The Macroeconomics of Itaipú: Measuring the Impact of a Large Fiscal Shock Javier Charotti University of Chicago Banco Central del Paraguay Constantino Hevia Universidad T. di Tella Andy Neumeyer Universidad T. di Tella Banco Central de la República Argentina August, 2017 Note: The opinions expressed in this presentation are those of the authors and do not necessarily reflect the views of the Central Bank of Argentina, nor those of the Central Bank of Paraguay or their boards of directors.
The Itaipú Hidroelectric Dam 1 / 44 Charotti-Hevia-Neumeyer Itaipú
Satellite view of Itaipu and Ciudad del Este 2 / 44 Charotti-Hevia-Neumeyer Itaipu
Satellite view of Itaipu and Ciudad del Este 3 / 44 Charotti-Hevia-Neumeyer Itaipu
The Itaipú Binacional Created in the 1970s by the governments of Paraguay and Brazil Each country owns 50 percent of the Binacional and its production Itaipú provides 17 percent of all of Brazil s energy consumption and 70 percent of Paraguay s Brazil buys Paraguay s excess production at negotiated prices The firm borrowed from Brazil to finance the construction of the dam Amount borrowed: about 27 billion USD (GDP in 1974 was 1.2 billion USD) Expected to repay in full by 2023 Construction period: 1974-1981 Itaipú begun exporting electricity in 1984 4 / 44 Charotti-Hevia-Neumeyer Itaipú
The Construction of Itaipú The course of the seventh biggest river in the world was shifted, as were 50 million tons of earth and rock. The amount of concrete used to build the Itaipú Power Plant would be enough to build 210 football stadiums the size of the Estádio do Maracanã. The iron and steel used would allow for the construction of 380 Eiffel Towers. The volume of excavation of earth and rock in Itaipu is 8.5 times greater than that of the Channel Tunnel and the volume of concrete is 15 times greater. Around forty thousand people worked in the construction. Ciudad del Este was built as a support town for the construction of the dam Itaipú is one of the most expensive objects ever built. 5 / 44 Charotti-Hevia-Neumeyer Itaipú
Itaipú as a Macroeconomic Impulse Investment in Itaipú Government Expenditure with no Taxes Percent of trend GDP 25 20 15 10 5 Itaipu investment Electricity exports Large transitory shock to government expenditure in non-tradable goods Almost 12% of GDP in 1979 Financing: No taxes FDI from Brazil, repaid in kind Paraguay exports electricity to Brazil after 1984 at negotiated prices. 17% of GDP in 1991 0 1970 1975 1980 1985 1990 1995 2000 Investment data on NT within Paraguay constructed from company records 6 / 44 Charotti-Hevia-Neumeyer Itaipú
An Itaipú Business Cycle? Does the Itaipú impulse explain the business cycle in the late 70s and 80s? What is the role of TFP or other impulses? Itaipú investment (percent of trend GDP) 12 10 8 6 4 2 Itaipú investment (left axis) Tradable output (right axis) Non-tradable output (right axis) GDP (right axis) 50 40 30 20 10 0-10 -20 Percentage deviation from trend 0 1970 1975 1980 1985 1990 1995 2000-30 7 / 44 Charotti-Hevia-Neumeyer Itaipú
The Relative Price of Non-Traded Goods Relative price Non-Tradable / Tradable (1975=1) 1.4 1.3 1.2 1.1 1 0.9 0.8 0.7 0.6 0.5 Consumer/Wholesale Construction/Manufactures Services/Goods CPI/(CPI * E) 1970 1975 1980 1985 1990 1995 2000 Note: Three measures of the relative price of non-traded goods. Consumer Price Index/Wholesale Price Index, Implicit Price Deflator of Construction over Manufacturing, Implicit Price Deflator of Services over Goods in national accounts excluding electricity 12 10 8 6 4 2 0 Itaipu investment (percent of trend GDP) 8 / 44 Charotti-Hevia-Neumeyer Itaipú
The Relative Price of Non-Traded Goods How does the increase in the demand for non-traded goods affect its relative price? Why does the relative price of construction and services fall? Relative price Non-Tradable / Tradable (1975=1) 1.4 1.3 1.2 1.1 1 0.9 0.8 0.7 0.6 0.5 Consumer/Wholesale Construction/Manufactures Services/Goods CPI/(CPI * E) 1970 1975 1980 1985 1990 1995 2000 12 10 8 6 4 2 0 Itaipu investment (percent of trend GDP) 9 / 44 Charotti-Hevia-Neumeyer Itaipú
A Model of the Macroeconomics Itaipú Small open economy with incomplete capital markets Two sector neoclassical growth model with sector specific capital Traded goods Non-traded goods Itaipú Investment is an exogenous process that uses non-traded goods in the resource constraint C N,t + I N,t + I Itaipu t = Y N,t Electricity exports to Brazil are an exogenous process akin to an endowment of traded goods, Q t, C T,t + I T,t + D t = D t+1 1 + r t + Y T,t + Q t Itaipú capital does not yield capital in other production functions Productivity shocks in each sector 10 / 44 Charotti-Hevia-Neumeyer Itaipú
Modeling Itaipú Shocks We model Itaipú shocks as an auto-regressive process [ ] [ ] [ ] I Itaipu t ρi 0 I Itaipu = t 1 + ɛ Q t 0 ρ It Q Q t 1 In the experiments we feed the model with the true realizations of ɛ It {Q t} shocks induce persistent movements in consumption, labor, and output. Incomplete capital markets Perfect foresight model: work in progress Allows agents to anticipate Itaipú shocks Only present value of {Q t} matters 11 / 44 Charotti-Hevia-Neumeyer Itaipú
Preferences and Technology Preferences E 0 β t U (C t, L t) C t is consumption and L t is labor. Cobb-Douglas preferences: t=0 [ C η t (1 L 1 η] 1 σ t) 1 U (C t, L t) =. 1 σ Consumption is a composite of tradable and non-tradable goods C t = [γ 1 θ C θ 1 θ N,t + (1 γ) θ 1 θ 1 ] θ θ 1 C θ T,t. 12 / 44 Charotti-Hevia-Neumeyer Itaipú
Preferences and Technology Production functions Tradables: Non-tradables: Y T,t = A 1 α T T,t K α T T,t L1 α T T,t Y N,t = A 1 α N N,t K α N N,t L1 α N N,t. Stocks of capital The stocks of capital are a (CRS) composite of capital goods made of tradable and non-tradable capital goods K T,t = ( ) KN,t T ωt ( ) K T 1 ωt T,t K N,t = ( ) KN,t N ωn ( ) K N 1 ωn T,t Once installed, capital is immobile and remains in the sector until it depreciates. 13 / 44 Charotti-Hevia-Neumeyer Itaipú
Preferences and Technology Capital accumulation Four types of capital goods which evolve as K i T,t+1 = (1 δ T )K i T,t + I i T,t Φ T (K i T,t+1, K i T,t) K i N,t+1 = (1 δ N )K i N,t + I i N,t Φ N (K i N,t+1, K i N,t) i = T, N i = T, N Φ T and Φ N are quadratic capital adjustment cost functions Investment Investment made out of tradable goods is allocated to increase the stock of capital used to produce tradables (I T T,t) or non-tradables (I N T,t), I T,t = I T T,t + I N T,t. Likewise for non-tradable investment goods I N,t = I T N,t + I N N,t. 14 / 44 Charotti-Hevia-Neumeyer Itaipú
Feasibility Tradable sector C T,t + I T,t + D t = Dt+1 1 + r t + Y T,t + Q t D t is one period debt (in tradables) acquired at time t 1, r t is the implicit interest rate on the bond price (charged by the rest of the world), and Q t is the flow of payments received from the electricity exported to Brazil. Interest rate is ( ) r t = r + φ r e D t+1 D GDP 1 Non-tradable sector C N,t + I N,t + I Itaipu t = Y N,t. Labor allocation L N,t + L T,t = L t. 15 / 44 Charotti-Hevia-Neumeyer Itaipú
Exogenous Processes Exogenous impulses evolve as an auto-regressive process I Itaipu t Q t A Nt A Tt = ρ I 0 0 0 0 ρ Q 0 0 0 0 ρ AT 0 0 0 0 ρ AN I Itaipu t 1 Q t 1 A Nt 1 A Tt 1 + ɛ t where ɛ t are independently and normally distributed. 16 / 44 Charotti-Hevia-Neumeyer Itaipú
Calibration Set σ = 2. Labor shares α T = 0.48 and α N = 0.62 (National Accounts) Foreign interest rate r = 0.05 and sensitivity parameter φ r = 0.001. Steady state Debt/Output ratio of 0.4. Depreciation rates δ N and δ T δ N = 0.027 (half-life structures=25 years) δ T = 0.094 (half-life equipment and machinery=7 years) Set ω N and ω T so that investment ratios Ī T T /Ī T N = 1 and Ī N T /Ī N N = 1. Set an arbitrary value for the relative price of non-tradables, p N, and choose ĀT /ĀN to match p N (Balassa-Samuelson). Elasticity of substitution T-NT θ = 0.4 (Neumeyer & Gonzalez-Rozada) Set γ to match share tradables on output ȲT /GDP = 0.52 (Nat. Acc.) Choose η so that L = 1/3 and Ā T so that GDP = 1. 17 / 44 Charotti-Hevia-Neumeyer Itaipú
Calibration Parameters of the exogenous stochastic process and capital adjustment cost For Itaipú investment and electricity exports, we run AR(1) regressions I Itaipu t = 0.97I Itaipu t 1 + ɛ It ; ɛ It N(0, 0.014 2 ) Q t = 0.99Q t 1 + ɛ Qt ; ɛ Qt N(0, 0.02 2 ) Use SMM to estimate productivity parameters ρ AT, ζ AT, ρ AN, ζ AN, and capital adjustment costs φ T and φ N to match 1 First order autocorrelation of tradable output 2 First order autocorrelation of non-tradable output 3 Volatility of tradable output 4 Volatility of non-tradable output 5 Volatility of construction (NT) relative to GDP 6 Volatility of machinery and equipment (T) relative to GDP 18 / 44 Charotti-Hevia-Neumeyer Itaipú
Summary of calibrated parameters Description Symbol Value Risk aversion σ 2 Exponent C in utility η 0.46 Share C T in C γ 0.417 Elasticity substitution T-NT θ 0.4 Foreign interest rate r 0.05 Labor share T α T 0.48 Labor share NT α T 0.62 Depreciation rate T δ T 0.094 Depreciation rate NT δ N 0.027 Share NT capital in K T ω T 0.67 Share NT capital in K N ω N 0.67 Persistence A T shock ρ AT 0.977 Volatility A T shock ζ AT 0.033 Persistence A N shock ρ AN 0.999 Volatility A N shock ζ AN 0.001 Persistence I Itaipu ρ I 0.97 Volatility I Itaipu ζ I 0.014 Persistence Q t ρ Q 0.99 Volatility Q t ζ Q 0.02 Capital Adjustment T φ T 0.008 Capital Adjustment NT φ N 0.214 19 / 44 Charotti-Hevia-Neumeyer Itaipú
Recovering Productivity Shocks Productivity shocks are unobservable: Labor and capital in each sector is unobservable. Employment data only on labor force. Tracks working age population. In the model, output is a function of productivity shocks in each sector, Itaipú shocks, other endogenous state variables, and parameters. ( ) YNt = f ( A Y Nt, A Tt, I Itaipu t, Q t, parameters ) Tt Recover productivity shocks (A Nt, A Tt ) using data for Y Nt, Y Tt, I Itaipu t, and Q t inverting the policy functions in f ( ) 20 / 44 Charotti-Hevia-Neumeyer Itaipú
Recovered Productivity Shocks and Output in Each Sector 50 40 Percentage deviation from trend 30 20 10 0-10 -20-30 Tradable output (data - excludes electricity) Tradable productivity (model implied) Non-Tradable output (data) Non-Tradable productivity (model implied) 1970 1975 1980 1985 1990 1995 21 / 44 Charotti-Hevia-Neumeyer Itaipú
Realization of shocks 25 20 15 Deviation from trend 10 5 0-5 -10-15 -20 Tradable productivity Non-Tradable productivity Itaipu investment Electricity exports 1970 1975 1980 1985 1990 1995 22 / 44 Charotti-Hevia-Neumeyer Itaipú
Macroeconomic Effect of Itaipu Experiments: 1 Feed the model with the shocks to I Itaipu t 2 Feed the model with the shocks to I Itaipu t and Q t. 3 Feed the model with all shocks: A Nt, A Tt, I Itaipu t, Q t 4 Feed the model with all but Q t Results: Output Output of non tradable good expands Output of tradable good contracts Labor Employment increases Labor is reallocated from traded to non-traded goods sector Prices Relative price of non-traded increases Relative price of non-traded decreases once productivity shocks are included Results depend on the elasticity of the labor supply Inelastic labor supply: large effect on output and small effect on prices Elastic labor supply: small effect on output and large effect on prices 23 / 44 Charotti-Hevia-Neumeyer Itaipú
Macroeconomic Effect of Itaipu Output: Only Investment Shock I itaipu t 50 40 Percentage deviation from trend 30 20 10 0-10 -20-30 Tradable output (data) Non-Tradable output (data) Tradable output (model) Non-Tradable output (model) 1970 1975 1980 1985 1990 1995 24 / 44 Charotti-Hevia-Neumeyer Itaipú
Macroeconomic Effect of Itaipu Labor: Only Investment Shock I itaipu t 30 25 Percentage deviation from trend 20 15 10 5 0-5 -10-15 -20 Tradable labor (model) Non-Tradable labor (model) Aggregate labor (model) 1970 1975 1980 1985 1990 1995 25 / 44 Charotti-Hevia-Neumeyer Itaipú
Macroeconomic Effect of Itaipu Output: Only Endowment Shock Q t 50 40 Percentage deviation from trend 30 20 10 0-10 -20-30 Tradable output (data) Non-Tradable output (data) Tradable output (model only endowment shock) Non-Tradable output (model only endowment shock) 1970 1975 1980 1985 1990 1995 26 / 44 Charotti-Hevia-Neumeyer Itaipú
Macroeconomic Effect of Itaipu Labor: Only Endowment Shock Q t 0 Percentage deviation from trend -5-10 -15-20 -25-30 Tradable labor (model only endowment shock) Non-Tradable labor (model only endowment shock) Aggregate labor (model only endowment shock) 1970 1975 1980 1985 1990 1995 27 / 44 Charotti-Hevia-Neumeyer Itaipú
Macroeconomic Effect of Itaipu Output: Investment and Endowment Shock: It itaipu and Q t 50 40 Percentage deviation from trend 30 20 10 0-10 -20-30 -40 Tradable output (data) Non-Tradable output (data) Tradable output (only I itaipu t and Q t ) Non-Tradable output (only I itaipu t and Q t ) 1970 1975 1980 1985 1990 1995 28 / 44 Charotti-Hevia-Neumeyer Itaipú
Macroeconomic Effect of Itaipu Labor: Investment and Endowment Shock: It itaipu and Q t 30 Percentage deviation from trend 20 10 0-10 -20-30 Tradable labor (only itaipu) Non-Tradable labor (only itaipu) Aggregate labor (only itaipu) Tradable labor (all shocks) Non-Tradable labor (all shocks) Aggregate labor (all shocks) 1970 1975 1980 1985 1990 1995 29 / 44 Charotti-Hevia-Neumeyer Itaipú
Macroeconomic Effect of Itaipu Labor: All shocks but Q t Percentage deviation from trend 30 25 20 15 10 5 0-5 -10-15 Tradable labor (all but Qt) Non-Tradable labor (all but Qt) Aggregate labor (all but Qt) 1970 1975 1980 1985 1990 1995 Dashed lines: model with all shocks 30 / 44 Charotti-Hevia-Neumeyer Itaipú
Model Validation I: Relative Prices Non-tradables Relative price Non-Tradable / Tradable (1975=1) 0.2 0.15 0.1 0.05 0-0.05-0.1-0.15-0.2-0.25 Construction/Manufactures Services/Goods Model only itaipu Model all shocks Model all but Qt 1970 1975 1980 1985 1990 1995 31 / 44 Charotti-Hevia-Neumeyer Itaipú
Model Validation I: Relative Prices Non-tradables Relative price Non-Tradable / Tradable (1975=1) 0.5 0.4 0.3 0.2 0.1 0-0.1-0.2-0.3-0.4-0.5 Consumer/Wholesale CPI/(CPI * E) Model only itaipu Model all shocks 1970 1975 1980 1985 1990 1995 32 / 44 Charotti-Hevia-Neumeyer Itaipú
Model Validation II: Productivity in Construction Note: Value added in construction over employment in construction. Census data. 33 / 44 Charotti-Hevia-Neumeyer Itaipú
Model Validation III: Solow Residuals Solow residuals in model generated data and in actual data Log-deviation from sample mean (percent) 30 20 10 0-10 -20-30 Model Model (constant labor) Data -40 1970 1975 1980 1985 1990 1995 34 / 44 Charotti-Hevia-Neumeyer Itaipú
Model Validation IV: Aggregate Investment Log-deviation from sample mean (percent) 80 60 40 20 0-20 -40 Data Model Model (no Qt) -60 1970 1975 1980 1985 1990 1995 35 / 44 Charotti-Hevia-Neumeyer Itaipú
Concluding Remarks The building of Itaipú is a natural experiment on the macroeconomic effects of government spending. Independent of cycle Large Transitory, but persistent, so that it has time to work Constructed data on Itaipu investment on non-tradable goods from company records Constructed a simple macroeconomic model to evaluate the effect of Itaipu on output, employment and prices. Preliminary results: There is a business cycle associated to Itaipu Output and employment increase Reallocation of resources towards NT sector Novel method to measure TFP shocks Model validation: relative prices, Solow residuals, investment, labor productivity in construction are consistent with the predictions of the model. 36 / 44 Charotti-Hevia-Neumeyer Itaipú
BACKUP SLIDES 37 / 44 Charotti-Hevia-Neumeyer Itaipú
Other Shocks?: Terms of Trade 50 Log-deviation from sample mean (percent) 40 30 20 10 0-10 -20-30 -40 1970 1975 1980 1985 1990 1995 2000 38 / 44 Charotti-Hevia-Neumeyer Itaipú
Consumption and capital: model only Q t 20 15 Percentage deviation from trend 10 5 0-5 -10-15 -20-25 -30 Tradable consumption (only Q t ) Non-Tradable consumption (only Q t ) Aggregate consumption (only Q t ) Tradable capital Non-tradable capital 1970 1975 1980 1985 1990 1995 39 / 44 Charotti-Hevia-Neumeyer Itaipú
Model Validation II: Impulse Response Functions Data and Model Insert graph with Impulse Response Functions Data and Model 40 / 44 Charotti-Hevia-Neumeyer Itaipú
Separate income from expenditure shocks in itaipu Insert graphs pf output, prices and employment with only expenditure and with only income shock from itaipu 41 / 44 Charotti-Hevia-Neumeyer Itaipú
Government Finance 42 / 44 Charotti-Hevia-Neumeyer Itaipú
Real Exchange Rate 43 / 44 Charotti-Hevia-Neumeyer Itaipú
Robustness check add terms of trade and government non-itaipu government expenditure shocks to the model 44 / 44 Charotti-Hevia-Neumeyer Itaipú