisis, while the fifth explores possibilities for improving the design of bank resolution strategies. The sixth section concludes.

CLASSIC APPROACH AND HORRENDOUS LOSSES

The Indonesian banking crisis was unusual in its scope, depth and severity.1

Nev-ertheless, it was diagnosed, monitored and managed in a conventional way. Established concepts and methods of assessing and categorising individual prob-lem banks were simply extended to the banking system as a whole. The under-lying conceptual background was anchored in the notion of solvency, which was the main guiding principle for a series of triages. A typical business approach to the measurement of bank losses was also adopted.

In practical terms, the main focus of banking supervisors is on banks’ solvency, i.e. their net worth or equity, defined as the surplus of assets over liabilities, and thus their ability to repay their depositors. Bank solvency is measured through business accounting standards and benchmarked against internationally accepted prudential norms, with a prominent role assigned to the capital ade-quacy ratio (CAR) as defined by the Basel Committee on Banking Supervision.

Even with system-wide banking distress, the basic concepts relied upon—in particular, that of solvency—remain the same. As was the case in Indonesia, the classic approach to dealing with a systemic crisis typically includes the following steps:

• first, bringing to light all banks’ hidden losses, in particular in their loan port-folios, through in-depth reviews relying on business accounting norms and applicable prudential standards;

• on this basis, assessing the adjusted solvency position of each bank, expressed as a single, synthetic figure or ratio (typically, the CAR);

• carrying out a triage among banks, and separating them into three groups: 1 banks that are solvent, or that have shareholders able and willing to cover any shortage in equity, for which no specific resolution measures are necessary; 2 banks that are facing severe solvency problems, with shareholders unable or unwilling to provide sufficient additional equity, and are slated for closure because their survival does not justify the amount of public resources that would be required; and

3 banks that face serious solvency problems, beyond the ability of their share-holders to provide support, but which deserve, for some reason, to be pre-served and brought back to an adequate solvency level with support from public resources;

• disposing of the banks slated for closure, while delivering the necessary injec-tion of equity to the problem banks selected for survival—typically by grant-ing them government bonds; and

• finally, recycling the non-performing assets of both the closed and the


(4)

itated banks with a view to maximising recoveries and offsetting part of the costs borne by the government.

Within this typical framework, the most salient feature of the Indonesian bank-ing crisis is the enormous amount of losses posted by the banks and eventually taken over by the general public. The banks were hit by a combination of repeated, violent shocks and by less dramatic, but equally destructive, shifts in the nature and quality of their assets and liabilities.

On the liability side of the balance sheet, each episode of currency depreciation mechanically increased the volume, expressed in domestic currency, of foreign currency denominated deposits and other liabilities, each time brutally shaking the financial structure of the banks. At the same time the domestic public and international investors were losing faith in the rupiah and selling it. Later they lost confidence in the country as a whole and began to liquidate their claims on Indonesia regardless of the currency of denomination. The banks were losing their resources. Credit lines were cut off by foreign banks, and letter of credit facilities progressively dried up.

Whenever doubts emerged about the condition of specific banks, runs devel-oped. By the end of 1997, more than half of the banks had already faced at least one episode of depositor runs. The runs were at times extremely brutal, as in the case of Bank Central Asia in May 1998, and at other times slow and lingering, leading to debilitating cases of haemorrhaging. The interbank market became segmented, inhibiting the recycling of liquidity among banks. Even the better banks knew they could face a severe run at any time, and so maintained a high level of liquidity. Banks requested, and obtained, emergency liquidity support from a central bank anxious to avoid further bank closures or collapses (Djiwan-dono 2004, in this issue). With the rise in interest rates, the tough competition among banks desperate for deposits, and the punitive rates imposed by the cen-tral bank for liquidity support for much of this period, the cost of bank resources increased sharply, dealing banks an additional and devastating blow.

On the banks’ asset side, the impact of macroeconomic shocks was less dis-cernible immediately, but equally devastating. There was a misleading sense of security from the fact that the banks had a slightly long net position in foreign currencies. However, while the liabilities in foreign currencies had to be repaid in full, and sooner than expected because of the loss of confidence by depositors and creditors, the assets in foreign currencies, mainly loans to domestic borrowers, were losing value with each episode of currency depreciation as borrowers’ repayment capacity deteriorated. Combined with the general climate of crisis and the accompanying political instability, this led to an erosion—and later a gener-alised collapse—of credit discipline; most borrowers just stopped servicing their loans, regardless of their ability to repay. The few who did continue to pay asked for, and obtained, concessional interest rates.

The rise in interest rates was applied asymmetrically between liabilities and assets, leading to deeply negative interest rate margins for the banks. With little revenue and high costs, the banks’ financial structure was severely undermined. Over a 15-month period, the well developed banking system of the fourth most populous country in the world, with 238 banks, 238,300 employees, assets


(5)

of $227 billion (roughly equivalent to the country’s GDP before the crisis), loans of $146 billion, deposits of $132 billion, and $17 billion of posted equity,2 was

ruined. It fell into deep insolvency, went on life support from the central bank for liquidity, and—with its credibility and reputation in tatters—became unable to play its intermediation role.

The bulk of banking sector losses materialised during this 15-month period between July 1998 and September 1999, reaching Rp 435 trillion (i.e. some $50 bil-lion, or 40% of GDP) (table 1). Although the crisis began in mid 1997, the bank-ing system continued to book net profits durbank-ing both the second half of 1997 and the first half of 1998. A sudden shift took place in the middle of 1998, however. Huge losses—some $5 billion—were posted by the banks during the third quar-ter. The booking of losses then reached a cataclysmic level of $15 billion during both the last quarter of 1998 and the first quarter of 1999, before diminishing gradually over the following quarters. In the five-quarter period under review, total losses reached some $50 billion.

Of these total losses of around $50 billion, 90% came from only two items: some $35 billion from provisions for depreciation of assets, and around $10 bil-lion from negative interest margins (table 2). When compared with these losses, the banks’ operating costs ($2.7 billion) appear quite moderate.

The government’s issue in January 1998 of a blanket guarantee3covering all

liabilities of the domestic banks meant that all banking sector losses above and beyond shareholders’ equity were allocated to the general public. No losses were suffered by depositors or other creditors. The government took over the banks’ losses in three ways, each involving the issue of new bonds. Some bonds were directly allocated to the banks as payment for new equity or in compensation for their assumption of closed banks’ liabilities, thus increasing the surviving banks’ assets. Others were given to the central bank in compensation for its earlier liq-uidity support to the commercial banks, thus reducing those banks’ liabilities.

In exchange, the government obtained two kinds of assets: first, ownership interests in the banks ‘taken over’ (in plain language, nationalised) by the Indo-nesian Bank Restructuring Agency, IBRA; second, assets transferred to IBRA from

TABLE 1 Banking Sector Losses

Quarter Q3 98 Q4 98 Q1 99 Q2 99 Q3 99

Losses for the period (Rp trillion) –65.8 –121.2 –135.3 –71.8 –40.8 Cumulative losses (Rp trillion) –65.8 –187.0 –322.3 –394.1 –434.9 Quarterly GDP (Rp trillion) 256.9 272.5 281.1 279.7 277.6 Losses : annualised GDP (%) 6.4 11.1 12.0 6.4 3.7 Cumulative losses : GDP (%) 6.4 17.5 29.6 36.0 39.7 Exchange rate (period average, Rp/$) 12,252 7,908 8,776 7,921 7,531 Losses ($ billion) –5.4 –15.3 –15.4 –9.1 –5.4 Cumulative losses ($ billion) –5.4 –20.7 –36.1 –45.2 –50.6

Source: Author’s estimates (see appendix).


(6)

the banks themselves and from the banks’ former shareholders, as part of settle-ment arrangesettle-ments in compensation for central bank liquidity support. The ini-tial agreements reached in September 1998 between the Habibie government and the bank owners turned out to be overly favourable to the latter. When it became clear that the pledged assets were far less valuable than originally estimated, the shareholder settlements were renegotiated in 2001.

IBRA ultimately assumed ownership rights over assets with a face value of Rp 534 trillion, including bank loans (most of them classified as total loss) of Rp 330 trillion, assets received as part of the shareholder settlements (Rp 112 tril-lion), and equity investments in recapitalised banks (Rp 92 trillion).

IBRA’s task is to extract as much value as possible from this mixed bag of assets and accordingly to reduce the final cost to the general public. As the coun-try’s largest creditor over corporate entities, IBRA is also deeply involved in dif-ficult corporate restructuring negotiations. Even after several years of operation, recoveries remained mediocre: Rp 118 trillion by August 2003—only 22% of face value. This is in line with the experience of similar asset recovery agencies else-where in the world. The final cost to the general public, after deduction of recov-eries, can thus be expected to remain considerable.

THE SNA DIFFERENCE

The SNA (1993 version) is the internationally recognised standard for the compi-lation of national accounts. It is distinctly different from business accounting in key respects.

Purpose and Sequence of Accounts

Business accounting is designed to assess the financial condition of individual productive units (including banks), measure their profits, and determine the

enti-TABLE 2 Income Statement—All Banks (July 1998 – September 1999)

Amount Share of Total Losses ($ billion) (%)

Net interest income –10.3 20.3 Net interest income in Rp –10.7 21.1 Net interest income in foreign exchange 0.4 –0.7 Net non-interest income –3.1 6.0

Gross earnings –13.3 26.4

Operating costs –2.7 5.3

Provisions against losses –34.9 69.1

Operating result –50.9 100.7

Other revenues and charges 0.4 –0.7

Net profit after tax –50.6 100.0

Source: As for table 1.


(7)

tlement to those profits of the respective interested parties (mainly the sharehold-ers and the tax authorities). It is relied upon to assess compliance with purpose-oriented norms—in particular, taxation rules and, for banks, prudential norms. It is by design exhaustive and meticulously precise: all transactions must be recorded with utmost exactitude. It focuses mainly on two concepts: solvency (how much net assets are held by the entity in question, as shown in a balance sheet) and profitability (how large a surplus available for distribution is being generated, as shown in an income statement).

National accounting has a far broader objective: to describe the functioning of the economic system as a whole, to trace the flows of resources between entities and sectors, and to provide a reasonably reliable basis for sound public policy decision making. It focuses on economic value, its creation through productive activities, and its uses through consumption and investment. It does not pay any special attention to ‘profit’, the closest SNA concept probably being operating surplus. Aiming to approach the elusive economic reality, the SNA is far less meticulously precise than business accounting. It accepts mere estimates, and tol-erates repeated subsequent adjustments. It also provides a more comprehensive sequence of accounts: current accounts (including production, generation of income, allocation of primary income, secondary distribution of income, and use of income accounts); a set of several accumulation accounts; and balance sheets. This reflects the larger number of concepts identified and measured by the SNA, each account bringing out one of these concepts.

Booking of Transactions

Whereas business accounting requires a double entry (debit and credit) for the booking of transactions, national accounting is based on quadruple entry book-ing: each transaction is booked four times (i.e. twice for each of the two parties involved). Quadruple entry imposes a discipline that makes it automatic to doc-ument and quantify not only what is happening in a given sector of the economy (for instance, the banking sector), but also the balancing items in the rest of the economy (for instance, the corporate sector). This also expresses the fact that the rule of symmetry is essential to national accounting, whereas it does not exist in business accounting, which focuses only on a single economic entity at a time.

The SNA approach is thus more conducive to the elucidation and quantifica-tion of the links between the banking and the corporate sectors. Specifically, whenever a bank books a reduction in the net value of its loan portfolio (with a double entry booking of a decrease in the value of assets, and an increase in losses), the SNA makes it an absolute requirement that there be, at the same time, a balancing item of strictly equal value somewhere else (with a double entry booking also).

Valuation Methods

Business accounting norms generally require booking of transactions at historical cost (notwithstanding exceptions, such as market valuation for some securities). It has mechanisms to reduce the value of assets through depreciation and provi-sioning—for instance, provisioning to take into account the loss of value of non-performing loans. On the other hand, in order to prevent abuses, the booking of increases in the economic value of assets is heavily restricted. Revaluation of


(8)

fixed assets, for instance, would normally create a tax liability in most countries, and tends to be avoided for that reason.

The prudence principle requires business accountants to err on the side of cau-tion in valuing assets. When there is uncertainty, the valuacau-tion should be conser-vative. For example, if proper loan documentation is missing in a bank, prudence requires that the loan portfolio of that bank be considered as impaired to some degree, regardless of the financial condition of the borrowers concerned. When banking distress develops, the expected recoverable value of the assets falls, while at the same time the level of uncertainty rises. Both factors contribute to the sharp increase in loss estimates typically observed during banking crises under classic business accounting.

National accounting follows a markedly different valuation approach. Its main objective is not prudence, but economic realism, so ‘national accounts are based on the assumption that assets and liabilities are continuously revalued at current prices’ (SNA 1993: para. 2.69). The practical consequences of these differences in approach are considerable in the context of a systemic banking crisis. On the basis of SNA valuation norms, the losses to the Indonesian economy related to the problems in the banking sector would have been significantly smaller than the losses to the banking sector alone. When international auditors reviewed the loan portfolios of a group of six large banks in 1998, whenever they were unable to find proper documentation and evidence of collateral, they categorised the corre-sponding loans as total losses and required a 100% provision, amounting in aggregate to billions of dollars. This is consistent with the norms of the auditing profession and the principle of prudence, but not with economic realism. In real economic terms, the fact that a bank has provided the funding for some asset (e.g. productive equipment held by a corporation) and later lost its rights to that asset because of untidy paperwork does not mean that this asset has become worthless. From the SNA’s broader perspective, the relevant issue is that of the capacity of the financed asset to be used for production, and thus to create economic value.

Interest Margin and Intermediation Services

The difference between interest collected from borrowers and interest paid to depositors (i.e. the net interest margin) is what allows banks to make a living. Therefore, business accounting and prudential supervision pay considerable attention to the interest margin, as a core item for the assessment of bank prof-itability and thus viability. Banks that feature a negative net margin are viewed as inefficient, loss-making institutions, slated for closure if they cannot improve their performance. The focus is on measuring the performance of individual banks separately from that of other types of economic entities.

Macroeconomic accounting takes a broader view, but also aims at measuring banks’ economic output through their interest margins, in spite of considerable conceptual and practical difficulties.4The starting point is the fact that banks

pro-vide a service to depositors and borrowers—financial intermediation—for which they do not charge explicitly. As a result, there are no receipts from ‘sales of finan-cial intermediation services’ that could be used directly to value these services, so banks would seem to have no economic output. The implicit charge for the serv-ice is incorporated into the interest rates, with the rates paid to depositors nor-mally lower than those charged to borrowers.


(9)

In an attempt to address this issue and to better reflect economic reality, the lat-est internationally sanctioned version of the SNA, issued in 1993 under the aus-pices of the United Nations, introduced the concept of financial intermediation services indirectly measured (FISIM). The FISIM concept requires that interest payments be split conceptually into two items, a commodity and a service. The commodity is the funding provided by depositors (through the banks) to borrow-ers, the price of which is the same for both borrowers and depositors, while the service is financial intermediation, the price of which may differ between these groups. In this conceptualisation, the cost of funds provided to banks is the same as the revenue gained from them: banks obtain no income from providing funds, since they do not ‘produce’ them, but only intermediate them. Their income is derived from the intermediation service.

This separation between the service provided—which by and large remains constant even during a crisis—and the commodity makes it clear that when banks feature a negative interest margin they in fact sell intermediation services at a negative price, i.e. they provide subsidies. When such negative value sales reach several points of GDP and are followed by a recapitalisation of the banks by the government, the banks appear to be playing the role of quasi-fiscal agen-cies allocating subsidies to the other sectors of the economy, as discussed in more detail below (see The Negative Output Aberrationbelow).

A NEW PERSPECTIVE ON THE CRISIS

A quite different impression of the banking crisis can be obtained by relying, to a large extent, on the analytical framework and the broad concepts provided by the SNA. The system is flexible, and may be implemented at different levels of aggre-gation. Thus it is possible to rely on the SNA as an analytical tool to review devel-opments affecting a single economic sector (e.g. banking) during a specific time period (e.g. the 15-month long peak crisis period). Indeed, an analysis relying on broad SNA principles provides original answers to several important questions about the crisis and reveals a significantly different economic landscape.

The Meaning of ‘Bank Losses’

The SNA provides provocative answers to the core question: how did the Indo-nesian banks manage to lose $50 billion in 15 months? The first answer from the SNA is that the concept of ‘bank losses’, derived from commercial accounting, is at the same time too vague and too narrow. Moreover, in so far as it tends to con-vey the impression of destruction of value, the term ‘bank losses’ may be mislead-ing.

The existence of ‘bank losses’, as understood in business accounting, means that, at the end of the period considered, based on the conventionally accepted evaluation norms of the business, the banks have no net surplus available for allocation to the shareholders and the tax authorities, but rather a deficit for which someone has to take responsibility. This information is certainly important for the interested parties, but from the macroeconomic point of view it has a far more limited value. It is relevant to assess the condition, and decide upon the fate, of individual corporate entities, including banks, but is less useful for the assess-ment of what happened in an economy as a whole.


(10)

The SNA provides not one but three concepts expressing the economic result of the activity of banks. To illustrate this point, the income statement of the Indo-nesian banks has been translated into an SNA-based sequence of accounts (table 3).5

The first concept offered by the SNA is value added, i.e. the sector’s contribu-tion to the overall creacontribu-tion of wealth in the country (GDP). The second is operat-ing surplus, reflectoperat-ing the result of the core business. And finally, after takoperat-ing into account all other transactions, the third concept makes apparent the bottom line, i.e. the change in net equity.

Table 3 shows that the dramatic $50 billion decline in bank equity over the 15-month period under consideration has two main origins: a sharp decrease in the volume of the assets (mainly write-downs and write-offs of customer loans),

TABLE 3 SNA-Based Presentation of the Income Statement—All Banks (July 1998 – September 1999)

Amount Share of ($ billion) Total Losses

(%)

1 Production

Output –8.3 16.4

FISIMa –10.3 20.3

FISIM—customer transactions in Rp –12.5 24.7 FISIM—all other transactions 2.2 –4.3

Non-FISIM output 2.0 –3.9

Intermediate consumption plus consumption of fixed capital –1.8 3.7

Value added—net –10.1 20.1

2 Generation of income

Compensation of employees –0.8 1.6

Operating surplus –11.0 21.7

3 Secondary distribution of income

Taxes on income –0.3 0.5

Savings –11.2 22.2

4 Changes in volume of assets

Losses on assets –38.4 75.8

of which

Losses on customer loans –33.2 65.7

5 Revaluation

Holding losses (on foreign currency denominated items) –1.0 2.0

Change in net equity –50.6 100.0

aFinancial Intermediation Services Indirectly Measured.

Source: Author’s estimates (see appendix).


(11)

amounting to $38 billion, and a negative output of financial intermediation serv-ices for customer transactions (i.e. taking deposits and granting loans) in domes-tic currency ($12 billion). By comparison, the cost of running the banks themselves—their intermediate consumption, consumption of fixed capital, and the compensation paid out to their staff—appears modest.

The Negative Output Aberration

When each component of the banks’ income statement is reclassified into the SNA’s sequence of accounts (table 3), a massive negative output of financial intermediation services, concentrated in the customer transactions denominated in rupiah, becomes apparent ($12.5 billion over 15 months). For the banking supervisor, this is no surprise: it is just another way to reflect the fact that banks suffered from deeply negative interest margins. But from the SNA point of view, this is a serious aberration.6

It is easily conceivable that an inefficient economic entity will generate nega-tive value added if its intermediate consumption is larger than its output. How-ever, how can output itself be negative? If a good has been produced, a service delivered, how can this physical productive activity be accounted for at less than zero? The illogicality of negative bank output is confirmed by looking at the bal-ancing items. Financial intermediation services—the output of banks—are a resource used in other sectors of the economy, as final consumption by house-holds and as intermediate consumption by corporations. Thus if the banks’ out-put of financial services is negative, as a result, the corresponding final consumption of the households will also be negative, as well as the intermediate consumption of financial services by the corporate sector. This has a discernible impact on GDP as a whole, since the contribution of the banks to Indonesian GDP in 1996, the last full year before the crisis, amounted to 4.1%.

As mentioned in the previous section, the concept of FISIM relied upon by the SNA splits interest flows into two components: the cost of a commodity (funding) and a charge for a service (financial intermediation), which includes collection, processing and distribution of the said commodity. The service is billed implicitly by banks through normally positive interest margins (namely, the difference between the higher rate charged to borrowers and the lower rate paid to deposi-tors), whereas the price of the commodity itself (the ‘reference rate’) is the same for both sellers and buyers.

When, as in Indonesia’s crisis, interest margins become deeply negative system-wide for an extended period, it appears that banks are distributing the commodity that is the basis of their trade at a subsidised price. This subsidy can be characterised as a third component of the interest flows, in addition to the cost of the commodity itself and that of the service. The subsidy is defined as a cur-rent transfer, and it can also be estimated.

A tentative quantification of the subsidy component incorporated in interest flows was carried out by the author. The key assumption is that the banks’ out-put of financial intermediation services, as measured through FISIM, remained unaffected by the crisis, except for the impact of changes in the volume of loans and deposits. This is predicated on the fact that the physical components of the delivery of financial intermediation services (maintaining a network of points of sale, processing transactions, and carrying out related administrative tasks)


(12)

remained in place for all but the closed banks. Based on historical GDP data, the normal level of FISIM was estimated and then extrapolated and adjusted for the crisis period. The subsidy component was determined as the difference between the normal, pre-crisis level of FISIM as adjusted, and the level as measured from the banks’ financial statements (see appendix). On this basis, an enhanced ver-sion of the SNA-based presentation of the banks’ income statement was pre-pared, in a preliminary attempt to better reflect the economic reality (table 4).

With this modification output becomes positive again, by construction, and the subsidy component explaining the negative interest margin is featured as a cur-rent transfer in the secondary distribution of income account. Table 4 indicates

TABLE 4 Enhanced SNA-Based Presentation of the Income Statement—All Banks (July 1998 – September 1999)

Amount Share of ($ billion) Total Losses

(%)

1 Production

Output 8.717.2

FISIMa 6.7 –13.2

FISIM estimate—customer transactions in Rp 4.5 –8.9 FISIM—all other transactions 2.2 –4.3

Non-FISIM output 2.0 –3.9

Intermediate consumption plus consumption of fixed capital –1.8 3.7

Value added—net 6.8 –13.5

2 Generation of income

Compensation of employees –0.8 1.6

Operating surplus 6.0 –11.9

3 Secondary distribution of income

Taxes on income –0.3 0.5

Transfers to customers (subsidy included in

negative net interest rates) –17.0 33.6

Savings –11.2 22.2

4 Changes in volume of assets

Losses on assets –38.4 75.8

of which

Losses on customer loans –33.2 65.7

5 Revaluation

Holding losses (on foreign currency denominated items) –1.0 2.0

Change in net equity –50.6 100.0

aSee table 3, note a.

Source: Author’s estimates (see appendix).


(13)

that banks continued to create net value throughout the crisis, which should not come as a surprise since they were still delivering the service that is their raison d’être, while their own consumption costs (of running the business) remained quite moderate. Even if their creation of value was substantially lower than before the crisis, the banks continued to post a positive operating surplus. Thus, the core banking activity remained ‘profitable’, in the sense that the banks still created economic value for the benefit of the other sectors of the economy, above and beyond the cost of running their business.

According to these estimates, the dramatic decline in equity that brought down the banking system came mainly from the transfer of real economic resources to other sectors of the national economy and/or to the rest of the world. A large frac-tion of what seems to have been interpreted under business accounting as a loss of economic value for the country, without further elucidation, is seen in fact to be only a redistribution of wealth among sectors. Most of the economic substance lost by the banks has been captured by the depositors and the borrowers who benefited from abnormally favourable interest rates, and also by the borrowers whose loans have been written down or off by the banks. Rough estimates based on the allocation of bank credit as of end 1998 suggest that 82% of the total (some $41 billion) went to the corporate sector, while the balance of 18% ($9 billion) ben-efited households. The vast majority of the beneficiaries appear to be residents, so that the direct, first-round transfers from the banks to the rest of the world seem to have been limited. The wealth redistribution—at least in the first round— took place mainly among Indonesian residents.

This might also explain why banks continued operating in spite of deeply neg-ative interest margins that caused them to incur escalating losses, instead of sim-ply shutting down, as standard microeconomic rationality would have dictated under such circumstances. Bank owners had nothing more to lose once their equity stakes were gone, early in the process. On the other hand, they retained full management control of their banks for a protracted period. For a long time too, central bank emergency liquidity support was accessible and plentiful. Its apparently high cost was theoretical, since the beneficiary banks were already bankrupt and knew that they would never repay the central bank. While this lasted, bank owners retained the opportunity to offer attractive rates to deposi-tors, to grant concessions to borrowers at their discretion, and even to extend new loans. Hence, to a large extent, these bank owners were in control of the huge wealth redistribution exercise that took place during 1998–99 at the general pub-lic’s expense.

To sum up, the SNA tells us that even if the banking sector’s core profitability had been substantially eroded, it survived throughout the peak crisis period. A large fraction of the so-called ‘bank losses’ were, in real economic terms, massive transfers from the banking sector to the domestic corporate sector and also to a lesser extent to household sector borrowers and depositors. Insofar as the bank-ing sector’s equity was subsequently rebuilt through injections of public funds paid for with government bonds, it appears that banks were temporarily trans-formed into government agencies distributing subsidies funded through public debt.


(14)

The Impact of the Crisis on the Corporate Sector

The Pre-crisis Situation. The SNA also sheds additional light on the corporate sec-tor’s position as the main direct beneficiary of the banking crisis. At the onset of the crisis, available estimates indicated that there was substantial equity in the corporate sector, around six times more than in the banking sector (table 5). Taken at face value, these data reflect a financial structure by and large adequate, for both banks and corporations. As reported, the equity position of the banks is compatible with international prudential norms.7Loans and deposits appear well

balanced, representing the bulk of assets and liabilities, as expected from a func-tioning banking system. Loans in foreign currencies, at only 25% of the total port-folio, seem reasonable for an open, export-oriented economy. They are commensurate with the volume of deposits in foreign currencies. As regards cor-porations, their leverage ratio, although a little high, is well in line with that of a number of other countries, including stable advanced economies.

The vulnerabilities are more apparent in relation to governance. Banks and cor-porations were closely related—the former in a relationship of subordination to the latter, often under common owners of large conglomerates. In this context, these owners had far more important stakes in their corporations than in their

TABLE 5 Estimated Financial Structure of Banks and Corporations (as of June 1997; $ billion)

Banks Corporations Banks and Corporations

Assets 226 310 536

of which

Loans 146

In rupiah 110 In foreign currencies 36

Liabilities 209 210 419

of which

Deposits 132

In rupiah 98

In foreign currencies 34

Equity 17 100 117

Assets : liabilities (%) 108 148 128 Leverage ratio (%) 12.3 2.1 3.6

Sources: Author’s estimates for the banks and, for the corporations, estimates prepared by the October 1997 IMF negotiation mission to Indonesia, based on data provided by the Jakarta branch of the Indonesian Chamber of Commerce and Industry, Kadin.


(15)

banks. The resources originally invested to provide capital to the banks had been recouped several times over through the extension of loans to related businesses and the provision of financial services on favourable terms. There was a strong incentive, in the event of severe financial distress, to sacrifice the banks—with a loss limited to a relatively small equity stake—and preserve the far more valuable corporations.

The abundant corporate equity could have been tapped at the onset of the cri-sis to absorb real sector losses. Instead, the losses were transferred to the banks, and then on to the general public, thus largely shielding corporate equity. IBRA, the bank restructuring agency, then tried to mitigate the final cost by extracting some economic value from the corporations but, coming into the process late and ill equipped, faced an uphill battle fraught with obstacles.

A Far Smaller Real Economic Impact. The SNA-based approach leads to the con-clusion that the real economic impact of the crisis on corporations was far smaller than business accounting indicates. Whereas the productive assets of the corpo-rations domiciled in Indonesia are recorded in rupiah in their balance sheets, a sizeable proportion of these physical assets are of foreign origin, have some for-eign content and/or are geared toward the production of tradable goods and services. These and other assets were also, in many cases, funded through loans denominated in foreign currencies. When the rupiah’s external value was sud-denly divided by three or four, the liabilities of the corporations denominated in foreign currencies were multiplied by three or four on the basis of applicable business accounting norms, while their productive assets, booked at historical cost in rupiah, remained unchanged. Large numbers of corporations instantly appeared to have become bankrupt, holding less in assets than in liabilities. How-ever, this was not the case in real economic terms.

As a consequence of the rules of business accounting, the impact of the depre-ciation of the rupiah on corporate balance sheets was taken into account asym-metrically: the liabilities (loans in foreign currencies) were revalued upwards, while the assets were not. These conventional business accounting rules are appropriate and prudent in a stable macroeconomic environment. However, when there is a major systemic upheaval they substantially distort economic real-ity. A valuation approach that gives priority to economic reality—broadly in line with the spirit, if not the exact letter, of the SNA principles—would provide a more accurate picture. If asset values were estimated, say, on the basis of the replacement price, the corporations’ losses would be far smaller, for the simple reason that they hold valuable physical assets that provide a natural hedge against both inflation and exchange rate depreciation.

More generally, there is also the issue of double counting of the impact of the crisis under business accounting. Because of the sharp depreciation of the rupiah, large sections of the corporate sector appeared insolvent, and stopped servicing their loans. Thus it was the banks’ turn to appear insolvent. The widespread view that both banks and their corporate borrowers were bankrupt as a result of the crisis is consistent with business accounting’s microeconomic purpose: to reflect accurately the situation of each economic unit separately, without taking into account the global picture. In this framework, the same wave of economic losses is shown to destroy first the corporate sector and then the banking system.


(16)

The SNA, with a different and broader approach (relying, in particular, on quadruple entry accounting and the principle of symmetry) concludes otherwise: at the macroeconomic level, economic losses should be taken into account only once, either in the corporate sector or in the banking sector, but not in both. In other words, it is necessary to pay only once for economic damage, not twice. If, as was the case in Indonesia, the general public is footing the bill by recapitalis-ing banks with government bonds, then the banks are in a position to write off their non-performing loans.8 Quadruple entry accounting and the symmetry

principle require a simultaneous cancellation, in the corporate balance sheets, of the corresponding liabilities toward the banks, thus restoring the corporate sec-tor’s solvency.

Real Economic Significance of the Transfers. The next analytical step is to identify the real economic significance of these massive transfers from the banks to the corporate sector. Either the transfers compensate for real, final losses that previ-ously had been hidden in corporate balance sheets (probably having accumu-lated silently over a number of years) or they represent windfall gains for the beneficiaries.

In the first case, as the SNA indicates through its sequence of accounts, this would mean that large amounts of resources provided by banks to corporations over the years, and supposedly used to increase the stock of fixed assets, were in fact used to fund intermediate consumption by firms. In other words, part of the intermediate consumption of the corporate sector was erroneously booked as investment. This would mean that the value added of the corporate sector, and thus the country’s GDP as a whole, was overstated during this period. The poten-tial magnitude of the distortion is such that this could lead to significant down-ward revisions of growth figures over a number of years.

The second possibility, namely windfall gains, would represent a clear misap-propriation of public funds. At first glance this might seem less likely than com-pensation of past real losses. However, upon closer examination, in the general climate of confusion and lack of transparency that prevailed during the crisis, it seems likely that at least isolated cases of such windfall gains have occurred.

At this stage, it is not possible to determine the respective proportions, in the transfers from the banks to the corporations, of the funding of ‘honest mistakes’ (real, final losses resulting from bad investment decisions) and of the inappropri-ate transfers of value. But a few points can be made in this regard. First, this dis-tinction is important, for reasons of both equity and economic efficiency. Second, it seems that this kind of forensic work has not been attempted before, and there is no doubt that it would meet serious conceptual and practical difficulties.

On the other hand, there is no reason why the reclassification of accounts into the SNA framework described here (tables 3 and 4) for the banking sector as a whole could not be carried out for selected individual banks and large delinquent borrowers. An appropriately calibrated multi-year set of accounts, defined jointly by a team of banking supervisors and national accountants, linking opening bal-ance sheets, current and accumulation accounts, and closing balbal-ance sheets, has the potential to trace, characterise and quantify the most significant transfers of value. Furthermore, an SNA-based custom-made analytical framework could be defined and experimented with. Beyond the specific case of Indonesia, the


(17)

results—and even the lack thereof—might yield valuable lessons with regard to crisis prevention and management.

In both cases, from an economy-wide point of view, the first round effect of the transfers from banks to corporations is neutral, with no net real economic impact. If they compensated for hidden losses suddenly brought to the open, the fact that the losses were previously hidden does not mean that they did not exist. Nor did the fact that they were made apparent increase their size: the aggregate amount of real wealth in the national economy remains unaffected. The same applies in case of a windfall transfer in favour of the recipient corporations. The corporate sector is better off, the banks are poorer, but the economy as a whole remains unaffected, as a first round effect. However, these transfers in favour of domestic corporations might have provided them with resources that were subsequently transferred abroad—for instance, by switching the ownership of a corporation from a resident to a non-resident, or to a resident’s offshore holdings. In the case of such second round transfers, there would be a real loss of wealth for the national balance sheet, in favour of the rest of the world.

IMPROVING BANK RESOLUTION STRATEGIES

As the previous section shows, the SNA has the potential to provide a distinctly different perspective on the Indonesian banking crisis. The availability of addi-tional and alternative information on the nature and the details of the crisis should provide opportunities to improve bank resolution strategies in the future. The best outcome would be to avoid the crisis altogether; this is the important issue of prevention, and will be discussed last. Short of preventing crises, the objective is to minimise their impact (mainly the fiscal cost) and accelerate the return to normalcy. There is also a need to improve the burden sharing arrange-ments and, in particular, to avoid leaving the general public to carry a dispropor-tionate share of the burden.

In practical terms, there are three main issues to address. The first is to contain and prevent further accumulation of economic losses. The second is to measure the damage accurately, including separating the losses of real economic substance from mere redistribution of wealth among sectors. The third is the damage allo-cation process. There is always a natural, spontaneous alloallo-cation of any loss of value that occurs in the economy. This allocation process has to be elucidated and checked for appropriateness, on both fairness and efficiency grounds. Then, if needed, corrective mechanisms have to be designed and implemented, relying on specific instruments.

Minimising the Costs

To control the overall cost, the first priority is to stop any further accumulation of losses. The SNA could provide a useful analytical framework to locate precisely the sources of real economic losses and to set the right priorities in tackling them systematically. The first step (in logical, not necessarily chronological, order) is to separate consumption from mere transfers of value. Consumption by the banks themselves happens to be relatively easy to deal with. It should be strictly con-tained. In practical terms, this means close scrutiny and tight controls over banks’ overhead expenses.


(18)

The next step is to focus on transfers of value out of the banks. These may be directed either to other sectors of the national economy or to the rest of the world. Priority should be given to containing any transfers of value from the domestic banks to the rest of the world, because these have a direct detrimental impact on the national net worth, and are usually more difficult to reverse afterwards. This means close review and, as far as possible and appropriate, containment of trans-fers of wealth abroad.

Still one step further is the review of transfers out of banks in the direction of other domestic sectors. Although these have no immediate impact on aggregate national net worth, a distinction should be made between those transfers that allow or facilitate consumption by their recipients and those that do not, so that priority can be given to containing the former. The former include, for instance, any flow of cash to a loss-making corporation (e.g. granting of new loans), or pay-ment of high interest rates, or paypay-ment of dividends. The latter would include, for instance, provisioning for an existing loan (a charge without any cash out-flow).

The process can be further refined by taking advantage of the capability of the SNA framework to be disaggregated at any level. The financial statements of individual banks and corporations of a significant size can be processed to pro-duce individual sequences of accounts based on the SNA. In the context of a cri-sis, this exercise could be applied to a limited number of systemically important banks and large borrowers. Full accuracy and minute detail are unimportant under such circumstances: broad orders of magnitude are sufficient for deciding if action is warranted. On this basis, the main sources of losses as detailed above can be identified and also broadly quantified, with a view to optimising the allo-cation of resources available to prevent the accumulation of further losses.

Under systemic crisis circumstances, auditors conducting special reviews of bank assets (separately from and in addition to regular annual audits) could be asked to rely on SNA concepts and norms in lieu of the usual prudential approach. They would, for the main borrowers, assess the economic value of the assets on the basis of replacement value, not historical cost, and separately iden-tify formal problems that have no impact on the substantial value of the assets held by the borrowers, but that might compromise the bank’s access to the assets funded (including lack of collateral and untidy loan files and legal documenta-tion). This alternative approach should lead to the identification of substantially smaller losses.

Optimising the Burden Allocation Process

If the first priority is to minimise the overall cost of the crisis, the second is to implement appropriate burden sharing arrangements. In the case of Indonesia, the vast majority of the costs were allocated to the general public. This is both unfair and economically inefficient, resulting in a public debt so large as to com-promise the medium-term economic future of the country. Moreover, by provid-ing wrong incentives it has also contributed to increases in the costs. One egregious example is that of the state-owned banks. Up to the middle of 1998, when the bank resolution strategy was still under discussion, the state-owned commercial banks hid their losses, pretending to be healthy and posting financial statements grossly out of line with economic reality. As soon as it became clear


(19)

that these banks would all be recapitalised with public money they began to book losses with abandon, as can be seen in their financial statements.

The SNA has the potential to provide practical solutions as to how the burden sharing exercise can be improved. Since the crisis was, to a large extent, a wealth redistribution exercise, there have been both winners and losers. The winners have captured some economic value, to the detriment of the losers. There does not seem to be any reason why the value captured could not be recycled, how-ever. The proper way to do so seems to be through taxation. Just as there is now, in a large number of countries, a value-added tax (VAT), based on a concept rooted in national accounting, there could be, during or following systemic bank-ing crises, a temporary value-captured tax (VCT), also based on the SNA analy-tical framework. Conceptual and pracanaly-tical difficulties have been overcome in the case of the VAT, and could be solved in the case of the VCT too.

The VCT should be temporary, and applicable only to a limited number of tax-able events based on specific definitions for the targeted transfers of value—such as the writing down or writing off of bank loans, payment by banks of high inter-est on their deposits or discounts on interinter-est billed by banks on their loans. There should also be a size threshold large enough to keep small depositors and busi-nesses outside the system and the number of concerned transactions at a manage-able level, in line with local administrative implementation capacities.

Symmetry is essential to the working of the system. Originators of the value transferred—in most cases, the banks—would book a VCT credit (equivalent to a claim on government) instead of posting losses, thus preserving their solvency. A VCT debit certificate of the same amount would be simultaneously issued and notified to both the debtor and the tax administration.

If the VCT debtor—in most cases, a large corporation—does not own sufficient assets to meet its fiscal liabilities, the resources provided by the bank will have been lost. The corporation will be clearly identified as bankrupt, and treated accordingly, with the tax administration as a creditor. If, on the other hand, the VCT liability corresponds to assets of some economic value, the tax administra-tion will be in a stronger posiadministra-tion to exercise its legitimate claim on those assets, thus minimising the final cost for the general public.

Prevention

Finally, as regards prevention policies, it is important to focus on the possibility of an accumulation in the economy, over a number of years, of undetected losses on bank loans. The place where these problems develop is the corporate sector, not the banks. The problem to address is a hidden accumulation of losses, which might be slow and remain undetected for years, when corporations book as investments what is in fact (intermediate) consumption, funded through bank loans. Banks have an important role to play in preventing the silent accumulation of such losses. Prudential reviews of loan portfolios should generally be able to detect them, although this will not always be the case. Too much faith might be put in historical costs and/or in the significance of the borrower’s repayment his-tory. Banking supervisors need to pay special attention to this issue.

An essential aspect of prevention is preparedness. Experience has shown that severe or systemic banking distress may appear in all categories of countries, regardless of their economic structure or level of development, but that their


(20)

occurrence in a given country is infrequent, often only once in a generation or two. As a result, it makes sense to separate preparedness into two components: at the level of each country, basic, low cost preparatory measures of a general nature and, at a multilateral level, the building up and maintenance of a permanent cri-sis management capability on which individual countries might rely, if they wish, as the need arises.

At country level, preparation should consist mainly of the drafting of a contin-gency plan and a review and adaptation of the relevant existing laws to ensure that they would be conducive to the implementation of an appropriate bank res-olution strategy in case of a crisis. At the multilateral level, the international financial institutions have a natural responsibility to maintain a permanent capa-bility to assist in the management by national authorities of a systemic banking crisis that might erupt in a specific country. This capability already exists to some extent, but the considerable potential of the SNA framework in this regard remains largely unexploited. Substantial work would be needed to take advan-tage of the SNA’s potential, but the rewards would probably be commensurate.

CONCLUSIONS

The coherent analytical framework provided by the SNA can be of help in discov-ering what happened with the $50 billion ‘lost’ by the Indonesian banks. The business notion of ‘loss’ is vague, inconsistent with macroeconomic concepts and even misleading. By contrast, an SNA-based review shows that most of this amount has not disappeared from the economy, but has been simply transferred out of the banks, mainly to the corporate sector.

The SNA approach provides a new perspective on the crisis, its origins, its nature and its repercussions. Ultimately the crisis was mainly a large-scale wealth redistribution exercise, neutral for the banks, beneficial for the corporations and, to a lesser extent, for household borrowers and depositors, but disastrous for the general public. The SNA has the potential to provide an analytical framework to identify, characterise and quantify these flows.

The SNA perspective also points to possible improvements in bank resolution strategies. Since the problem is mainly that of an opaque and unintended large-scale redistribution of wealth, improved resolution approaches should include bringing full transparency to the process, quantifying the redistribution on the basis of SNA concepts, and then recycling the captured wealth through adequate taxation of the beneficiaries.

Such resolution strategies could not have been implemented in Indonesia at the time of the crisis, however, because of a lack of operational readiness that per-sists to this day. A number of critical assumptions still need to be vetted, and seri-ous practical obstacles overcome. The SNA framework, notwithstanding its huge promise, would still need to be reviewed, adjusted and developed before it could play any innovative role in managing banking crises.

Ultimately there is one general conclusion: the SNA has promising potential to assist in the analysis of a large-scale financial disaster of macroeconomic impor-tance, thanks to its conceptual integrity as a global and coherent framework. For the sake of the rapidly globalising economy, where systemic banking crises seem to multiply, this promising potential should not be left untapped any longer.


(21)

NOTES

* The views expressed in this article are those of the author and should not be inter-preted as reflecting the views of the International Monetary Fund.

1 The Indonesian banking crisis has been covered, for example, in Djiwandono (2004, in this issue); Enoch et al. (2001); Fane and McLeod (2002); and Lindgren et al. (1999). 2 Data as of end June 1997.

3 Indonesia did not have a formal deposit insurance scheme when the crisis began; it was in the process of designing such a scheme at the time of writing.

4 The current approach is the result of a compromise that emerged progressively among the international community of national accountants, after decades of debates, tensions and reversals in practices. For an elucidation of the issues and the historical back-ground to the discussions, see Vanoli (2002: 199–203). For the details of the outcome, see SNA (1993), Production Account, The Output of Particular Industries, Financial Intermediaries, beginning at paragraph 6.120.

5 For this exercise, write-downs on loans (i.e. provisions for expected future losses) have been treated similarly to write-offs (i.e. final losses), as negative changes in the volume of assets. This is not in line with the current treatment of non-performing loans by the SNA (an issue currently being discussed in international expert circles), but reflects the economic reality as seen from the banking supervisor’s point of view.

6 Note that the data published by the Central Statistics Agency show a positive overall contribution of the banking sector to GDP throughout the crisis years: 4.0% in 1997 and 1998, and 2.2% in 1999. Elucidation of this discrepancy is beyond the scope of this paper (see appendix).

7 There are strong suspicions that this equity position, as declared by the banks, did not reflect their real financial position, however, because of under-provisioning for non-performing loans.

8 For a discussion of the treatment of non-performing loans in macroeconomic accounts, see Bloem and Gorter (2001).

REFERENCES

Bloem, Adriaan M., and Cornelis N. Gorter (2001), ‘The Treatment of Nonperforming Loans in Macroeconomic Statistics’, IMF Working Paper No. 01/209, December. Djiwandono, J. Soedradjad (2004), ‘Liquidity Support to Banks during Indonesia’s

Finan-cial Crisis’, Bulletin of Indonesian Economic Studies40 (1), in this issue.

Enoch, Charles, Barbara Baldwin, Olivier Frécaut and Arto Kovanen (2001), ‘Indonesia: Anatomy of a Banking Crisis—Two Years of Living Dangerously—1997–99,’ IMF Work-ing Paper No. 01/52, International Monetary Fund, WashWork-ington DC, May.

Fane, George, and Ross H. McLeod (2002) ‘Banking Collapse and Restructuring in Indo-nesia, 1997–2002’, Cato Journal22 (2): 277–95.

Frécaut, Olivier (2002), ‘Banking System Losses in Indonesia—Looking Out for Fifty Bil-lion U.S. Dollars—Can the SNA Help?’, Paper prepared in June for the 27th General Conference of the International Association for Research in Income and Wealth (Stock-holm, Sweden, August 2002), parallel session 4B (www.eco.nyu.edu/iariw).

Lindgren, Carl-Johan, Tomas J.T. Balino, Charles Enoch, Anne-Marie Gulde, Marc Quintyn and Leslie Teo (1999), ‘Financial Sector Crisis and Restructuring—Lessons from Asia,’ IMF Occasional Paper No. 188, International Monetary Fund, Washington DC. SNA (1993), System of National Accounts 1993, Prepared under the auspices of the

Inter-Secretariat Working Group on National Accounts; Commission of the European Com-munities; International Monetary Fund; Organization for Economic Cooperation and Development; United Nations; World Bank; Brussels/Luxembourg, New York, Paris, Washington DC.

Vanoli, André (2002), Une Histoire de la Comptabilité Nationale [A History of National Accounting], Editions La Découverte, Collection Manuels Repères, Paris.


(22)

APPENDIX: SOURCES AND PROCESSING OF DATA

The core data relied upon in this paper are the losses of the Indonesian banks dur-ing the 15-month period from July 1998 to September 1999. These loss estimates, totalling Rp 434.9 trillion (table 1), were prepared by the author as part of his assignment as a banking sector expert to the IMF team in charge of monitoring the banking crisis. The overall accuracy of the aggregate loss estimate was con-firmed subsequently by comparing it with publicly available data—namely, the variation in the net equity position of the banking system (a decrease of Rp 457.6 trillion, taking into account equity injections in the form of government bonds), as it appears in Bank Indonesia’s monthly publication, Indonesian Financial Statis-tics.

Details about these data and their processing are provided in three technical appendices to the author’s original June 2002 paper (Frécaut 2002) introducing the thesis summarised in the present paper.

• Appendix II of the 2002 paper provides details of the loss estimates and of the accuracy assessment.

• Appendix III elucidates, through two detailed tables, the transposition of the components of the banks’ income statements into the SNA’s accounts and headings. For instance, the different categories of interest received and paid (in the income statement) have been netted and each result assigned as FISIM out-put to the SNA’s production account.

• Appendix IV describes the steps followed (including the actual numbers) to extract the subsidy element in the negative interest margins as part of the pro-posed alternative definition of FISIM. Based on historical GDP data, the nor-mal level of FISIM has been estimated, and then extrapolated and adjusted for the crisis period. The subsidy element embedded in interest rates during the crisis has then been obtained as the difference between ‘normal’ and measured FISIM.

The whole exercise is a preliminary attempt at a new approach to the elucida-tion of a particularly severe systemic banking crisis. The focus has been limited to the identification of plausible orders of magnitude, for bank losses only. At this early stage, no detailed and systematic checks have been conducted on the valid-ity of the other data and, in particular, no reconciliation has been attempted between the data obtained as described above and actual national accounts data for the crisis period.


(1)

results—and even the lack thereof—might yield valuable lessons with regard to crisis prevention and management.

In both cases, from an economy-wide point of view, the first round effect of the transfers from banks to corporations is neutral, with no net real economic impact. If they compensated for hidden losses suddenly brought to the open, the fact that the losses were previously hidden does not mean that they did not exist. Nor did the fact that they were made apparent increase their size: the aggregate amount of real wealth in the national economy remains unaffected. The same applies in case of a windfall transfer in favour of the recipient corporations. The corporate sector is better off, the banks are poorer, but the economy as a whole remains unaffected, as a first round effect. However, these transfers in favour of domestic corporations might have provided them with resources that were subsequently transferred abroad—for instance, by switching the ownership of a corporation from a resident to a non-resident, or to a resident’s offshore holdings. In the case of such second round transfers, there would be a real loss of wealth for the national balance sheet, in favour of the rest of the world.

IMPROVING BANK RESOLUTION STRATEGIES

As the previous section shows, the SNA has the potential to provide a distinctly different perspective on the Indonesian banking crisis. The availability of addi-tional and alternative information on the nature and the details of the crisis should provide opportunities to improve bank resolution strategies in the future. The best outcome would be to avoid the crisis altogether; this is the important issue of prevention, and will be discussed last. Short of preventing crises, the objective is to minimise their impact (mainly the fiscal cost) and accelerate the return to normalcy. There is also a need to improve the burden sharing arrange-ments and, in particular, to avoid leaving the general public to carry a dispropor-tionate share of the burden.

In practical terms, there are three main issues to address. The first is to contain and prevent further accumulation of economic losses. The second is to measure the damage accurately, including separating the losses of real economic substance from mere redistribution of wealth among sectors. The third is the damage allo-cation process. There is always a natural, spontaneous alloallo-cation of any loss of value that occurs in the economy. This allocation process has to be elucidated and checked for appropriateness, on both fairness and efficiency grounds. Then, if needed, corrective mechanisms have to be designed and implemented, relying on specific instruments.

Minimising the Costs

To control the overall cost, the first priority is to stop any further accumulation of losses. The SNA could provide a useful analytical framework to locate precisely the sources of real economic losses and to set the right priorities in tackling them systematically. The first step (in logical, not necessarily chronological, order) is to separate consumption from mere transfers of value. Consumption by the banks themselves happens to be relatively easy to deal with. It should be strictly con-tained. In practical terms, this means close scrutiny and tight controls over banks’ overhead expenses.


(2)

The next step is to focus on transfers of value out of the banks. These may be directed either to other sectors of the national economy or to the rest of the world. Priority should be given to containing any transfers of value from the domestic banks to the rest of the world, because these have a direct detrimental impact on the national net worth, and are usually more difficult to reverse afterwards. This means close review and, as far as possible and appropriate, containment of trans-fers of wealth abroad.

Still one step further is the review of transfers out of banks in the direction of other domestic sectors. Although these have no immediate impact on aggregate national net worth, a distinction should be made between those transfers that allow or facilitate consumption by their recipients and those that do not, so that priority can be given to containing the former. The former include, for instance, any flow of cash to a loss-making corporation (e.g. granting of new loans), or pay-ment of high interest rates, or paypay-ment of dividends. The latter would include, for instance, provisioning for an existing loan (a charge without any cash out-flow).

The process can be further refined by taking advantage of the capability of the SNA framework to be disaggregated at any level. The financial statements of individual banks and corporations of a significant size can be processed to pro-duce individual sequences of accounts based on the SNA. In the context of a cri-sis, this exercise could be applied to a limited number of systemically important banks and large borrowers. Full accuracy and minute detail are unimportant under such circumstances: broad orders of magnitude are sufficient for deciding if action is warranted. On this basis, the main sources of losses as detailed above can be identified and also broadly quantified, with a view to optimising the allo-cation of resources available to prevent the accumulation of further losses.

Under systemic crisis circumstances, auditors conducting special reviews of bank assets (separately from and in addition to regular annual audits) could be asked to rely on SNA concepts and norms in lieu of the usual prudential approach. They would, for the main borrowers, assess the economic value of the assets on the basis of replacement value, not historical cost, and separately iden-tify formal problems that have no impact on the substantial value of the assets held by the borrowers, but that might compromise the bank’s access to the assets funded (including lack of collateral and untidy loan files and legal documenta-tion). This alternative approach should lead to the identification of substantially smaller losses.

Optimising the Burden Allocation Process

If the first priority is to minimise the overall cost of the crisis, the second is to implement appropriate burden sharing arrangements. In the case of Indonesia, the vast majority of the costs were allocated to the general public. This is both unfair and economically inefficient, resulting in a public debt so large as to com-promise the medium-term economic future of the country. Moreover, by provid-ing wrong incentives it has also contributed to increases in the costs. One egregious example is that of the state-owned banks. Up to the middle of 1998, when the bank resolution strategy was still under discussion, the state-owned commercial banks hid their losses, pretending to be healthy and posting financial statements grossly out of line with economic reality. As soon as it became clear


(3)

that these banks would all be recapitalised with public money they began to book losses with abandon, as can be seen in their financial statements.

The SNA has the potential to provide practical solutions as to how the burden sharing exercise can be improved. Since the crisis was, to a large extent, a wealth redistribution exercise, there have been both winners and losers. The winners have captured some economic value, to the detriment of the losers. There does not seem to be any reason why the value captured could not be recycled, how-ever. The proper way to do so seems to be through taxation. Just as there is now, in a large number of countries, a value-added tax (VAT), based on a concept rooted in national accounting, there could be, during or following systemic bank-ing crises, a temporary value-captured tax (VCT), also based on the SNA analy-tical framework. Conceptual and pracanaly-tical difficulties have been overcome in the case of the VAT, and could be solved in the case of the VCT too.

The VCT should be temporary, and applicable only to a limited number of tax-able events based on specific definitions for the targeted transfers of value—such as the writing down or writing off of bank loans, payment by banks of high inter-est on their deposits or discounts on interinter-est billed by banks on their loans. There should also be a size threshold large enough to keep small depositors and busi-nesses outside the system and the number of concerned transactions at a manage-able level, in line with local administrative implementation capacities.

Symmetry is essential to the working of the system. Originators of the value transferred—in most cases, the banks—would book a VCT credit (equivalent to a claim on government) instead of posting losses, thus preserving their solvency. A VCT debit certificate of the same amount would be simultaneously issued and notified to both the debtor and the tax administration.

If the VCT debtor—in most cases, a large corporation—does not own sufficient assets to meet its fiscal liabilities, the resources provided by the bank will have been lost. The corporation will be clearly identified as bankrupt, and treated accordingly, with the tax administration as a creditor. If, on the other hand, the VCT liability corresponds to assets of some economic value, the tax administra-tion will be in a stronger posiadministra-tion to exercise its legitimate claim on those assets, thus minimising the final cost for the general public.

Prevention

Finally, as regards prevention policies, it is important to focus on the possibility of an accumulation in the economy, over a number of years, of undetected losses on bank loans. The place where these problems develop is the corporate sector, not the banks. The problem to address is a hidden accumulation of losses, which might be slow and remain undetected for years, when corporations book as investments what is in fact (intermediate) consumption, funded through bank loans. Banks have an important role to play in preventing the silent accumulation of such losses. Prudential reviews of loan portfolios should generally be able to detect them, although this will not always be the case. Too much faith might be put in historical costs and/or in the significance of the borrower’s repayment his-tory. Banking supervisors need to pay special attention to this issue.

An essential aspect of prevention is preparedness. Experience has shown that severe or systemic banking distress may appear in all categories of countries, regardless of their economic structure or level of development, but that their


(4)

occurrence in a given country is infrequent, often only once in a generation or two. As a result, it makes sense to separate preparedness into two components: at the level of each country, basic, low cost preparatory measures of a general nature and, at a multilateral level, the building up and maintenance of a permanent cri-sis management capability on which individual countries might rely, if they wish, as the need arises.

At country level, preparation should consist mainly of the drafting of a contin-gency plan and a review and adaptation of the relevant existing laws to ensure that they would be conducive to the implementation of an appropriate bank res-olution strategy in case of a crisis. At the multilateral level, the international financial institutions have a natural responsibility to maintain a permanent capa-bility to assist in the management by national authorities of a systemic banking crisis that might erupt in a specific country. This capability already exists to some extent, but the considerable potential of the SNA framework in this regard remains largely unexploited. Substantial work would be needed to take advan-tage of the SNA’s potential, but the rewards would probably be commensurate.

CONCLUSIONS

The coherent analytical framework provided by the SNA can be of help in discov-ering what happened with the $50 billion ‘lost’ by the Indonesian banks. The business notion of ‘loss’ is vague, inconsistent with macroeconomic concepts and even misleading. By contrast, an SNA-based review shows that most of this amount has not disappeared from the economy, but has been simply transferred out of the banks, mainly to the corporate sector.

The SNA approach provides a new perspective on the crisis, its origins, its nature and its repercussions. Ultimately the crisis was mainly a large-scale wealth redistribution exercise, neutral for the banks, beneficial for the corporations and, to a lesser extent, for household borrowers and depositors, but disastrous for the general public. The SNA has the potential to provide an analytical framework to identify, characterise and quantify these flows.

The SNA perspective also points to possible improvements in bank resolution strategies. Since the problem is mainly that of an opaque and unintended large-scale redistribution of wealth, improved resolution approaches should include bringing full transparency to the process, quantifying the redistribution on the basis of SNA concepts, and then recycling the captured wealth through adequate taxation of the beneficiaries.

Such resolution strategies could not have been implemented in Indonesia at the time of the crisis, however, because of a lack of operational readiness that per-sists to this day. A number of critical assumptions still need to be vetted, and seri-ous practical obstacles overcome. The SNA framework, notwithstanding its huge promise, would still need to be reviewed, adjusted and developed before it could play any innovative role in managing banking crises.

Ultimately there is one general conclusion: the SNA has promising potential to assist in the analysis of a large-scale financial disaster of macroeconomic impor-tance, thanks to its conceptual integrity as a global and coherent framework. For the sake of the rapidly globalising economy, where systemic banking crises seem to multiply, this promising potential should not be left untapped any longer.


(5)

NOTES

* The views expressed in this article are those of the author and should not be inter-preted as reflecting the views of the International Monetary Fund.

1 The Indonesian banking crisis has been covered, for example, in Djiwandono (2004, in

this issue); Enoch et al. (2001); Fane and McLeod (2002); and Lindgren et al. (1999).

2 Data as of end June 1997.

3 Indonesia did not have a formal deposit insurance scheme when the crisis began; it was in the process of designing such a scheme at the time of writing.

4 The current approach is the result of a compromise that emerged progressively among the international community of national accountants, after decades of debates, tensions and reversals in practices. For an elucidation of the issues and the historical back-ground to the discussions, see Vanoli (2002: 199–203). For the details of the outcome, see SNA (1993), Production Account, The Output of Particular Industries, Financial Intermediaries, beginning at paragraph 6.120.

5 For this exercise, write-downs on loans (i.e. provisions for expected future losses) have been treated similarly to write-offs (i.e. final losses), as negative changes in the volume of assets. This is not in line with the current treatment of non-performing loans by the SNA (an issue currently being discussed in international expert circles), but reflects the economic reality as seen from the banking supervisor’s point of view.

6 Note that the data published by the Central Statistics Agency show a positive overall contribution of the banking sector to GDP throughout the crisis years: 4.0% in 1997 and 1998, and 2.2% in 1999. Elucidation of this discrepancy is beyond the scope of this paper (see appendix).

7 There are strong suspicions that this equity position, as declared by the banks, did not reflect their real financial position, however, because of under-provisioning for non-performing loans.

8 For a discussion of the treatment of non-performing loans in macroeconomic accounts,

see Bloem and Gorter (2001). REFERENCES

Bloem, Adriaan M., and Cornelis N. Gorter (2001), ‘The Treatment of Nonperforming Loans in Macroeconomic Statistics’, IMF Working Paper No. 01/209, December. Djiwandono, J. Soedradjad (2004), ‘Liquidity Support to Banks during Indonesia’s

Finan-cial Crisis’, Bulletin of Indonesian Economic Studies40 (1), in this issue.

Enoch, Charles, Barbara Baldwin, Olivier Frécaut and Arto Kovanen (2001), ‘Indonesia: Anatomy of a Banking Crisis—Two Years of Living Dangerously—1997–99,’ IMF Work-ing Paper No. 01/52, International Monetary Fund, WashWork-ington DC, May.

Fane, George, and Ross H. McLeod (2002) ‘Banking Collapse and Restructuring in

Indo-nesia, 1997–2002’, Cato Journal22 (2): 277–95.

Frécaut, Olivier (2002), ‘Banking System Losses in Indonesia—Looking Out for Fifty Bil-lion U.S. Dollars—Can the SNA Help?’, Paper prepared in June for the 27th General Conference of the International Association for Research in Income and Wealth (Stock-holm, Sweden, August 2002), parallel session 4B (www.eco.nyu.edu/iariw).

Lindgren, Carl-Johan, Tomas J.T. Balino, Charles Enoch, Anne-Marie Gulde, Marc Quintyn and Leslie Teo (1999), ‘Financial Sector Crisis and Restructuring—Lessons from Asia,’ IMF Occasional Paper No. 188, International Monetary Fund, Washington DC. SNA (1993), System of National Accounts 1993, Prepared under the auspices of the

Inter-Secretariat Working Group on National Accounts; Commission of the European Com-munities; International Monetary Fund; Organization for Economic Cooperation and Development; United Nations; World Bank; Brussels/Luxembourg, New York, Paris, Washington DC.

Vanoli, André (2002), Une Histoire de la Comptabilité Nationale [A History of National

Accounting], Editions La Découverte, Collection Manuels Repères, Paris.


(6)

APPENDIX: SOURCES AND PROCESSING OF DATA

The core data relied upon in this paper are the losses of the Indonesian banks dur-ing the 15-month period from July 1998 to September 1999. These loss estimates, totalling Rp 434.9 trillion (table 1), were prepared by the author as part of his assignment as a banking sector expert to the IMF team in charge of monitoring the banking crisis. The overall accuracy of the aggregate loss estimate was con-firmed subsequently by comparing it with publicly available data—namely, the variation in the net equity position of the banking system (a decrease of Rp 457.6 trillion, taking into account equity injections in the form of government bonds), as it appears in Bank Indonesia’s monthly publication, Indonesian Financial Statis-tics.

Details about these data and their processing are provided in three technical appendices to the author’s original June 2002 paper (Frécaut 2002) introducing the thesis summarised in the present paper.

• Appendix II of the 2002 paper provides details of the loss estimates and of the accuracy assessment.

• Appendix III elucidates, through two detailed tables, the transposition of the components of the banks’ income statements into the SNA’s accounts and headings. For instance, the different categories of interest received and paid (in the income statement) have been netted and each result assigned as FISIM out-put to the SNA’s production account.

• Appendix IV describes the steps followed (including the actual numbers) to extract the subsidy element in the negative interest margins as part of the pro-posed alternative definition of FISIM. Based on historical GDP data, the nor-mal level of FISIM has been estimated, and then extrapolated and adjusted for the crisis period. The subsidy element embedded in interest rates during the crisis has then been obtained as the difference between ‘normal’ and measured FISIM.

The whole exercise is a preliminary attempt at a new approach to the elucida-tion of a particularly severe systemic banking crisis. The focus has been limited to the identification of plausible orders of magnitude, for bank losses only. At this early stage, no detailed and systematic checks have been conducted on the valid-ity of the other data and, in particular, no reconciliation has been attempted between the data obtained as described above and actual national accounts data for the crisis period.