Structural Transformation and Growth

Size: px
Start display at page:

Download "Structural Transformation and Growth"

Transcription

1 Structural Transformation and Growth Diego Restuccia University of Toronto and NBER University of Oslo August 4-8, 27 Restuccia Macro Growth and Development University of Oslo / 49

2 Overview Structural transformation and aggregate productivity Basic facts and framework Importance of agriculture in rich/poor gaps A model of structural transformation Productivity, structural change, and aggregate outcomes What accounts for productivity differences? Restuccia Macro Growth and Development University of Oslo 2 / 49

3 Basic Facts of Structural Transformation As income/productivity rises over time for individual countries: Share of employment in agriculture falls Share of employment in industry first rises then falls Share of employment in services rises As the level of development is higher across countries Share of employment in agriculture is smaller Share of employment in industry features an inverted-u shape Share of employment in services is higher Restuccia Macro Growth and Development University of Oslo 3 / 49

4 Share of Employment Historical Share in total employment Agriculture Log of GDP per capita (99 international $) Belgium Spain Finland France Japan Korea Netherland Sweden United Kingdom United States Restuccia Macro Growth and Development University of Oslo 4 / 49

5 Share of Employment Historical Manufacturing Share in total employment Log of GDP per capita (99 international $) Belgium Spain Finland France Japan Korea Netherland Sweden United Kingdom United States Restuccia Macro Growth and Development University of Oslo 5 / 49

6 Share of Employment Historical Share in total employment Services Log of GDP per capita (99 international $) Belgium Spain Finland France Japan Korea Netherland Sweden United Kingdom United States Restuccia Macro Growth and Development University of Oslo 6 / 49

7 Share of Employment (WDI 6-) Share in total employment Agriculture Log of GDP per capita (99 international $) Restuccia Macro Growth and Development University of Oslo 7 / 49

8 Share of Employment (WDI 6-) Manufacturing Share in total employment Log of GDP per capita (99 international $) Restuccia Macro Growth and Development University of Oslo 8 / 49

9 Share of Employment (WDI 6-) Share in total employment Services Log of GDP per capita (99 international $) Restuccia Macro Growth and Development University of Oslo 9 / 49

10 Sectoral Labor Reallocation Over Time Turkey Mexico.9.9 Agriculture Industry Services Agriculture Industry Services Spain US.9 Agriculture Industry Services.9 Agriculture Industry Services Restuccia Macro Growth and Development University of Oslo / 49

11 Sectoral Labor Reallocation Over Time Greece Agriculture Industry Ireland Services Spain Canada Restuccia Macro Growth and Development University of Oslo / 49

12 Basic Framework Two goods produced in sectors: agriculture (a) and non-agriculture (n) Technologies Y a = A a N a Y n = A n N n Preferences and endowments: representative household, one unit of productive time each period, population size normalized to, per period utility (for now increasing in each argument) u(c a, c n ) Competitive markets for inputs and outputs, relative price of agriculture p a and wage w. Restuccia Macro Growth and Development University of Oslo 2 / 49

13 Equilibrium A competitive equilibrium is a set of prices (p a, w) and allocations (c a, c n ) for households and (N a, N n ) for firms such that: (i) Given prices (p a, w), (c a, c n ) solve the household s problem max u(c a, c n ) s.t. p a c a + c n = w (ii) Given prices (p a, w), (N a, N n ) solve the firm s problems max p a A a N a wn a ; max A n N n wn n (iii) Markets clear: = N a + N n ; c a = Y a ; c n = Y n Restuccia Macro Growth and Development University of Oslo 3 / 49

14 Characterization From the firm s problems w = A n ; p a = A n A a. Case A: Homothetic preferences u(c a, c n ) = a log(c a ) + ( a) log(c n ). Then, from the household s problem p a c a = aw; c n = ( a)w. Therefore, no structural change over time and across countries N a = a; N n = ( a). Restuccia Macro Growth and Development University of Oslo 4 / 49

15 Characterization Case B: Non-homothetic preferences u(c a, c n ) = a log(c a ā) + ( a) log(c n ). Then, from the household s problem p a c a = p a ā + a(w p a ā); c n = ( a)(w p a ā). A bit more algebra implies, N a = ( a) ā A a + a. Restuccia Macro Growth and Development University of Oslo 5 / 49

16 Characterization Special case (when a ) c a = ā N a = ā A a. Implications: Over time: growth in labor productivity in agriculture A a reduces the share of employment in agriculture N a Across development: countries with higher labor productivity in agriculture A a have lower share of employment in agriculture N a Taking logs implies that growth (differences) in agricultural labor productivity imply a change (difference) of the opposite sign in the share of employment in agriculture Simple structure represents a remarkable good fit with data (over time and across countries) Restuccia Macro Growth and Development University of Oslo 6 / 49

17 Characterization Case C: Non-unitary elasticity of substitution u(c a, c n ) = [ac ρ a + ( a)c ρ n] /ρ, ρ [, ]. Then, from the household s problem A bit more algebra implies, ac ρ a = ( a)[w p a c a ] ρ p a. N a = ( ) ( ) ρ. + a ρ A n ρ a A a Restuccia Macro Growth and Development University of Oslo 7 / 49

18 Characterization N a = ( ) ( ) ρ. + a ρ A n ρ a A a Implications ρ =, N a = a (homothetic case) ρ <, N a increases when growth in a higher than n (complementary goods) ρ >, N a falls when growth in a higher than n (substituable goods) More relevant for structural change between manufacturing and services. Generally productivity growth faster in manufacturing than services. Restuccia Macro Growth and Development University of Oslo 8 / 49

19 Summarizing Productivity key for structural transformation. What accounts for productivity differences and growth? We will come back to this. Application: Structural change and aggregate outcomes. Mainly drawn from Duarte and Restuccia (2), The Role of the Structural Transformation in Aggregate Productivity, Quarterly Journal of Economics, Volume 25 (), February, pp Restuccia Macro Growth and Development University of Oslo 9 / 49

20 Aggregate Productivity Time paths of aggregate labor productivity (GDP per hour) of individual countries relative to the United States show a variety of experiences: catch-up, slow-down, stagnation, and decline Restuccia Macro Growth and Development University of Oslo 2 / 49

21 Relative GDP per Hour some countries Ireland Austria Spain Italy Sweden Denmark Canada New Zealand Restuccia Macro Growth and Development University of Oslo 2 / 49

22 Structural Transformation and Aggregate Productivity The structural transformation may matter for aggregate productivity if there are systematic differences in sectoral productivity across countries Duarte and Restuccia (2) ask whether sectoral productivity differences can account for the structural transformation and the time paths of aggregate productivity across countries Restuccia Macro Growth and Development University of Oslo 22 / 49

23 Duarte and Restuccia (2) Consider a simple general equilibrium model of the structural transformation and calibrate it to the experience of the United States Use the model to: Measure sectoral labor productivity differences across countries at a point in time Assess the quantitative role of sectoral labor productivity differences on the structural transformation and aggregate productivity outcomes across countries Restuccia Macro Growth and Development University of Oslo 23 / 49

24 Preview of Results Find large and systematic differences in sectoral labor productivity across countries both at a point in time and over time Productivity differences are larger in agriculture and services than in industry Over time productivity gaps substantially reduced in agriculture and industry but not services Differences in sectoral productivity account for the processes of structural transformation and for aggregate productivity experiences across countries: catch-up, slow-down, stagnation, and decline Restuccia Macro Growth and Development University of Oslo 24 / 49

25 Data Output per hour, obtained by combining data from Heston et al. (26) PWT6.2 and Groningen Growth and Development Center Disaggregated output per hour and hours worked for agriculture, industry, and services from a variety of sources Panel data for 29 countries, from 956 to 24 for most countries Data are trended using the Hodrick-Prescott filter (λ = ) Restuccia Macro Growth and Development University of Oslo 25 / 49

26 Some Facts Common process of structural transformation across countries Falling share of hours in agriculture Hump-shaped pattern of the share of hours in industry Increasing share of hours in services Lag differences in economic structure across countries Substantial dispersion in growth rates of sectoral labor productivity Restuccia Macro Growth and Development University of Oslo 26 / 49

27 Sectoral Labor Productivity Growth Services Industry Agriculture VEN BOL VEN BOL VEN BOL DNK ESP ITA NLD FRA BEL AUT ARG CANAUS CHL GBR FIN IRL KOR SWE NOR USA GRC JPN BRA PRT CRI NZL MEXCOL TUR KOR BEL AUT GBR NOR IRL SWE FIN JPN NLD AUS ARG CAN CHL DNK FRA ESP ITA PRT TURGRC NZL COL USA BRA CRI MEX JPN BEL AUT DNK FRA IRL NLDGBR NOR ITAPRT KOR AUS USA TURGRC ESP FIN NZL ARG CAN COLCHL SWE CRI MEX BRA Annualized Growth Rate of Aggregate Labor Productivity Restuccia Macro Growth and Development University of Oslo 27 / 49

28 Model General equilibrium model of the structural transformation following Rogerson (28) Three sectors: Agriculture, Industry, and Services The model has two forces of structural transformation: Non-homothetic preferences Differential productivity growth between industry and services Restuccia Macro Growth and Development University of Oslo 28 / 49

29 Production In each period, three goods are produced: agriculture (a), industry (m), and services (s) Technologies Y i = A i L i, i {a, m, s} A i labor productivity in sector i L i total labor hours in sector i Labor and goods markets are competitive Restuccia Macro Growth and Development University of Oslo 29 / 49

30 Preferences Representative household with preferences over consumption: u(c a,t, c m,t, c s,t ) = a log(c a,t ā) + ( a) log(c t ) Preferences are non-homothetic: Subsistence level of agricultural consumption ā Income elasticity of services consumption greater than one, c t = [ bc ρ m,t + ( b)(c s,t + s) ρ] ρ, s >, b (, ), ρ < Restuccia Macro Growth and Development University of Oslo 3 / 49

31 Equilibrium A competitive equilibrium is a set of prices {p a, p m, p s } and allocations {c a, c m, c s } for the household and {L a, L m, L s } for firms such that: Given prices, firm s allocations {L a, L m, L s } maximize profits Given prices, household s allocations {c a, c m, c s } maximize utility subject to the budget and non-negativity constraints Markets clear: Labor market: L a + L m + L s = L Goods markets: c a = Y a, c m = Y m, c s = Y s Restuccia Macro Growth and Development University of Oslo 3 / 49

32 Characterization Prices of goods: p i = w A i Preference for agricultural goods: L a = ( a) ā ) + a (L + sas A a Consumption of industry and services: ( [ ( ) b ρ ] ) Am ρ + L m = L L a + s b A s A s Restuccia Macro Growth and Development University of Oslo 32 / 49

33 Calibration Strategy: Calibrate a benchmark economy to be consistent with the process of structural transformation in the United States We focus on data for the United States from 956 to 24 Period = year Restuccia Macro Growth and Development University of Oslo 33 / 49

34 Calibration II Parameter Value Target U.S. Data A i,56. Normalization {A a,t } t= { } Productivity growth in agriculture {A m,t } t= { } Productivity growth in industry {A s,t } t= { } Productivity growth in services a. Long-run share of hours in agriculture ā. Share of hours in agriculture 956 s 9 Share of hours in industry 956 b.4 Share of hours in industry ρ -.5 Aggregate productivity growth Restuccia Macro Growth and Development University of Oslo 34 / 49

35 Hours across Sectors United States.7 Share of Hours.5.3 Ag. Data Ag. Model Ind. Data Ind. Model Svc. Data Svc. Model Years Restuccia Macro Growth and Development University of Oslo 35 / 49

36 Quantitative Experiment Step : use the model to measure sectoral relative labor productivity in the first period across countries. Three targets: shares of hours in agriculture and industry relative aggregate labor productivity Step 2: use data on growth rates of sectoral labor productivity in each country model implies time paths for sectoral hours and aggregate productivity for each country Step 3: counterfactual exercises Restuccia Macro Growth and Development University of Oslo 36 / 49

37 Relative Sectoral Productivity first period Agriculture Industry Services Quintile of Aggregate Productivity (first year) Restuccia Macro Growth and Development University of Oslo 37 / 49

38 Rel. Sect. Prod. first and last period Agriculture.5 First year Last year Industry Services Quintile of Aggregate Productivity (first year) Restuccia Macro Growth and Development University of Oslo 38 / 49

39 Structural Transformation across Countries Model generates time paths for shares of hours and output across sectors, aggregate labor productivity, and relative prices for each country Overall, the model reproduces the salient features of the data across countries Restuccia Macro Growth and Development University of Oslo 39 / 49

40 Model vs Data Changes Data Share of Hours in Agriculture AUS BEL NZL GBR AUT USA NLD DNK ARG FRA SWE CAN CHL VENOR JPN COL CRI ESP ITA IRL PRT MEX FIN BRA BOL TUR GRC KOR 4 2 Share of Hours in Industry 2 TUR IRL KOR CRI GRC BOL ITA PRTBRA COL JPN ESP FIN VEN CHL 2 NOR CANDNK ARG AUT USAFRA NZL SWE AUS NLD BEL GBR Data Share of Hours in Services NZL FIN GRC ESP BRA NORPRT ITA BOL FRA ARG DNK BEL AUS AUT CAN GBR SWE KOR NLD VEN COL MEX IRL TUR JPN CHLCRI USA Model Relative Aggregate Productivity 6 NOR AUT IRL ITA BEL 4 JPN ESP FIN KOR PRT GRC 2 DNK NLD GBR SWE TUR BRA CHL AUS COL USA MEX CAN BOL CRI NZL ARG 2 VEN Model Restuccia Macro Growth and Development University of Oslo 4 / 49

41 Model vs Data in Relative Prices 4 Relative Price of Agriculture 2 Data 2 4 ESP BOL DNK ITA VEN GRCARG JPN CAN FIN TUR KOR BRA COL CRIMEX SWE AUS CHL FRA NZL BEL GBR AUT NOR NLD USA PRT IRL Relative Price of Services Data 4 2 BOL 2 VEN BEL PRT GRC ITA SWE AUTFIN USA JPN NZL TUR AUS FRAGBR ESP CRI ARG CAN MEX DNKBRA NLD COL NOR CHL IRL KOR Model Restuccia Macro Growth and Development University of Oslo 4 / 49

42 Counterfactuals We construct counterfactuals to assess the quantitative role of sectoral productivity differences on the structural transformation and aggregate productivity experiences across countries We focus on two sets of counterfactuals: illustrate the mechanics of sectoral productivity growth differences explain aggregate productivity experiences across countries Restuccia Macro Growth and Development University of Oslo 42 / 49

43 Mechanics Sectoral Productivity Growth Counterfactual Counterfactual Counterfactual Counterfactual 5 () γ a = Δ(L a /L) Δ(L m /L) Δ(L s /L) Δ(y/y US ) Model 2 2 (2) γ m = Δ(L a /L) Δ(L m /L) Δ(L s /L) Δ(y/y US ) Model (3) γ s = Δ(L a /L) Δ(L m /L) Δ(L s /L) Δ(y/y US ) Model (4) γ i =γ US Δ(L a /L) Δ(L m /L) Δ(L s /L) 5 Δ(y/y US ) Model Restuccia Macro Growth and Development University of Oslo 43 / 49

44 Sectoral Productivity Patterns and Cross-Country Experiences We assess the role of specific labor productivity patterns across sectors in explaining cross-country aggregate productivity experiences. role of substantial catch-up in agriculture and industry across countries role of low productivity and lack of catch-up in services Restuccia Macro Growth and Development University of Oslo 44 / 49

45 Change in Relative Aggregate Productivity All Catch-up Decline countries countries countries Model Counterfactual: () γ i = γi US (a) Agriculture (b) Industry (3) Catch-up in Services Restuccia Macro Growth and Development University of Oslo 45 / 49

46 Importance of Industry Catch-up Counterfactual 3 2 VEN MEX CRI TUR BOL BRA KOR AUS NZLCHL COL ARG CAN GRC PRT ITA FIN FRA DNKJPN IRL ESP NLD GBR SWE AUTNOR BEL Model Restuccia Macro Growth and Development University of Oslo 46 / 49

47 Importance of Services Argentina Colombia Greece Norway Australia 98 2 Costa Rica Italy Portugal Canada Spain Mexico Sweden Chile Finland Netherlands Turkey Restuccia Macro Growth and Development University of Oslo 47 / 49

48 Importance of Services II Relative Labor Productivity in Services last year.2 ARG GRC CHL PRT VEN NOR AUT FRA DNK BEL GBR NLD CAN SWE AUS ITA FIN NZL JPN ESP IRL COL TUR MEX CRI BRA KOR BOL.2 Relative Aggregate Labor Productivity last year Restuccia Macro Growth and Development University of Oslo 48 / 49

49 Summarizing: Sectoral Perspective There are large and systematic differences in sectoral productivity across countries A sectoral analysis is important in understanding aggregate outcomes low productivity in agriculture key productivity catch-up in industry explains about 5 percent of the gains in aggregate productivity across countries low productivity and lack of catch-up in services explains slowdown, stagnation, and decline in relative aggregate productivity Productivity patterns (levels and growth) key in accounting for structural transformation and aggregate outcomes What then accounts for these productivity differences? Restuccia Macro Growth and Development University of Oslo 49 / 49