Structural Transformation, Education and Growth

Transcription

1 Structural Transformation, Education and Growth Pedro Cavalcanti Ferreira (EPGE-FGV) Alexander Monge-Naranjo (St. Louis Fed & Wash U.) Luciene Pereira (EPGE-FGV) Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 1 / 39

2 Introduction General Question: Do all countries go through the same process of Structural Transformation (ST)? Many recent papers: Common features across most countries. Focus in preferences and technology. This paper: A crucial difference between countries with different growth/development success. Focus in policies on human capital. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 2 / 39

3 Introduction Common (& commonly studied) facts of ST: Initially: Shift of labor from aggriculture (A) to manufacturing (M) and services (S). This initial process leads to increased aggregate productivity. Long standing puzzle: why poor countries allocate labor to A. At a later stage: shift from M to S; decline in overall share of M. S end up dominating total VA and labor. Overall income and growth performance of country driven by S. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 3 / 39

4 Introduction Lesser known: Wide differences in skill-intensity in S. Success growth stories: S in high skill sectors. Services: designers, researchers, chefs, social workers, inv. bankers. Innovation/adoption/skills in S: drive countries to grow. Examples: Developed countries (late); South Korea. Not so successful stories: S in low skill sectors. Services: street vendors, handymen, domestic labor, and moneylenders Low skill accumulation/innovation S: ceiling for growth. Examples: Brazil; other Latin American after 1980s. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 4 / 39

5 Introduction Policies? Trade/industrial policies: Usual story when comparing South East Asia vs. Latin America Perhaps, but: S are mostly non-tradeables. 1980s onward: Convergence of policies & divergence of incomes. Human Capital Policies: New workers: If unskilled: low (measured) productivity growth in S. If skilled: high (measured) productivity growth in S. Can also look at implications for demographic change. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 5 / 39

6 This Paper: A Simple Quantitative Model: Education and Fertility: Quantity/Quality of Children. Parents choose number & skills of children (as in Becker). Structural Transformation: Sectors and Skills: Agriculture: low skills only. Manufacturing: high skills only. Services: low and high skills. Exogenous sectoral productivities. Non-homothetic preferences. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 6 / 39

7 This Paper: Education/Demographic Policies: Two Policies: Child labor (allowed or not) Schooling subsidies (funded with labor taxes). Redistribution with endogenous types. Calibration: Two Countries: South Korea and Brazil, Korea: fast growth after 1980; Services: high producitivity/skills Brazil: slow growth after 1980; Services: low producitivity/skills. Model: education policies explain a big chunk of differences. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 7 / 39

8 Related Literature Structural transformation: Duarte and Restuccia (2010); Herrendorf, Rogerson and Valentinyi (2013), Ferreira and Silva (2014), McMillan and Rodrik (2011). Souh Korea: Betts, Giri and Verma (2013), Kim and Topel (1995). Demographic Transition and Human Capital: Fertility and Education: Becker (1960), Becker, Murphy, and Tamura (1990), De la Croix and Doepke (2013), Doepke (2004). DGE models: Erosa, Koreshkova and Restuccia (2010); Restuccia and Vandenbroucke (2013). Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 8 / 39

9 Brazil and Korea: ST Similarities Brazil: Allocation of Labor Korea: Allocation of Labor Sectoral allocation of Labor Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 9 / 39

10 Brazil and Korea: Education Trends Differences Average years of schooling, (economically active population) Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 10 / 39

11 Brazil and Korea: Fertility Differences Total Fertility Rates Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 11 / 39

12 Brazil and Korea: Sharp Differences in Income (a) Brazil (b) South Korea Output per worker and per person. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 12 / 39

13 The Model Demographics: Two period lived OLG; t = 0, 1, 2... Lifetime decisions: Preferences: 1st period: children: work or not; attend school or not. 2nd period: adults: labor market; number and skill of children. Altruistic Parents: current utility and utility of offspring. Non-homothetic preferences wrt to three goods: Production Sectors and Skills: Agriculture: low skills only. Manufacturing: high skills only. Services: low and high skills. Exogenous sectoral productivities. Exogenously given policies. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 13 / 39

14 The Model Demographics:Adult types: 1 Skilled (H), if they went to school. 2 Unskilled (L), if they didn t. Child labor = No school. N H : number of skilled adults. N L : number of unskilled adults. Production Sectors and Skills: Agriculture: low skills only Y A = Z A L AL. Manufacturing: high skills only Y M = Z M L MH.. Services: low and high skills Y S = Z S (L SH ) α (L SL ) 1 α. Exogenous sectoral productivities: Z j = (1 + γ j )Z j. State of the Economy: X {Z A, Z M, Z S, N H, N L }. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 14 / 39

15 The Model Production: All sectors j = A, M, S, are perfectly competitive: max p j (X ) Y j (l jh, l jl ) w H (X )l jh w L (X )l jl, l jh,l jl Free entry + FOC: w L (X ) = p A (X )Z A = p S (X )Z S (1 α) w H (X ) = p M (X )Z M = p S (X )Z S α and L AH = L ML = 0. [ ] LSH (X ) α L SL (X ) [ ] LSL (X ) 1 α L SH (X ) Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 15 / 39

16 The Model Households: Preferences: An adult s utility: V (u, n H, n L ) = U + β(n H + n L ) ε [n H V H + n L V L], where [v(c A )(c M ) b (c S + c S ) (1 b)] 1 σ U(c A, c M, c S ) = v(c A ) = { 1 σ 0 if c A < c A. 1 if c A c A, here ε > 0 and σ, b (0, 1). Time Allocation: Work and raising children. Raising a child: each child takes a fraction φ > 0 of time. If skilled child: addtional φ H > 0 of units of skilled (teacher s) time. If unskilled: if allowed to work, 0 φ L < φ units of unskilled labor. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 16 / 39

17 The Model Government Policies: Schooling subsidies & child labor restrictions. Education subsidies: The government subsidizes a fraction δ of the educational costs. Financing: proportional income tax τ(x ), s.t. budget balance. (δ, τ (X )) is a regressive tax program (taxes paid even if unskilled). Child labor policies: Limit on the number of hours a child can work. Equivalent to reductions on returns of child labor to φ g L : 0 φg L < φ L. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 17 / 39

18 Equilibrium Definition Let j {A, M, S} index the sectors and i, k {L, H} index the skill/school levels of the population. Given an initial state X 0 R 5 +, exogenous growth rates {γ A,γ M,γ S } and government policies {δ, φ g L }, an equilibrium in this economy is: (a) a low of motion Λ : R 5 + R 5 +, for the state X ; (b) price and wage functions p j : R 5 + R +, and w i : R 5 + R +, (c) labor allocations L j,i : R 5 + R +, and consumption and fertility decisions, c i,j : R 5 + R + and n i,k : R 5 + R + ; and (d) a tax function τ : R 5 + [0, 1], such that: (i) for any X in the current period, the state X in the next period is given by X = Λ (X ); (ii) given prices {w i ( ), p j ( )}, (1) the allocations {L j,i ( )} solve the firms problem; (2) the allocations {c i,j ( ),n i,k ( )} solve the household problem; (3) the goods and labor markets clear; (4) the budget constraint of the government balances (i.e. the tax rate is given by (1); and (iii) the transition Λ ( ) is given by the growth rates by {γ A,γ M,γ S } and the fertility and education decisions {n i,k ( )}. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 18 / 39

19 Equilibrium An adult of type i {H, L}: V i (X ) = subject to: max c A,c M,c S,n L,n H 0 { } U + β(n H + n L ) ε [n H V H + n L V L], p j c j + (1 δ)φ H w H (X )n H j {A,M,S } (1 τ(x )) {w i (X ) [1 φ(n H + n L )] +φ g L w L(X )n L }. For each adult, the values V H and V L given by the law of motion of X. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 19 / 39

20 Equilibrium Let n i,j (X ): number of children of type j that parents of type i have. Population law of motion: [ N H N L ] = [ nh,h (X ) n L,H (X ) n H,L (X ) n L,L (X ) ] [ NH N L ]. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 20 / 39

21 Equilibrium Total supply of skilled labor: L H (X )= [ 1 (φ + φ H )n H,H (X ) φn H,L (X ) ] N H φ H n L,H (X )N L. Total supply of unskilled labor: L L (X )= [ 1 φn L,H (X ) φn L,L (X ) ] N L +φ g L [n H,L(X )N H +n L,L (X )N L ]. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 21 / 39

22 Equilibrium Market-clearing labor markets: Skill workers: option to be unskilled: w H (X ) w L (X ). Unskilled labor market clearing: L MH (X ) + L SH (X ) L H (X ) (with= if w H (X ) > w L (X )), Skilled labor market clearing: L AL (X ) + L SL (X ) = L L (X ) + [L H (X ) L MH (X ) L SH (X )]. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 22 / 39

23 Equilibrium Market-clearing labor markets: As long as L L > c A Z A (N L + N H ), unique L H,M : [ 1 + ] α (1 b) L H,M = L H + c S (N L + N H ) [L H L H,M ] 1 α b and all {w i (X ), p j (X )} can be solved. [ Z S L L c ] 1 α. A (N L + N H ) Z A Budget Constraint of the Government: Balanced every period: τ(x ) = δφ H N H (X ) L H (X ) + φ H N H + L L(X ) w L(X ) w H (X ), (1) Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 23 / 39

24 Some Simple Analytics Let the costs of each type of child for parents i = L, H: Skilled child: Unskilled child: p i H = φw i + φ H (1 δ) w H ; p i L = φw i φ g L w L. An alternative way of formulating the problem: Adults chose The total expenditure E on child rearing ( E = p H + p L ); The fraction f that will be spent with skilled children Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 24 / 39

25 Some Simple Analytics The intratemporal problem: Given any w and E: Ū (w E ) = max c j U (c), s.t.: p j c j w E. j {A,M,S } The intertemporal problem: Given w, choose E w, f [0, 1]: { ( f V i = max Ū (w i E ) +βe 1 ε E,f ph i + (1 f ) ) [ fv H pl i ph i + (1 f )V L ]} pl i Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 25 / 39

26 Some Simple Analytics Assumed preferences: high-low skilled children are perfect substitutes. indifference curves (n H, n L ) are straight lines. Proposition For all {E, f }, that attains the maximum in (1), the solution is f = 0 or f = 1, i.e., the problem of adult has only corner solutions. So adults of each type choose only to have one type of child, that is, there are never both skilled and unskilled children within the same family. Only corner solutions: f i = 0 or f i = 1. n i,h, n i,l by comparing solutions with only one type of child. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 26 / 39

27 Some Simple Analytics Proposition An adult is indifferent between both types of children if, and only if the costs and utilities of children satisfy the following condition: V S p 1 ε S = V U p 1 ε U (2) If an adult is indifferent, the total expenditure on children does not depend on the type of child chosen. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 27 / 39

28 Some Simple Analytics If w H > w L : skilled children relatively cheaper for skilled parents: φw H + φ H (1 δ) w H φw H φ g L w L < φw L + φ H (1 δ) w H φw L φ g L w L. Parents of different skills cannot simulateneously be indifferent between the two types of children. In our calibration: equilibrium with intergenerational upward mobility. High-skilled parents: only high-skilled children. Low-skilled parents: indifferent between the two types of children. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 28 / 39

29 Impact of Policies Partial equilibrium: Given { w t i, pt j, τt } t=0. Relaxing child labor: higher φ g L : Higher V L, higher n L,L, partly due to higher V L. Higher subsidies to eduction: higher δ: Higher V H, higher n H,H, partly due to higher V H. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 29 / 39

30 Impact of Policies Redistributive: Given Relaxing child labor: higher φ g L : { } wi t, pt j, but τ balancing GBC: t=0 Higher V L, higher n L,L, partly due to higher V L and lower τ. Higher V H, higher n H,H, partly due to and lower τ. Higher subsidies to eduction: higher δ: Higher V H, higher n H,H, despite higher τ. Lower V L, lower n L,L, because higher τ (and hence lower V L ). Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 30 / 39

31 Calibration Preferences:. Parameter Value Target/Source β Doepke (2004) σ 0.5 " ε 0.5 " b 0.15 Herrendorf et al. (2011) c A by country/year share labor in agriculture c S by country/year match C S =Y S Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 31 / 39

32 Calibration Technology, Production, Education: Parameter Value Target/Source Korea Brazil α 0.40 PNAD φ Doepke(2004) φ H 0.04 " δ " φ g L, " φ g L, " Sectoral productivity paths: set as in Duarte and Restuccia (2010). Skilled labor defined as "some secondary education" (otherwise too few in 1960). Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 32 / 39

33 Results Growth rates: Calibrated model: large gaps in the Korea vs. Brazil growth rates. Growth: Output per worker Brazil Korea avg PWT: avg % 210% Model: 36% 232% Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 33 / 39

34 Results: Allocation of workers across sectors Initial: ; Final: Brazil Korea Variable Initial Final Initial Final Data Model Data Model Data Model Data Model N H L AL /L L L MH /L H L SL /L L L SH /L H Overall: model matches share of skilled labor in Korea and Brazil for both periods. Overestimates skilled workers (b.c.only skilled labor in M) Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 34 / 39

35 Results: Demographic Transitions Total Fertility rate Brazil Korea Data Model Data Model Model close to the data in Korea. Underestimates initial fertility in Brazil. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 35 / 39

36 Counterfactuals: Education Policies and Growth Brazil Korea Growth Growth Benchmark 36% Benchmark 232% Brazil: Korean policies 57% Korea: Brazilian policies 112% Korea with Brazilian policies: growth less than half. Brazil with Korean polices: growth 60% more. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 36 / 39

37 Counterfactual Policies: Fertility & Labor Allocation Brazil Korea Variable Benchmark Korean Policies Benchmark Brazilian Policies Initial Final Initial Final Initial Final Initial Final TFR N H L AL /L L L MH /L H L SL /L L L SH /L H Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 37 / 39

38 Counterfactual Policies: Fertility & Labor Allocation Brazil with Korean policies: Fertility rate first increases and then decreases. Large increase (70%) in skilled workers in second period. Korea with Brazilian policies: Fertility rate would remain high (1.9 instead of 1.2). Skilled workers would decreases by almost 50%. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 38 / 39

39 Concluding Remarks We incorporate fertility and education decisions to a structural transformation model. Calibrated the model to Korea and Brazil. Model matches growth paths, fertility and labor allocation across skills and sectors. We use the model to understand impact of education policies. Model helps explain the differences between Korea and Brazil. Education subsidies and restrictions on child labor: help explain Koreans growth, human capital accumulation and structural transformation. Extension: Three skill levels, i {U, L, H} and truly high skill services. Ferreira, Monge-Naranjo, Pereira. () Structural Transformation & Education 39 / 39