A RMSM-X model for Turkey

Policy, Research, and External Affairs WORKINQ PAPERS P Macroeoonomic Adjustment and Growth Country Economics Department The World Bank August 1990 WPS 486 A RMSM-X Model for Turkey Luc Everaert Fernando Garcia-Pinto and Jaume Ventura The theoretical design of a RMSM-X model, its interaction with a debt module, and the construction of a consistent historical data set is applied to Turkey. ,k, The Policy. Research, and Extenal Affairs Complex distributes PRE Working Papers todisserninaLethefindings of work in progress and to enoourage the exchange of ideas among Bank staff and all others interested in development issues These papers carry the names ol the authors, reflect only their views, and should be used and cited accordingly. The findings, interprtations, and conclusions are the authors' own. T'hey should not be attributed to the World Bank, its Board of Directors, its management, or any of its member countries.  Pollcy, Research, and Erlernal Affairs MUacroeconomic Adjustment and Growth WPS 486 Thispaper- ajointproduct of the Macroeconomic Adjustmcnt and Growth Division, Country Economics Departmcnt and the Country Operations Division, Country Departmcnt 1, Europe, Middle East, and North Africa Regional Office - is part of a larger effort in PRE to assist in the design and analysis of macroeconomic policies. Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington DC 20433. Pleasc contact Sanjev Aggarwal, room, N I1-019, extension 39176(59 pages plus 105 pages of appendices). To improve the Bank's macroeconomic model- targets on economic variables are given and ing capabilities, the Country Economics Depart- policy variables are solved for (the normative ment is developing a continuum of macro closure). Under both closures, a second choice, models referred to as RMSM-X and RMSM-XX. depending on whether an external credit con- These models share a common accounting straint or target is binding or not, is imple- framework that ensures economic consistency mented. among economic sectors. The interaction of the projection model and a RMSM-X is the simplest model, with an debt module is explained in detail. The debt elementary economic structure. The RMSM-XX module, which in the future should become more richly specifies the behavioral links among automatically linked to the DRS, allows the user economic variables. to experiment with different forms of debt restructuring in a simple manner. The debt Everaert, Garcia-Pinto, and Ventura show in module also allows to calculate the supply detail how to specify the budget constraints and schedule for foreign credit and to project in market clearing conditions in a RMSM-X model detail (by creditor) debt stocks, capital flows, for Turkey. They include six sectors: the and interest payments. Government, the State Economic Enterprises, the Central Bank, the domestic banking system, the Finally, since the model is based on the nonfinancial private sector, and the foreign concept of a consistent flow of funds among all sector. The different markets consist of a the specified sectors, it is necessary to build a domestically produced and exportable good, an consistent historical data set for at least the base importablc, a money market, a domestic credit year. Appendix I explains how such a set of market, a quasi-market for Central Bank Credit, consistent macroeconomic data was constructed. and a foreign asset market. This model can be used to project the behavior of these sectors in a The RMSM-X model presented in this paper simple manner, linked through the various will be extended to include more estimated markets. behavioral relations (RMSM-XX) for future operational work on Turkey. Applications of the They explain four possible closurcs of the RMSM-X model have also been developed for model. One choice depends on whether policy Colombia, Zimbabwe, Chile, and the Philip- variables are exogenous (the positive closure) or pines. KThe PRE Working Paper Series disseminates thc findings of work under way in the Bank's Policy, Research, and External Affairs Complex. An objective of the scries is to get these findings out quickly, evcn if presentations are lcss than fully polished. The findings, interpretations, and conclusions in these papers do not necessarily represent official Bank policy. Produced by thc PRE Dissemination Centcr  TABLE OF CONTENTS I. AN OVERVIEW OF THE SYSTEM II. THE RNSM-X MODEL The Consistency Framework The Behavioral Structure Closing the Model III. THE DEBT MODULE Historical Data and Assumptions Existing Debt and Debt-Restructuring New Debt Total Debt IV. CONCLUDING REMARKS BIBLIOGRAPHY APPENDIX 1: Creating a Consistent Data Base APPENDIX 2: Historical Data aud Output of the Model We are grateful to Ismail Arslan, Cavit Dagdas, Shideh Hadian, Mehmet Polat, Paulo Vieira da Cunha and Steve Webb for their contributions and to Yavuz Arinsoy, Vittorio Corbo, Nihal Ergun, John Holsen and Luis Serven for their useful comments.  The predominance of adjustment problems in LDCs since the early 1980. has prompted the need for extensive use of adjustment lending by the World Bank. In order to assist the design and analysis of macroeconomic policies, CECMG has initiated a major effort to enhance the macroeconomic modelling capabilities of the Bank. A continuum of macro models is being developed which are referred to as RMSM-X and RMSM-XX. These models share a common accounting framework that ensures economic consistency among the sectors of the economy. The level of sophistication of the behavioral structure is what distinguishes the different classes of models. The RMSM-X stands as the simplest model, with an elementary economic structure. The RMSM-XX is a step forward in the sense that it includes a richer specification of the links among economic variables. The present pape:-r, which is the result of a joint effort by staff of EMlCO and CECMG, presents the Turkey application of the RMSM-X model. This is only the first stage of a larger project involving the construction of a RMSM-XX model for Turkey. Given the macroeconomic management problems facing Turkey at present and the Bank's heavy involvement in all sectors of its economy, the development of these analytical tools is quite timely. They will be used extensively in EMlCO's future economic work. The RMSM-X model presented here builds on Holsen (1989a, 1989b) and Serven and Ventura (1989b). Other applications of this model for Colombia, Zimbabwe, Chile and Philippines are: Easterly et al. (1990); Kahdr et al.(1989); Serven (1990); and Riveros et 1  al. (1989), respectively. The paper is organized as follows. First, we rresent an overview of the system defined by the RMSM-X model, the Debt module (DM) and the data base. Second, we provide a detailed explanation of the theoretical model. We carefully specify its underlying economic structure in terms of both budget constraints for the different economic agents and market specification. We also discuss the implementation of alternative closure rules. Third, we present the Debt Module. As explained below, the DM allows us to calculate the supply schedule for foreign credit and to project in detail debt stocks, capital flows and interest payments by creditor. Since the RMSM-X model is based upon the concept of a consistent flow-of-funds, a prerequisite for the empirical application of the model is the construction of a consistent historical data set at least for the base year. Appendix 1 explains how we constructed such a set of consistent macroeconomic data. Appendix 2 presents a complete set of historical data as well as the output of the model, including the debt module. 2  I. AN OVERVIEW OF THE SYSTEM The Turkey Model is an integrated system that includes four components: Historical Data, the RMSM-X macroeconomic model, the Debt Module and Standard Tables. All these components have been organized in several, linked JAVELIN-models. Figure 1 gives a schematic view of their organization. Technically, the transfer of data among the different JAVELIN models is done with the special feature called the "Import Data Building Block." This reduces the effort of transferring the data between models to a few keystrokes. FIGURE 1: DATA FLOW CHART HISTORICAL| | DATA 3  In t.s HISTORICAL DATA model we copy each of the original data tables obtained from Turkish sources in a separate worksheet. We design formulas to map the original data in the squired flow of funds format. In this way future updates, revisions, or additional historical data collection are rendered automatic. It is sufficient to add or change the data in the original worksheet and the model will automatically (re-) calculate a consistent flow of funds. Separating the data collection and transformation from the actual projection model not only has the advantage of saving space and increasing speed but it avoids the potential confusion that might arise from the difference between the historical closure of the model (where the private sector is the residual, and the simulation closure (solving for selected endogenous variables). Appendix I explains in detail the construction of the historical flow of funds and the mapping of actual data into it. The RMSM-X projection model only requires historical data for a few periods in order to provide initial values for the simulations. In most cases, the base year data are sufficient since the economic relations between two variables seldom exceed a one year lag. Sometimes if parameters are calibrated on historical data it may be useful to add a few more historical periods in order to check the stability of these parameters. A detailed description of the RMSM-X model is provided in the next section of the paper. External debt and creditworthiness are key variables for the country as well as for the Bank's work. Therefore, we decided to 4  introduca separately a DEBT MODULE that details the interaction between the country and the foreign credit market. The particular details of this model are discussed in section TII. The basic input into the DEBT model is the existing pipeline debt by creditor as reported in the Bank's Debt Report System (DRS) and the terms and conditions of new debt (maturity and grace periods, interest rate and time profile of gross disbursements). At this stage the transfer of data between the DRS and JAVELIN is not yet automatic but this problem will be resolved in the near future. The nature of the transfer of data between the DEBT model and the RMSM-X model depends on the circumstances facing the country. If the country faces a foreign borrowing constraint the total available credit is sent from the DEBT model to RMSM-X as well as the interest rate. No data return from RMSM-X to the ,EBT model. If the country's foreign horrowing is not constrained, the DEBT model provides the interest rate to the RMSM-X. The RMSX-X model returns total demand for credit which is then matched by the DEBT model's pipeline and a gap. The gap is distributed across foreign creditors in the DEBT model, based on a set of assumptions. Finaily, we design standard output required for various purposes in a STANDARI) TABLES model. This includes the standard attachment to CSPs, CEMs and other Bank documents. The STANDARD TABLES model imports the data from the three other models. Part of the historical data come from the HISTORICAL DATA model and the DEBT model. Projections come from the RMSM-X model and the DEBT MODULE. 5  II. THE RMSM-X MODEL In this section we describe the RMSM-X model for Turkey. We start by presenting the cortnistency framework. Then, we append a simple behavioral structure to produce projections. Finally, we discuss alternative closure rules. 11.1. The Consistency Framework The RMSM-X model assures consistency in the projections by requiring that the budget constraints for the six eco.oamic sectors are satisfied. Each budget constraint consists of two statements of the type: CURRENT INCOME - CURRENT EXPENDITURE = NET SAVINGS NET SAVINGS = NET ACCUMULATION OF WEALTH The first statement is the current account of the sector, while the second is the capital account. These two equations can be reduced to a single expression: CURRENT INCOME - CURRENT EXPENDITURE = NET ACCUMULATION OF WEALTH In the rest of this su>Fsectionj we present the budget constraints for each of the economic agents or sectors. We omit the time subscript for current end-of-period stocks and for flows occurring during the current period. All budget constraints are defined in nominal terms. The symbols used are exp'Lained in Table 1. 6  Non-Financial Public Sector The economic importance of the non-financial SEEs in Turkey leads us to distinguis" them from other components of the non- financial public sector. Hence, we proceeded to decompose the non- financial public sector in two: budget (b) and non-financial SEEs (0). The former includes the central government, local governments, social security, extra-budgetary funds and revolving funds. As it will be shown shortly, the financial public sector, composed of the central bank and the financial-SEEs, has been incorporated into the financial sector. The current accounts of the budget and non-iinancial SEEs can be written as follows: (2.1) OFIb + P&Lc + TI + TDo + TDp + ET *fb SUB Tbo Tbp - iRCRb 1 - ic*Bb1l - Eei* F*b-1l PC Cb b (2.2) FIo + Tbo - TDo - iR*CRo_l- ic 'Bo-l Eoi**F So Equation ;2.1) defines budgetary savings as the sum of factor income, distributed central bank profits, tax revenues, and transfers from abroad, minus transfers to domestic sectors, interest payments on domestic and net foreign-currency denominated debt and consumption of the budget. Note that, although the budget owns some financial institutions other than the central bank, it does not receive a share of the profits and losses from the banking system. These are all distributed to the private sector. 7  TABLE 1s DEEINITIONS OF VARIABLES IN BUDGET CONSTRALINTS Variables with an asterisk are defined in US$. The rest of the variables are express^.i in local currency at current prices except for those variables marked with (#) which are defined in constant terms. B Bonds C Consumption (#) CR Credit from the central bank Cu Currency in circulation DD Demand deposits E* AoA..rage exchange rate F Pet foreign-currency denominated borrowing FI Factor income I* Investment (#) i Nominal foreign interest rate iD Nomi.aal interest rate on deposits iC Nominal interest rate on credits i Nominal rate of rediscount IA Imports (#) KT Capital transfers NW Net worth OFI Other factor income P&i Distributed profits PN Profit remittances abroad R Foreign reserves RR Legal reserves S Savings SUB Subsidies T* Net current transfers T Net transfers from abroad TD Direct taxes TI Indirect taxes VA* Value added WR* Workers remittances from abroad X Exports (#) Sector-specifii. variables and intersectoral flows are represented by the following suffixes at the end of each variable: b Budgetary government c. Central bank d Banking system 0 Other non-financial public sector (SEEs) g Consolidated non-financial public sector p Private sector m Consolidated monetary sector f Foreign sector t Total 8  Similarly, the banking sector does not pay taxes. We are implicitly assuming that these are paid by the private sector on its behalf. Equation (2.2) defines savings of the non-financial SEEs as the difference between factor income plus transfers from the budget and taxes plus interest payments on its domestic and net foreign-currency denominated debt (hereafter foreign debt). The capital accounts for the budget and non-financial SEEs are given bys: (2.3) Sb PIIb + KTbo + KTbd + KTbp - ABb - EeF*b - ACRb (2.4) so P.I10 + KTop - KTbo - AB0 EAF*0 - ACRQ Equation (2.3) simply states that budgetary savings plus domestic and foreign borrowing are invested and distributed as capital transfers to the =est of domestic sectors except for the central bank. Equation (2.4) shows that non-financial SEEs inve.tment and capital transfers to the private sector are finasiced through its own savings, capital transfers from the budget and domestic and foreign borrowing. It is important to note that changes on the net foreign position of the budget and non-financial SEEs, AF*b and AF*O, do not coincide with recorded balance of payments capital flows to these sectors. This is due to the existence of both cross-currency 1 Throughout the paper, we will use the following conventions: AX a X-X_1; X = AX/X 9  effects and foreign exchange transactions among domestic sectors. The same applies to changes in the net foreign position of the other domestic sectors. In Appendix I we provide a detailed account of how these effects were calculated. Non-Financial Private Sector In our model, the private sector incorporates all domestic economic agents not included elsewhere. It would have been desirable to decompose the private sector in firms and households. There is no doubt that the economic behavior of these groups responds to different incentives, and that modelling them together may obscure some interesting issues. Unfortunately, the lack of data made the task of distinguishing between firms and households impossible. The current account of the private sector is specified as follows: (2.5) VAp + Tb + E(T *fp+WR*) + i *Bp_. + iDD-D1 + P&Ld - TD - ES(i* .F *_+PR*) - PcCp =p p P p Equation (2.5) defines private sector savings as the excess of income, transfers and net interest receipts over consumption. Then, the capital account for the private sector is given by: (2.6) Sp = p1eIp+ ACU + ADD+ AB - E*AF p- KT - KT KT -EOADFI 10p p pp op bp dp 10  This equation explains that private investment, as well as increases in domestic lending and mcney holdings, are financed through savings, foreign borrowing (bond and equity) and capital transfers from other domestic sectors. Financial Sector The financial sector has been divided in the central bank and the banking system, which consists of financial SEEs and private financial institutions. This distinction makes it easier to distinguish among policy variables, e.g. central bank credit, and intermediate variables, e.g. the money supply. The current accounts of the central bank and banking system are given by: (2.7) iRCRt_l + Esi*(R*c-lF *c-l) P&Lc = c (2.8) iCeBd-l - iR CRd-1 - iDD*DD_1 - Eei *F*d-1 - P&Ld = NW Equations (2.7) an (2.8) define the central bank's and banking system's savings (increase in their net worth), respectively. In both cases, savings are equal to the excess of net interest receipts over distributed profits. Note that the banking system does not receive interest on their deposits in the central bank. This is consistent with Turkish financial regulations. The capital accounts of the financial sectors can be written as: (2.9) ANWc = ACRt + Eh(AR*c-AF*c) - AH 11  (2.10) NWd = ACUd+ ARR + ABd - ADD - ACRd- (KTbd-ETdp) - EoAF d Equation (2.9) states that changes in the net worth of the :entral bank and base money creation must equal credit creation and reserve accumulation. The base money consists of currency in circulation in the hands of both the private sector and the banking system (vault cash), and bank reserves. Accordingly, the implicit assumption is that the public sector does not hold money. Equation (2.10) forces the changes in banking sector's assets to be equal to changes in liabilities plus savings and net capital transfers. The former consist of vault cash, reserves in the central bank and domestic credit (net of time deposits). The latter consists of demand deposits, credit from the central bank, foreign borrowing, and savings plus net transfers. Rest of the World To complete the specification of the open economy model we include a foreign sector containing the rest of the world. The budget constraint of the foreign sector is nothing but the balance of payments. The current account is: (2.11) pIMIM-poX+Eo[i *(F*t_l-R )+PR*)-E(T fb+Tfp+WR*) = Sf Equation (2.11) defines foreign saving as the excess of imports and factor payments (interest and profits) over exports, transfers from abroad and worker remittances. The counterpart of this foreign savings is shown in the capital account as direct 12  foreign investmnt plus total debt accumulation minus the increase in c.ntzal bank's foreign reservest (2.12) Sf - B ADFI* 4 Ea(F*t-R*c) The Plow of Funds Presentation An alternativ, presentation of the budget constraints is given by the following expressions: CURRENT SOURCES - CURRENT USES CAPITAL SOURCES - CAPITAL USES Figure 2 presents the budget constraints following this alternative procedure. In the figure, we have two different matrices, one for the current account and one for the capital account. The CURRENT ACCOUNT MATRIX shows how the different sectors finance their respective current expenditure and savings. The CAPITAL ACCOUNT MATRIX gives us a picture of how the sectors finance their capital expenditures. The functioning of both matrices is parallel. Rows represent incomings and columns outgoings, i.e. the rows give us the"sources"of funds for a particular sector while the columns show the "uses" of funds. For example, the intersection of the Budget column with the Private Sector row in the current account gives us the amount of funds that are at the same time current uses for Budget and current sources for the Private Sector. In our frameowrk, these are transfers and interest payments. Total 13  FIGURE 2: SOURCES AND USES OF FUNDS MATRIX CURRENT ACCOWiT Gov rnsnt ou0hr Private Central Banking Belnce of Production Total Budget Public Sector Bank S1.t Payenta Account Sources TI tver,nmnt ID* lo PL E*Tf*b udget OFIb Vier Tho Flo ubl ic --tor CppEM .nk anking 1COBb tCSood iCeBpd -lance of I - l d syn nt I E*I-OF*b E-T-oF o E4100FP EOT*FWc Ei*TF*ed _ ~-Xt Consumption Cb Cp hI,11 Mid St and Savings Sb S, Sp AccountI ota I CAPITAL ACCOUNT Governument Other Private Central Banking Balance of Savings Total Budget Public Sector Bank System Paymnt. Account Sources :vernsnt Ibp CRb 5bd E*'Fb Sb udget__ _ _ ___ _ _ _ _ _ ub;,i KTbo Dop 60 03od E061o SO rivet. KTbp KTop AB E*AF Sp _ - _tar _ _ _ _ _ _ _ _ _ E 7 _D_ Central AP EO*F0AW Bank _=d nking KTbd D CEd ElFd = r ance ot E.hR% St Inveetmnnt lb I= Account TotalI Uses 14  sources are represented in the final column of a sector and must equal total uses which are represented in final row of the same sector. In this way, the flow of funds framework assures consistency among the data. II.2 The Behavioral Structure The flow of funds is a useful accounting framework that helps to organize the data in a consistent manner. If we want to use this information for policy analysis, the data framework must be linked to a model that contains the behavioral and technical relationships among variables. In this section we develop a simple behavioral model and append it to the consistency framework described above. I.1.2 The Real Economy For simplicity, we assume that the domestic economy produces only one composite good that can be used for domestic consumption and investment, or sold abroad. The condition of equilibrium in the goods market is: (2.13) Y + IM = Cp + Cb + Ip + Ib + Io + X where C, I, Y, X and IM denote consumption, investment, output, exports and imports of goods and services, respectively. All variables are measured in real terms. In order to provide a complete description of the real side of the model, we must specify the supply-side of the economy and the expenditure functions. 15  The Supply Side On the supply-side of the model, we assume that the economy is operatir. under a fixed coefficients production function and that capital is the constraining factor. Therefore, (2.14) AYF+l 0 BIt where yF is potential output and B is the incremental output- capital ratio or the inverse of the ICOR, corrected by the depreciation rate. It is total investment in the economy. Equation (2.14) is the growth equation. It states that the change in potential income is given by both the amount of investment of the previous period and the efficiency of this investment, as measured by the parameter S. The simplifying assumption that we are making is that investment has the same efficiency across sectors. This assumption can be easily removed by having different 4s for the different economic sectors. Equation (2.15) simply defines total investment as the sum of each sector's investment: (2.15) It I j - b,o,p. Finally, current income and gross output are obtained as: (2.16) Y - YF*cu where cu is the rate of capacity utilization. 16  The Expenditure Functions On the expenditure side, we must specify the consumption, investment, exports and imports equations. we will assume the following: (2.17) Cp - C$Yd (2.18) Ip - op*Y (2.19) 10 - 0*o@FIo (2.20) X - (1+exoq+py *Y )X_1 (2.21) IN - 1MF + IN, + Iv + ING (2.22) I - (1+COq+pC*Ct)oIMC_ (2.23) . (1+eI*q+pI*It)*IX _1 (2.24) IOK - ( A+evq pVY)*IMV l where q is the real exchange rate and Yd is disposable income2; c is the propunsity to consume; Y is the level of income of Turkey's export partners, respectively. as and ps refer to real exchange rate and foreign-income, consumption, investment and income elasticities. Finally, IMC, I', INV and ING refer to consumption, 2 Yd can be obtained from (2.5). Yd - (VAp+Tbp+E@(T*fp+WR* -PR* )+rCBp_l+rDD.DD-1 +P&Ld-TD p -E-r *OF* JPC Note that interest payments have been corrected of advanced capital payments due to inflation. Hence r is tjhe real interest on the asset i. In the case, of foreign debt r is the nominal foreign interest rate corrected for foreign inflation. 17  investment, intermediate and non-monetary gold imports. Ct denotes total consumption: (2.25) Ct = Cp + Cb Equation (2.17) states that consumption demand is linearly dependent on disposable income. Equation (2.18) explains that investment demand is a exogenously given fraction (oap) of GDP. Equation (2.19) projects investment of non-financial SEEs as a share (ao) of their factor income. It is a shortcoming of this model the fact that c and a p do not depend upon the real interest rate and inflation. The former would provide a measure of the opportunity cost of accumulating real assets versus financial assets and/or consumption. The latter would account for the effect of the inflation tax on both investment and consumption. Equation (2.20) assumes export growth to be a function of the growth rate of Turkish export markets and changes in the real exchange rate. Equation (2.21) defines total imports as the sum of consumption, investment, intermediate and non-monetary gold imports. The first three are then projected in (2.22)-(2.24) as a function of total consumption, total investment and GDP at factor cost, respectively. They also depend upon the real exchange rate. Imports of non-monetary gold are projected exogenously. Prolecting Prices and Nominal Variables Once real variables have been calculated, the RMSM-X model computes prices and nominal variables. The former follow these 18  simple rules: (2.26) P =(PE'PE1) (2.27) PI = (1-X)*P + XOPIM (2.28) PIM EOpIM (2.29) PX P where PE is the end-of-period GDP deflator, and pi represents the period average deflator of the expenditure component i, with i = I, IM, X. E is the period average nominal exchange rate, and X and -IM indicate the share of imports in investment demand, and the foreign prices of imports, respectively. Once these prices are obtained, it is straightforward to project nominal variables as follows: (2.30-2.36) NZ = p.JZ Z = Y,IpIIbIIoICbIXIIM' Total nominal consumption is obtained as follows: (2.37) NCP + NCb = NY + NIM - NIp - NIb NIO - NX Equation (2.37) constitutes the "national accounts" of the RMSM presentation. It portrays goods market equilibrium in nominal terms. Finally, the consumption deflator is obtained by dividing total nominal consumption by total real consumption: (2.38) PC = (NCp+NCb)/(CP+Cb) 19  In this way, mathematical consistency among prices and real and nominal variables is achieved.3 1.1.2 The Asset Markets As opposed to the RMSM model, which only includes the real side of the economy, the RMSM-X model integrates both real and monetary aspects. Therefore, we introduce a menu of four assets in the model. Money is defined as currency plus demand deposits, which are held by the private and banking sectors. The foreign asset can be held by all sectors. The domestic bond includes bank credit as well as public debt. Finally, the central bank extends credit to the budget, the non-financial SES and the banking system. The Money Market The RMSM-X model considers the different components of money as perfect substitutes. Therefore, the equilibrium condition in the money market is given bys (2.39) Ms = Md Money demand is projected with this simple rule: (2.40) Md - kpeY 3 As opposed to economic consistency. The latter demands to have as many prices and material balance equations as goods exist in the economy. 20  where k is the --exogenously given -- inverse of the velocity of circulation. Money supply is determined as: (2.41) MN - 'r*H where H is the base money and i- is the money multiplier. These are defined ass (2.42) H - CUp + CUd + RR (2.43) - a (cc+l)/(cc+re) cc and re are the currency to deposits and reserves to deposits ratios. These ratios are given exogenously. Thus, (2.44) cc - (CUp+CUd)/DD (2.45) ra - RR/DD Finally, the fraction of currency in circulation held by the private sector is given bys (2.46) - CUp/(CUp+CUd) where 0 is a given parameter. Foreign Credit Market We assume that there is only one type of foreign-currency- denominated asset. This allows us to state the following equilibrium condition in the foreign credit market: (2.47) F*bd + F* d + * pd + 2 + (F*c-R*c) d = F*t  Equation (2.47) states that the sum of the net demands for foreign credit of each of the national sectors must equal the total supply of foreign credit. The demand for foreign credit of the private and banking sectors are given by: (2.48) F pd OF*poY/EE (2.49) F dd = pF(l-re)*DD/EE where PF and OF are fixed parameters and EE is the end of period nomirnal exchange rate. The relationship between the end of period and period average exchange rates is given by: (2.50) E = (EE*EE-1)4 Note that *F and PF in (2.48-2.49) do not depend upon the relative rates of return on the different assets (including the rate of currency depreciation). This is an unrealistic assumption that we are forced to make in the absence of econometric estimates of asset demand equations. On the supply side two assumptions are possible. First, the country is credit constrained. In this case, F *ts and i* would be exogenously determined: (2.51) F t = t (2.52) i* = 1* F t and I would be calculated in the debt module. This case has been traditionally labeled the "availabilities" model in 22  operational work in the Bank. The other possible assumption is that the country can borrow in the international market at an interest rate that can be fixed or increasing. This interest rate would be calculated as follows: (2.52') i* = i eF e/F*t + i* n(F t-F e)/F*t where i*et and F e are the interest rate and amount of already e~~~~~~~ contracted (existing) credit, and i n is the interest rate on new credit. These variables are obtained in the debt module. This case has been traditionally called the "requirements" model. Domestic Assets There are two domestic assets other than money: the domestic bond and central bank credit. The domestic bond, issued by the budget, other public and/or private sectors, is held by the private and/or banking sectors. This market could be disaggregated into credit provided by the financial system and domestic public debt sold to the private non-financial sector. While this further distinction may appear very appropriate, one should be aware that it will be entirely irrelevant for practical purposes unless both assets are assumed imperfect substitutes from the viewpoint of at least one economic sector. In addition, this imperfect substitutability should be explicitly embedded somewhere in the model. Of course, this would require the specification of distinct supply and/or demand rules for each of the assets. This disaggregation would introduce unnecessary complications when we 23  cannot back it with meaningful behavioral assumptions. The condition of equilibrium in the domestic bond market is given by the following equation: (2.53) Bbs + Bos = gpd + Bdd Equation (2.53) states that the net supply of the domestic bond by the budget and other public sector, must equal the net demand by the private and banking sectors. We enter the interest rate on bonds exogenously: (2-54) ir = This assumption could be interpreted as having a perfectly elastic supply or demand for credit. This would be an unrealistic assumption and the user must carefully project the interest rate consistently with the evolution of demand and supply. The banking sector's demand for bonds is calculated as: (2.55) Bd = pB(l-re)DD where PB is a given parameter. The other domestic asset consists of the central bank credit. The equation that must be satisfied in order to assure equilibrium in the central bank's credit market is: (2.56) CRt" = CRbd + CRod + CRdd Equation (2.56) states that overall credit extended by the central bank is distributed among the budget, the other public 24  sector and the banking system 11.3. Closing the xodel If we consider each budget constraint as a single equation4 and substitute the behavioral relationships into the budget constraints and the market equilibrium conditions, we obtain a system of eleven equations. This is the compact form of the model. Given that the sum of all budget constraints is equal to the sum of the excess demands of all markets, one equation is linearly dependent on all others. Consequently, we can solve for a set of ten endogenous variables. The appropriate selection of the set of ten endogenous (or residual) variables depends upon the purpose of the simulation exercise to be undertaken. Nonetheless, the mathematical structure of the model imposes one restriction on the set of variables chosen. This condition is that each of the eleven equations must contain at least one endogenous variable. If this condition is not satisfied, the system cannot be solved. Note that this requirement applies to the eleven equations above. The choice of the set of endogenous variables determines whether the solution of the model is recursive, simultaneous, or if it has both simultaneous and recursive blocks. For simplicity, we will only consider those sets of endogenous variables that allow a recursive solution to the model. The purpose of this restriction 4 That is, if savings variables are always obtained as residuals. 25  is to reduce the software requirements. The RMSM-XX version of the Turkey model will extend the class of models to those requiring a simultaneous solution. There are many closure rules that are meaningful from the economic point of view. In the application for Turkey we have chosen to implement four alternative possibilities that are shown in Table 2. The four closure rules result from a two-dimensional classification. TABLE 2: CLOSURE RULES FOREIGN CREDIT CONSTRAINT Yes No Financial Normative Normative Programming & & USE OF THE Availabilities Requirements MODEL Positive Positive Projections & & Availabilities Requirements On the one hand, the Turkey RMSM-X can be used both for assessing the effects on the target variables of alternative macroeconomic programs and for obtaining the values of the policy variables that would be consistent with a set of exogenously given targets. In the first case, the policy variables would be determined exogenously, and the model would give us their most likely effects on the target variables. This closure rule defines a positive model which is useful to make projections of the "most 26  likely scenario", or to analyze the effects of proposed policies. In the second case, the purpose is not to find out the most likely path for a number of variables, but to determine which are the values of the instrumental variables that would be consistent with the desired levels of the objective variables. This closure defines a normative model. This closure rule will be preferred when the model is used to design a feasible financial or macroeconomic program. On the other hand, the model can be used with or without a binding upper bound on foreign credit. If the RMSM-X is used without a credit constraint or target, we will follow Bank's convention and call it a "requirements" model. In this case, the debt module would provide the RMSM-X with a credit supply schedule. The foreign credit market would be solved in the RMSM-X together with the rest of the macroeconomic model. If we use the model with a binding credit constraint, we label it the 'availabilities" model. Now the debt module would provide us with the amount of foreign credit that has been targeted or is available, and the RMSM-X would calculate the implications for macroeconomic policy or the effects on the target variablea. Before turning to the description of each model closure, we must warn the reader that the recursive nature of the model does not allow for the explicit consideration of all the relevant economic relationships among variables. As a result, it becomes necessary to check some of these relationships 'ex-post'. For example, the model does not directly relate consumption and 27  investment to the real interest rate or the velocity of circulation to inflation and interest rates. All these must be checked 'ex- post'. If these tests are not satisfactory, another iteration, reconsidering some of the assumptions and/or targets, is needed. In the rest of this subsection we first define the projection rules for some exogenous variables. Then we consider a positive and a normative closure for the model. We describe first the "requirements" version of both normative and positive closure rules. Then, we show how these closures are modified under the "availabilities" case. Projecting Exogenous Variables We project the following variables according to the rules: (2.57) FIo = fiop*Y (2.58) OFIb = ofiepOY (2.59) VAp = p*Y - FIo - OFIb - TI + SUB (2.60) TI = t yY + tIMOIM (2.61) TDp = tp*VAp (2.62) TDO = to*FIO (2.63) SUB = t5eY Then we project transfers among domestic sectors as follows: (2.64-2.70) X = (1+w)*XX1; X = TboITbptKTbo,KTbd,KTbp,KTop,KTdp. and transfers among domestic and the foreign sector: 28  (2.71-2.73) X - (1+4w)*X_1; X = T fb,T fpoWR- Finally, there are some variables that we project exogenously. These ares DFI , PR , P&Lcl P&Ld, iDD and iR. The Normative Model The purpose of this model closure is to find the fiscal, monetary and exchange rate policies that are consistent with a given set of macroeconomic policy objectives. The first step is to set up tarlets for the inflation rate (X), full employment growth rate (g), the real exchange rate (q), and foreign reserves as a certain number of months of imports- (res). This allows us to calculate p, YF, E and R *d: (2.74) p - (1+)*p_1 (2.75) YF - (l+g)*YF (2.76) E - q(p/pIM *) (2.77) R d * reso[(IM/E)/12] Once we have set these targets, the model solves for the values of policy variables that would be consistent with this objectives, given our assumptions. Figure 3 shows our proposed closure rule for a normative type model closure.5 We start with the goods market. Note first, that the growth equation, (2.14), and the definition of total investment, (2.15), determine the S5 This type of closure is consistent with the Fund's financial programming approach as presented in IMF Institute (1981). 29  FIGURE S: THE NORMATIVE AND REQUIREMENTS MODEL Fix Targets: g, W, q, res. Goods Balance of money Marle Payments Market Cb AFt H Foreign-Asset Central Bank Market Balance Sheet f ---- F*g{ I CRt < Public Sector Budget S [It] | CR_ _ Central ,c-CRg< Bank <- CRd < Banking System Credit Balance Sheet residual CRO Market t B9 |estic Bond Bp <C Market Secto;r-3 30  values of IT and Ib that are consistent with our growth target. Then capacity utilization is determined exogenously so that current output can be obtained. Once Y is known, all expenditure items can be projected except for Cb, which is the residual on the goods market equilibrium equation. In the balance of payments, the residual item is the amount of foreign borrowing, F*t. The projection riles for the rest of the variables have already been discussed. Once we have obtained the supply of foreign credit, we solve the foreign asset market. The demand for foreign credit of the private and banking system is projected with the help of equations (2.48) and (2.49). These projections together with the supply of foreign credit obtained in the balance of payments and the target for reserves determine the amount of foreign borrowing by the public sector, (F*b+F*o+F*c). The distribution of credit between F*b, F*0 and F*c is determined exogenously. In the money market, the money supply is determined by the exogenously given targets for inflation and growth. Given the paths of k, re and cc, the money market equilibrium equation yields the value of H that is compatible with the given inflation and growth targets. The central bank budget constraint is closed once we know the level of reserves, the amount of foreign borrowing and the base money. The residual is the level of domestic credit, CRt. The budget constraint of the banking system is solved for the banking system's demand for central bank credit. Then, the central bank credit market, solves for the amount of credit that is given 31  to the public sector, (CRb+CRo). CRO, is the residual of the non- financial SEEs budaet constraint. Consequently, credit to the budget, CRb, is obtained residually. The private sector's budget constraint determines Bpo Finally, in the domestic bond market, public sector's domestic borrowing, (Bb+Bo), is determined. As was the case in the foreign asset market, we distribute the amount of borrowing between SEEs and budget exogenously. The Positive Model In contrast to the previous model, the purpose now is to find the effect on the target variables of given fiscal, monetary and exchange rate policies. This model closure could be especially useful when a particular policy package needs to be evaluated. First we must determine the path of fiscal policy variables. Once this has been done, the values for the nominal exchange rate and monetary policy must also be entered.6 These, together with the remaining assumptions, will determine the growth rate, inflation, real exchange rate and foreign reserves. Figure 4 details the solution structure of this second case. Comparing figures 4 and 3, the complete symmetry between the present closure and the previous one becomes obvious. Note, however, that there are only eight endogenous variables in 6 As it stands now, monetary policy is entered in terms of base money. we could alternatively define it in terms of money supply. The change required in the model would be trivial. 32  II- q 4' IV I I N .99' 21 U w p A, On rq; '9 1* S .93 [ .9 1' 0 1 4. Jr - sip p gI pp.9 -9' pg 4- c 2 . ii N, gY .9  Figure 4. This is because all public sector variables have been determined previously. Any two variables in the budgetary government and non-financial SEEs budget constraints could be seen as the missing residuals. In the goods market, budget investment and consuLnption are now projected exogenously. Therefore, potential output growth is no longer determined exogenously, but by using the growth equation. A tempting closure would have been to leave the real exchange rate as the adjusting va.iable. The problem with such a solution is its simultaneous nature. Hence, we are forced to choose an expenditure component to close the goods market. Given the positive nature of this model, government variables have already been determined. It seems appropriate to leave as the endogenous variable the level of private investment, and hence future growth, compatible with domestic and foreign savings that result from the assumed policies. Therefore, the private investment equation, (2.18), is not used. In the money market, the money supply is determined by the exogenously projected base money. Therefore, the price level is the variable that adjusts money demand to the fixed supply. There a.e no changes in the balance of payments. In the foreign asset market, the balancing item is now the stock of foreign reserves. As in the normative model, the central bank budget constraint determines overall domestic credit. However, in the central bank credit market, the residual is now CRd instead of CRb. The budget constraint of the banking system determines its demand for the domestic bond -- that is, its total credit supply. 34  Therefore, equation (2.61) is not used under the positive closure rule. Finally, the domestic bond market is now closed with the private sector's domestic financing, Bp. Constrained Foreign Borrowing The closure rules just presented implicitly exclude the possibility of a foreign borrowing constraint. We define the latter as a binding upper bound on F* . In other words, the current account deficit resulting from (2.1l) and (2.12) is not feasible. In this case, F*t would be exogenously determined. In the normative model the residual of the balance of payments would not longer be AF*t but the change in reserves of the central bank. In the positive model F*t would be replaced by F * as the closing t ~~~~~~p variable for the foreign asset market. Figures 5 and 6 show the normative and positive models under a foreign credit constraint. 35  V1U3o 2. 13 NOUI&?1V AND AVAMAILrMLZI3 MOmU Viz S rsets.s. r q, rem. 1 1. 1~ Goods e of NY Market [jsmat market Wle I Z~~~~~~ Cb _ Iorei-Asset Catral 'l Macktt |balance 8 r t*g I I A wt e .1m Ihee _~~~~~~ Public Sector Budget $nl Central Bank V - ca k. q,.g Systm Cedt lie Sheet I residual Ca* Market IS Domwtl sm op "ri tc Ka sector 36  FIG=U GS TII POSITIV AND AVAILABILITIES MODELS Fix fiscal. metarr nd exchange rate policy . . ~~~~~~1 r., Goods Froregn Balance of Market AsstMake Payents