Material on Growth Part II: Productivity and Technology

Transcription

1 Material on Growth Part II: Productivity and Technology David N. Weil Brown University and NBER Prepared for AEA Continuing Education, January 2014

2 Taking Stock We see these huge income dierences among countries We see a good degree of historical persistence We have all sorts of theories Maybe we can learn more by looking directly at income dierences Key tool will be Development Accounting (cross-sectional analogue of Solow-style growth accounting)

3 Productivity and Factor Accumulation

4 Implementing Development Accounting We need a production function (functional form and parameters) measurements of the relevant factors of production (physical and human capital)

5 Measuring Physical Capital Perpetual Inventory Method K i,t+1 = (1 δ)k i,t + I i,t K i,1970 will be some sort of informed guess, like 2 Y i,1970 δ most frequently taken to be 6% potential problems: mis-measurement of investment (see below) cross-country dierences in depreciation rates (i.e. old cars in developing countries)

6 Measuring Human Capital Data on educational attainment of working age population widely available (Barro and Lee, 2013) Mincerian approach allows for conversion into years of schooling

7 The Mincerian Approach Aggregate Production: Y i = A i K α i H 1 α i Human Capital Aggregate: H i = L j=1 h ij where i indexes countries and j indexes individuals Return to one unit of human capital: w i = dy i dh i Wages for individual: ln(w i,j ) = ln(w i ) + ln(h i,,j ) + η i,j Human capital and schooling: ln(h i,j ) = φ(s i,j ) = βs i,j = (1 α)a i ( Ki H i ) α We can recover β by regressing individual log wages on years of schooling (s i ) within a country Once we know β, we can construct H i based on aggregate data on schooling

8 Two Approaches to Human Capital in the Production Function Approach Mincer Mankiw, Romer, Weil (1992) Production function Y=F(K, hl) Y=F(K, H, L) in general form Production function Y = AK α (hl) 1 α Y = AK α H β L 1 α β in CD, CRS form Salient human capital and human capital and Characteristic raw labor are raw labor really the same thing are dierent things return to human return to human capital As H increases, return capital is invariant to quantity to human capital declines big advantage can learn mapping dynamics are much easier from education to easier to to analyze if h from micro data we treat H like K nagging worry try to move a piano Hard to believe that with one Ph.D. vs raw labor's share four uneducated guys of income is constant

9 Production Setup Well known property of Cobb-Douglas production function that if factors are paid their marginal products, the capital share of national income is invariant to the K/L ratio Holds for US time series Gollin (2002) and Bernanke and Gurkaynak (2002) nd something similar in cross section We take this as permission to use Cobb-Douglas

10

11 Development Accounting Results Productivity dierences are really large!

12 Assessing the Importance of Factor Accumulation vs. Productivity I will follow Caselli (2005). Other related treatments are Klenow and Rodriguez-Clare (1997) and Hall and Jones (1999). production function in per-worker terms: dene y = Ak α h 1 α y KH = k α h 1 α This is the part of output due to factor accumulation. So... y = A y KH

13 Measures of Success of the Factor-Only Model Variance decomposition: var(ln(y)) = var(ln(a)) + var(ln(y KH )) + 2cov(ln(A), ln(y KH )) Success 1 = var(ln(y KH)) var(ln(y)) Success 2 = y 90 KH /y 10 KH y 90 /y 10 var(ln(y)) y 90 /y var(ln(y KH )) ykh /ykh 7 success success ==> to some extent Caselli's approach overstates the role of factor accumulation because of induced factor accumulation to see this: imagine that all countries had the same saving rate

14 Biases in These Calculations Pritchett CUDIE is not Capital (2000) Investment overstated in poor countries; gaps in physical capital are thus larger than indicated in the data. Education quality (Hanushek) Rich countries have higher test scores; gaps in human capital are thus larger than indicated in the data. In both cases, productivity gaps are over-stated. Neither of these will overturn the basic conclusion of development accounting that productivity dierences are very important

15 Weil Framework

16 Productivity Growth and Cross-Sectional Dierences We are comfortable with thinking about productivity growth over time in the leading countries as being due to technological progress (i.e. Jones analysis) we observe new inventions that clearly raise productivity productivity growth in catching up countries is some combination of technological progress and eciency improvements not clear why we should believe that eciency is really constant in lead countries like US. could be rising (less discrimination leads to more ecient allocation of labor: Hsieh, Hurst, Jones, and Klenow, 2013: 0.2% per year rise in labor productivity since 1960) could be falling (growth of government, larger tax wedges (Prescott), more regulation, etc.) What about productivity dierences among countries at a point in time?

17 A Simple Framework for Examining Productivity Dierences Productivity = Technology Efficiency A = T E A India,2013 A US,2013 = T India,2013 T US,2013 E India,2013 E US,2013, Let g be the growth rate of technology in the US and let B be the technology gap between India and the US, measured in years T US,2013 = T US,2013 B(1 + g) B = T India,2013 (1 + g) B T India,2013 T US,2013 = (1 + g) B

18 Calculation of Eciency Gap India vs. US E India,2013 E US,2013, = A India,2013 A US,2013 T India,2013 T US,2013 = A India,2013 A US,2013 (1 + g) B Data : A India A US = 0.31 g = 0.54% (this is US average ) Years India Lags T India /T US E India /E US US in Technology (B)

19 Evidence on Technological Backwardness This suggests that poor countries can't be that far behind technologically; gap in A must be due to eciency. caveat: Appropriate Technology (Basu and Weil, 1999; Acemoglu and Zilibotti, 2001). technological progress is not a general shift in the production function, but is specic to certain factor intensities. And maybe certain institutions as well.

20 Neutral Technological Change

21 Capital Biased Technological Change

22 What is this Eciency Thing Anyway?

23 A Tentative Taxonomy of Ineciency 1 unproductive activities (rent seeking, theft, hiring guards, insurrection) 2 idle resources (cyclical unemployment, featherbedding) 3 misallocation of factors among sectors (formal-informal, regional, family farms, nance in the US) 4 misallocation of factors among rms (Hsieh and Klenow, 2007) 5 technology blocking (patent trolls, margarine)

24 A Word from Our Sponsor Economic Growth David N. Weil Third Edition

25 What Underpins Low Eciency in Poor Countries? institutional framework (security of property, enforcement of laws) trade restrictions (legal or physical) barriers to mobility (of factors) monopolies government ownership of rms poorly functioning nancial system tax wedges etc. This is economics of Adam Smith rather than Malthus, Solow, or Technology (Arrow, P. Romer, Aghion and Howitt, Jones, Grossman and Helpman).