Institutional Development, Organizational Capitaland Misallocation

Ping WangDepartment of Economics

Washington University in St. Louis

February 2017

1

A. Introduction

It is not until recent that macroeconomists have devoted effort toward understandingthe role of institutions and organizations played in the process of economicdevelopment. This relatively thin but important literature includes:

! Institutional development: " classic: North (1990), Rogoff (1990)" voting and political equilibrium: Grossman-Helpman (1993), Perotti (1993),

Alesina-Spolaore (1997), Bolton-Roland (1997), Saint-Paul (2002),Rotemberg (2003)

" new wave: Acemoglu-Robinson (2000, 2001, 2008), Acemoglu-Johnson-Robinson (2001, 2002, 2005), Galor-Moav-Vollrath (2009), Palivos-Wang-Yip(2010)

! Organizational capital: " classic: Lucas (1978), Prescott-Visscher (1980), Jovanovic (1982)" new wave:

- Atkenson-Kehoe (2005, 2007): theory and measurement- Extensions: Samaniego (2006) and Burstein & Monge-Naranjo (2009) –

organizational capital and productivity- Grossman-Helpman (2002, 2004), Grossman & Esteban Rossi-Hansberg

2

(2008), Lai-Riezman-Wang (2009), Riezman-Wang (2009): organizationof trade

- Ghatak-Morellib-Sjostromc (2007), Newman (2007), Antunes-Cavalcanti-Villamil (2008), Jiang-Wang-Wu (2010): entrepreneurialcapital

- Lu-Peng-Wang (2012): organization of production and trade! Integrating institutions and organizations – misallocation, growth and

development: Hsieh-Klenow (2009), Uras (2010), Hopenhyen (2012), Lee-Hsieh-Wang (2014), Uras-Wang (2014), Buera-Shin (2015), Hsieh-Lee-Uras-Wang (inprogress)

B. The Importance of Institutions and Organizations

! Institutional factors " affect laws and regulations under which households and firms function" shape the incentives individuals have for various decision-making

! Organizational structures" affect the operations of firms and government agencies" influence the efficiency of production and the effectiveness of public policy

3

C. Societal Hierarchy and Institutions: Acemoglu and Robinson (2008)

! While Acemoglu-Johnson-Robinson (2005) provide convincing empiricalevidence and informal arguments on the role of institutions played in thedevelopment of Western European Atlantic traders, a formal modelingframework has been absent

! This paper makes crucial progress toward filling the gap by constructing a modelto study how changes in political institutions can lead to subsequent changes ineconomic institutions

1. The Model

! Total population L within which there is a small elite (E) group of size M with theremaining as general citizen (C)

! Preference: , linear over private/public goods, h {E, C}

! In each period, only 2 types of public goods are provided:" gt+j = e (elite type) => GE

t+j = γE > 0, GCt+j = 0

" gt+j = c (citizen type) => GEt+j = 0, GC

t+j = γC > 0

4

! Ricardian technology: each citizen owns one unit of labor (supplied inelastically),capable of producing A units of good

! Institutions: τt {e, c} (pro-elite or pro-citizen)" τt = c => wc = A, Rc = 0" τt = e => we = λ(1-δ)A, Re = (1-λ)(1-δ)AL/M

where δ = inefficiency loss due to labor repression under e-institution" labor wage and elite rent differentials under the two institutions:

- Δw = wc - we = [1-λ(1-δ)]A > 0- ΔR = Re - Rc = (1-λ)(1-δ)AL/M > 0- since L/M is very large, ΔR >> Δw

! Political regimes s {N, D} (nondemocracy/monarchy or democracy)! De facto political power depends on the investment in power-gaining:

" elite:

" citizen: , with

- ω iid, drawn from a given distribution F with support (ω, ), ω < 0, andsingle-peaked density

- η measuring citizens’ de jure power in democracy" indicator of power π {e, c}: π = e iff , and = c otherwise

5

! Timing of events:" the group in power at t decides gt {e, c}" each elite i E and each citizen i C choose their investment in gaining

power and PEt is determined

" ω is drawn from F and PCt is determined

" if π = e, a representative elite chooses current institution and future regime(τt, st+1); otherwise, a representative citizen chooses

" given τt, Rt and wt are determined and consumption takes place! Symmetric Markov perfect equilibrium (MPE):

" equilibrium strategies are mappings from payoff-relevant states s {N, D}" all agents in the same group behave symmetrically

! Symmetry implies:"" PC = " pr(e in power) =

! Backward induction within each stage implies the following best responses:" g(N) = e, g(D) = c" τ(e) = e, τ(c) = c" s(e) = N, s(c) = D

6

! Values under N (using one-step-ahead deviation principle a la Fudenberg-Tirole1994):" elite:

" citizen:

where is theconditional probability for e in power given all elite members choosing θE

and all other citizens choosing θi

! First-order conditions for power-spending θ under N:" elite:" citizen:

where and measure value differentials between two political regimes

! Values under D and the associated first-order conditions:" elite:

(FOC)

7

" citizen:

(FOC)

2. Results

! Power-gaining investment: Any symmetric MPE involves: " θC(N) = θC(D) = 0: this is because the elite group has much larger gains from

power than citizens (ΔR >> Δw), implying two of the 4 FOCs hold forinequality

" {θE(N), θE(D)} solve the remaining 2 FOCs:--

! Condition R: The additional rent by elite from labor repression is sufficientlylarge such that

! State Dependence: Under Condition R, a symmetric MPE features:" Markov regime switch with the society fluctuating between {N,e} and {D,c}

- with the regime probabilities p(N) > p(D) if - with invariance p(N) = p(D) if

8

! Condition I: s.t.

(interior power-gaining investment)! Condition D: Democracy creates a substantial advantage in favor of citizens s.t.

η > -ω! Nondemocracy as Absorbing State: Under Conditions I and P, there exists a

symmetric MPE in which p(N) (0, 1) and p(D) = 0! Comparative Statics: Under Condition R with , a symmetric

MPE features:" equilibrium power-gaining investments {θE(N), θE(D)} are increasing in ΔR,

β and η, and decreasing in M" the equilibrium probability for the elite to be in power is increasing in ΔR, β,

η, and " more patient (β) or greater de jure power advantage for citizens (η) causes

the elite to have greater incentive to invest in power-gaining and raises thelikelihood of labor repression institution

! Meeting the facts: M was sufficiently large while η was sufficiently low in UK andthe Netherlands, thereby destroying the elite incentive to invest in its de factopower and leading to the eventual establishment of the democracy regime

9

D. Human Capital Institutions, Land Inequality and the Emergence of the GreatDivergence: Galor-Moav-Vollrath (2009)

! The Great Divergence: the ratio of per capita real GDP between the richest andpoorest regions increased from 3 in 1820 to 18 in 2001 (Maddison 2001)

! Main idea: link human capital promoting institutions (public schooling/childlabor regulations) to the emergence of the Great Divergence

1. The Model

! Agriculture production (CRS in workers and land): " (raw) labor demand: " land demand:

! Manufacturing production (CRS in physical/human capital): " (effective) labor demand: " capital demand:

! Two-period lived OG with pop(generation) = 1 and household preference: (i.e., an individual household values only 2nd-period

consumption and bequest to descendant)

10

! Bequest tax: at rate τt, to finance public education et! Intergenerational human capital transmission: , strictly increasing

and strictly concave in e, satisfying h(0) = 1 and Inada conditions! Household budget constraint: , where ! Household optimization:

" consumption-bequest allocation: " indirect utility: , where

2. Equilibrium

! Aggregate output: ! Aggregate bequest: ! Public education: ! Equilibrium capital evolution:

" physical capital:" human capital: , where θ = manufacturing labor

! Equilibrium sectoral outputs:" agricultural sector: " manufacturing sector:

11

! Equilibrium capital-effective labor ratio: , which is

strictly increasing in y and strictly decreasing in (θ, τ)! Equilibrium labor allocation:

" factor price equalization:

" MPL equalization: FL(X, 1- θt+1) = , implying:

- the RHS is:# strictly increasing/concave in y and strictly decreasing in θ# strictly increasing/concave in τ if α < 1/2

- the solution 1-Lt+1 = is:# strictly increasing in y# strictly decreasing in (X, τ)

! Aggregate output-maximizing tax rate τt*:

" (FOC): , implying:- τt

* = τ*(yt)- τ*(yt)yt is strictly increasing in y

" w (ρ) is strictly increasing (decreasing) in τ for τ (0, τt*)

" θ, yM and (1-τ)R are all strictly increasing in τ for τ (0, τt*)

" τt* is optimal to individuals with low landownership xi

12

3. Political Equilibrium

! Political mechanism: changes in the existing educational policy require theconsent of all groups

! Landownership: suppose that a fraction λ of young individuals in period 0 arelandlords, each owning an equal fraction 1/λ of the aggregate stock land X (i.e.,per landlord land holding is X/λ) and being endowed with b0

L units of output! Key result: a critical income s.t. yt > , τt

L = τt*, with

increasing in X and decreasing in λ, satisfying:" => with no land inequality, human capital promoting

institutions, τtL = τt

*, emerges at date 0" (extremely high land inequality results in τt

L = 0)! Process of development: a nation’s output per capita evolves according to:

" and both ψ0 and ψj are:- strictly increasing and strictly concave in y with - strictly increasing in X

" economic growth is higher under human capital promoting institution

13

! Switch to the human capital promotion regime:

! Main Finding: countries with higher land inequality will implement soonerhuman capital promoting institutions and experience higher economic growth

! Empirical test: historical evidence in the U.S. during the high school movementover 1880-1920 suggests that the Northeast and the Pacific regions had lowerland inequality and higher high school graduation rates than the South region

14

E. Organizational Capital: Theory and Measurement: Atkeson and Kehoe (2005)

! Organizational capital is an important part of intangible capital! Organizational capital can be tied to the life cycle of a plant:

" variable profit of a plant of age s: " cost of the fixed factor: wm" organization rent:

" free entry condition:

" cross-section aggregate organization rent:

" if MPL rises with plant age (learning by doing), then older plants will belarger and hire more labor than younger ones

" thus, organizational capital is summarized by the plant-specific productivity(fs

) as well as the age of the plant (s)" letting variable profit to grow at a constant rate γ > 1 (i.e., ), we can

then use free entry condition to obtain:

" thus, , where

15

1. The Basic Model

! Preference: U =

! Budget constraint:

! Production: " F is CRTS" z = aggregate technology" v = span of control parameter determining the return to scale (Lucas 1978)

! Organization capital (A, s): a plant with organization capital (A, s) at t hasstochastic organization capital (Aε, s+1) at t+1

! Time-to-build: a plant built in t-1 can start operating in t! Frontier knowledge: productivity τt, adopted by all new plants, implying a new

plant built in t-1 will have organization capital (τt, 0) at t! Plant optimization:

" variable profit: " fixed cost of hiring a manager (one per plant, fixed supply): wm

" Bellman:

16

! Plant operating decision xt(A, s) (=1 if operating, =0 otherwise)! Plant establishment decision, determined by the value of a new plan:

0, which pins down the measure of managers

! Measure of operating plants: , with the distribution

evolving as:

! Factor market clearing: " capital:" labor:" manager:

! Goods market clearing: , where aggregate output is given

by

! Plant size: , = = aggregate specific productivity

! Equilibrium allocation: and ! Equilibrium output: = ! Equilibrium variable profit:

17

2. Generalization: Monopolistic Competition

! The competitive final good output: , implying the demand

schedule for intermediate goods: ! Supply of intermediate goods: , with the powers

adjusted to include the markup accrued from local monopoly power! All other setups remain the same

3. Calibration Analysis

! Use standard macroeconomics and firm-distribution parameters and set themarkup parameter to θ = 0.9 and the span of control parameter to γ = 0.95

! The rates of job turnovers can then be computed (based on the definition byDavis-Haltiwanger-Schuh 1996):

Data Model

Overall job creation rate 8.3 10.2

Overall job destruction rate 8.4 10.2! Mean and standard deviation of shocks to ln(n):

18

! Firm age and average productivity

! Measurement of organizational capital and growth accounting" physical capital income share: θγα = 19.9%

19

" labor income share: θγ(1-α) = 65.1%" managerial and organization rent share: 1-θγ = 15%

- by using the expression for wm, managerial rent share is: 11.7%- organization rent share is: 3.3%

" Varying v = θγ by 5 percentage points, we obtain:________________________________________________________________

________________________________________________________________

20

F. Organizational Capital and Productivity Slowdown: Samaniego (2006)

! Link organizational capital to productivity growth to explain establishmentlifecycles and productivity slowdown of the 1970s and the 1980s based onchanging patters of technological adoption of fast-depreciating IT capital, whoseshare in equipment investment rose from 7% to 56%

! Average age of capital (equipment vs. structures)

" average age of equipment capital rose over 1970-92" this, together with the fact of fast depreciation of the IT capital, implies that

the adoption of new capital decelerated over the period" over the same period, Hobijn-Jovanovic (2001) find that young firms

outperformed old ones

21

1. The Model

! Production of a plant of age τ and experience a: , where γnand γs measure general and plant-specific technologies, Ω(a) measuresaccumulated plant-level expertise, and αk + αn < 1

! Organizational capital: π = = plant-specific component of productivity! Updating cost from age τ to age ν: ! Measure of plant follows a process μt+1 = Γ(μt), relying on entry/exit/updating! Plant values:

" growing factor on BGP: " plant value deflated by γy: " continuation value:" updating value:

" optimal updating rule:

" value function:" properties:

- updating follows (S,s)-rule, with full updating once decided to do so- adoption lags decrease with age

22

! Firm profit Π will be redistributed to households (measure one)! Managerial supply: a household invests e to generate φ(e) managerial units paid

at p! One manager per plant: managerial units φ and new plant establishment

investment (q) are homogeneous! Household optimization given state X = (μ, K):

max

s.t.

2. Equilibrium and Productivity Slowdown

! Stationary recursive competitive equilibrium with Γ(μ*) = μ*

! Productivity slowdown due to an organizational shock:" after t* = 1973, learning accumulated before is no longer compatible with

new technologies born since t*

" plants established before t* must suffer by starting from the lowest rung ofthe learning ladder with:

23

! Responses to the organizational shock:" given fixed prices,

- slower updating: adoption lags increase for all plant types- faster turnovers (entry/exit): the ratio of values of the incumbent to the

new entrant drops- age-biased updating: young plants update before older ones

3. Calibration Analysis

! Lifecycle dynamics:

24

! Impulse responses to the organizational shock" output and productivity

25

" plant dynamics and values

26

" plant investment and age

27

G. Institutional and Organizational Barriers to Productivity Growth: Hsieh and Klenow (2009)

! Cross-country differences in income and TFP are large and widened (see a nicesurvey in the North-Holland Handbook of Economic Growth by Caselli 2005

! Restuccia-Rogerson (2008) argue tht misallocation of resources across firms canhave large effects on aggregate TFP

! Lewis (2004, McKinsey Global Institute) argues that many institutions andpolicies can result in resource misallocation

! This paper ties all such bolts and nuts together

1. The Basic Model: Monopolistic Competition

! In addition to production efficiency differences as in Melitz (2003), firms alsoface different output and capital distortions

! A single final good is produced with a basket of industry goods, taking a Cobb-Douglas form: , with , where industry s’s output is a CES

aggregate of Ms differentiated products: , with

28

! Profit of firm i in industry s yields: " τYsi and τKsi measure output and capital distortions tied to economic

institutions and policies- τYsi captures entry barriers, good market imperfections, income taxes,

and/or transport costs- τKsi capture capital barriers, credit market imperfections, capital taxes

and/or intermediation costs" profit maximization implies:

- MRTS = relative cost:

- competitive profit:

" induced demand for labor:

" firm output:

" marginal revenue product of labor:

" marginal revenue product of capital:

! Industry factor demand:

29

" labor: , with

" capital: , with

" aggregate factor demand: and ! Final sector:

" aggregate output:

" cost minimization implies: , where

! Measurement of TFP:" physical productivity of firm i in industry s:

" revenue productivity of firm i in industry s: , or,

, which increases in both distortions,

implying that those facing larger barriers are smaller than the optimal sizeand hence have higher marginal products (under diminishing returns)

30

" industry TFP: , with

" if TFPQ (A) and TFPR are jointly log-normally distributed, then:

, that is, greater dispersion of marginal

products worsens the extent of misallocation, thus lowering industry TFP

2. Applications: China/India versus U.S.

! Calibration: based on the theory developed above, we can back out the twodistortion measures as well as firm-level productivity:

" capital distortion:

" output distortion:

" firm productivity: , with set as one to infer

Ps from observed value PsYs

31

! Sources of TFPR variation within industries

! TFP gains from equalizing TFPR within industries" China: 115.1% in 1998 86.8% in 2005" India: 100.4% in 1987 127.5% in 1994" U.S.: 36.1% in 1977 42.9% in 1997

! TFP gains from equalizing TFPR relative to 1997 U.S. gains" China: 50.5% in 1998 30.5% in 2005" India: 40.2% in 1987 59.2% in 1994

32

! TFP by ownership in China and India

33

! China and India have lower TFPQ and higher TFPR than the U.S.:

34

! China and India have overly concentrated plan size distribution than the efficientone

35

! Experienced and larger firms in the U.S. have lower TFPR (less barriers)" in India, the results are opposite (need theory to explain)" in China, experienced and small firms have lower TFPR (need theory)