Estimating quasi-fiscal deficits in a consistency framework : the case of Madagascar

Policy Resoarch WORKING PAPERS Private Soctor Development Economic. Africa Technical Department The World Bank rebruary 1993 WPS 1 105 Estimating Quasi-Fiscal Deficits in a Consistency Framework The Case of Madagascar Philippe Le Houerou and Hector Sierra Toassessfully theeffectofadjustmentprogramsanddevelopment strategies, it is essential that the fiscal deficit include quasi-fiscal deficits - the losses of public financial institutions such as the central bank. A flow-of-funds format may be the best approach for doing so. as this case shows. PocyReuchWomingPapmd;dztflhnsdrwcof wc k iDpmnda Weugetthcuhangcofzide mangnB stffand allehm3adin deWicimcThm.eppapcdisW bedbydc Rccarb Advis;yStaff.can theamesfft5 auro,fflcct 1ydrirvsndsbcdbwcdbeuadckaardiy.1coyl.n edinguim apdoum.adcdnchmaartheauthoaz own.leysh*Od not be atutied to the Wedd Bnk. its Board of Dihto, its mnagamnt. armny of its maeb countdra  Policy 8 aeuch Private Sector Development Economic WPS 1105 This paper - a product of the Private Sector Development Economics Division, Africa Technical Department-ispartofa largereffortinthe departmentto assess the macroeconomic impactofquasi-fiscal deficits. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Nydia Velasco, room J3-283, extension 34346 (Febriary 1993, 26 pages). In practice, conventional measures of the dscal methodological and practical problems of deficit exclude the activities of public financial treating quasi-fiscal deficits, Le Houerou and institutions. As a result, fiscal policies may be Sierra propose using a flow-of-funds format, applied inappropriately when these institutions which in principle could be standardized across - especially the central bank - run large losses countries. (the quasi-fiscal deficit). Using Madagascar as an example, they show The macroeconomic effects of the quasi- that the public sector deficit is significantly fiscal deficit are similar to the effects of the undervalued if quasi-fiscal deficits are not deficit from other public entities - and should considered. therefore be included in the public deficit. They contend that such deficits must be Conceptual and practical difficulties have so taken into account in assessing the success of far precluded a definition of quasi-fiscal deficits adjustment problems and development strategies that is operationally use:Dd and comparable supported by the IMF and the World Bank. across countries. After studying the hrPolicy Research WorkingDPiaessmesdisseiinates thermdingsofwork uderway in theBankr Anobjectiveoftheseries is to get these findings out q!iiclcly. even if presentations are less than fully polishedL The findings, interpretations, and conclusions in these papers do not necessarily represent official Bank policy. Produced by the Policy Research Disseniination Center  ESTIMATING QUASI-FISCAL DEFICITS IN A CONSISTENCY FRAMEWORK: THE CASE OF MADAGASCAR Philippe Le Houerou Hector Sierra the World Bank Washington DC * The Yiews expressed here are those of the authors and should not be attributed to the WoriG Bank or any of its affiliated institutions. The authors wish to express appreciation to Mario Blejer and Paul Popiel for their insightful comments.  I. Introduction. In many developing countries the Central Bank performs "quasi-fiscal" activities not specifically conmected with monetary policy. The losses incurred by these activities--known as the "quasi-fiscal" deficit--may be huge, as shown by the experience of many Latin American countries during the debt crisis of the i980s. In the last five yeaxs or so. a considerable amount of effort has been devoted to define adequate ways to measure the quasi-fiscal deficit. It is now widely agreed that it has macroeconomic effects similar to the deficit of other public ent! `s. and therefore should be included in the overall fiscal deficit (see Anand and Van Wijnbergen (19881, Robinson and SteUla [19881, Blejer and Cheasty [19911, Rezende-Rocha and Saldanha [ 19921.) The correct measure of the quasi-fiscal deficit is then a prerequisite for diagnosing economic problems and finding appropriate fiscal policies to address them. In most practical applications, however, conventional deficit measures include only the central government. Even in comprehensive measures the public financial institutions are often excluded because of their special role as financing agents. One problem is the practical difficulty of measuring the amount of the losses incurred by these institutions. Quasi-fiscal deficits are often omitted or blend with other operating losses, unless they are too large to ignore (Teijeiro (19891.) Another problem is the country- specific nature of the quasi-fiscal deficits, which so far has precluded a general definition applicable across countxies and institutions. Our main objective is to present a stylized accounting methodology to help examine the underlying principles behind quasi-fisca' deficits, and to compute them in an operationaUy meaningful way. Our approach is based on the flow-of-funds format, which essentially consists of a series of identities that reflect the fact that what is a use for one sector, is also a source for another sector. This consistency framework reconciles historical data on income. expenditue. saving, investment and financiai flows of different sectors in the economy (Holsen (19891.) In this paper we will concentrate on the quasi-fiscal deficit of the Central Bank. The methodology, however, may be applied to public financial institutions in general. The flow-of-funds approach has two main advantages. First, it emphasizes the macroeconomic nature of quasi-fiscal deficits. To be able to assess correctly the impact of quasi-fiscal deficits, the analyst must specify precisely the linkages between the main economic agents. and treat the deficits as an integrated whole. The framework will make more transparent the link between the quasi-fiscal deficit and inflation and the crowding-out of the private sector. Second, the methodology implies full consistency of treatment of the deficits of the non-financial public sector and the financial institutions being examined. This is an important requirement for the proper consolidation. and symmetry of treatment, of transactions between the government and public financial institutions. AdditionaUy, the flow-of-funds methodology allows the possibility of obtaining estimates of some flows residually, such as Central Bank losses, adjusted such that the other sectors are consistent within the macroeconomic framework. The latter is specially important for countries, like most in Africa. for which data is scarce. Throughout the paper, we test our methodology with data collected for Madagascar. We examine the historical period from 1986 to 1991. During this period. the Central Bank incunred considerable losses. We show that the public sector deficit is significantly undervalued if quasi-fiscal deficits are not considered. For the Malagasy economy, any reaionship between fiscal deficit. credit to the private sector and inflation must take into account the results of the Centrl Bank. The case of Madagascar. however, is not different from most African countries, where the tremendous pressure on the govemments to cut 2  fiscal deficits while maintaining non-viable public enterprises. for political and social reasons, often results in the accunulation of the less-visible quasi-fiscal deficits. Conventional deficit measures most likely will underestimate the fiscal deficit. and therefore result in an erroneous assessment of its sustainability. The paper is organized into six sections, including the present introduction. In the next section we revise conceptual aspects of QFDs. Here. we examine in some detail the main theoretical and practical issues that make the computation of quasi-fiscal deficits difficult. In section Three we examine briefly the economic environment of Madagascar. In section Four. we develop an analytical framework, based on the flow-of-funds approach. within which we may examine the impact of quasi-fiscal activities. In section Five, we use the framework to estimate the size of the quasi-fiscal deficits for the period 1986 to 1991. We then "correct" the fiscal deficit by taking into account the quasi-fiscal deficit. In section Six, we establish our conclusions. II. What are Quasi-Fiscal Deficits?: Main Conceptual Issues In a broad sense. quasi-fiscal deficits (QFDs) are expenses or losses incurred by pdblic financial institutions. They reflect a subordination of these institutions--in particular of the Central Bank--to the ministry of finance. Quasi-fiscal activities are diverse: they include the management of explicit or implicit subsidies, transfers to public non-financial or financial institutions outside the central government, the provision of preferential credit. the bailout of ailing industries. etc. QFDs gmw at times of domestic financial and extemal payment crises due to support to distressed financial institutions and domestic debtors burdened by extemal liabilities denominated in extemal currencies. Indeed, they can reach massive prportions during years of fiancial and extemal crises in countries with weak financial institutions. QFDs of the Central Bank amounted to a cumulative 55 % of GDP during 1982-85 in Argentina, and to 41 % of GDP during the same period in Chile (Easterly and Schmidt-Hebbel (19911.) It has frequently been argued that quasi-fiscal operations are similar to other budgetary activities and should be included in a comprehensive measure of the public sector balance (Blejer and Cheasty [19911.) Regularly, institutions such as the Central Bank undertake activities such as the provision of subsidized credit to particular sectors. In many countries these activities may be undertaken directly by the agencies in the central govenmment. We agree with Robinson and Stella (19881 that "it may be difficult to see why an activity administered by the Central Bank may be different in an economic sense from one administered by other government agencies. and therefore. why, if the latter are in a measure of the central govemment deficit, the latter should be excluded." Certainly, the experience of the Latin American countries illustrate how misleading public sector deficits can be if not corrected for QFDs. A key "working assumption" in the computation of QFDs is that the Central Bank as well as other public financial institutions should be financially sound. and that--in the absence of quasi-fiscal activities- should not incur losses on their core operations. For example, the Central Bank will have access to zero cost fnancing (monetary base), and financing below market rates (unremunerated deposits of commercial banks), while being able to invest these funds at market rates. Moreover, profits may originate from financial investments directly financed bv the liquid net worth of the Central Bank. Thus, under "normal" conditions, the Central Bank should report profits or at least zero losses. This assumption has motivated researchers to "amalgamate" Cental Bank losses with the fiscal deficit. as the more practical way to capture the QFD. For example, Robinson and Steila [19881 propose 3  that (i) Central Bank operational losses be incorporated into the public sector deficit by the addition of a transfer from govenmment to the Central Bank. financed by credit from the Central Bank. and (ii) an estimate of the size of Central Bank quasi-fiscal activities falling outside the profit-and-loss account should be made, and then amalgamated into the adjusted fiscal deficit. As it is stated in Blejer and Cheasty [1991.p.16631, this definition "would mix net worth with cash concepts, but would have value as a supplementary indicator showing the approximate impact of Central Bank quasi-fiscal activities on the overaU public sector balance.' According to Robinson and Stella. Central Bank losses should be included in the fiscal deficit, whether or not they arise from QF activities. Their argument is based on an existing asymmetry. Frequently, the Central Bank transfers all of its profits to the government as excess fiscal earnings. In this case. the fiscai deficit will already reflect the impact of QF activities. However, current losses do not elicit a transfer from tie government to cover the losses, so the measured deficit does not rise. While the consolidation of the Central Bank and fiscal deficits may be the more practical approach. it is essential to identify all the revenues and expenditures arising from QF activities. Isolating these activities in the accounts of public financial institutions would make their costs more transparent. thus aiding scrutiny by the authorities. This transparency is essential for the design of sound policy prescriptions Even if the public financial institution is making a profit, leaving such activities in the accounts will understate the gross level or revenues and expenditures. frequently taken as a proxy for govenmuent intermediation in the economy. One practical problem, however, is the precise identification of "normal". as opposed to "quasi- fiscal" activities. For example, the "normal" monetary activities of the Central Bank include currency issue, banking regulation and supervision, the aggregate conaol of credit. the clearance of balances between banks. and custody of the govenmment reserves. If the Central Bank is used to bailout commercial banks and/or enterprises (public and/or private) experiencing fmiancial difficulties, this is clearly a quasi- fiscal activity. Ideally, then, govemment accounts should incorporate all quasi-fiscal revenues and expenditures, leaving Centrl Bank accounts covering only "normal" monetary activities. Nevertheless, clear distinctions may be difficult. For example, direct government credit and bond rediscounting, geierally considered monetary activities, will take on a quasi-fiscal dimension if performed at subsidized rates. Central Banks usually provide exchange rate guarantees as a way to facilitate fortign borrowing by domestic residents. If a premium is charged. the Central Bank may make a profit in return for reducing the insured's risk. If, however, the Centra Bank is forced to assume the extemal transfer portion of private sector debt, even when it was not guaranteed by the govemment, the resulting losses are of a QF naure. Different accounting conventions between the govenmment and the Central Bank n'ay also complicate the identification of the "real" QFD. Usually, the conventional govemment deficit is based on a cash-accounting systemq. Central Bank accounting systems. on the other hand, typicaUy follow the normal business practice UL being on an accrual basis. Inflation may be another major source of distortion in the computation of the QFD (see Teijeiro (19891, and Rocha and Saldanha (1992].) So far. most country studies of QFDs have relied on the "orthodox" criteria to compute the QFD. For example. in Onandi and Viana (19871, and Velazquez [1991], an operation of the Central Bank is of a QF nature if it is not related with the classical or "orthodox" functions of the Central Bank. Onandi and Viana (19881 examine the impact of quasi-fiscal activities in Uruguay. According to these authors, "Traditionally, a series of Central Bank operations are performed outside of the govenmment budget, but their fiscal impact can not be ignored. Thus, we understand an cperation of the Central Bank to be of a quasi-fiscal nature if it is not related to the classical (orthodox) functions of the Central Bank" (authors' 4  translation.) According to Velazquez ( 19911. ".. the eamings or losses incunred as a consequence of non- monetary activities of the Central Bank, should be coffectly incorporated in the measure of the fiscal deficit" (authors' translation.) Clearly, this approach discriminates against "purely monetary" activities performed at subsidized rates. Furthermore. precisely what Central Bank activities may be considered as "orthodox" is still a controversial issue (nevertheless, see de Kock [19741 for a list of activities generaly considered to be properly within jurisdiction of the Central Bank.) Another approach is followed in Piekarz (19871, wihere a more general framework is provided to examine the issue of QFDs from the Argentine perspective. He argues that QFDs should be defined Is the difference between the earnings and expenditures not considered in the fiscal defic:t. This is basically the difference between income from net foreign and domestic assets. and net expenditures on interest payments, and/or other adjustments resulting from the instruments used in the control of monetary expansion. Piekarz excludes instruments like unmemunerated reserve requirements and open market operations with govemment securides. In the case of the latter, he argues that they properly belong to the Treasury, not the Central Bank, computing their interest payments as public expenditure in the conventional fiscal deficit definition of the non-financial public sector. In the case of the fonmer, they are rigorously part of the inflationary tax and/or seigniorage. More recently, Giorgio [19911 applied the Piekarz approach to examine QFDs in Argentina for the period 1977-89. It is our view that any effort to provide a generic definition of the QFD will inevitably be partial and incomplete. Quasi-fiscal activities vary from country to country, and have an effect in both the profit- and-loss and balance sheet accounts, making almost impossible to assess their impact through one sinble deficit measure. The situation is further complicated in practice by the differing accounting systems used in the government and other public financial institutions. Still, in most cases the QFD can be traced to "clear-cut" QF activities, such as subsidized lending and bailout of commercial enterprises. In a few instances, the distinction between "normal" and QF activities may be subjective A useful criteria is to include all activities that may be undertaken direcdy by central government agencies. In this way, the cost of quasi-fiscal activities may be taken as a proxy for the level of govemment internediation in the economy. It is also important to provide reference levels. above or below which a "normal" activity will become a quasi-fiscal activity. NormaUy, these will arise from targeted private sector or government lending operations from the Central Bank. A convenient benchmark is to use the cost of net lending as the sum that would have to be paid to a private commercial bank to undertake the lending itselfWj. Govemment accounts should then incorporate revenues and expenditures of all quasi-fiscal activities, leaving the accounts of public financial institutions covering only their "normal" activities. This should make more transparent the accountability of the govemment. Here, we follow Robinson and Stella [19881 by adjusting ex-post the fiscal deficit by the addition of a transfer from the govemment to the Central Bank--to cover the cost of QF activities-financed by credit from the Central Bank. As we shall see, this has the advantage of estimating explicitly the cost of QF activities in terms of "crowding-out" of the private sector. To summarize, there is currently a lack of genera agreement as to what is the best methodology to identify and compute the QFD. at least one that may be applied across countries and institutions. A 1 So far. however. there is not a genemlly accepted objective metbod for eshitg the subsidy value in official direct lending prhgns. nainly because the difficulty of prcisely establishing the privae utes tast would have ben paid by borowen in private market without govemment intervention (see eg. Wattlewonh (19881.) 5  single definition, no matter how comprehensive, wil not capture the complex and country specific nature of the relationships between the centra goverment and other public institutions that give rise to the QFD. Still. given the potential prctical relevance of the QFD for the whole economy, it is necessary to develop a general framework within which one may examine its macroeconomic implications. In this section we provided genera guidelines that wiU be usefu when we try to estimate the QFD for Madagascar. IIL The Case of Madagascar General Background The recent economic history of Madagascar is, to a large extent, representative of a number of African countries. In the 1970s. Madagascar moved from a relatively open economy to a tightly regulated, command economy. The role of the public sector greatly increased. following the nationalization of foreign-owned assets and the introduction of pervasive economic controls. Import substitution and industrialization became centra objectives, with agriculture relegated to a supporting role. In the late 1979s this strategy culminated in an ambidous public investment program cominosed of large, capital- intensive and economically non-viable projects. As a result, many new public enterprises generatng negative value-added were created. To fmance this strategy, the extenal deut burden increased significantly, constraining economic gmwth of the country to this day. Mhis policy- together with declining terms of trade and stagpant revenues. led to a widening of extemal and intemal disequilibruia, high inflation, and a contraction un real GDP of 11 percent between 1980 and 1982. In 1982-85. the Govenmment carried out a genemally successful stabilization prgram with EMF support. The program succeeded in halting the economic decline, and GDP grew by 1.4 percent on average between 1983 and 1987. On the extemal side, while exports declined in nominal dollar terms and interest payments soared, there was a major comraction in imports. which dropped by more than 40 percent in real terms during the firut half of the l980s. These developments, together with increasing extemal aid flows. helped improve significantly Madagascar's balance of payments. In paallel, the Government implemented a program of fiscal austerity. Fiscal deficit was reduced from its peak 14 percent of GDP in 1980 to 3.5 percent in 1985, through a drastic reduction of public expenditures. During the second part of the decade. the authorities consolidated this substntial fiscal adjustment effort and fiscal deficit was kept within a 3-4 percent range of GDP. Economic recovery, however, was hampered by structur constrains. Among the most import were an overvalued currency, widespread quantitative import restrictions and price controls. a bloated and inefficient parastatal sector, a badly deteriorated and obsolete infrastucture, poor public services, and pervasive govenmment interference in viually all aspects of economic life. To address these structua constraints, the Govemment launched a structural adjustment program in the second half of the 1980s. Most price controls were removed in 1985-89. and the public investment program was progressively rationalized. With regads to foreign exchange and tade libeialization, the Govenmment devalued the Malagasy Franc (FMG) by a cumulative 53 percent in trade-weighted foreign currency terms in 1987, and in 1988 the FMG was futher depreciated by a total 14 percent. As a result, by 1989 the real effective exchange rate had fallen to about 40 percent of its 1982 level. This major 6  Table I Madagascar: Selected Indicators (Annual Percentage Change, 1986-1991) 1986 1987 1988 1989 1990 1991 National Accounts GDP Market Prices 2.0 1.2 3.4 4.1 3.1 -6.9 GDP Deflator 14.2 23.0 21.2 12.0 11.5 14.1 Private Consumption -2.6 -3.3 -1.8 1.2 4.1 -2.1 Govenmment Constunption -7.1 5.6 -9.9 8.1 -7.0 -6.7 Total Investment 7.8 16.6 33.9 4.8 28.0 -53.0 Exports of G&NFS 0.3 18.2 -0.6 24.0 11.1 -4.2 Imports of G&NFS -6.0 -1.4 -10.3 6.2 26.3 -22.1 Foreign Sector Terms of Trade (deterioration -) 15.7 -22.1 -9.1 -12.0 -24.0 -5.8 Devaluation Effective Exchange 10.9 34.3 23.5 -1.5 -13.5 10.0 Rate (FMG/US$) Total Foreign Debt (% of GDP) 82.2 121.2 118.8 116.1 99.6 117.2 Debt Service Ratiol]/ 31.6 37.9 49.0 53.1 42.1 68.0 of which Govenmment 28.0 27.4 29.6 16.1 17.7 18.1 Foreign Reserves 2.8 3.8 4.9 5.2 1.4 1.6 (months of imports) i/ After debt reshebduling and write-offs. as a proportion of exports of goods and services. exchange rate adjustment enabled the Govenmment to eliminate quantitative restrictions on imports in 1988, to launch a four-year program of tariff reform, and to introduce an automatic allocation system of foreign exchange for merchandise imports thrugh an Open General Licensing system. In addition, export restrictions were substantially reduced, all state agricultural monopolies were eliminated (except for vanilla), and export taxes were abolished on all but the three traditional products (i.e.. coffee, cloves. vanilla). In the financial sector, where all banks were state-owned and were experiencing severe financial problems as well as weak and wowening portfolios, the Government also undertook a major reform. Specifically, by end-1989 the portfolios of the three state-owned banks had been cleaned-up of most of non-performing and of most doubtful assets. In 1980, a fully private new bank with majority foreign capital started operations and, in eady 1991. one of the three state-owned bank was privatized, while a second obtained minority private participation. and two private foreign banks opened representations. In 1990. to encourage fuither competition in the financial sector, the Central Bank began replacing direct quantitative control mechanisms with a more flexible system based on the use of indirect market instnments. Overall. despite slower than expected progress in "process-oriented" policy improvements (i.e., budgetary, civil service and public enterprise reforms), the cumulative implementation of sectoral and macroeconomic reforms implemented since 1985 was far reaching, and the difference between the policy 7  landscape in 1985 and that prevailing in 1991 was striling. Beginning in 1988, economic results were encouraging (see Table 1.) The country experienced posidve per capita-albeit modest--GDP growth in three consecutive years (1988-90) for the first time in a generation. with tangible evidence of increased private investment. especially in labor-intensive. export-oriented ventures. Financial Sustainability of the Economic Recovery Notwithstanding the encouraging results of 1988-90, the supply response to the structural reform program implemented since 1985 has not reached its potential and may not be sustainable. In addition to major physical (infrastucture) bottlenecks. the inadequacy of the regulatory and legal framework to a well- functioning maricet economy, and declining terms of trade. the private sector-led economic recovery in Madagascar is severely constrained by lack of credit and the heavy debt burden of the country. The latter two constraints are two sides of the same coin and are both reflected in the country's precarious public finance. Despite the above-mentioned success in reducing the central government's fiscal deficit, the latter is increasingly misleading to assess the public sector need of financial resources. and therefore. the financial resources available for the private sector. In paraUel with the rcduction of the central government deficit, the Centrl Bank (CB) has generated increasing "net other items" reflecting increasing losses in the 1980s. These losses can be principally traced as a result of the transfer to the CB of the responsibility for servicing the country's rescheduled extemal debL Furthermore, these losses were not compensated by transfers nor interest payments from the govemment to the CB. The govenmment has been using subsidized credit from the CB to finance its deficit. Untir 1991, advances2 from the CB to the treasury were virtually interest free (0.5 percent) until 1991. On the other hand, treasury's deposits at the CB were not remunerated. lhe evolution of the size of the net subsidy from the CB to the treasury was therefore determined by the evolution of CB net credit to govemmenL As it is shown in Table 2, the implicit interest rates paid by the govemment on its net domestic debt are well below commercial interest rates, which have fluctuated between 15 and 25 percent per year. The increase in the implicit interest rate paid by the govemment to the CB in 1989 and 1990 reflected simply the substantial reduction in the CB's net credit to government (from 17 percent of GDP in 1986 to 5 percent in 1990). In tum, this reduction reflected essentially (i) the low govenmment payments of extemal debt matLuities (see below) together with (ii) the accumulation of counterpart funds of extemal aid flows. Conversely, in 1991, net credit to govenmment increased. The increase of the implicit interest rate paid by the govemment to the CB reflected the decision of the Govenmment in mid- 1991 to remunerate treasury deposits (to) and bormwing (from) the Central Bank at money matiet rate. 2 Advances from the CB to the treasury are statutorily limited to a ceiling of 15 percent of ordinary budget revenues of the previous fiscal year and should be statutorily reimbursed widhin six months of the closing of fiscal year in which these advances are made. In practice however. these statutory provisions have been consistently violated. The actual ratio at the end of 1990, for example was about 290 permcnt of ordinary budget revenues of the previous year: even net of government deposits, the ratio of advances was 78 percent of the previous year's ordinary budgetary receipts. 8  Table 2 Average Interest Rates Paid on Domestic and Foreign Debt 1986 1987 1988 1989 1990 1991 Implicit interest re on net 2.0% 2.1% 2.6% 6.0% 7.0% 8.8% CB credit to govenmnent Average Interest Rates on Foreign Debt Central Bank 6.8% 7.0% 7.1% 7.5% 6.3% 9.5% Govenmment 5.0% 5.5% 3.9% 2.3% 2.2% 2.6% In the mid-1980s, the CB losses baMlooned as a result of previous arrangements whereby the responsibility for servicing the rescheduled external debt was transferred to the CB. After seven Paris Clubs and four London Clubs, in 1991 CB debt outstanding represented 46 percent of total debt and 67 percent of the country's debt service. The interposition of the CB between domestic debtors. namely the govemment and public enterprises, and foreign creditors has resulted in three types of losses for the CB: (i) As some intial debtors (public enterprises) defaulted, the CB is stuck with a portfolio of irrecoverable claims: (ii) While the original debtors reimbursed the Centrl Bank in FMG according to the initial contract, CB was committed to service the debt on the basis of successive rescheduling agreements which, in lengthening maturities, were also increasing the total burden of the interests; (iii) Subsequent exchange rate devaluations gave rise to valuation losses in the books of the CB. The CB statutes were amended at the time of the arrangements to state explicitly that the government would guarantee any resulting valuation losses. When extemal liabilities are amortized, CB realizes its accrued valuation losses. However, this guarantee has so far been ignored and the losses have accrued in a valuation account which the CB nLgards as a claim on government. As a "'rnsequence. the "revaluation account" records a mnix of accrued and cash results fnxm servicing of extemal debt (as well as a number of sundry operations and transactions). As a rra'ilt of this extemal debt arrangement, the govemrnment has significantly reduced its extenal debt service (and therefore its need of domestic financing) at the expense of the CB. By 1991, the long- tenn foreign liabilities of the CB (i.e.. the country's rescheduled external debt) represented 180 % of its total assets (see Table 5). Clearly, the past and future losses generated by such an agreement should be viewed as part of the cost of a debt rescheduling and, therefore, should be part of a deficit measure, especially if the direct impact on the government's deficit was to reduce debt service payments. Furthennore. as sOjwn in Table 2, while the government paid an average 2.9 percent interest in 1991 on its foreign debt. the CB paid 9.5 percent This reflects the fact that after shifting the rescheduled part of the extemal debt to the CB, the govenmment was left with essentiaUy the more concessional (non- reschedulable) part of the debt 9  Although the source of the CB losses have been identifed. the full extent of the QFD cannot be readily assessed from the financial data available (i.e.. the CB balance sheet and income stazement). Since corrections for QF activities are done retroactively, we must take into account accumulated debt service payments from debt rescheduling agreernents and subsidized credit dating back to 1986. Also, when computing the CB losses. it is imporant to clarify the accounting practice regardng the valuatik.n account (see above). IV. Analytical Framework In order to examine the issues presented In this paper, wG require the aid of schematic representations of the budget and balance sheet of the CB. and the govenment accounts. The need, however, of an economy-wide analytical framework is readiy apparent. To a great extent. the QFD of the Central Bank reflects the level of "financial repression." and the distortions generated in the Malagasy economy due to heavy govemment intervention in the investment and saving process. The main sectors examined are then the central government. Private Sector, Central Bank. commercial banks, and the balance of payments. The budget constraints for all sectors may be represented in a matrix foiaZ. commonly referred to as the flow-of-funds matrix (See eg. Holsen (19891.) Table 3 shows flow-of-funds matrices for Madagascar for the year 1991, both for the current and capital accounts. This table will be the basis of our discussion in the next pages. The overarching concept is that each sector in the framework must satisfy the following budget constraint: Sources of Funds = Uses of Funds Sources = Current Income + Accumulation of Liabilities Uses = Current Expenditures + Aceumulation of Assets An important distinction is made between the current account and the capital account. The link between the current and capital account is each sector's saving. Generly, saving will be defined as the cunrent account excess of sources over uses of funds above the line in each sector. In the case of the Central Bank and Private Banks, the current accounts would normally be identfied with the profit-and-loss statements, and the capital accounts with the balance sheet. Cunenu savings is then identified with the changes in net worth. For each of the sectors, the row and column represents the expendire and income breakdowns respectively. The balancing item for each sector's curent account is saving, shown in he lower left-hand comer of the matrix. The equality of the capital account row and column says that the sum of their total uses of financing is equivalent to the total of their sources of finmcing, with the latter including own saving. To put it another way, the excess of investment over saving is equal to the net bonowing from domestic and extemal sources for each sector. 10  Table 3. Madagascar: 1991 Matrix of Sources and Uses of Funds (Billion FMG) Current CreDal Private CenI ComtewiaI Balabn of Productwn Account Govrngint sfec,tor Bak Bank Payments Account TotAL Central TD 64.51 Ecb 0.01 7;f 0.01 77 29641 455.51 Government .,7R 56.51 COo 38.21 -Sub 0.0 ,. ~ ~I i ' Private TJp 64.91 1 OPCeb 16.11 OPCb 0.0 jTp 116.0j GMf 4561.6 488L.91 Sector |DDO 0.01 WR 0.0j iT'TDp 0.0 i PRfp 123J I Central ;iCCRcb ;4.81 iL'L.Vcb 4.9 iL'LNb 0.0i iRESm 7.5 47.2i Bank Commercial iCCRb 8.91 iC'LNb 1031 i Balance iFFg 55.71 iFFp 7.1 iiFFeb '020|1 IhMt 116R3 606 of Tgf 0.01 PRpf 3.51 -Xt 8492 payments Tpf 0.0O I C h S 51901cp 4t327 ; 'Account Sg l. I Sp 188.41 Seb -189I Sb 1132 Sf 321.6 Total 455.8 488L. 47.21 11.32 6.6j 57I3 Capitai Central Piivate CeotraI Cotmenal BHahne of Production Account Government sector Bant Banb Pawmenta Aecount Jotal Ceontral dHgp 9.3 dCReb -5.3j dCRb -35.2 dfg 190.7 Sg 1.1 213J . Govetnment KOG 525 Private iKT;p 28.01 dLNcb 1.0 dLNb 1002 1dFp 34.6 Sp 188. 377SB Sector ' DMI 2.' 077-F 0.0 ICntral dDEPeb -105.81 dCU 7Z.5 dRR 93.1 dlm 83.21 Scb -18Qt -46.81 Commecial dDEPb 14.71 dDD 1065 dLNb -433 dFb -12.1 Sb 1132 20001 . BFob | * ~~~ ~~~drD 31.0 . Balbnce dRESm 0.8 dDEPf 41.6 Sf 321. 364.1 ;Payments:| l chto !gt J32 Acceunat ~ I762 ~pl * f _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . Total 731 ! _-4.8| 00.0j 3$10 4352 11  Table 4 Government Accounts (as a Percentage of GDP) 1986 1987 1988 1989 1990 1991 Total Revenue 12.7 15.4 13.8 15.4 16.2 10.5 Tax Reveiue 11.8 14.1 12.5 10.2 10.3 7.4 Non-Tax Revenue 0.2 0.6 0.6 1.7 1.5 1.2 Foreign Grants 0.7 0.7 0.7 3.5 4.4 1.9 Total Expenditure 16.0 18.9 17.3 19.5 17.1 15.8 Consumption 7.7 7.5 6.8 6.9 6.2 6.0 Investment 5.4 7.0 6.9 9.7 7.9 5.7 Interest Payments 1.6 2.1 2.1 1.6 1.5 2.0 Domestic Debt 0.4 0.4 0.3 0.5 0.5 0.9 Foreign Debt 1.2 1.7 1.8 1.1 1.0 1.1 Other Expenditure 1.3 2.3 1.5 1.3 1.5 2.1 Govenmuent Deficit -3.3 -3.5 -3.5 -4.1 -0.9 -5.3 Govenmuent Financing 3.3 3.5 3.5 4.1 0.9 5.3 Net Extemal Borwwing 2.3 3.7 3.9 4.6 2.1 3.9 Drawings 4.6 6.6 7.2 6.6 4.1 6.0 Amortization -2.3 -2.9 -3.2 -1.9 -1.9 -2.0 Net Domestic Credit 0.8 -0.4 -0.6 -0.7 -1.2 1.2 Non-Bank Financing 0.2 0.3 0.2 0.2 0.0 0.2 We will only describe in detail the budget con.traints for the central govemment and the Central Bank. The algebraic representation of the budget constraints for aU sectors is presented in the Appenchx, together with a description of the variables used. 1. The Central Government Under this heading, we wiU include only the fisa authorities. Ideally, we would include the complete non-fuiancial public sector. That is. aU public sector except for the Central Bank and publicly- owned financial institutions. However, the lack of information in the case of Madagascar wil not allow us to use this concept. The rest of the non-financial public sector wiU automaticaUy be included in the private sector, which is calculated as the residual. 12  The cunrent account of the central govenument can be described as follows: DT + T/ + NR + Ed, + ECOG - SUB ( Tl+ icCRl+ i.E-F.-F + P C + Sg Equation (1) defines the savings of the central govemnment (S.) as the sum of direct (Dl) and indirect (TI) taxes, non-tax revenue (NTR), current official grants (E-COG). minus subsidi. s (SUB). pivate transfers (Tp), interest payments on Central Bank credit (ic-CR,b.1), interest on commercial bank credit (ic'CR ,q), interest payments on foreign debt (E.F.,), and govemment consumption (pc Cg). If the Central Bank reports profits. part of this may be transferred to the govemment as excess fiscal earnings. If on the other hand. the Central Bank incurs operating losses, the govemment may cover part of this through a transfer. Net Central Bank current transfers are represented by E