Policy Research Working Paper 8965 The World Bank Macro-Fiscal Model Technical Description Andrew Burns Benoit Campagne Charl Jooste David Stephan Thi Thanh Bui Macroeconomics, Trade and Investment Global Practice August 2019 Policy Research Working Paper 8965 Abstract This paper outlines the structure and economic founda- parameters are country specific and estimated at the country tion of the World Bank’s macroeconomic and fiscal model level. Forecasts across countries are live-linked, with the (MFMod). MFMod consists of individual country models export market growth of each country calculated as a trade- for 181 countries. The models are used by country econo- weighted average of imports of each of its trading partners. mists within the World Bank’s Macroeconomics, Trade and Remittance inflows and outflows are balanced across coun- Investment Global Practice to (i) generate country forecasts tries through a similar mechanism. Other cross-country and (ii) simulate various policies. Each model has a similar linkages come through the real effective exchange rate and structure and functional form, with variation reflecting data export and import prices, which are a function of world availability and economic specialization (notably for oil commodity prices and local cost considerations. exporters). Although the functional forms are similar, the This paper is a product of the Macroeconomics, Trade and Investment Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at aburns@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team The World Bank Macro-Fiscal Model Technical Description Andrew Burns, Benoit Campagne, Charl Jooste, David Stephan, Thi Thanh Bui1 JEL Codes: C10, C50 Keywords: Economic modeling, Econometric specification 1 The views in this paper reflect those of the authors and not necessarily that of the World Bank. For any queries reach out to aburns@worldbank.org. The above authors represent the team currently most active in model development, the current version of the model itself is a product of many years of work by a much larger group of people, within and without the Bank. Introduction ...................................................................................................................................................................3 Supply Side Equilibrium .................................................................................................................................................4 GDP Demand Side ..........................................................................................................................................................6 Real Household Consumption ...................................................................................................................................6 Government Consumption and Investment in the national income accounts .........................................................8 Private Gross Fixed Investment .................................................................................................................................9 Exports of Goods and Services ................................................................................................................................10 Imports ....................................................................................................................................................................12 Inventories and the statistical discrepancy .............................................................................................................12 Price Determination ....................................................................................................................................................13 Relative Price Determination ..................................................................................................................................14 Keyfitz Prices ...........................................................................................................................................................14 Import Deflator .......................................................................................................................................................15 Export Deflator ........................................................................................................................................................15 Exchange Rates........................................................................................................................................................16 Real and Nominal effective exchange rates ............................................................................................................16 Labor Market ...............................................................................................................................................................17 Labor supply ............................................................................................................................................................17 Employment ............................................................................................................................................................17 Wages ......................................................................................................................................................................18 Fiscal Accounts ............................................................................................................................................................18 Government Expenditure ........................................................................................................................................19 Government Revenue .............................................................................................................................................20 Fiscal Balances and Debt .........................................................................................................................................21 Balance of Payments ...................................................................................................................................................22 Current Account ......................................................................................................................................................22 Capital and Financial Account .................................................................................................................................24 Overall Balance of Payments and Change in Reserve Assets ..................................................................................24 Monetary Policy ...........................................................................................................................................................24 Production Accounts ...................................................................................................................................................25 Real sectoral Value-added.......................................................................................................................................25 Sectoral Value-added deflators ...............................................................................................................................26 Concluding Remarks ....................................................................................................................................................26 References ...................................................................................................................................................................28 Appendix A: Conversions and Identities ......................................................................................................................30 Appendix B: Data .........................................................................................................................................................31 Appendix C: Detailed Variable List ...............................................................................................................................32 2 Introduction The World Bank’s Macroeconomic and Fiscal Model (MFMOD) is a structural econometric model, similar in basic formulation to the classic Klein- or Cowles Commission-type (Fair, 1992) models. Like Financial Programming (and Computable General Equilibrium) models (cf. Mikkelsen, 1998), structural models reproduce the flow of funds across the whole economy by mapping out the main identities of the national accounts, balance of payments, labor markets and financial sectors. In contrast to financial programming models, macrostructural models make a concerted effort to estimate the economic and behavioral determinants of economic variables. In models like MFMOD, these structural relationships are developed to be both consistent with economic theory and the observed dynamics of the economy. While such models fell into disfavor among some academic researchers, they continue to be widely used in ministries of finance (Rudebusch, 2002) and by professional forecasters. Moreover, noted academics (Clements and Hendry, 1999; Blanchard, 2018) argue that macroeconometric models are important tools for forecasting and policy analysis. Indeed, they are the preferred instrument for many tasks due to their ability to incorporate a detailed mix of judgement and the rich set of transmission channels they embody, which allows policy impacts on the broader economy to be followed at a deeper level of disaggregation than in other macroeconomic models (Saxegaard, 2017). Most recently, macro- structural models have enjoyed something of a resurgence among Central Banks. For example, recent reviews of forecasting and macroeconomic analysis for the Bank of England (Pagan, 2003) and the Reserve Bank of Australia (Pagan and Wilcox, 2016) recommend that these central banks supplement their existing tools with macro-structural models because of their ability to support a hybrid judgment and model-based forecasting process – versus less flexible but theoretically pure model-centric approaches. The European Central Bank has also recently returned to a macro- structural model as its main policy analysis tool (Constancio, 2017). MFMOD is similar in structure and design to global models such as OECD's Interlink model (Richardson, 1998; Daalsgard, André, Richardson, 2001), the UN's Project Link model, the Federal Reserve's FRB/US model (Brayton, Laubach and Reifschnedier, 2014), and the ECB’s multi-country model ECB-MC, as well as those used by the Congressional Budget Office in analyzing policy in the United States (CBO, 2013). A wide range of private-sector forecasters also use similar models (IHS Global Insight, DRI, Oxford Economics and a variety of investment banks). MFMOD is currently estimated for 181 individual countries (developing and developed). Most parameters in the model are estimated using the error correction approach of Wickens and Breusch (1988). The equations’ equilibrium or steady state conditions are derived to be consistent with economic theory. In contrast, short-run dynamics are data-driven, with estimated parameters reflecting the actual behavior of the economy (see accompanying paper on the calibration and 3 estimation of MFMod (Burns et al., 2019)). As a result, the speed of adjustment of each country- specific model to its economically-determined long-term equilibrium as well as the steady-state to which each return depend mainly on the actual historical behavior of the economy. In contrast, in DSGE models, this steady-state is typically derived using micro-founded parameters which are often calibrated, notably when data are limited as in developing economies. In both DSGE and macro-structural models the overall time-variant equilibrium growth path of the economy is determined by potential output, which is calculated using the production-function methodology as a function of total factor productivity, the level of capital and labor. This paper describes the May 2019 version of the model, used in the Spring 2019 forecasting round. The following section provides a technical overview of the model. Supply Side Equilibrium � ): The core of each country model is a notion of potential output ( � ̃ ∗ 1−   =      −1 (1) 1564 ∗ ∗ ∗ = �(1 − ) ∗ ∗ � ̃ is trend total factor productivity (TFP); ∗ where is structural employment; is the natural rate ∗ of unemployment; ∗ is the equilibrium labor force participation rate; 1564 is the working age population; and −1 is the capital stock installed at the end of period and where which represents time on an annual basis. 2 Trend TFP (see Box 1 for details on the calculation of trend TFP and the capital stock), the structural unemployment rate and structural labor force participation rate are all computed using an HP filter. To deal with the end-point problem, the historical time series are extended 15 years in the future using the average growth rate or level as appropriate, prior to running the filter, over the recent period. In a subset of countries where employment and labor force participation data are unavailable the working-age population (15-64) is used as a proxy for trend employment. In these countries, because the labor input is larger, the level of the TFP measure will be smaller than it would have been if (unobserved) employment was used as the labor input. The growth rate of trend TFP will not be affected unless the unobserved trend employment to participation rate changes or the unobserved natural unemployment rate is changed. 2 Unless otherwise noted, all variables are expressed in constant price terms. Levels of the variables are denoted by upper case letters while lower-case letters denote the log of levels (i.e. = log( ) ). The first difference operator is ∆. 4 Over the forecast horizon, the growth rate of trend TFP, the natural rate of unemployment and the natural participation rate are assumed to be exogenous and are held constant at their latest estimated structural rates. From the above equations, it follows that the only factors that will lead to changes in potential output are those that directly impact the capital stock, equilibrium employment or total factor productivity. Shocks to demand that do not affect these three factors do not affect potential output. Instead they will change the output gap (the difference between actual and potential GDP expressed as a percent of potential GDP) (see equation 2). The output gap measures the extent to which resources within the economy are under- or over-employed. Higher (lower) levels of demand will increase (lower) the output gap. A positive or negative output gap is the principle equilibrating force in the model. The output gap feeds back into the model via the Phillips curve through inflation. A positive (negative) output gap raises (reduces) inflationary pressures, which in turn reduce (increase) real incomes and decrease (raise) profits. The first effect serves to reduce aggregate demand, while the second induces additional investment and supply. Combined, these effects serve to progressively return the economy toward equilibrium, the level of potential output. �   −     =   � �  100 (2) � For oil-producing countries the production function is modeled using non-oil GDP rather than total GDP. In general, the non-oil GDP output gap is found to provide a better reflection of demand-sensitive inflationary pressures in oil-producing economies. Box 1: Construction of the Supply Side in the Macro-Fiscal Model In the absence of reliable data on the capital stock, the capital stock is estimated through a highly simplified implementation of the perpetual inventory method following Burns et al. (2014). The capital stock (K ) at any given period is estimated by adding the fixed capital formation (I ) (while also allowing for depreciation and scrapping of capital at a given rate (δ ) over the previous period).    =    − 1 (1  −    ) + A problem with this formulation is that in the absence of an initial estimate of the capital stock this method will underestimate the capital stock in early periods, because implicitly at the beginning of the data period the capital stock is zero, whereas it is almost certainly much higher. To overcome this problem, later-period estimates of the capital stock (at which point the influence of the missing initial capital stock should be smallest) are used, in combination with 5 GDP data, to estimate an equilibrium capital-output ratio. This ratio is then used to estimate the capital stock at a given start-point using output data at t=0, allowing more accurate estimation of the capital stock during earlier periods. ∑ =1(1 − ) −1 (0)   =     � − (1 − ) � 0 For most countries structural employment is derived as the product of the working-age population (15+), the equilibrium labor-force participation rate and (1-the natural rate of unemployment). Once the capital stock and the labor force have been estimated, instantaneous total factor productivity ( ) is calculated by inverting the production function using output data. Income shares (α) are taken from the estimates reported in the Penn World Tables.   =  1−α ∗α K t−1   Lt ̃ ), and the time-varying natural unemployment rates are estimated using a Hodrik- Trend TFP ( Prescot Filter with smoothing parameter λ=100. In each case, the end-point problem is alleviated by extending the historical series to 2100 at the average level or growth rate of the most recent 10 years. Typically, the values of all three variables are set to the most-recent level of the estimated trend series. GDP Demand Side GDP by expenditure is the sum of its components:   =     +     +     +     −     +     +   (3) The model’s demand side breaks out private consumption (C ), government consumption (G ), investment (I ), which is the sum of private investment (IP ) and government investment (IG), exports (X ) of goods and services imports (M ) of goods and services, changes in inventories (ΔInv ), and a variable for statistical discrepancy (D ). Real Household Consumption In the model, real private consumption is modeled as an error-correction equation (see Box 2) of real household disposable income and, when data are available, real financial wealth. The real interest rate also enters the consumption function and acts as a constraint on borrowing. If the economy overheats, rising interest rates constrain borrowing, which will reduce household consumption. 6 − + − − △ =   − �− − � � − ( − ) � ��+ − − (4) + − △ � � + △ � �− ( − ) + where is private consumption; are government transfers and, if relevant, remittances; WEALTH is financial wealth of domestic agents; PtC : the private consumption deflator, and other sources of household income (government transfers and remittances). Disposable income is defined as labor income (net of direct taxes) = − Where is the nominal wage bill in the economy (the sum of the government and private wage bills) and are direct taxes. Income from capital is not directly measured but is subsumed in the wealth measure (which proxies for both income from wealth and consumption from changes in wealth). In the model, financial wealth is defined as the sum domestically-owned capital stock (foreign or resident) and domestically-owned bonds (foreign or resident). Yet, in the absence of data on both the location of ownership of assets and the decomposition of financial assets, wealth will be imperfectly proxied in the model as the sum of net foreign assets, the domestic capital stock and domestic government debt. 3 In the long-run the coefficients on income and wealth are constrained to sum to unity. This constraint determines the long-run saving rate as a function of the ratio of wealth to income. In the short-run, the coefficients on labor income and wealth are unconstrained. The consumption function illustrates the importance of prices in restoring equilibrium in the model. First, when consumer prices rise in the model, as in the case when the economy overheats (through a Phillips-curve mechanism described below), the real labor income and the real return on financial wealth are reduced. As such, consumption also decreases. The real interest rate is derived from the household’s first order condition and resembles the Euler equation that dictates intertemporal consumption choices. An increase in interest rates today, as for example through higher lending rates, will reduce consumption today. 3 The idea of Ricardian equivalence, popularized in Barro (1974), suggests that public debt should not feature as net wealth for households because the gross asset of debt is offset by an equal liability of the future taxes that will have to be paid to service that debt (see Mankiw and Elmendorf (1999) for a theoretical and empirical review of the evidence for Ricardian equivalence). Others argue that in a world with overlapping generations, or myopia, distortionary taxes, or imperfect financial markets, an increase in debt will be perceived as an increase in wealth for some households resulting in increased consumption. The modeling here is agnostic on the debate, allowing for the possibility that Ricardian equivalence does not hold. 7 Box 2: Error-correction models Error-corrections models are a standard way of modeling non-stationary but cointegrated variables X and Y (see for instance Wickens and Breusch, 1988). For instance, the consumption function shows the importance of prices in restoring equilibrium in the model. Consumer prices will rise in the model when the economy overheats (through a Phillips-curve mechanism described below) reducing real labor income and financial wealth and hence consumption. This relationship can be modeled using an error-correcting model (ECM), of which the simplest version is in the following form: Δ = − [−1 − −1 ] + Δ−1 + Δ Where lowercase variables are in log-terms. As such, the ECM relates the growth rate of variable Y to variable X, and some other variables Z that might not be cointegrated with Y. The term [−1 − −1 ] captures the long-run cointegrated relationship, while Δ−1 + Δ measures the short- term impacts (here, autoregressive persistence and another variable). The model is said to be error-correcting because it ensures that the long-term relationship between Y and X holds, i.e., the cointegrating relationship is always verified in the long-term. The cointegrated relationship is derived from the long run equation −1 = −1 + , (which presents an elasticity of y with respect to x of 1) rewritten as −1 − −1 because the expected value of (the error term in the LR equation) is zero. Written in a one period lag, the […] expression measures the distance of the equation from equilibrium at t-1, and the parameter θ measures the speed of adjustment in the absence of additional shocks. Mathematically, θ must be between (0,1) for the equation to converge to the long-run. The inverse of θ approximates the speed of adjustment, for example with annual data, if θ=0.2 then it will take roughly five years for the gap to disappear. The short-term parameters operate only in the moment and drop out when those variables reach a steady state (when Δ−1 and Δ Δ converges to constant values) or play a temporary role as along the equilibrium balanced growth path of the model. A final implication of this particular ECM form is that in the long term, the growth rate of Y will equal the growth rate of X or that the ratio of Y and X converge to a fixed point (for example household consumption will converge to a fixed share of GDP). Government Consumption and Investment in the National Income Accounts In the model, nominal government consumption in the national income accounts (NIA) is mapped to nominal public expenditures on wages and goods and services in the government accounts, 8 while nominal investment in the NIA is mapped to nominal capital expenditures in the fiscal accounts. For many developing countries the government spending aggregates in the national accounts do not equal government spending in the fiscal accounts (due to, for example, differences in accounting methodologies). The model allows for these historical differences to exist but forces the proportional gap between the two measures to be preserved in the forecast period by setting the rate of growth of nominal government spending in the national income accounts equal to the rate of growth of the government spending in the fiscal accounts over the forecast period. In the model, real government consumption and investment as recorded in the NIA are set equal to the corresponding nominal government expenditures deflated by the corresponding deflator from the NIA. Private Gross Fixed Investment The stock of capital in the economy accumulates as follows: 4 = (1 − )−1 + (5) Dividing both sides by −1 gives = (1 − ) + −1 −1 And re-arranging − 1 + = −1 −1 The assumption of a long-run balanced growth path implies that in the long-run the rate of growth of the capital stock will equal the rate of growth of output ( ), or� = . Δ − 1� = −1 −1 Re-writing further gives: + = −1 −1 = (6) ( + ) 4 Capital stock in the model represents total capital stock. However, one may also write capital stock as the sum of the private and the public capital stocks. The model response to a public investment shock will differ under various functional forms. In this model, an increase in public investment will crowd out public investment via changes in the cost of capital. If public capital stock, however, enters as a separate argument in the production function (see Baxter and King, 1993), then public capital stock acts as a technology shifter which crowds in private investment. 9 Using our production function, the demand for capital can be derived from the first-order conditions for profit maximization as: (1 − ) ∗ −1 = (7) Combining this condition with equation (7), investment in the long-run solves to: ( + )(1 − ) ∗ ∗ = (8) Notice that (1 − ) ∗ = which is the gross operating surplus (our proxy for profits). Equilibrium investment therefore rises with an increase in equilibrium profits and expected growth ( + ) while falling with a higher cost of capital ( ). Private investment is modeled with an error-correction form using the equilibrium condition in equation (9). In the short-term, deviations in investment from equilibrium are driven by movements in the real rental-rate of capital and short-run accelerator effects from changes in demand. The equation for private investment can thus be written as: ∗ (−1 + −1 )(1 − )−1 −1 ∆ = �−1 − log � � � + + ∆ − ∆� / � (9) −1 Where is the user cost of capital and the private investment deflator; is the depreciation rate; and is the growth rate of potential GDP. The real rental rate evolves according to a user-cost of capital type formulation (see Jorgensen, 1996): ( + + − ) (10) = 1 − ̂ Where the user cost of capital depends on the nominal interest rate ( ), the effective direct income tax rate (̂ ) and the private investment deflator. The user cost also includes other costs a portmanteau for the equity premium but also other factors that affect the cost of capital, such as banking-sector regulations, banking-sector concentration and the cost of recovering collateral. In history, these costs are calculated as a residual to ensure that the user cost identity in equation (11) holds while in the forecast period the term is exogenous and is typically held constant at its most recent estimated value. Exports of Goods and Services Exports of goods and services depend on each country’s trading partners’ demand for imports and the competitiveness of that country’s exports. In the model the growth rate of demand for 10 domestic exports ( , ) are proxied by the weighted sum of the growth rate of import demand from their trading partners (∑ ≠  ). The size of the export market (, ) is then , recomputed using this growth rate and normalizing it to the domestic export volumes in the base year 2010 (2010). , ,   =  � (11)  ≠  , , = (1 + ),−1 and ,2010 = 2010 where gX is the growth rate of variable X, ω ij : i the weight of country j ’s share imports in the exports of the domestic economy (with the weights calculated using the average rates from COMTRADE data for 2009-11) and M tj KD : is the volume of country j ’s total imports in constant price dollars at time t. We assume that small open economies are price takers. This implies that the price of domestic exports ( ) corresponds to the world price ( ) of the export basket. In this context, the decision to export or to sell domestically from the point of view of a local producer will depend on the relative price of the world price ( ) and the domestic sale price (an increase in constitutes are reduction in the relative price of export goods and a movement downwards along the export supply curve). The price-maker variant of equation (13) expresses the price of exports relative to the world price (in this case proxied by the US price in local currency). For price-makers, an increase in the export price relative to the world price reduces export volumes. The demand for exports is therefore modeled as: (12) −1 ∆ = − �−1 − −1 − � ( � + ∗ � + 2 ∆ ) −1 + 2 ∆ � � + Where the term in square brackets is the error correction term, which serves to move exports back towards steady-state levels over time. A homogeneity restriction on the export market variable is imposed to ensure that in the long-run and in the short-run, if relative prices are constant, that domestic exports grow in line with world demand. The homogeneity restriction is a strong assumption that has been violated for long periods of time in many countries, notably during the 1990s when structural changes in global trade patterns saw the export market shares of developing countries rise substantially. 11 To account for this a linear time-trend is included in the historical period, allowing for a variation in the long-term export share. Over the forecast period, the increment in the value of the time- trend is gradually lowered towards zero so that in the very long run equilibrium growth moves in line with global demand. For oil-exporting countries, exports are disaggregated into oil and non- oil exports. Non-oil exports are modeled in an identical way to total goods and services in equation (13). For oil exports, it is assumed that countries are price takers on global markets and that they will export all production that is not consumed domestically. Therefore, real oil exports are determined by an identity as the difference between production ( ) and apparent domestic consumption of oil ( ) (consumption of oil and changes in stocks). If the domestic country produces 10 bn barrels in a given year and stores or directly consumes 2 bn then 8 bn will be exported. Oil production is modeled as an exogenous variable, dictated by political and engineering factors. Imports Imports reflect both final and intermediate demand for goods and services (both consumption and capital goods) as well as the relative price of domestic goods compared to imported goods. −1 (13) ∆ = − �−1 − −1 + � � + ∗ � + 2 ∆ − ∆ � � −1 + The term in square brackets represents the error correction term. A homogeneity restriction is imposed on domestic demand to ensure that in the long-run imports remain as a steady share of GDP (constant import-penetration ratio). The second term � � accounts for changes in −1 −1 the relative price of imported to domestically produced goods. If domestic prices fall relative to import prices, then the import share will decline. As in the export equation, a linear time-trend is included in the long-run to account for historical changes in the share of imports in GDP that would violate the homogeneity restriction. In the short-run the coefficient on domestic demand (and similarly on the short-run relative price of imports to domestic goods) is freely estimated. Inventories and the Statistical Discrepancy While inventories do change in the historical period, forecasting changes in stock building is notoriously difficult. As a result, this variable is left exogenous with the advice that unless analysts have strong reasons to change it in the short-run they should hold both its nominal and real values as a constant share of GDP during the forecast period. Similarly, both the nominal and real statistical discrepancies are exogenous variables in the model and in the forecast period analysts are suggested to hold them constant as a share of GDP at the level of the most recent observation. 12 Price Determination Inflation is modeled using a modified Phillips curve specification, incorporating both demand-pull (summarized by the output gap) and cost-push (derived from marginal costs) pressures. The key price in the model is the implicit deflator for gross domestic product at factor cost ( ), which serves as a proxy for producer prices. △ = (1 − 1 ) △ −1 + 1 ((1 − 2 ) + 2 Δ ) + 1 + (14) Where are inflation expectations; is the output gap; and is the marginal cost of production. In the model, expectations include both an implicit forward-looking component (which in the model is exogenous and equal to the historical average inflation rate or the central bank target rate) and a backward-looking component that incorporates a degree of stickiness in price changes. The weight of each component in the model is determined by the parameter 1 . Notice, however, that equation (15) is written in growth terms (no ECM terms) – this due to the existence of a price-wage loop where cointegration exists between wages and prices. In MFMOD, wages are written in ECM form, thus accounting for the common stochastic trend between the two variables. Nominal marginal costs are determined by firms’ cost minimizing behavior (derived from the production function): 1 1 1− 1 (15) = � � + ln �� � � � � + (1 − ) ln � ⏟ ⏟ ) � + ln( 1 − ��������������� = = Where is the labor share of income; is the nominal wage rate; is the nominal rental rate of capital and is total factor productivity. The output gap is included in the Phillips curve (equation 15) to account for short-run deviations from equilibrium to respond to demand pull pressures. Thus, any excess demand over supply will generate an increase in prices, which erodes real incomes and increases the cost of capital. Prices will continue to rise until the economy moves back into equilibrium (the output gap closes). In the long-run, growth in marginal costs (mc) will equalize the growth in prices (see equation 16) as factor costs will also follow price inflation. In addition, the central bank interest rate rule will impose that growth in prices equalizes to expectations in the long-run. All in all, factor costs will therefore also grow along inflation (expectations) in the long-run. The output gap term also serves as a proxy for the time varying markup (see Gali (2005) for an exposition on how the output gap 13 and the gap between the marginal rate of substitution and the marginal rate of capital are related). 5 Relative Price Determination The different demand side deflators (private and government consumption, and investment) are determined by a mix of the producer price and import prices. The relative importance of each component in both the short- and long-runs is determined econometrically. In addition, effective indirect taxes (primarily tax on the sales of goods and services ∆ ) are also included. To account for both prices of locally produced and imported goods and services in the consumption basket, in equilibrium the private consumption deflator will grow at a rate equal to the weighted average of the rate growth of producer prices and imported goods plus any change in indirect tax rates (usually zero in equilibrium). The inclusion of the output gap proxies the markup of the retailer. The level of prices is indeterminate. ∆ = − �−1 − 1 −1 − (1−1 )−1 + log(1 + −1 )� + 3 ∆ (16) + (1 − 3 )∆ + ∆ + + The investment and government consumption deflators are modeled in an identical fashion (albeit with different estimated coefficients) to the private consumer deflator in equation (17) thereby ensuring consistency across domestic deflators in the economy. Keyfitz Prices To determine export and import deflators, the model uses a weighted average of the global (USD) price ( ) of 32 internationally traded commodities that each country imports and exports. The weights w j X and w j M are given by the share of each commodity in total domestic exports and imports of goods in the base period (the average observed between 2009-2011). These accounting constructs are called Keyfitz prices in the model. The Keyfitz prices for oil exporters exclude oil prices and oil exports from the Keyfitz price calculation given that the oil export deflator is directly a function of international oil prices. 5 One may compute the markup () following Roeger (1995) by equating the primal Solow residual ( ) with its dual ( ): (Δ + ) − (Δ + Δ ) − (1 − )(Δ + Δ ) ( − 1) = [(Δ + Δ ) + (Δ + Δ )] 14 (17) ,   = � 2010 (18) ,   = � 2010 Changes in these Keyfitz prices reflect the change of import and export prices that would be observed if domestic prices for imported goods were equal to world prices in local currency and if these prices were fully passed through. These indices are used as inputs in the equations for import and export price deflators. Import Deflator In the model, the price of imported goods is a function of the world price of goods (proxied by our Keyfitz variable converted to local currency terms). The Keyfitz price does not include services. In the model, the price of imported services is proxied by domestic consumption price deflator on the assumption that imported services are priced to market. Overall, the import price of goods and services equation is expressed as: ∆ = − �−1 − −1 −1 − (1 − )−1 � + ∆−1 (19) + (1 − )∆ + Where is the import Keyfitz price expressed in local currency terms; and is the consumer price. In both the short- and long-runs a restriction is imposed that the weight of the world and local prices sum to one. This means that if all prices rise by 1 percent in the model, the import price will also rise by 1 percent. Export Deflator Like the price deflator of imported goods and services, the export deflator of goods and services is a weighted average of world export goods (commodity) prices ( ) and the domestic consumption price which proxies both cost pressures and the pricing of the export of services: ∆ = − �−1 − −1 −1 − (1 − )−1 � + ∆−1 (20) + (1 − )∆ + Where is the export Keyfitz price expressed in local currency terms; and is the price of domestic consumption. In both the short- and long-runs a restriction is imposed that the weight 15 of the world and local prices sum to one. This means that if all prices rise by 1 percent in the model, the export price will also rise by 1 percent. Exchange Rates In MFMOD different economies follow different exchange rate regimes: pegged (where the value of the currency is set to follow directly the exchange rate of another country); floating (where the exchange rate is determined by market forces); and mixed regimes like managed float regimes; and crawling peg regimes. In countries where the exchange rate is (at least partially) determined by market forces, the equilibrium exchange rate evolves according to the uncovered interest parity condition. In such a setting, capital will flow in (out) of the country to the extent that there the domestic interest rate exceeds (is less than) the foreign risk-free interest rate (the US rate in our setting). The change in the exchange rate is written as: (21) (1 + ) ( ) = + (−1 ) + � � (1 + ) In economies with a floating exchange rate, differences between domestic and foreign inflation will transmit to the exchange rate via the monetary policy reaction (either a Taylor rule or a money supply rule, where interest rates become a behavioral equation reflecting the estimated parameters of money demand). In economies with fixed regimes the exchange rate is linked directly to the appropriate base currency. In this context, the UIP condition will become the monetary policy rule (instead of the Taylor rule described infra) as the central bank is now constrained to adjust its main policy rate according to changes in the world interest rate to ensure that the peg holds (independence of the central bank is foregone). Depending on the degree of management in a mixed regime’s float, the currency may be either exogenous or endogenous following an estimated relationship that approximates the rule in place. Real and Nominal Effective Exchange Rates Real and nominal effective exchange rates (REER and NEER) are calculated using the trade- weighted average of each countries’ bilateral-exchange rates (deflated by the consumer price deflator in the case of REERs). Trade weights are given by the average of the share of partner countries in the bilateral import and export matrices using weights calculated from COMTRADE data as described above for the export market variable. 16 ,   =   100 � ≠   , =   100� ≠ where ω ji : country I’s share in country j ’s total imports and the real exchange rate for each country is defined as taking each nominal exchange rates and deflating by the consumer price deflator of each country: = �  =      Labor Market Labor Supply Labor supply is determined by the participation rate ( ), which is assumed to follow a simple autoregressive process, trending toward the structural participation rate discussed earlier in the section on supply-side equilibrium: ∗ = + (1 − ) −1 . 6 Employment In the short-run the level of employment is determined by short-run changes in demand and supply of labor, where equilibrium labor demand is determined by producer real wages adjusted for productivity growth. The change in the output gap measures the degree to which temporary or short-term workers fill in excess demand by firms to manage production over the economic cycle. Finally, employment growth is corrected for steady-state growth of the labor-force. ∗ ] ∗ ∗ (22) △ = 1 [−1 − −1 + 2 △ + 2 �△ log � � −△ log � ∗ � � +△ 6 This assumes that labor supply is inelastic to wages net of taxes. Alternatively, one may write out the marginal rate of substitution between leisure and consumption which renders the labor supply condition. The labor supply elasticity in this case would be at the extensive margin. The marginal rate of substitution and the marginal product of labor then jointly determine the non-accelerating wage rate of unemployment. 17 Wages In the model wages adjust to equilibrate supply and demand in the labor market. In the long-run, when the unemployment rate is equal to its structural rate, nominal wages will grow in line with the marginal product of labor. In the short-run, some persistence is allowed in wage growth (in a similar vein to consumer prices) with the estimated pass-through of consumer prices and productivity gains into nominal wages typically being less than one. This pass-through is determined by the parameter . In the short-run, when unemployment is higher (lower) than its structural rate, the growth in wages will fall (rise) by the parameter 1 . (23) ⎛ ∗ ⎞ ∗ ∗ ∆ = + ⎜−1 − � � � � −1� − −1 ⎟ + ∆−1 + (1 − )(Δ + (Δ − Δ )) ��������� ⎝ ⎠ ∗) + 1 ( − ∗ Where ∗ ∗ �. = log � Fiscal Accounts The level of disaggregation of the fiscal accounts in the model is based on the availability of data and is generally based on GFS 2014 definitions. Some of the fiscal data are available only for a relatively brief period (2010 to 2017), which makes econometric identification of parameters difficult. In some of these cases, coefficients drawn from the literature or expert judgment were imposed. In countries where the national authorities follow a different accounting scheme and where data are available, they are used. The theoretical basis for the modeling of government spending is less well established than for firm and household behavior. Fiscal policy theory provides for many kinds of rules to govern an optimal fiscal policy, but seldom are these rules observed in practice. Estimating equations that describe how fiscal spending and revenues react to economic events is particularly difficult as the legislative basis for those reactions is constantly changing. While such “regime changes” if unidentified would bias parameter estimates, in many cases the changes in rules often are induced by changes in economic performance and/or fiscal conditions and may therefore be themselves endogenous. 18 The equations (reaction functions) estimated in the model and discussed below are not prescriptive, rather they reflect the behavior of government spending on average over the estimation period. As such they must be used with care, especially in cases where important regime change has occurred. Often, when analyzing fiscal policy or alternative rules, the analyst will want to exclude these equations (or re-specify them), for example by exogenizing spending variables when analyzing a spending proposal or altering revenue functions when new tax rates or rules are being contemplated. Government Expenditure Total government expenditure is divided between nominal government consumption of goods and non-factor services ( ), government compensation of employees ( ), acquisition of non-financial capital ( ), transfers ( ), other expenditure ( ) and cost of financing ( ).   =   + + + + + + (24) In the long-run, government expenditures on goods and services (which includes the wage bill) and in capital goods are assumed to grow at the same pace as nominal GDP, while in the short- term some persistence of expenditure is allowed (controlled by the parameter 3 ). For example, the equation for expenditure on goods and services is: ∆   =   1   +   2 �−1 − −1 �   +   3 ∆−1 + (1 − 3 )∆ +   (25) Government cost of finance is a simple function of the debt level and the effective interest rate paid on the debt. Going further, on a case by case basis, the potential pro- or counter-cyclicity of fiscal policy can be considered by also introducing the output gap into equation 27, although in general this is not done in the standard MFMod model. For countries that have domestic and foreign debt servicing variables, the cost of financing is then the sum of those variables. The historical cost of finance rate is computed as the effective interest paid on the debt, that is the ratio of current interest payments ( ) to previous period debt (−1 ), ie. = /−1 ∗ 100. On the forecasting horizon, the cost of financing is expressed as an identity equal to an interest rate differential (spread) ( ) on a base rate. For external debt, we use the United States Treasury Bill rate ( ) for the base rate. For domestic debt, the cost of financing assumes a spread relative to the domestic monetary or lending interest rate. 19   =   +     =  −1     (26) ∗ + ∗ (27)   =   −1 + −1   =     +  ,  (for external debt) (28b)   =     +  ,  (for domestic debt) (29c) Government Revenue Government revenues are divided (depending on data availability) between direct taxes ( ), taxes on goods and services ( ), social insurance contributions ( ), taxes on international trade ( ), grants ( ) and other revenue (ℎ ).   =   + + + + + ℎ (30) Direct tax revenues, social security contributions and indirect taxes are mapped to their relevant tax bases. In the case of indirect taxes, the tax base is assumed to be total consumption, for direct revenues (such as personal and corporate income tax) it would be nominal GDP or the wage bill depending on the nature of the tax; for social security contributions the tax base is the wage bill (average wage multiplied by employment); and the base for taxes on international trade is nominal imports. For each tax, an effective rate ( ) of tax is calculated, defined as revenue divided by the assumed base. Thus, for k = Drct, Idrct, Soc, Cust, we can express the evolution of fiscal revenues as:   = (31) In the projection period, the effective rate is held constant (approximating a business as usual scenario). As a result, tax revenues simply grow at the rate of the tax base (). However, the analyst can change the effective tax rate to simulate the impacts of alternative tax-policy scenarios. This tax rate can also be endogenized on a case by case basis to consider the potential pro- or counter-cyclicality of fiscal policy. For all other revenues, a simple assumption is made that they grow at the rate of nominal GDP. 20 In countries where the legal tax rate is known, tax revenues can be decomposed by taking account of the tax base ( ) multiplied by a coverage rate (), which summarizes information on rebates, avoidance, evasion and exemptions. The coverage rate will vary with the output gap to capture the role of automatic stabilizers: = 1 + 2 We can thus rewrite equation (31) as: = , = , (1 + 2 ) Fiscal Balances and Debt The fiscal block includes several key fiscal indicators, which are derived from modeled revenue and expenditure variables and simple identities. The overall government balance ( ) is an identity derived by subtracting total expenditure from total revenues.   =     −   (32) The total financing requirement (FRT) is equal to the negative of the overall balance, which may be met through a combination of local and foreign bonds and / or reduction in gross assets. = − (33) The government may then decide how much new debt to issue locally ( ) and abroad ( ): = + (34) Domestic ( ) and foreign debt ( ) are then equal to the previous stock of debt plus the new financing: = −1 + (35) = −1 + (36) In addition, the stock of debt may change due to debt revaluation and other discrepancies, that arise either because of “below the line expenditures” or liabilities that are realized (such as 21 contingent liabilities). As such, an additional ‘revaluation’ term is included to account for this. Historically it is calculated as: = − −1   + (37) In the forecast period the revaluation term is exogenous. The primary balance ( ) is an identity derived by subtracting total government expenditure net of interest payments from total revenues. )   =     −   (   −   (38) Balance of Payments Current Account The current account of the balance of payments in the model is expressed in US dollar terms (CD). The current account balance ( ) is the sum of: i) the trade balance ( ); and ii) the balance for factor service incomes and transfers (remittances, government interest payments and all other non-trade flows) ( ).   =     +   (39) The USD value of exports and imports of goods and services as recorded in the balance of payments are determined in the model entirely by the value of import and exports (volumes times prices) from the national income accounts expressed in USD (multiplied by the exchange rate), after accounting for historical differences between the NIA and BOP measures of exports and imports. Historically, the same-currency value of exports (and imports) in the national accounts do not exactly equal the value of exports (and imports) in the balance of payments accounts due to differences in accounting methodologies. The model preserves these historical differences and keeps the ratio of BOP and NIA differences in the forecast period constant by forcing the growth rate of the nominal exports and imports of goods and services in the balance of payments to be identical to the exchange rate adjusted growth rate of nominal exports and imports calculated in the national income account block during the forecast and simulation period. , , , ,= ℎ (40) , = , ∗� , � ,= ℎ 22 , , , ,= ℎ (41) , = , ∗� , � ,= ℎ The balance for factor service incomes and transfers is the sum of remittances, government interest payments on external debt, and other balances. (42)   =    +  + The remittance balance is the difference between remittance inflows ( ) and outflows ( ).   =     −   (43) Remittance inflows ( ) are determined by income growth of the remitter. Nominal income growth in each remittance-sending country (j ) is weighted by the population of migrants from the modeled economy (country i ) living in country j as a share of the total number of emigrants from country i . Higher nominal income growth in countries where emigrants reside therefore leads to higher remittance inflows to the modeled economy, and vice versa. The calculation of the growth of potential remittance inflows is as follows:   =  � (44)  ≠  where ω ij is country j ’s share in the number of total emigrants from country i and Y tj KD : is the country j ’s nominal GDP in current dollars in time t and g X denotes the growth rate of variable X. The rate of growth of actual remittances may differ from the values that derive from the above accounting identity and therefore actual remittances are estimated in a simple equation relating actual remittances to the rate of growth this weighted average of incomes in remitting countries: △ log( )   =   1   +   2 △ log( )  (45) Changes in remittance outflow (BM REMT CD ) are treated more simply and assumed to grow in line with nominal GDP (in USD terms) of the home country. Government interest payments on external debt are determined in the fiscal section of the model (while being converted into USD for the balance of payments). Finally, the all “other balances” category includes factor services and aid flows and is modeled to grow in line with nominal GDP (in USD terms). 23 Capital and Financial Account The financial and capital account is comprised of the Capital Account Balance (BF KACC CD ), Foreign Direct Investment (BF FDI C D ), Portfolio (BF PF CD ) and Other Financial Account (BF OTH CD ) flows.   =   +   +     +   (46) In the current model, the capital account, FDI and Portfolio Investment are assumed to grow in line with nominal GDP (in USD). All Other Financial Investments are split into the amortization and disbursement of external debt of the private and public sectors. The net external financing of the government sector is determined in the fiscal sector while private financing grows in line with nominal GDP. The residual other financial investment also grows in line with nominal GDP. Overall Balance of Payments and Change in Reserve Assets The overall balance of payments position is calculated as follows:   =   ,   + , + (47) where refers to the net errors and omissions of the Balance of Payments. The external financing of the Balance of Payments position is through change in reserve assets. , = − (48) For some countries, the external financing of the BOP is further disaggregated into changes in reserve assets and the net use of IMF credits. Monetary Policy Monetary policy for economies that follow an inflation targeting regime is expressed via a Taylor rule that relates the central bank policy rate to inflation expectations, deviations from target, and output deviations from potential. This interest rate acts on the government bond interest rate, which consequently affects consumption and investment decisions. Thus, as prices operate above target, the central bank acts to stabilize inflation. Also, when the economy overheats, the central bank moves to cool it down: = −1 + (1 − )� + + 1 ( − ) + 2 � + (49) 24 Where is the central bank monetary policy rate, is a parameter that governs the persistence of monetary policy decisions, ( + ) is the natural real rate of interest plus the inflation target. In the model the inflation target and inflation expectations are assumed to be equal. Other domestic interest rates in the model, such as government interest payments on domestic debt and lending/borrowing rates that form part of the cost of capital calculation, are assumed to move in line with movements in the monetary policy rate. While changes in the two rates are assumed equal in the forecasting period, the levels may remain different due to spreads between the different rates (which are assumed exogenous and typically held constant throughout the forecast period). For non-inflation targeting countries the main monetary policy tool is money supply, where the interacting interest rates yield a money demand equation. The money demand ( ) equation in the model is standard, with money demand expressed as an increasing function of real incomes and a decreasing function of the interest rate: 7 − = − + + (50) Real money demand thus varies positively with domestic demand, negatively with current interest rates. An increase in aggregate demand shifts the money demand curve upwards reflecting higher borrowing. An increase in contemporaneous interest rates constitutes a market tightening policy and hence a fall in credit extension and a decline in money demand. The velocity of money is an identity and calculated as: = , where in equilibrium money supply equals money demand ( = ). Since money is used as an instrument, the interest rate becomes a reaction function instead, where an implicit interest rate response can be computed. The interest rate moves with the output gap and deviations of inflation growth from target money supply growth: 1 − = + + � ) + ( − + � is the target money supply growth rate and and are the behavioral parameters in Where equation (50). Production Accounts Real Sectoral Value-Added 7 McCallum (2012) argues that instead of stipulating a money demand equation one may want to close the model with a money rule: ∆ = + 1 −1 + 2 + 3 ∆ + . 25 In the absence of data on sectoral factor inputs and costs from which cost minimizing supply decisions could be modeled, forecasts in the production block are driven by the model’s demand side (expenditure accounts). In most countries in MFMOD, GDP at factor costs consists of agriculture ( ), industry ( ) and services ( , that might include a statistical discrepancy on the value-added side). In the model GDP at factor prices ( ) is determined by an identity as GDP at market prices ( ) less indirect taxes and subsidies ( ): = − (51) Net indirect taxes and subsidies are exogenous in the model, with the advice to hold them constant as a share of GDP in the forecast period. For the agricultural and industrial sectors, each sector grows in line with potential GDP in the long- run. In the short-run growth is driven as an empirically estimated function of growth in gross domestic expenditure and exports as follows. For = , : ∗ △ = 1 �−1 − −1 � + 2 △ ( + ) + (1 − 2 ) △ −1 (52) The service sector is modeled as a balancing identity to ensure that GDP measured at producer prices is equal to GDP at market prices minus net taxes on production. = − − (53) Sectoral Value-Added Deflators Price deflators for each production sector are modeled to converge to the overall production deflator. In the short-run, prices are mapped to the overall deflator with some persistence, governed by the parameter 2 . ∗ △ = 1 � −1 − −1 � + 2 △ −1 ∗ + (1 − 2 ) △ (54) Concluding Remarks The MFMod system is a macro-structural model of the global economy comprised of 184 separate country models. In general, the functional forms of individual equations in the model are comprised of two parts, a long-run equilibrium condition consistent with economic theory and a short-run component that is more idiosyncratic and reflective of real-world (as opposed to theory- world) dynamics. While functional forms are similar across countries, parameters are estimated at the country-level to reflect observed behavior. 26 The MFMod system is the product of many years of work by a large and ever-changing team of economists at the World Bank. As such, it is very difficult to ascribe intellectual ownership of the model to any one group of individuals. The authors of this paper represent the current generation of model authors, but the model itself reflects the work of many others. The MFMod model serves not only as the main forecasting tool at the World Bank, but also as the starting point of the many customized models the modeling team builds for both internal and external clients. Not only are these models informed by the thinking and the decisions that went into MFMod, but so too MFMod benefits from the innovations that go into these models. Like any model, MFMod is a living and changing entity. No model is ever finished. The documentation here describes the version of the model used in the forecasting round ending April 4, 2019. Areas for future work include: a more systematic evaluation of theoretical parameters and the basis for their imposition; improvements in the mechanism by which changes in sector- level value added are mapped back to the changes in the demand side; a more comprehensive accounting of income from capital and wealth in consumer demand; introduction of pro- and/or counter-cyclical behavior in government revenue and expenditure equations; and introduction of endogenous mechanisms by which the accumulation of domestic debt and external financing conditions influence domestic conditions, notably interest rates. 27 References Barro, R. J. (1974). Are government bonds net wealth? Journal of Political Economy, 82(6), pp. 1095-1117. Brayton, F., Laubach, T. and Reifschneider, D. (2014). The FRB/US Model: A tool for Macroeconomic Policy Analysis. FRB: Fed Notes: 2/6/2015. Blanchard, O. (2018). On the future of macroeconomic models, Oxford Review of Economic Policy, Volume 34, Issue 1-2, Spring-Summer, Pages 43–54 Burns, A., Janse Van Rensburg, T., Dybczak, K. and Bui, T. (2014). Estimating potential output in developing countries. Journal of Policy Modelling, 36(2014), pp. 700-716. Burns, A., and Jooste, C. (2019). Estimating and calibrating MFMod: A panel data approach to identifying the parameters of data poor countries in the World Bank’s structural macro model. Forthcoming. Clements, M.P. and Hendry, D. (1999). Forecasting non-stationary economic time series. The MIT press, Cambridge, MA. CBO. (2013). How CBO Analyzed the Macroeconomic Effects of the President's Budget. A CBO Paper. July 2013. Available at: https://www.cbo.gov/publication/42972. Constancio, Vitor. (2017). Developing Models for Policy Analysis in Central Banks. Speech. Downloaded Sept. 24, 2016 from http://www.ecb.europa.eu/press/key/date/2017/html/ecb.sp170925.en.html Daalsgard, T., André, C. and Richardson, P. (2001). Standard Shocks in the OECD Interlink Model. OECD Economics Department Working Papers. No. 306. OECD Publishing. Fair, R.C. (1992). The Cowles Commission Approach, Real Business Cycle Theories, and New Keynesian Economics. NBER Working Paper Series. No, 3990. Gali, J. Gertler, M. and Lopez-Salido, J.D. (2005). Markups, gaps, and the welfare costs of business fluctuations. Economics Working Papers 836, Department of Economics and Business, Universitat Pompeu Fabra Jorgensen, D. (1996). Investment – Vol. 2: Tax policy and the cost of capital. MIT Press. URL: http://mitpress.mit.edu/catalog/item/default.asp?sid=1C99FCC8-C26B-4CC6-BE21- CA1CDAFDE8AF\&ttype=2\&tid=8190 Mankiw NG, & D. Elmendorf D. (1999). Government Debt. In Handbook of Macroeconomics. North Holland. McCallum, B.T. (2012). The role of money in New-Keynesian models. Central Bank of Peru Working Paper No October 2012. Mikkelsen, J.G. (1998). A model for financial programming. IMF Working Paper WP/98/80. International Monetary Fund. Pagan, Adrian R. (2003) Report and Modelling and Forecasting at the Bank of England. Bank of England Quarterly Bulletin, Spring. 28 Pagan, A. and Wilcox, D. (2016) External Review- Reserve Bank of Australia Group Forecasts and Analysis. Richardson, P. (1998). The structure and simulation properties of OECD’s INTERLINK model, OECD Economic Studies, No. 10. Rudebusch, G. D. (2002). Macroeconomic Models for Monetary Policy. FRBSF Economic Letter, Number 2002-11, April 19. Roeger, W. (1995). Can imperfect competition explain the difference primal and dual productivity measures? Estimates for U.S. manufacturing. Journal of Political Economy, 103, pp. 316-330. Saxegaard M. (2017). The Use of Models in Finance Ministries – An Overview. Finansdepartementet, Arbeidsnotat 2017/1. Wickens, R.M. and Breusch, T.S. (1988). Dynamic specification, the long-run and the transformation of reformed regression models. The Economic Journal, 98, pp. 189-205. 29 Appendix A: Conversions and Identities This section presents additional equations used in the model for required conversions and balances. The equations presented here are used for presentation and calculation of output variables, using the behavioral and economic forecasting variables discussed above. Throughout, the following superscripts are used to denote currency and volume units. KD : constant US dollars, using a 2010 base year CD : current US dollars KN : constant local currency unit CN : current local currency unit. Balance of Payments Merchandise exports and imports in the balance of payments grow in line with the growth rate of overall goods and services trade allowing for a non-unity elasticity. △ log( )   =   △ log( ) (55) △ log( )   =   △ log( ) (56) Non-factor service exports and imports in the balance of payments are calculated as the residual when merchandise exports and imports are subtracted from total exports and imports.   =     −   (57)   =     −   (58) GDP in dollar term GDP, expressed in current US dollars, is calculated in terms of current local currency units divided by the US dollar exchange rate. (59)   =   � GDP expressed as constant dollars is calculated from GDP in constant local currency units, using a local currency price deflator (base year 2010) and the US exchange rate in 2010. (60)   =     Per capita GDP in constant dollar or local currency unit terms is calculated simply by dividing output in constant dollar or local currency unit terms by the population. (61)   =   � Appendix B: Data For the most part, variable mnemonics in MFMOD follow the protocols of the World Bank World Development Indicators database. 8 Typically, each variable is comprised of 14 characters: CCCAAMMMNNNNUC CCC is the three letter ISO mnemonic for the country – AUS for Australia. AA refers to the major accounting framework from which the variable is derived. i.e. NY – National income accounts NE – National expenditure accounts NV _Value added accounts BX – Balance of payments exports GG – General government fiscal accounts BF – Balance of Payments Financial Account MMM refers to the major category for that variable, for example GDP for gross domestic product or CON for consumption NNNN refers to the minor category that describes that variable, i.e. MKTP – market prices FCST -factor cost PRVT – private GOVT – government TOTL = total POTL – potential Etc. U indicates the unit of measure, K for volumes or constant prices C for values or current prices X for prices, interest rates or tax rates C indicates the currency in which the unit is expressed N – for local currency D – for dollars P – for PPPs 8 https://datahelpdesk.worldbank.org/knowledgebase/articles/201175-how-does-the-world-bank-code- its-indicators Appendix C: Detailed Variable List Mnemonics Variable Description BFBOPTOTLCD Overall Balance of Payments BFCAFCAPTCD Capital Account Balance, US$ mn BFCAFFFDICD Net Foreign Direct Investment, US$ mn BFCAFFINTCD Financial Account Balance, US$ mn BFCAFFINXCD Financial Account (excluding reserves) Balance, US$ mn BFCAFFPFDCD Net Portfolio Investment, Debt, US$ mn BFCAFFPFECD Net Portfolio Investment, Equity, US$ mn BFCAFFPFTCD Net Portfolio Investment, US$ mn BFCAFFSRCCD BOP Financing Sources BFCAFNEOMCD Net Errors and Omissions, US$ mn BFCAFOOTHCD Net Other Other Investment (Other Investment less Net Govt Ext. Borrowing) BFCAFOTHRCD Net Other Investment, US$ mn BFCAFRACGCD Change in Reserve Assets, US$ mn BFCAFTOTLCD Capital and Financial Account Balance, US$ mn BFFINAGOVCD External Govt Amortization BFFINAOTHCD Non-Government Amortization BFFINCABDCD Current Account Deficit BFFINDGOVCD External Govt Disbursements BFFINDOTHCD Non-Government Disbursement BFFINFGAPCD External Financing Gap BFFINIMFCCD IMF Credit (NET) BFFINMLTACD Medium and Long-term Debt Amortization (Excl. IMF) BFFINMLTAGCD Medium and Long-term Debt Amortization (Govt) BFFINMLTAPCD Medium and Long-term Debt Amortization (Private) BFFINMLTDCD Medium and Long-term Debt Disbursements (excl. IMF) BFFINMLTDGCD Medium and Long-term Debt Disbursements (Govt) BFFINMLTDPCD Medium and Long-term Debt Disbursements (Private) BFFINNEOMCD Net Errors and Omissions, US$ mn Gross External Borrowing BFFINNETDCD Net Foreign Financing, USD$ mn BFFINOTHRCD Net All Other BOP Financing Flows BFFINOTSRCD Other Short-term Capital Outflows BFFINREQMCD BOP Financing Requirement BFFINSRCSCD BOP Financing Source of Funds BFFINSRTACD Short-term Debt Amortization BFFINSRTDCD Short-term Debt Disbursements BFFINTGOVCD Net External Financing Govt (USD) BFFINTOTACD Short and Medium-LT Debt Amortization BFFINTOTDCD Short and Medium-LT Debt Disbursements BMFSTCABTCD Imp., Factor Services and Transfers (BOP), US$ mn BMFSTINTECD Imports, External Debt Interest Payments BMFSTOTHRCD Imp., Other Factor Services and Transfers (BOP), US$ mn BMFSTREMTCD Imp., Remittances (BOP), US$ mn BMGSRGNFSCD Imp., GNFS (BOP), US$ mn 32 BMGSRMRCHCD Imp., MRCH (BOP), US$ mn BMGSRNFSVCD Imp., NF SERV (BOP), US$ mn BMOTHCACD Imp., Other Current Account, US$ mn BNCABFLGPCD Current Account Balance (local definition - IMF), US$ mn BNCABFLGPCD_ Current Account Balance (local definition - IMF), % of GDP BNCABFLGPCN Current Account Balance (local definition - IMF), LCU BNCABFUNDCD Current Account Balance, US$ mn BNCABFUNDCD_ Current Account Balance (% of GDP) BNCABFUNDCN Current Account Balance, LCU BNCABLOCLCD Current Account Balance (local definition), US$ mn BNCABLOCLCD_ Current Account Balance (local definition), % of GDP BNCABLOCLCN Current Account Balance (local definition), LCU BNFSTCABTCD Net Exp., Factor Services and Transfers., GNFS (BOP), US$ mn BNFSTOTHRCD Net Exp., Other Factor Services and Transfers (BOP), US$ mn BNFSTREMTCD Net Exp., Remittances (BOP), US$ mn BNGSRGNFSCD Net Exp., GNFS (BOP), US$ mn BNGSRMRCHCD Net Exp., MRCH (BOP), US$ mn BNGSRNFSVCD Net Exp., NF SERV (BOP), US$ mn BNOTHCACD Net Exp., Other Current Account, US$ mn BXFSTCABTCD Exp., Factor Services and Transfers (BOP), US$ mn BXFSTOTHRCD Exp., Other Factor Services and Transfers (BOP), US$ mn BXFSTREMTCD Exp., Remittances (BOP), US$ mn BXGOILMRCHCD Exp., Oil Exports (BOP), US$ mn BXGSRGNFSCD Exp., GNFS (BOP), US$ mn BXGSRMRCHCD Exp., MRCH (BOP), US$ mn BXGSRNFSVCD Exp., NF SERV (BOP), US$ mn BXNOILMRCHCD Exp., Non-Oil Exports (BOP), US$ mn BXOTHCACD Exp., Other Current Account, US$ mn EXTLSHAREXN Share of New Debt that is Externally Issued FIRESTOTLCD Reserve Assets, US$ mn FMLBLBASECN Base Money FMLBLCRNGCN Credit to Non-Government FMLBLMTWOCN Nominal Base Money (Mil. LCU) FMLBLMTWOKN Real Base Money (Mil. Real LCU) FMLBLPOLYXN Key Policy Interest Rate FPCPITOTLXN Consumer Price Index GDPFCSTCAPCD GDP Factor Cost per capita GDPPCKD GDP per capita, 2000 US$ mn GDPPCKN GDP per capita, 2005 LCU mn GGBALCYCLCD General Government Revenue, Cyclical Balance USD GGBALCYCLCN General government cyclical balance (millions of LCU) GGBALEXGRCN Government Balance, excl. grants GGBALOTHRCN General Government Revenue, Additional Financing Requirements GGBALOVRLCD General Government Revenue, Deficit, US$ mn GGBALOVRLCD_ General Government Revenue, Overall Balance (% of GDP, USD) 33 GGBALOVRLCN General Government Revenue, Deficit, LCU mn GGBALPREXCN General Government Revenue, Primary Balance, excl grants GGBALPRIMCD General Government Revenue, Primary Balance, USD GGBALPRIMCN General Government primary balance, millions of local curency units GGBALSTRLCD General Government Revenue, Structural Balance (USD) GGBALSTRLCN General government structural balance (millions of LCU) GGDBTDOMTCN General government gross domestic debt millions lcu GGDBTEXTLCD General government gross debt external debt millions USD GGDBTEXTLCN General government gross debt external debt millions lcu GGDBTTOTLCD General Government Gross Debt (USD) GGDBTTOTLCN General government gross debt millions lcu GGDBTVALDCN Government debt revaluation (LCU), domestic GGDBTVALECN Government debt revaluation (LCU), external GGDBTVALTCN Government debt revaluation (LCU) GGEXPCAPTCN General government expenditure on capital expenditure (millions lcu) GGEXPCRNTCN General government Current Expenditures (millions lcu) GGEXPCYCLCN Cyclical component of general government expenditures (millions LCU) GGEXPGNFSCN General government expenditure on goods and services (millions lcu) GGEXPINTDCN General government interest payments on domestic public debt (millions lcu) GGEXPINTECD General government interest payments on external public debt (millions USD) GGEXPINTECN General government interest payments on external public debt (millions lcu) GGEXPINTPCN General government interest payments on public debt (millions lcu) GGEXPKINVCN General Government Capital Investments (millions LCU) GGEXPKTRNCN General Government Capital Transfers (millions LCU) GGEXPOTHRCN General government expenditures, other (includes transfers), (millions LCU) GGEXPSTRLCN Structurally adjusted general government expenditures (millions LCU) GGEXPTOTHCN General Government Expenditure, All Other Current Transfers GGEXPTOTLCN General government total expenditure (millions lcu) GGEXPTOTSCN General Government Expenditure, Other Social Transfers (millions LCU) GGEXPTPNSCN General Government Expenditure, Pensions (millions LCU) GGEXPTRNSCN General Government Expenditure, Current Transfers (millions LCU) GGEXPTSOCCN General Government Expenditure, Social Assistance (millions LCU) GGEXPWAGECN General government expenditure on compensation expenditure (millions lcu) GGFINDOMTCN General Government Net Domestic Financing (millions LCU) GGFINDOTHCN General Government, Other Domestic Financing (millions LCU) GGFINDPRVCN General Government, Domestic Financing from Privatization (millions LCU) GGFINEAMTCN General Government, Amortization External (millions LCU) GGFINEAMTXN General Government, Amortization External Rate GGFINEDSBCN General Governent, Disbursement External (millions LCU) GGFINEOTHCN General Government Revenue, Other External Financing (millions LCU) GGFINEXTLCN General Government Revenue, Net External Financing (millions LCU) GGFINFGAPCN Government Financing Gap (millions LCU) GGFINREQMCN Government Financing Requirement (millions LCU) GGFINTOTLCN Total Financing (External and Domestic) (millions LCU) GGREVCOMMCN General Government Revenue, Commodity Related Revenues (millions LCU) 34 GGREVCOMMXN Commodity Revenue Tax Rate GGREVCYCLCN Cyclical component of General government revenues (millions LCU) GGREVDRCTCN General government revenues, direct taxes (millions LCU) GGREVDRCTXN Direct Revenue Tax Rate GGREVGNFSCN Taxes on Goods and Services (millions LCU) GGREVGNFSXN Goods and services Tax Rate GGREVGRNTCN General Government Revenue, Grants, LCU (millions LCU) GGREVIDRTCN General government revenues, indirect taxes (millions LCU) GGREVIDRTXN Implied Indirect Average Tax Rate GGREVNONTCN General Government Revenue, Non-Tax Revenues (millions LCU) GGREVNONTXN Non-Tax Revenue Rate GGREVOTHRCN General government revenues, other (includes privatization), (millions LCU) GGREVOTHRXN Other Revenue Rate GGREVSSOCCN General Government Revenue, Social Security Contributions (millions LCU) GGREVSSOCXN Social Security Contributions Tax Rate GGREVSTRLCN Structurally adjusted General government revenues (millions LCU) GGREVTAXTCN General Government Revenue, Total Tax Revenue (millions LCU) GGREVTAXTXN Tax Revenue Rate GGREVTOTHCN General Government Revenue, Other Taxes (millions LCU) GGREVTOTHXN General Government Revenue, Other Taxes Rate GGREVTOTLCN General government revenues, total (millions lcu) GGREVTOTLXN Implied Average Tax Rate GGREVTRDECN General Government Revenue, Trade Traxes (millions LCU) GGREVTRDEXN Trade Taxes Rate INTR Implicit interest rate in government debt (percentagte points) INTRDDIFF Domestic Interest Rate Spread Over Policy Rate INTRDIFF Differentital between local and US implicit interest rate INTRDXN Implicit interest rate in government domestic debt (percentagte points) INTREDIFF External Interest Rate Spread Over Policy Rate INTREXN Implicit interest rate in government external debt (percentagte points) INTRXN Interest Rate (Implicit) LMEMPSTRLCN Structural Employment LMEMPTOTLCN Employment LMLBFTOTLCN Labor Force (15+) LMPRTTOTLCN Participation Rate LMUNRSTRLCN Unemployment Rate, Structural LMUNRTOTLCN Unemployment Rate NECONGOVTCD Govt. Cons., US$ mn NECONGOVTCN Govt. Cons., LCU mn NECONGOVTKD Govt. Cons., 2000 USD mn NECONGOVTKN Govt. Cons., 2000 LCU mn NECONGOVTXD Implicit USD defl., Govt. Cons., 2000 = 1 NECONGOVTXN Implicit LCU defl., Govt. Cons., 2000 = 1 NECONPRVTCD Pvt. Cons., US$ mn NECONPRVTCN Pvt. Cons., LCU mn 35 NECONPRVTKD Pvt. Cons., 2000 USD mn NECONPRVTKN Pvt. Cons., 2000 LCU mn NECONPRVTXD Implicit USD defl., Pvt. Cons., 2000 = 1 NECONPRVTXN Implicit LCU defl., Pvt. Cons., 2000 = 1 NEER Nominal Exchange Rate (Trade Weighted) NEEXPGNFSCD Exp., GNFS (NIA), US$ mn NEEXPGNFSCN Exp., GNFS (NIA), millions LCU NEEXPGNFSKD Exp., GNFS (NIA), 2000 US$ mn NEEXPGNFSKN Exp., GNFS (NIA), 2000 LCU mn NEEXPGNFSXD Exp., GNFS (NIA), US$ Deflator (2010=1) NEEXPGNFSXN Exp., GNFS (NIA), LCU Price defl. 2000 = 1 NEEXPGOILCN Exp., Oil (NIA), millions LCU NEEXPGOILKN Exp., Oil (NIA), 2000 LCU mn NEEXPGOILXN Exp., Oil (NIA), LCU Price defl. 2000 = 1 NEEXPNOILCN Exp., Non-Oil (NIA), millions LCU NEEXPNOILKN Exp., Non-Oil (NIA), 2000 LCU mn NEEXPNOILXN Exp., Non-Oil (NIA), 2000 Deflator NEGDETOTTKD Gross Domestic Expenditure (GDE), USD NEGDETOTTKN Gross Domestic Expenditure (GDE), LCU NEGDETTOTCN Total demand, current LCU NEGDETTOTKN Total demand, constant LCU NEGDIFGOVCN Public Fixed Domestic Inv., LCU mn NEGDIFGOVKN Public Fixed Domestic Inv., LCU constant price NEGDIFGOVXN Public Fixed Domestic Inv., deflator NEGDIFPRVCN Private Fixed Domestic Inv., LCU mn NEGDIFPRVKN Private Fixed Domestic Inv., LCU constant price NEGDIFPRVXN Private Fixed Domestic Inv., deflator NEGDIFTOTCD Fixed Domestic Inv., US$ mn NEGDIFTOTCN Fixed Domestic Inv., LCU mn NEGDIFTOTKD Fixed Domestic Inv., 2000 USD mn NEGDIFTOTKN Fixed Domestic Inv., 2000 LCU mn NEGDIFTOTXD Fixed Domestic Inv., USD Price defl., 2000 = 1 NEGDIFTOTXN Fixed Domestic Inv., LCU Price defl., 2000 = 1 NEGDIKGOVKN Capital stock, Government LCU NEGDIKPRVKN Capital stock, Private LCU NEGDIKSTKKD Capital stock, USD NEGDIKSTKKN Capital stock, LCU NEGDISTKBCD Change in stock, USD mn NEGDISTKBCN Change in stock, LCU mn NEGDISTKBKD Change in stock, 2000 USD mn NEGDISTKBKN Change in stock, 2000 LCU mn NEGDISTKBXN Change in stock, LCU Price defl., 2000 = 1 NEGDITOTLCD Gross Domestic Inv., USD mn NEGDITOTLCN Gross Domestic Inv., LCU mn NEGDITOTLKD Gross Domestic Inv., 2000 USD mn 36 NEGDITOTLKN Gross Domestic Inv., 2000 LCU mn NEGDITOTLXN Gross Domestic Inv., LCU Price defl., 2000 = 1 NEIMPGNFSCD Imp., GNFS (NIA), USD mn NEIMPGNFSCN Imp., GNFS (NIA), LCU mn NEIMPGNFSKD Imp., GNFS (NIA), 2000 US$ mn NEIMPGNFSKN Imp., GNFS (NIA), 2000 LCU mn NEIMPGNFSSCCN Inter SACU imports NEIMPGNFSXD Imp., GNFS (NIA), USD Price defl. 2000 = 1 NEIMPGNFSXN Imp., GNFS (NIA), LCU Price defl. 2000 = 1 NEKRTTOTLCN Rental Rate of Capital (Derived) NEWRTTOTLCN Implicit Wage Rate (Total Wage Bill / Employment) NVAGRTOTLCN Value Added Agriculture Local Currency units Values NVAGRTOTLKN Value Added Agriculture Local Currency units Volumes National base year NVAGRTOTLXN Value Added Agriculture Local Currency units Implicit Price deflator NVGDPNOILCN Non-Oil GDP, Local Currency NVGDPNOILKN Non-Oil GDP, Local Currency Volumes NVGDPNOILXN Non-Oil GDP, Deflator NVINDNOILCN Value Added Non-Oil Local Currency units Values National base year NVINDNOILKN Value Added Non-Oil Local Currency units Volumes National base year NVINDNOILXN Value Added Non-Oil Local Currency units Deflator NVINDTOTLCN Value Added Industry Local Currency units Values NVINDTOTLKN Value Added Industry Local Currency units Volumes National base year NVINDTOTLXN Value Added Industry Local Currency units Implicit Price deflator NVOILCONSKN Domestic Oil Consumption NVOILPRODKN Oil Production NVOILTOTLCN Value Added Oil Local Currency units Values National base year NVOILTOTLKN Value Added Oil Local Currency units Volumes National base year NVOILTOTLXN Value Added Oil Local Currency units Deflator NVSRVTOTLCN Value Added Services Local Currency units Values NVSRVTOTLKN Value Added Services Local Currency units Volumes National base year NVSRVTOTLXN Value Added Services Local Currency units Implicit Price deflator NYGDPDISCCD GDP Disc., USD mn NYGDPDISCCN GDP Disc., LCU mn NYGDPDISCKD GDP Disc., 2000 USD mn NYGDPDISCKN GDP Disc., 2000 LCU mn NYGDPFCSTCD GDP Factor Cost USD NYGDPFCSTCN GDP Factor Cost Local Currency units Values NYGDPFCSTKN GDP Factor Cost Local Currency units Volumes National base year NYGDPFCSTXN GDP Factor Cost Local Currency units Implicit Price deflator NYGDPGAP_ Output Gap (% of Potential GDP) NYGDPMKTPCD GDP, Market Prices, US$ mn NYGDPMKTPCN GDP, Market Prices, LCU mn NYGDPMKTPCP Model Param. NYGDPMKTPCP NYGDPMKTPKD GDP, Market Prices, 2000 US$ mn NYGDPMKTPKN GDP, Market Prices, 2000 LCU mn 37 NYGDPMKTPKP Model Param. NYGDPMKTPKP NYGDPMKTPXD GDP, Marker Prices, USD Price defl., 2000 = 1 NYGDPMKTPXN GDP, Marker Prices, LCU Price defl., 2000 = 1 NYGDPPOTLKD Potential Output, 2000 US$ mn NYGDPPOTLKN Potential Output, constant LCU NYGDPTFP Total factor productivity NYTAXNINDCN Net Indirect Taxes Local Currency units Values NYTAXNINDKN Net Indirect Taxes Local Currency units Volumes National base year NYTAXNINDXN Net Indirect Taxes Local Currency units Implicit Price deflator NYYWBTOTLCN Total Wage Bill NYYWBTOTLCN_ Labor Share of Income PANEUATLS Official exchange rate (LCU per EURO, period avg) PANEUATLSK Official exchange rate, Real (LCU per EURO, period avg) PANUSATLS Exchange rate LCU / US$ PANUSATLSK Official exchange rate, Real (LCU per US$, period avg) PMKEY Keyfitz Price Imports PSTAR Marginal Cost of Production PXKEY Keyfitz Price Exports PXKEYEX Keyfitz Export Price Excl. Oil REER Real Exchange Rate (Trade Weighted) REMT_IN Remittances Inflow Indicator REMT_OUT Remittances Outflow Indicator SPPOP1564TO Labor force (15-64) SPPOPTOTL Population (mn) SPPOPWORK Population 15+ TOT Terms of Trade VELOCITY Velocity of Money WLDFALUMINUM Price of aluminum (USD) WLDFBANANA_US Price of banana (USD) WLDFBEEF Price of beef (USD) WLDFCOAL_AUS Price of coal (USD) WLDFCOCOA Price of cocoa (USD) WLDFCOFFEE_COMPO Price of coffee (USD) WLDFCOPPER Price of copper (USD) WLDFCOTTON_A_INDX Price of cotton (USD) WLDFCRUDE_PETRO Price of oil (USD) WLDFGOLD Price of gold (USD) WLDFGRNUT_OIL Price of groundnut (USD) WLDFIRON_ORE Price of iron ore (USD) WLDFISTL_JP_INDX Price of steel (USD) WLDFLEAD Price of lead (USD) WLDFLOGS_MYS Price of logs (USD) WLDFMAIZE Mai price of ze (USD) WLDFNGAS_EUR Price of natural gas (USD) WLDFNICKEL Price of nickel (USD) 38 WLDFORANGE Price of orange (USD) WLDFPALM_OIL Price of palm oil (USD) WLDFPLYWOOD Price of plywood (USD) WLDFRICE_05 Price of rice (USD) WLDFRUBBER1_MYSG Price of rubber (USD) WLDFSAWNWD_MYS Price of sawnwood (USD) WLDFSILVER Price of silver (USD) WLDFSORGHUM Price of sorghum (USD) WLDFSOYBEAN_MEAL Price of soybean (meal) (USD) WLDFSOYBEAN_OIL Price of soybean (oil) (USD) WLDFSOYBEANS Price of soybeans (USD) WLDFSUGAR_WLD Price of sugar (USD) WLDFTEA_AVG Price of tea (USD) WLDFTIN Price of tin (USD) WLDFTOBAC_US Price of tobacco (USD) WLDFWHEAT_US_HRW Price of wheat (USD) WLDFWOODPULP Price of woodpulp (USD) WLDFZINC Price of zinc (USD) MUV Price of manufacturing goods (USD) XMKT Weighted Trading Partner Demand 39