Lecture 3: Quantifying the Role of Credit Markets in Economic Development

Transcription

1 Lecture 3: Quantifying the Role of Credit Markets in Economic Development Francisco Buera UCLA January 18, 2013

2 Finance and Development: A Tale of Two Sectors Buera, Kaboski & Shin 2011

3 Development Facts 1. Huge differences in economic development across countries. 2. Development explained by TFP differences.

4 Hall and Jones (1999)

5

6 Development Facts 1. Huge differences in economic development across countries. 2. Development explained by TFP differences. 3. Poor countries are particularly unproductive in manufacturing sector.

7 Sectoral Productivity

8 Productivity Interpretation of Relative Prices Y j = A j Kj L 1 j ) p j p j 0 = A j 0 A j Herrendorf and Valentinyi (2007), Hsieh and Klenow (2007)

9 Hsieh and Klenow (2007)

10 Hsieh and Klenow (2007)

11 Hsieh and Klenow (2007)

12 Hsieh and Klenow (2007)

13 Hsieh and Klenow (2007)

14 Hsieh and Klenow (2007)

15 Early on, investment seemed more distorted Hsieh and Klenow (2007)

16 Hsieh and Klenow (2007)

17 Development Facts 1. Huge differences in economic development across countries. 2. Development explained by TFP differences. 3. Poor countries are particularly unproductive in manufacturing sector.

18 Development Facts 1. Huge differences in economic development across countries. 2. Development explained by TFP differences. 3. Poor countries are particularly unproductive in manufacturing sector. 4. Large differences in scale across sectors.

19 Scale and Finance Needs by Sector Workers per Establishment Workers per Enterprise External Dependence Capital Share US OECD US US Manufacturing Services

20 Scale by Sector Workers per Estab. Workers per Enterp. Ext. Fin. Dep. Ratio Capital Share (lr)2-4 U.S. OECD avg. U.S. U.S. Mfg. Cons. (m) Serv (s) Equip. Inv. (e) Constr. Inv. (c) Trad. (m+e) Non-trad. (s+c) Source for Capital Share: Valentinyi and Herrendorf (2008)

21 Development Facts 1. Huge differences in economic development across countries. 2. Development explained by TFP differences. 3. Poor countries are particularly unproductive in manufacturing sector. 4. Large differences in scale across sectors.

22 Development Facts 1. Huge differences in economic development across countries. 2. Development explained by TFP differences. 3. Poor countries are particularly unproductive in manufacturing sector. 4. Large differences in scale across sectors. 5. Underdeveloped nancial/credit markets in less developed countries.

23 Private Credit to GDP vs. Per-Capita Income Private Credit to GDP IND KEN BRA FRA GRE USA ,000 20,000 30,000 40,000 50,000 60,000 per capita income (US$, PPP)

24 Models of Financial Frictions Literature Review: Theoretical Literature Individual poverty traps: Banerjee & Newman (1993), Galor & Zeira (1993), Buera (2007) non-convexities + financial frictions individual poverty traps Aggregate poverty traps/ Multiple equilibria: Multiple stationary wealth distributions: Banerjee & Newman (1993), Galor & Zeira (1993), Matsuyama (2000, 2006) Multiple stationary equilibrium w/ unique wealth distribution given prices: Piketty (1997) Unique equilibria, Kuznets dynamics (slower convergence?) Richer models tend to have unique equilibria: Mookerjee & Ray (2002, 2003), Mookerjee & Napel (2007) Explain joint dynamics of development & inequality: Aghion & Bolton (1996), Lloyd-Ellis & Bernhardt (2001)

25 Models of Financial Frictions Literature Review: Quantitative Literature Transitional Dynamics: Gine & Townsend (2004), Jeong & Townsend (2007) (exogenous savings) Buera & Shin (forthcoming), Kahn & Thomas (2011) (forward-looking savings, self-financing) Stationary Equilibrium/Long-run Effects: Caselli & Gennaioli (2005), Amaral & Quintin (2008), Greenwood et al. (2008) (exogenous savings/ no self-financing) Buera et al. (2011), Buera and Shin (2011), Midrigan & Xu (2012) (forward-looking savings, self-financing)

26 Goal of the Paper Construct a quantitative model: where scale is the main difference across sectors; that matches key features of size distribution of establishments across sectors (average size) and within sectors (thick right tail). Quantify the effect of credit frictions on: per-capita income, sectoral TFP, establishment size distribution, K/Y ratios.

27 Preview of Results Financial frictions reduce per-capita output by as much as 50%; lower the relative TFP of manuf. sectors; increase the relative price of manuf. sector, explaining 80% of the relative price-income relationship; decrease K/Y ratios, when severe decrease scale in service sector relative to manuf. sector.

28 Model Two sectors: p = (p S ; p M ), with different xed costs, S < M. Heterogenous entrepreneurial ability/productivity and wealth. Endogenous credit frictions: limited enforcement.

29 Model: Plant Technology Lucas (1978), Rossi-Hansberg and Wright (2006) Fixed cost S < M (in units of sector output) Period technology: f (z; k; l) = zk l z: entrepreneurial productivity k: capital input l: labor input + < 1

30 Model: Preferences Households maximize u (c t ) = 1 1 X 1 U (c) = E 0 t u (c t ) t=0 c 1 " S;t + (1 ) c 1 " M;t 1 1 "

31 Model: Timing Sector and Occupation Choice

32 Model: Timing Sector and Occupation Choice

33 Model: Individual Problem Workers' Bellman Equation v w (a; z) = max c;a 0 0 u (c) + E zv a 0 ; z 0 pc + a 0 w + (1 + r) a

34 Model: Individual Problem Entrepreneurs' Bellman Equation v j (a; z) = max c;a 0 ;k;l u (c) + E zv a 0 ; z 0 pc + a 0 p j f (z j ; k; l) Rk wl (1 + r)p j j + (1 + r) a k k j (a; z; )

35 Model: Endogenous Rental Limits where max c;a 0 ;l u (c) + E z v a 0 ; z 0 v j;def v j;def = max c;a 0 ;l u (c) + E zv a 0 ; z 0 pc + a 0 (1 ) [p j f (z j ; k; l) wl + (1 )k]

36 Model: Endogenous Rental Limits max u(c) + βe z v ( a,z ) v j,def p j f (z j,k,l) Rk wl (1 + r)p j κ j + (1 + r)a (1 φ)[p j f (z j,k,l) wl + (1 δ)k] k k j (a,z;φ)

37 Rental Limit k(a,z,φ)/w z max z a/w

38 Stationary Competitive Equilibria G (a; z), policies o (a; z), c (a; z), a 0 (a; z), k (a; z), l (a; z) and prices w, r, p such that: Allocations solve individuals' problems given prices; Labor, credit and goods markets clear; G (a; z) satises G (a; z) = (w;r;p) [G (a; z)]

39 Pareto Distribution of Productivity z j z (+1) j, z S? z M Thick right tail within each sector.

40 Pareto Distribution of Productivity z j z (+1) j, z S? z M Thick right tail within each sector. Cobb-Douglas benchmark.

41 First Best Benchmark: Results Size Distribution of Establishments Sector j: h i Pr ~lj > l = l (^zj ) l (1 )

42 First Best Benchmark: Results Size Distribution of Establishments Sector j: h i Pr ~lj > l = l (^zj ) l (1 ) Average employment per establishment l j : lmls = p M M + w p S 0 S + w

43 Perfect Credit Benchmark: Results Sectoral (Net) Production Function Y j (K j ; L j ; N) A j N 1= ++1= K ++1= j L ++1= j

44 Perfect Credit Markets 5 log(a) (Wealth) 0 Workers Entrepreneurs log(z) (Entrepreneurial Ability)

45 Imperfect Credit Markets: Partial Equilibrium log(a) (Wealth) Workers Entrepreneurs log(z) (Entrepreneurial Ability)

46 Imperfect Credit Markets: General Equilibrium log(a) (Wealth) Workers Entrepreneurs log(z) (Entrepreneurial Ability)

47 Imperfect Credit Markets: Extensive & Intensive Margin 5 Unconstrained log(a) (Wealth) 0 5 Workers Constrained log(z) (Entrepreneurial Ability)

48 Illustrating the Distribution of Productivity 5 log(a) (Wealth) 0 5 z 90 z 95 z 99 z max log(z) (Entrepreneurial Ability)

49 Illustrating the Distribution of Wealth Conditional on Productivity log(a) (Wealth) a 99.9 z a 99 z a 90 z a 50 z a 89 z log(z) (Entrepreneurial Ability)

50 Poverty Trap 5 log(a) (Wealth) log(z) (Entrepreneurial Ability)

51 Poverty Trap (cont d) 5 log(a) (Wealth) log(z) (Entrepreneurial Ability)

52 Policy Functions 10 8 a /w z min z 90 z 95 z 99 z max a/w

53 Policy Functions (cont d) 2 1 log(a ) z min z 90 z 95 z 99 z max log(a)

54 Dynamic of Capital Input 400 Poor (a 0 =0) 400 Rich (Top 5% a 0 ) z k/w 200 z k/w 200 z 99 z z 95 0 z 95 z Years z Years

55 Empirical Strategy 1. Choose technology (; ; j ) and productivity process (; ) to match US data on the size distribution and dynamics of establishments and income concentration. 2. Choose nancial frictions () to match cross-country variation in external nance to GDP. 3. Use cross-country data on the size distribution of establishments to test additional implications of theory.

56 Calibration Target Moments US Data Model Parameter Top 10% employment share η = 4.84 Top 5% earnings share α + θ = 0.79 Average scale in services κ S = 0.00 Average scale in manufacturing κ M = 4.68 Exit rate γ = 0.89 Manufacturing Share of GDP ψ = 0.91 Interest rate β = 0.92

57 Calibration: Establishment Size Distribution

58 Quantitative Strategy Target US Data Model Parameter top 10% employment share η US = 4.84 top 5% income share α+θ = 0.79 Exit rate γ = 0.89 Interest rate β = 0.92 Target Indian Data Model Parameter top 10% employment share η IND = 5.56 Ext. fin./gdp φ IND = 0.08

59 Contract Enforcement Data Country Time to recover Cost Recovery rate (years) (% of estate) (cents on the dollar) U.S.A France Greece Brazil China India Kenya

60 Recovery Rate of Loans vs. Per-Capita Income recovery rate (cents on a dollar) CHN KEN IND BRA FRA GRE USA UAE ,000 20,000 30,000 40,000 50,000 60,000 per capita income (US$, PPP)

61 Time to Recover vs. Per-Capita Income 8 IND Time to recover (years) KEN BRA CHN GRE FRA USA UAE ,000 20,000 30,000 40,000 50,000 60,000 per capita income (US$, PPP)

62 Private Credit to GDP vs. Per-Capita Income Private Credit to GDP IND KEN BRA FRA GRE USA ,000 20,000 30,000 40,000 50,000 60,000 per capita income (US$, PPP)

63 Preview of Results Financial frictions reduce per-capita output by as much as 50%; lower the relative TFP of manuf. sectors; increase the relative price of manuf. sector, explaining 80% of the relative price-income relationship; decrease K/Y ratios, when severe decrease scale in service sector relative to manuf. sector.

64 Per-Capita GDP, TFP, K/Y

65 Misallocation of Capital and Talent

66 Relative Prices Relative TFP

67 Establishment Productivity and Size

68 Additional Testable Implications Signicant scale differences across sectors Sector-level scales are differentially affected by nancial frictions.

69 Scale Differences: US v. Mexico Comparable industry classication (NAICS), at least for manufacturing. US: Economic Census 2002 Mexico: 2004 Economic Census (non-xed establishments/rms not included) ENAMIN 2002 (all small establishments)

70 Scale Differences: US v. Mexico

71 Conclusions Financial frictions are quantitatively important (factor of 2) for GDP/capita

72 Conclusions Financial frictions are quantitatively important (factor of 2) for GDP/capita Scale differences help understand the impact of nancial frictions on sectoral productivity. biggest TFP effects on large scale/manufacturing sector distorts relative prices and capital accumulation entry and self-nance are quantitatively important Size distribution varies systematically across countries and across sectors.

73 Alternative Specications for Scale Differences 1. Setup cost instead of xed cost ) larger effects

74 Alternative Specications for Scale Differences 1. Setup cost instead of xed cost ) larger effects 2. S + S < M + M ) different factor shares across sectors

75 Alternative Specications for Scale Differences 1. Setup cost instead of xed cost ) larger effects 2. S + S < M + M ) different factor shares across sectors 3. Monopolistic competition with different demand elasticities across sectors

76 Alternative Specications for Scale Differences 1. Setup cost instead of xed cost ) larger effects 2. S + S < M + M ) different factor shares across sectors 3. Monopolistic competition with different demand elasticities across sectors 4. One sector model ) smaller effects

77 Alternative Specications for Scale Differences 1. Setup cost instead of xed cost ) larger effects 2. S + S < M + M ) different factor shares across sectors 3. Monopolistic competition with different demand elasticities across sectors 4. One sector model ) smaller effects 5. Two-period model ) exaggerates effects