A SEQUENTIAL DYNAMIC CGE MODEL FOR POVERTY ANALYSIS Nabil Annabi*, John Cockburn* and Bernard Decaluwé* Preliminary Draft1 (May, 2004) * CIRPEE, Laval University, http://www.cirpee.org 1 This draft was prepared for the Advanced MPIA Training Workshop in Dakar, Senegal, June 10-14, 2004. For more information see the PEP Network website http://www.pep-net.org Annabi, N. Cockburn, J and Decaluwé, B. (2004) 1. Introduction Much current debate focuses on the role of growth in alleviating poverty. However, the majority of CGE models used in poverty and inequality analysis are static in nature. The inability of this kind of model to account for growth effects makes them inadequate for medium run analysis of the poverty impacts of economic policies. They exclude accumulation effects and do not allow the study of the transition path of the economy where short run policy impacts are likely to be different from those of the medium run2. To overcome this limitation we propose a simple dynamic CGE model for poverty analysis. Dynamic general equilibrium models can be classified as truly dynamic (“intertemporal”) or sequential dynamic (“recursive”) models. Truly dynamic models are based on optimal growth theory where the behaviour of economic agents is characterized by perfect foresight. They know all about the future and react to future changes in prices. Households maximize their intertemporal utility function under a wealth constraint to determine their consumption schedule over time. Investment decisions by firms are the result of cash flow maximization over the whole time horizon. However, the application of this kind of model to the analysis of poverty and inequality is not straightforward and remains in our research agenda for the future. In this document, we develop a sequential dynamic CGE model for poverty analysis. This kind of dynamics is not the result of intertemporal optimization by economic agents. Instead, these agents have myopic behaviour. A sequential dynamic model is basically a series of static CGE models that are linked between periods by an exogenous and endogenous variable updating procedure. Capital stock is updated endogenously with a capital accumulation equation and population (total labour supply) is updated exogenously between periods. It is also possible to add updating mechanisms for other variables such as public expenditure, transfers, technological change or debt accumulation. But the crucial questions to deal with in this kind of model concern the distribution of new investments between sectors and the capital stock calibration. These will be discussed below. This document is organized as follows. We begin by a short overview of the static version of the model in section 2 and present its extension to include sequential dynamics in section 3. Sections 4 and 5 describe briefly the calibration and the resolution of the model. Finally, we present a complete list of the equations and the GAMS codes in Appendix I and II. 2 Annabi (2003) uses an intertemporal model to study the effects of trade liberalization in Tunisia. Simulation results show that the static version of the model underestimates trade liberalization’s impacts on production and welfare, since it excludes accumulation effects. 2 Annabi, N. Cockburn, J and Decaluwé, B. (2004) 2. Static module The static part of the model is based on the EXTER+ model3. As the main objective of the present document is to introduce sequential dynamics we present only a brief description of production, consumption, international trade and government behaviour. 2.1 Production We adopt a multi-stage production function. In a first stage, sectoral output is a Leontief function of value added and total intermediate consumption. Value added is in turn represented by a CES function of labour and capital in the non-agricultural sectors (industry and services), and a CES function of land and a composite factor in agriculture. The latter is also represented by a CES function of primary factors: agricultural capital and labour. Value added in the public sector is generated by labour alone. In the different production activities we assume that a representative firm remunerates factors of production and pays dividends to households. 2.2 Consumption Household demand is represented by a linear expenditure system (LES) derived from the maximization of a Stone–Geary utility function. The model includes four household categories: rural poor, urban poor, rural rich and urban rich. Minimal consumption levels are calibrated using estimates of the income elasticity and the Frisch parameters. 2.3 Foreign Trade We assume that foreign and domestic goods are imperfect substitutes. This geographical differentiation is introduced by the standard Armington assumption with a constant elasticity of substitution function (CES) between imports and domestic goods. On the supply side, producers make an optimal distribution of their production between exports and domestic sales according to a constant elasticity of transformation (CET) relation. 2.4 Government The government receives direct tax revenue from households and firms and indirect tax revenue on domestic, imported and exported goods. Its expenditure is allocated between the consumption of goods and services, public wages and transfers. The model accounts for indirect or direct tax compensation in the case of a tariff cut. 3 For a general presentation of EXTER+ model see Cockburn et al., (2004). 3 Annabi, N. Cockburn, J and Decaluwé, B. (2004) 2.5 Equilibrium General equilibrium is defined by the equality (in each period) between supply and demand of goods and factors and the investment-saving identity. In a future version, the model will be extended to take into account unemployment and endogenous labour supply. Finally, the nominal exchange rate is the numéraire. 3. Dynamic module: EXTER+SD model The EXTER+SD (Sequential Dynamic) model is formulated as a static model, which is solved sequentially over time. In every period the capital stock is updated with a capital accumulation equation. Total labour supply increases at the same rate as exogenous population growth. We also account for debt accumulation. We use the static model, index all variables in time (t) and introduce the equations presented in Table 14: Table 1: Additional equations for the sequential dynamic model 1. 2. KDit +1 = (1 − δ ) KDit + Indit Capital Accumulation LSt +1 = ( 1 + ng ) ⋅ LSt Labour force growth 2 3. 4. ⎛r ⎞ ⎛r ⎞ Ind it = γ 1i ⎜ it ⎟ + γ 2i ⎜ it ⎟ KDit ⎝ Ut ⎠ ⎝ Ut ⎠ U t = Pinvt ⋅ ( ir + δ ) min tr , h , t +1 Investment demand Capital user cost = ( 1 + ng ) ⋅ C 5. C 6. ITt = Pinvt ⋅ ∑ Indit min tr , h ,t LES minimal consumption Investment equilibrium i where, KDit Indit LSt SGt CABt rit PCit : Capital stock : Investment by destination : Total labour supply : Government savings : Foreign savings : Return to capital : Consumption price DGt DFt δ ng ir : Public debt µi ITt : Share parameter (free parameter) : Foreign debt : Capital depreciation rate (free parameter) : Population growth rate (free parameter) : Interest rate (free parameter) : Nominal total investment 3.1 Capital accumulation Equation 1 describes the law of motion for the sectoral capital stock. It supposes implicitly that the stocks are measured at the beginning of the period and that the flows are measured at the end 4 See Appendix I for the complete list of equations. 4 Annabi, N. Cockburn, J and Decaluwé, B. (2004) of the period. We note that in some cases the law of motion can be applied to the total capital stock KSt , rather than at the sectoral level. 3.2 Growth in labour supply With the introduction of equation 2, total labour supply LSt becomes an endogenous variable, although it simply increases at the exogenous rate ng , which is simultaneously the population growth rate and the labour force growth rate. Note that the minimal level of consumption within the LES function also increases at the same rate ng (equation 5). 3.3 Investment demand As we mentioned in the introduction the question is how will new investment be distributed between different sectors? This can be done either through a distribution function or an investment demand function. Note that the investment here is not the investment by origin (product) that we have already in the static model, but it is investment by sector of destination. The investment demand function we use here is similar to the one proposed by Bourguignon, Branson and De Melo (1989). In equation 3, γ1i and γ 2i are positive parameters calibrated on the basis of the investment elasticity and the investment equilibrium equation (equation 8). The investment rate is increasing with respect to the ratio of the rate of return to capital and its user cost. In introducing investment by destination, we respect the equality condition with total investment by origin in the original social accounting matrix (equation 6). Given the investment demand equation, 2 ⎛r ⎞ ⎛r ⎞ Ind it = γ 1i ⎜ it ⎟ + γ 2i ⎜ it ⎟ KDit ⎝ Ut ⎠ ⎝ Ut ⎠ the elasticity of the investment rate with respect to the ratio of the rate of return of capital to its user cost is: d ⎛⎜ ⎝ ⎞ ⎛ rit ⎞ ⎜ U ⎟ KDit ⎟⎠ t⎠ ⋅ ⎝ =ν rit ⎞ Indit ⎛ ⎛ ⎞ d⎜ ⎟ ⎜ KDit ⎟⎠ ⎝ Ut ⎠ ⎝ Indit which implies that: 5 Annabi, N. Cockburn, J and Decaluwé, B. (2004) γ 2i 2 −ν = ν − 1 γ ⎛ rit ⎞ 1i ⎜ U ⎟ t⎠ ⎝ The parameters γ1i and γ 2i are therefore positive only if the value of the elasticity of Indit with KDit rit is between 1 and 2. If ν = 1.5 we obtain the following relation: Ut respect γ 2i = γ 1i rit Ut this relation, taken together with the investment demand equation, makes it possible to calibrate γ 1i and γ 2i . In the literature on dynamic CGE models, we note some alternative assumptions for the investment demand function or the distribution of total investment. Abbink, Braber and Cohen (1995), present an applied sequential dynamic CGE model for Indonesia. They determine the shares of investment by destination as follows: θi 0 ⎛⎜ PRit ⎞ ⎟ ⎝ APRt ⎠ θit = PR ∑θi 0 ⋅ ⎛⎜⎝ APRitt ⎞⎟⎠ i where θi 0 , PRit and APRt are respectively the base run sectoral investment shares (which are assumed equal to the base year sectoral shares in total capital remuneration), sectoral profit rates and the average profit rate defined as follows: θi 0 = ri 0 KDi 0 ∑ ri 0 KDi 0 i rit ⋅ KDit − δ ⋅ KDit ⋅ Pinvt KDit ⋅ Pinvt KDit ⋅ Pinvt APRt = ∑ ⋅ PRit i ∑ j KD jt ⋅ Pinvt PRit = Finally, capital accumulation is written as: KDit +1 = (1 − δ ) KDit + θit ⋅ ITt 6 Annabi, N. Cockburn, J and Decaluwé, B. (2004) ITt where is the total volume of investment. The second element on the right hand side of this expression represents investment by sector of destination. Jung and Thorbecke (2003) use the following investment demand function: ⎛ ⎞ Indit KINCit = Ai ⋅ ⎜ ⎟ KDit ⎝ Pinvt ⋅ KDit ⋅ ir ⎠ βi with KINCit = rit KDit this equation could be simplified to: ⎛ ⎞ Indit rit = Ai ⋅ ⎜ ⎟ KDit ⎝ Pinvt ⋅ ir ⎠ Where Ai ,r and βi βi represent a scale parameter, the interest rate and the investment elasticity, the latter of which is fixed to 1. 3.4 Alternative version with debt accumulation An alternative or extended version of this sequential dynamic model can be obtained by adding the following equations for public and foreign debt: DGt +1 − DGt = irDGt − SGt DFt +1 − DFt = irDFt + CABt The variation in public debt is equal to the interest payments on this debt less public savings. In the same manner, the variation in foreign debt is equal to interest payments on this debt plus the current account balance (foreign savings): In this version, we assume that firms earn the interest paid on bonds issued by the government to finance its deficit. The government is the only agent who can borrow abroad and therefore they pay the interest on foreign debt, as well as on their public debt. In this second case, we obtain: Firm income: ( YFt = 1 − λ − λ ROW Public savings: ) ∑ r KD it it + irDGt i SGt = YGt − Gt − TGt − irDGt − irDFi 7 Annabi, N. Cockburn, J and Decaluwé, B. (2004) The current account balance: CABt = et ∑ PWM i M it + λ ROW ∑ rit KDit + DIV ROW − et ∑ PWEi EX it + irDFt i i i If initial debt is nil, the debt equations may be rewritten as: DGt +1 = DGt − SGt DFt +1 = DFt + CABt Debts are exogenous in the first period. Labour supply and capital demand are endogenous, except in the first period. It is interesting to note that, with the introduction of the debt equations, we can change the traditional closure rules of static CGE models. For example, we can fix the ratio of public or foreign debt with respect to GDP, instead of fixing the public or foreign deficits. 4. Calibration With the introduction of the investment demand and the accumulation equations we must adjust the calibration of the capital stock to make it consistent. For this purpose we make assumptions on the sectoral growth rates of capital, as data is not generally available on investment by destination in developing countries. Some possible assumptions include: • • The investment rate is equal to the production growth rate. The investment rate is equal to the sum of the population growth rate and the depreciation rate5. In our application we fixed the sectoral growth rates of capital at 5 percent in order to calibrate the capital stock6. We call this “method 1”. The second possible way of calibrating the capital stock (method 2) consists of making an assumption about the rate of return to capital and calibrating its stock from the operating surplus (OS) in the social accounting matrix. Table 3 presents the advantages and disadvantages of each method. 5 The latter assumption implies the steady state condition for capital accumulation, Indit Kdit = ( ng + δ ) that is used in intertemporal models. Note that in the sequential dynamic models it is not possible to normalize the return to capital otherwise one could obtain unrealistic growth trends. 6 8 Annabi, N. Cockburn, J and Decaluwé, B. (2004) Table 3: The capital stock calibration methods Advantages Method 1 (assumed sectoral growth rates) Method 2 (assumed sectoral rates of return) Disadvantages - Accumulation profile more homogenous. - If OSi are not well estimated, ri could differ substantially between sectors. - If OSi < 0 fl ri < 0: acceptable. - If Indi < 0 fl Ki < 0: not acceptable. - Base run Ri are equal (long run equilibrium). - If the OSi are not well estimated, Ki (and Indi/Ki) could differ substantially between sectors. - If Indi < 0 fl Ki > 0. - If the OSi < 0 fl Ki < 0: not acceptable. 5. Model Execution The model is solved simultaneously as a system of non linear equations. To illustrate, we simulate the impact of a total tariff cut in the industrial sector and present some results7. The values of the new (free) parameters of the model are presented in the following table: Table 2: Free parameters δ ir 0.025 ν ng 1.50 0.05 0.03 In static CGE models, counterfactual analysis is made with respect to the base run that is represented by the initial SAM. But in dynamic models the economy grows even without a policy shock and the analysis should be done with respect to the growth path in the absence of any shock. Note, in graph 1, that the economy grows even without a tariff cut. Graph 2 shows the evolution of the ratio of the industrial capital stock in the presence of the policy shock with respect to capital stock without the shock (“business as usual” or Bau). 7 This section will be developed in the future version of this document. 9 Annabi, N. Cockburn, J and Decaluwé, B. (2004) - Graph 1: Industrial capital stock with and without shock 45 Base run = 1 40 35 30 25 20 1 2 3 4 5 6 7 8 9 10 Years Shock Bau - Graph 2: Ratio of industrial capital stocks with and without shock 1.1 1.09 1.08 Base run = 1 1.07 1.06 1.05 1.04 1.03 1.02 1.01 1 1 2 3 4 5 6 7 8 9 10 Years IND 10 Annabi, N. Cockburn, J and Decaluwé, B. (2004) References Annabi N. (2003), Libéralisation Commerciale en Tunisie: Une analyse à l’Aide de Modèle d’Équilibre Général Calculable Dynamique. Ph.D. dissertation, University of Paris I (Sorbonne). Abbink G. A., Braber M. C. et Cohen S. I. (1995), A SAM-CGE demonstration model for Indonesia: A Static and Dynamic specifications and expirements. International Economic Journal, Vol 9 (3), p.15-33. Agénor, P-R. (2003), The Mini-Integrated Macroeconomic Model for Poverty Analysis. The World Bank. Ballard, C. L., Fullerton, D., Shoven, J. B. and Whalley, J. (1985), A General Equilibrium Model for Tax Policy Evaluation. The University of Chicago Press. Beghin, J., Dessus, S., Roland, H. D. and Van der Mensbrugghe, D. (1996), General Equilibrium Modelling of Trade and The Environment. OECD, Technical Paper No. 116. Bourguignon, F., Branson, W. H. and de Melo, J. (1989), Macroeconomic Adjustment and Income Distribution: A Macro-Micro Simulation Model. OECD, Technical Paper No.1. Cockburn, J., Decaluwé, B. and Robichaud, V. (2004), Trade Liberalization and Poverty: A CGE Analysis of the 1990s Experience, (forthcoming). Decaluwé, B., Martens, A. and Savard, L. (2001), La Politique économique du développement et les modèles d’équilibre général calculable. Les Presses de l’Université de Montréal. Dervis, K., de Melo, J. and Robinson S. (1982), General Equilibrium Models for Development Policy, Cambridge University Press. Jung, H.S. and Thorbecke, E.(2003) The Impact of Public Education Expenditure on Human Capital, Growth, and Poverty in Tanzania and Zambia: A General Equilibrium Approach. Journal of Policy Modeling, 25: 701–725 Van der Mensbrugghe, D. (2003), LINKAGE. Technical Reference Document, The World Bank. Schubert, K. (1993), Les modèles d’équilibre général calculable : une revue de la littérature. Revue d’Économie Politique, 103 (6) : 775-825. 11 Annabi, N. Cockburn, J and Decaluwé, B. (2004) Appendix I: Equations of EXTER+SD Model structure The subscript presenting the time periods is introduced only when necessary. Production 1. ⎡ CI j VA j ⎤ XS j = min ⎢ , ⎥ ⎣ io j v j ⎦ 4 ( ) −ρnag −ρnag ⎤ KL ⎡ KL KL α nag LDnag + 1 − α nag KDnag 2. VAnag = Anag KL ⎣ 3. VAAGR = AtrCL ⎡ αCL CF −ρCL ⎣ 4. KL ⎦ + ( 1 − αCL ) LAND −ρCL −ρagr ⎤ KL ⎡ KL KL α LD −ρagr + ( 1 − α agr CF = Aagr ) KDagr ⎣ agr agr ⎦ KL −1 KL −1 ⎤ ⎦ KL ρnag −1 ρCL KL ρagr 2 1 1 5. VAntr = LDntr 1 6. CI j = io j XS j 4 7. DI tr , j = aijtr , j CI j 8. ⎛ 1 − αCL ⎞ LAND = ⎜ CL ⎟ ⎝ α ⎠ 9. ⎛ αtrKL ⎞ LDtr = ⎜ KL ⎟ ⎝ 1 − αtr ⎠ 10. LDNTR = 12 σCL σtrKL ⎛ rc ⎞ ⎜ rl ⎟ ⎝ ⎠ ⎛ rtr ⎞ ⎜ w⎟ ⎝ ⎠ σCL CF 1 KDtr 3 σtrKL PNTR XS NTR − ∑ PDtr DI tr ,NTR tr w 1 Income and savings R L YH h = λW h ⋅ w∑ LD j + λ h ∑ rtr KDtr + λ h ⋅ rl ⋅ LAND j 11. tr 4 + PINDEX ⋅ TGh + DIVh 12. YDH h = YH h − DTH h 4 13. SH h = ν ⋅ ψ h ⋅ YDH h 4 14. YF = λ RF ∑ rtr KDtr + λ LF ⋅ rl ⋅ LAND 1 tr 12 Annabi, N. Cockburn, J and Decaluwé, B. (2004) 15. SF = YF − ∑ DIVh − e ⋅ DIV ROW − DTF 1 h 16. YG = ∑ TItr + ∑ TIEtr + ∑ TIM tr + ∑ DTH h + DTF 1 ∑ TGh 1 tr tr 17. SG = YG − G − PINDEX tr h h 18. TI tr = txtr ( Ptr XStr − PEtr EX tr ) + txtr ( 1 + tmtr ) e PWM tr M tr 3 19. TIM tr = tmtr e PWM tr M tr 3 20. TIEtr = tetr PEtr EX tr 3 21. DTH h = tyhhYH h 4 22. DTF = tyf ⋅ YF 1 Demand 23. CTH h = YDH h − SH h 4 ⎛ 24. PC tr C tr , h = PC tr C trMIN , h + γ tr , h ⎜ CTH h − ⎝ ∑ PC trj trj ⎞ C trjMIN ,h ⎟ ⎠ 12 25. G = XS ntr Pntr 1 26. INVtr = µtr IT PCtr 3 27. DITtr = ∑ DI 3 j j Prices Pj XS j − ∑ PCtr DI tr , j 28. PV j = 29. rnag = tr VAj PVnagVAnag − w LDnag KDnag rc ⋅ CF − w LDAGR KDAGR PVAGRVAAGR − rl ⋅ LAND 31. rc = CF 30. rAGR = 4 2 1 1 32. PDtr = ( 1 + txtr ) PLtr 3 33. PM tr = ( 1 + txtr ) ( 1 + tmtr ) e ⋅ PWM tr 3 e ⋅ PWEtr 1 + tetr 3 35. PCtr Qtr = PDtr Dtr + PM tr M tr 3 36. Ptr XStr = PLtr Dtr + PEtr EX tr 3 34. PEtr = 13 Annabi, N. Cockburn, J and Decaluwé, B. (2004) ∏ ⎛⎜⎝ 37. Pinv = PCtr ⎞ µtr ⎟⎠ tr µtr 1 ∑ δ PV 38. PINDEX = i 1 i i International Trade 39. XStr = BtrE ⎡⎣ βtrE EX trκtr + ( 1 − βtrE E ⎡ ⎛ PEtr ⎞ ⎛ 1 − βtrE 40. EX tr = ⎢ ⎜ ⎟⎜ E ⎣ ⎝ PLtr ⎠ ⎝ βtr ) 1 E Dtrκtr ⎤⎦ κtr E 3 τtrE ⎞⎤ ⎟ Dtr ⎠ ⎦⎥ 3 −1 41. Qtr = AtrM ⎡⎣ α trM M tr−ρtr + ( 1 − α trM ) Dtr−ρtr ⎤⎦ ρtrM M M ⎡ ⎛ PDtr ⎞ ⎛ α trM ⎞ ⎤ 42. M tr = ⎢ ⎜ ⎟⎜ M ⎟ ⎣ ⎝ PM tr ⎠ ⎝ 1 − α tr ⎠ ⎦⎥ 3 σtrM 3 Dtr CAB = ∑ PWM tr M tr + λ ROW ∑ rtr KDtr e + λ LROW rl ⋅ LAND e tr 43. tr 1 + DIV ROW − ∑ PWEtr EX tr tr Equilibrium 44. Qtr = DITtr + ∑C tr ,h + INVtr 3 h 45. LS = ∑ LD 1 j j 46. IT = ∑ SH h + SF + SG + e ⋅ CAB 1 h ⎛⎛ ⎞ ⎡ PCOtr ⎤ EVh = ⎜ ⎜ CTH h − ∑ PCtrj CtrjMIN ,h ⎟ ∏ ⎢ trj ⎝⎝ ⎠ tr ⎣ PCtr ⎥⎦ 47. ⎛ ⎞⎞ − ⎜ CTHOh − ∑ PCOtrj CtrjMIN ,h ⎟ ⎟ trj ⎝ ⎠⎠ γ tr ,h 4 Dynamic Equations 48. KDtr ,t +1 = (1 − δ ) KDtr ,t + Indtr ,t 49. LSt +1 = ( 1 + ng ) ⋅ LSt min 50. Ctrmin , h , t +1 = ( 1 + ng ) ⋅ Ctr , h ,t 3 1 12 14 Annabi, N. Cockburn, J and Decaluwé, B. (2004) 2 ⎛R ⎞ ⎛R ⎞ 51. = γ 1tr ⎜ tr ,t ⎟ + γ 2tr ⎜ tr ,t ⎟ KDtr ,t ⎝ Ut ⎠ ⎝ Ut ⎠ 52. U t = Pinvt ⋅ ( ir + δ ) Ind tr ,t 53. ITt = Pinvt ⋅ 3 1 ∑ Indtr ,t 1 tr Total: 151 Endogenous variables Number of variables Ctr ,h : Household h's consumption of good tr (volume) CF : CI j : Composite agricultural capital-labor factor (volume) 1 Total intermediate consumption of activity j (volume) 4 Household h's total consumption (value) 4 Demand for domestic good tr (volume) 3 CTH h : Dtr : DI tr , j : DITtr : DTF : DTH h : e : EVh : EX tr : G : INVtr : IT : LD j : M tr : Pi : PCtr : PDtr : PEtr : PINDEX : Pinv : PLtr : PM tr : PV j : Qtr : rtr : rl : Intermediate consumption of good tr in activity j (volume) 12 12 Intermediate demand for good tr (volume) 3 Receipts from direct taxation on firms' income 1 Receipts from direct taxation on household h's income 4 Exchange rate 1 Equivalent variation for household h 4 Exports in good tr (volume) 3 Public expenditures 1 Investment demand for good tr (volume) 3 Total investment 1 Activity j demand for labour (volume) 4 Imports in good tr (volume) 3 Producer price of good i 4 Consumer price of composite good tr 3 Domestic price of good tr including taxes 3 Domestic price of exported good tr 3 GDP deflator 1 Price index of investment 1 Domestic price of good tr (excluding taxes) 3 Domestic price of imported good tr 3 Value added price for activity j 4 Demand for composite good tr (volume) 3 Rate of return to capital in activity tr 3 Rate of return to agricultural land 1 15 Annabi, N. Cockburn, J and Decaluwé, B. (2004) rc : SF : SG : SH h : TI tr : TIEtr : TIM tr : VA j : Rate of return to composite factor 1 w : XStr : YDH h : YF : YG : YH h : LS : KDtr : CAB : Firms' savings 1 Government's savings 1 Household h's savings 4 Receipts from indirect tax on tr 3 Receipts from tax on export tr 3 Receipts from import duties tr 3 Value added for activity j (volume) 4 Wage rate 1 Output of activity tr (volume) 3 Household h's disposable income 4 Firms' income 1 Government's income 1 Household h's income 4 Total labour supply (volume) 1 Demand for capital in activity tr (volume) 3 Current account balance 1 Indtr ,t : Demand for capital in activity tr (volume) 3 Ut : Capital user cost 1 CtrMIN ,h Minimum consumption of good tr by household h : 12 Total: 151 Exogenous variables Number of variables DIVh : DIV ROW : LAND : PWEtr : PWM tr : TGh : XS NTR : Dividends paid to household h 4 Dividends paid to the rest of the World 1 Land supply (volume) 1 World price of export tr 3 World price of import tr 3 Public transfers to household h 4 Output of activity NTR (volume) 1 Total: 17 Parameters 16 Annabi, N. Cockburn, J and Decaluwé, B. (2004) Production functions Aj : Scale coefficient (Cobb-Douglas production function) aijtr , j : Input-output coefficient αj : Elasticity (Cobb-Douglas production function) io j : Technical coefficient (Leontief production function) vj : Technical coefficient (Leontief production function) CES function between capital and labor AtrKL : Scale coefficient αtrKL ρtrKL σtrKL : Share parameter : Substitution parameter : Substitution elasticity CES function between composite factor and land AtrCL : Scale coefficient α Ctr L : Share parameter ρCL tr σCL tr : Substitution parameter : Substitution elasticity CES function between imports and domestic production AtrM : Scale coefficient αtrM : Share parameter ρtrM σtrM : Substitution parameter : Substitution elasticity CET function between domestic production and exports BtrE : Scale coefficient βtrE κtrE τtrE : Share parameter : Transformation parameter : Transformation elasticity LES consumption function γ tr ,h : Marginal share of good tr Tax rates tetr : Tax on exports tr 17 Annabi, N. Cockburn, J and Decaluwé, B. (2004) tmtr : txtr : tyhh : tyf : Import duties on good tr Tax rate on good tr Direct tax rate on household h's income Direct tax rate on firms' income Other parameters δj : Share of activity j in total value added λ hL : Share of land income received by household h λ LF : Share of land income received by firms λ LROW : Share of land income received by foreigners λ hR : Share of capital income received by household h λ RF : Share of capital income received by firms λ ROW : Share of capital income received by foreigners λW h Share of labor income received by household h : ψh : µtr : ng : δ: γ 1tr : γ 2tr : ir : Propensity to save Share of the value of good tr in total investment Population growth rate Capital depreciation rate Parameter in the investment demand function Parameter in the investment demand function Real interest rate sets i, j ∈ I = { AGR,IND,SER,NTR } tr ∈ TR = { AGR,IND,SER } nag ∈ NAG = { IND,SER } h ∈ H = { RP,UP,RR,UR } t , t ∈ T = {1, 2,",10 } All activities and goods (AGR: agriculture, IND: industry, SER: services, NTR: non-tradable services) Tradable activities and goods Non-agricultural Tradable activities and goods Households (RP: rural poor, UP: urban poor, RR: rural rich, UR: urban rich) Periods 18 Annabi, N. Cockburn, J and Decaluwé, B. (2004) Appendix II: GAMS Code $TITLE $STITLE SEQUENTIAL DYNAMIC CGE MODEL FOR POVERTY ANALYSIS EXTER+SD $ontext Model of a small open economy with government producing 4 goods using 3 factors owned by 4 households and recursive dynamics THIS VERSION OF MAY,2004, This copy could be used freely as long as proper reference is made to the source of the model (c) COPYRIGHT: Nabil ANNABI, John COCKBURN AND Bernard DECALUWÉ $offtext *-------------------Variables definition---------------------------* VARIABLES *=============== *Prices *=============== w(T) r(TR,T) rl(T) * rf rc(T) P(I,T) PD(TR,T) PV(I,T) PL(TR,T) PC(TR,T) PM(TR,T) PE(TR,T) PWM(TR,T) PWE(TR,T) PINDEX PINV e(T) *=============== *Production *=============== XS(I,T) VA(I,T) DI(TR,J,T) CI(I,T) Wage rate Rate of return to capital in sector TR Rate of return to agricultural land Uniform return to capital return to composite factor Producer price of good I Domestic price of good TR including tax Value added price for sector I Domestic price of good TR excluding tax Price of composite good TR Domestic price of imported good TR Domestic price of exported good TR World price of import TR (foreign currency) World price of export TR (foreign currency) Producer price index Price index of investment Exchange rate Production of sector I Value added in sector I (volume) Intermediate consumption of good TR in sector J Total intermediate consumption of sector I *=============== *Factors *=============== KD(TR,T) LD(I,T) LS(T) LAND CF(T) *=============== *Demand *=============== C(TR,H,T) CTH(H,T) Sector TR demand for capital Sector I demand for labour Total labour supply Agricultural land Composite agricultural capital-labor factor Household H consumption of good TR (volume) Household H total consumption (value) 19 Annabi, N. Cockburn, J and Decaluwé, B. (2004) INV(TR,T) IND(TR,T) IT(T) ITVOL DIT(TR,T) G(T) D(TR,T) Q(TR,T) Investment in good TR (volume) INvestment by destination Total investment (value) Total investment (volume) Intermediate demand for good TR Total public consumption (value) Demand for domestic good TR Demand for composite good TR *=============== *International trade *=============== M(TR,T) Imports of good TR EX(TR,T) Exports of good TR CAB(T) Current account balance *=============== *Income and savings *=============== YH(H,T) YDH(H,T) YF(T) YG(T) SH(H,T) SF(T) SG(T) DIV(H,T) DIV_ROW(T) TG(H,T) TI(TR,T) TIM(TR,T) TIE(TR,T) DTH(H,T) DTF(T) nu(T) adj(T) savadj(T) Household H income Household H disposable income Firms income Government income Household H savings Firms savings Government savings Dividends paid to capitalist households Dividends paid to foreigners Public transfers to households Receipts from indirect tax Receipts from import duties Receipts from tax on exports Receipts from direct taxation on household H income Receipts from direct taxation on firms income Adjustment variable for hh savings Adjustment variable for indirect taxes Adjustment variable for investment and saving *=============== *Others *=============== EV(H,T) LEON(T) OMEGA Equivalent variation for household H Walras law verification variable Objective variable ; *-------------------------Equations definition------------------------* EQUATIONS *=============== *Production *=============== SUPPLY(I,T) VAD1(NAG,T) VAD2(T) VAD3(NTR,T) ECF(T) CIEQ(I,T) DIEQ(TR,J,T) LDEM1(TR,T) LDEM2(NTR,T) LANDEM(T) Production function for sector I Value added in non-agricultural sectors Value added in agricultural sectors Value added in non-tradable sectors Composite agricultural labor-capital factor Total intermediate consumption for sector I Intermediate consumption of good TR by sector J Labour demand for tradable sectors Labour demand for non-tradable sectors Agricultural land demand *=============== *Income and savings 20 Annabi, N. Cockburn, J and Decaluwé, B. (2004) *=============== INCH(H,T) INCDH(H,T) INCF(T) INCG(T) SAVH(H,T) SAVF(T) SAVG(T) *=============== *Taxes *=============== Household income (workers) Household H disposable income Firms income Government income Household H savings Firms savings Government savings INDTAX(TR,T) IMDUTY(TR,T) EXTAX(TR,T) DIRTAXH(H,T) DIRTAXF(T) *=============== *Demand *=============== Receipts Receipts Receipts Receipts Receipts CTHEQ(H,T) CONSH(TR,H,T) CONSG(T) INVEST(TR,T) INVDF(TR,T) INVVOL(T) INTDEM(TR,T) KACCUM(TR,T) DEMOG(T) from from from from from indirect taxes on TR import duties tax on exports household taxation firm taxation Household H total consumption Household H consumption of good TR Public consumption Investment in good Total investment in volume Intermediate demand *=============== *Prices *=============== PRVA(I,T) Value added price RETK1(NAG,T) Rate of return to capital from non-agricultural sectors RETCF(T) Return to composite capital-labor factor in agricultural sectors RETK2(AGRS,T) Rate of return to capital from agricultural sectors PRDL(TR,T) Domestic price PRM(TR,T) Import prices PRE(TR,T) Export prices PRC(TR,T) Composite price (tradables) PRP(TR,T) Producer price (tradables) PII(T) Price index for investment AVPRI(T) Producer price index *=============== *International trade *=============== CET(TR,T) EXPORTS(TR,T) ARMING(TR,T) IMPORT(TR,T) CURACC(T) *=============== *Equilibrium *=============== DOMABS(GOOD,T) LEQUI(T) INVDEM(T) ISEQUI(T) Relation between D and EX Export supply CES between imports and domestic good Import demand Current account Domestic absorption (goods) Labour market equilibrium Investment-savings equilibrium *=============== *Others *=============== EVEQ(H,T) WALRAS(T) Calculation of EV Verification of the Walras law 21 Annabi, N. Cockburn, J and Decaluwé, B. (2004) OBJ Objective function ; *=============== *Production *=============== XS(I,T) =E= VA(I,T)/v(I); VA(NAG,T) =E= A_KL(NAG)*(alpha_kl(NAG)*LD(NAG,T)**(-rho_kl(NAG)) +(1-alpha_kl(NAG))*KD(NAG,T)**(-rho_kl(NAG))) **(-1/rho_kl(NAG)); VA("AGR",T) =E= A_CL*(alpha_cl*CF(T)**(-rho_cl)+(1-alpha_cl) *LAND(T)**(-rho_cl))**(-1/rho_cl); CF(T) =E= A_KL("AGR")*(alpha_kl("AGR")*LD("AGR",T)**(-rho_kl("AGR")) +(1-alpha_kl("AGR"))*KD("AGR",T)**(-rho_kl("AGR"))) **(-1/rho_kl("AGR")); VA(NTR,T) =E= LD(NTR,T); CI(I,T) =E= io(I)*XS(I,T); DI(TR,J,T) =E= LAND(T) aij(TR,J)*CI(J,T); =E= (((1-alpha_cl)*rc(T))/(alpha_cl*rl(T)))**sigma_cl*CF(T); LD(TR,T) =E= ((alpha_kl(TR)/(1-alpha_kl(TR)))**(sigma_kl(TR)) *(r(TR,T)/w(T))**sigma_kl(TR))*KD(TR,T); LD(NTR,T) =E= (P(NTR,T)*XS(NTR,T)-SUM(TR,DI(TR,NTR,T)*PC(TR,T)))/w(T); *=============== *Income and savings *=============== YH(H,T) =E= lambda_w(H)*w(T)*SUM(I,LD(I,T))+lambda_r(H) *SUM(TR,r(TR,T)*KD(TR,T))+lambda_l(H)*rl(T)*LAND(T) +PINDEX(T)*TG(H,T)+DIV(H,T); YDH(H,T) =E= YH(H,T) - DTH(H,T); YF(T) =E= lambda_rf*SUM(TR,r(TR,T)*KD(TR,T)) +lambda_lf*rl(T)*LAND(T); YG(T) =E= SUM(TR,TI(TR,T))+SUM(H,DTH(H,T))+SUM(TR,TIE(TR,T)) + SUM(TR,TIM(TR,T))+DTF(T) ; SH(H,T) =E= nu(T)*psi(H)*YDH(H,T) ; SF(T) =E= YF(T) - SUM(H,DIV(H,T)) - DTF(T) - e(T)*DIV_ROW(T) ; SG(T) =E= YG(T) - G(T) - SUM(H,TG(H,T))*PINDEX(T); *=============== *Taxes *=============== TI(TR,T) =E= (adj(T)+tx(TR)*(1+adj(T)))*(P(TR,T)*XS(TR,T)-PE(TR,T)*EX(TR,T))+(adj(T) +tx(TR)*(1+adj(T)))*(1+tm(TR))*e(T)*PWM(TR,T)*M(TR,T); TIM(TR,T) =E= tm(TR)*PWM(TR,T)*e(T)*M(TR,T); TIE(TR,T) =E= te(TR)*PE(TR,T)*EX(TR,T); DTH(H,T) =E= tyh(H)*YH(H,T) DTF(T) =E= tyf*YF(T) ; ; 22 Annabi, N. Cockburn, J and Decaluwé, B. (2004) *=============== *Demand *=============== CTH(H,T) =E= YDH(H,T)-SH(H,T); C(TR,H,T)*PC(TR,T) =E= C_MIN(TR,H)*PC(TR,T)+gamma(TR,H) *(CTH(H,T)-SUM(TRJ,C_MIN(TRJ,H)*PC(TRJ,T))); G(T) INV(TR,T) =E= XS("NTSER",T)* P("NTSER",T); =E= mu(TR)*IT(T)/PC(TR,T) ; ITVOL(T)*PINV(T) =E= IT(T); IND(TR,T)/KD(TR,T) =E= KD(TR,T+1) DIT(TR,T) SAVADJ(T)*SP(TR)*EXP(R(TR,T)/PINV(T)); =E= (1-DEP(TR))*KD(TR,T)+ IND(TR,T) ; =E= SUM(J, DI(TR,J,T)) ; *=============== *Prices *=============== PV(I,T) =E= (P(I,T)*XS(I,T)-SUM(TR,DI(TR,I,T)*PC(TR,T)))/VA(I,T); R(NAG,T) =E= (PV(NAG,T)*VA(NAG,T) - w(T)*LD(NAG,T) )/KD(NAG,T) ; R("AGR",T)*KD("AGR",T) =E= RC(T)*CF(T)-w(T)*LD("AGR",T); RC(T) =E= (PV("AGR",T)*VA("AGR",T) - rl(T)*LAND(T))/CF(T); PD(TR,T) =E= PL(TR,T)*(1+tx(TR))*(1+adj(T)); PM(TR,T) =E= (1+TX(TR))*(1+adj(T))*(1+tm(TR))*e(T)*PWM(TR,T); PE(TR,T) =E= PWE(TR,T)*e(T)/(1+te(TR)); PC(TR,T) =E= (PD(TR,T)*D(TR,T)+PM(TR,T)*M(TR,T))/Q(TR,T); P(TR,T) =E= (PL(TR,T)*D(TR,T) + PE(TR,T)*EX(TR,T))/XS(TR,T); PINV(T) =E= PROD(TR$(INVO(TR) NE 0), (PC(TR,T)/mu(TR))**mu(TR)); PINDEX(T) =E= SUM(I,PV(I,T)*delta(I)); *=============== *International trade *=============== XS(TR,T) =E= B_E(TR)*(beta_e(TR)*EX(TR,T)**kappa_e(TR) +(1-beta_e(TR))*D(TR,T)**kappa_e(TR))**(1/kappa_e(TR)); EX(TR,T) =E= Q(TR,T) =E= A_M(TR)*(alpha_m(TR)*M(TR,T)**(-rho_m(TR)) +(1-alpha_m(TR))*D(TR,T)**(-rho_m(TR)))**(-1/rho_m(TR)); M(TR,T) =E= ((alpha_m(TR)/(1-alpha_m(TR)))**(sigma_m(TR)) *(PD(TR,T)/PM(TR,T))**sigma_m(TR))*D(TR,T); ((PE(TR,T)/PL(TR,T))**tau_e(TR)*((1-beta_e(TR))/beta_e(TR)) **tau_e(TR))*D(TR,T); CAB(T) =E= SUM(TR,PWM(TR,T)*M(TR,T))+ lambda_row*SUM(TR,r(TR,T)*KD(TR,T))/e(T) +lambda_lrw*rl(T)*LAND(T)/e(T)+DIV_ROW(T)-SUM(TR,PWE(TR,T)*EX(TR,T)); *=============== *Equilibrium *=============== Q(GOOD,T) =E= SUM(H,C(GOOD,H,T))+DIT(GOOD,T)+INV(GOOD,T); 23 Annabi, N. Cockburn, J and Decaluwé, B. (2004) LS(T) LS(T+1) =E= =E= SUM(I,LD(I,T)); (1+GN)* LS(T) ; IT(T) =E= PINV(T)*SUM(TR,IND(TR,T)) ; IT(T) =E= SUM(H,SH(H,T)) + SF(T) + SG(T) + CAB(T)*e(T); *=============== *Others *=============== EV(H,T) LEON(T) =E= PROD(TR,(PCO(TR)/PC(TR,T))**gamma(TR,H)) *(CTH(H,T)-SUM(TRJ,C_MIN(TRJ,H)*PC(TRJ,T))) -(CTHO(H)-SUM(TRJ,C_MIN(TRJ,H)*PCO(TRJ))); =E= Q("SER",T)- SUM(H,C("SER",H,T))-DIT("SER",T); OBJ.. OMEGA =E= 1; *------------------------Initialization----------------------------* w.L(T) r.L(TR,T) rl.l(T) rc.l(T) P.L(I,T) PD.L(TR,T) PV.L(I,T) PL.L(TR,T) PC.L(TR,T) PM.L(TR,T) PE.L(TR,T) PWM.L(TR,T) PWE.L(TR,T) PINDEX.L(T) PINV.L(T) e.L(T) XS.L(I,T) VA.L(I,T) DI.L(TR,J,T) CI.L(I,T) KD.L(TR,T) LD.L(I,T) LS.L(T) LAND.L(T) CF.L(T) C.L(TR,H,T) CTH.L(H,T) INV.L(TR,T) IT.L(T) ITVOL.L(T) DIT.L(TR,T) G.L(T) D.L(TR,T) Q.L(TR,T) M.L(TR,T) EX.L(TR,T) CAB.L(T) YH.L(H,T) YDH.L(H,T) YF.L(T) YG.L(T) SH.L(H,T) SF.L(T) SG.L(T) DIV.L(H,T) DIV_ROW.L(T) = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = wo; ro(TR); rlo; rco; PO(I); PDO(TR); PVO(I); PLO(TR); PCO(TR); PMO(TR); PEO(TR); PWMO(TR); PWEO(TR); PINDEXO; PINVO; eO; XSO(I); VAO(I); DIO(TR,J); CIO(I); KDO(TR); LDO(I); LSO; LANDO; CFO; CO(TR,H); CTHO(H); INVO(TR); ITO; ITVOLO; DITO(TR); GO; DO(TR); QO(TR); MO(TR); EXO(TR); CABO; YHO(H); YDHO(H); YFO; YGO ; SHO(H); SFO; SGO; DIVO(H); DIV_ROWO; 24 Annabi, N. Cockburn, J and Decaluwé, B. (2004) TG.L(H,T) TI.L(TR,T) TIM.L(TR,T) TIE.L(TR,T) DTH.L(H,T) DTF.L(T) nu.FX(T) adj.L(T) *OTHERS LEON.L(T) OMEGA.L = = = = = = TGO(H); TIO(TR); TIMO(TR); TIEO(TR); DTHO(H); DTFO; = 1; = 0; = = 0; 1; *----------------------------Closure--------------------------------* * Exchange rate is the numeraire, capital is sector specific, fixed public * expenditure (in volume). LAND.FX(T) = LANDO; ITVOL.L(T) = ITO/PINVO; DIV.FX(H,T) = DIVO(H); DIV_ROW.FX(T) = DIV_ROWO; TG.FX(H,T) = TGO(H); XS.FX("NTSER",T) = XSO("NTSER"); PWM.FX(TR,T) = PWMO(TR); PWE.FX(TR,T) = PWEO(TR); e.FX(T) = eo; CAB.FX(T) = CABO; SG.FX(T) = SGO; adj.L(T) = 0; *---------------------------savadj.L(t) = 1 ; LS.L(T) = LSO; KD.L(TR,T) = KDO(TR); IND.L(TR,T) = INDO(TR) ; LS.FX(T1) KD.FX(TR,T1) = LSO; = KDO(TR); * Simulation *tm(TR) = 0; *---------------------------Model execution---------------------------* OPTION NLP=CONOPT3; *OPTION NLP= MINOS5; option limrow=1; MODEL EXTERPSD OPEN ECONOMY WITH GOVERNMENT EXTERPSD.HOLDFIXED = 1 ; EXTERPSD.WORKSPACE = 100 ; SOLVE EXTERPSD MAXIMIZING OMEGA USING NLP; /ALL/; 25