Managers and Productivity Differences Nezih Guner, Andrii - - PowerPoint PPT Presentation

Managers and Productivity Differences

Nezih Guner, Andrii Parkhomenko and Gustavo Ventura RIDGE, 2014

Motivation

Motivation

• Understanding cross-country income differences.

Motivation

• Understanding cross-country income differences.

Bulk of work indicates that productivity differences are key.

Motivation

• Understanding cross-country income differences.

Bulk of work indicates that productivity differences are key.

• Growing body of work emphasizes differences in management

practices as a sources of productivity differences across countries.

• Bloom and Van Reenen (2011), Bloom, Sadun and Van

Reenen (2013).

Motivation

• Understanding cross-country income differences.

Bulk of work indicates that productivity differences are key.

• Growing body of work emphasizes differences in management

practices as a sources of productivity differences across countries.

• Bloom and Van Reenen (2011), Bloom, Sadun and Van

Reenen (2013).

• How to interpret differences in management practices?

What we do

We use large data sets from high-income countries. We document:

What we do

We use large data sets from high-income countries. We document:

1 Age-earnings profiles of managers are steeper than those for

non-managers in most countries.

What we do

We use large data sets from high-income countries. We document:

1 Age-earnings profiles of managers are steeper than those for

non-managers in most countries.

2 Earnings of managers grow faster with age than earnings of

non managers in countries with higher GDP per worker.

What we do

• We develop a life-cycle, span-of-control model to interpret

these facts:

What we do

• We develop a life-cycle, span-of-control model to interpret

these facts:

• managers invest in their managerial abilities.

What we do

• We develop a life-cycle, span-of-control model to interpret

these facts:

• managers invest in their managerial abilities.
• Goods are used to produce new skills. Goods and current skills

are complementary in skill production.

What we do

• We develop a life-cycle, span-of-control model to interpret

these facts:

• managers invest in their managerial abilities.
• Goods are used to produce new skills. Goods and current skills

are complementary in skill production.

• Steepness of age-earnings profiles depends on the incentives of

managers to invest.

What we do

• We develop a life-cycle, span-of-control model to interpret

these facts:

• managers invest in their managerial abilities.
• Goods are used to produce new skills. Goods and current skills

are complementary in skill production.

• Steepness of age-earnings profiles depends on the incentives of

managers to invest.

• Distribution of managerial abilities is endogenous and affected

by distortions.

What we do

• We develop a life-cycle, span-of-control model to interpret

these facts:

• managers invest in their managerial abilities.
• Goods are used to produce new skills. Goods and current skills

are complementary in skill production.

• Steepness of age-earnings profiles depends on the incentives of

managers to invest.

• Distribution of managerial abilities is endogenous and affected

by distortions.

• Amplification.
• Differences in management quality reflect the fact that some

individuals have (endogenously) better managerial skills.

What we do

Calibrate this model to U.S. economy.

What we do

Calibrate this model to U.S. economy.

• What is effect of cross country differences in exogenous

aggregate productivity levels? Effects on relative managerial incomes, plant size and productivity measures.

What we do

Calibrate this model to U.S. economy.

• What is effect of cross country differences in exogenous

aggregate productivity levels? Effects on relative managerial incomes, plant size and productivity measures.

• Connect with literature on misallocation. What is role role of

distortions (implicit output taxes)?

• Restuccia and Rogerson (2008) and Guner, Ventura and Yi

(2008) and others.

What we do

Calibrate this model to U.S. economy.

• What is effect of cross country differences in exogenous

aggregate productivity levels? Effects on relative managerial incomes, plant size and productivity measures.

• Connect with literature on misallocation. What is role role of

distortions (implicit output taxes)?

• Restuccia and Rogerson (2008) and Guner, Ventura and Yi

(2008) and others.

• Reallocation effect: resources (capital and labor) will move

from more productive to less productive establishments.

What we do

Calibrate this model to U.S. economy.

• What is effect of cross country differences in exogenous

aggregate productivity levels? Effects on relative managerial incomes, plant size and productivity measures.

• Connect with literature on misallocation. What is role role of

distortions (implicit output taxes)?

• Restuccia and Rogerson (2008) and Guner, Ventura and Yi

(2008) and others.

• Reallocation effect: resources (capital and labor) will move

from more productive to less productive establishments.

• Skill investment effect: managers choose to invest less to

improve their skills.

Managers and Non-Managers: Data

• Use large cross sectional data sets for a group of high income

countries.

• 20 countries in our sample

Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Iceland, Ireland, Israel, Italy, Luxembourg, Netherlands, Norway, Spain, Sweden, Switzerland, UK, US.

Managers and Non-Managers: Data

• Use large cross sectional data sets for a group of high income

countries.

• 20 countries in our sample

Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Iceland, Ireland, Israel, Italy, Luxembourg, Netherlands, Norway, Spain, Sweden, Switzerland, UK, US.

• Divide individuals according to reported occupation.
• Who is a manager?

Who is a Manager?

US

Chief executives and public administrators, Financial managers, Human resources and labor relations managers, Managers and Specialists in marketing, advertising, and public relations, Man. in educ. and related fields, Man. of medicine and health occupations, Postmasters and mail superintendents, Managers of food services and lodging occupations, Managers of properties and real estate, Funeral directors, Managers of service organizations, Managers and administrators

Age-Earnings Profiles

Estimate age effects in earnings. We estimate

Age-Earnings Profiles

Estimate age effects in earnings. We estimate xi

a,y,e = α + β0a + β1a2 + γ1y + γ2ei + εi a,y,e,

• xi

a,y,e : log-earnings of individual i, age a in year y with

education e.

• a : age
• y : year dummy. e : education dummy

Estimate this equation for managers and non-managers separately. Focus on age-earnings profiles implied by β0 and β1.

1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 Age

Figure 3. Age-earnings profiles in the US US (managers) US (nonmanagers)

1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 Age

Figure 1. Age-earnings profiles for managers Canada US Belgium Spain

1 1.1 1.2 1.3 1.4 1.5 1.6 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 Age

Figure 2. Age-earnings profiles for non-managers Canada US Belgium Spain

Relative Managerial Earnings

Focus on ’slope’ of earnings growth of managers to non-managers to output per worker in prime-working years.

Relative Managerial Earnings

Focus on ’slope’ of earnings growth of managers to non-managers to output per worker in prime-working years. WE FIND: Managerial income grows faster with age than non-managerial income in richer countries.

Relative Managerial Earnings

Focus on ’slope’ of earnings growth of managers to non-managers to output per worker in prime-working years. WE FIND: Managerial income grows faster with age than non-managerial income in richer countries. Finding similar to Lagakos, Moll, Porzio, Qian and Schoellman (2014).

Relative Managerial Earnings

Focus on ’slope’ of earnings growth of managers to non-managers to output per worker in prime-working years. WE FIND: Managerial income grows faster with age than non-managerial income in richer countries. Finding similar to Lagakos, Moll, Porzio, Qian and Schoellman (2014). Finding robust to consideration of different subsamples of our data, cohort instead of year effects and presence of education controls.

Australia Austria Belgium Canada Denmark Finland France Germany Iceland Ireland Israel Italy Luxembou Netherlands Norway Spain Sweden Switzerland UK US_Census

−.2 .2 .4 Log (Income Growth, 50−54/25−29) 11 11.2 11.4 11.6 Log (GDP per Worker)

Sample: benchmark. Slope: 0.569. Corr: 0.489

GDP per Worker and Relative Income Growth

Model

Model

• Life-cycle economy. j = 1, ...R, ...N.
• Heterogenous individuals, all born with an initial endowment
• f managerial ability, z1(t) = Gz(t)z.
• z is drawn from a cumulative distribution F(z)
• Gz(t) grows at rate 1 + gz
• Individuals can be workers or managers. Workers earn wage w.
• Managers operate Lucas (1978) technology. They rent capital

and labor services, produce output and collect managerial rents.

• Occupational choice takes place at start of life, and it is

irreversible.

Model

• Workers and managers decide how much to consume and how

much to save.

Model

• Workers and managers decide how much to consume and how

much to save.

• Managers also decide how much capital and labor to rent

each period and how much to invest in their skills.

Model

• Workers and managers decide how much to consume and how

much to save.

• Managers also decide how much capital and labor to rent

each period and how much to invest in their skills.

• All agents are born without any assets.
• There are no borrowing constraints.

Technology

• Production in each plant

y(t) = A(t)z1−γ(kαn1−α)γ

• A(t) = A GA(t)− economy-wide productivity level, GA(t)

grows at rate 1 + gA

• γ− span of control parameter
• Managers can enhance future skills

zj+1(t + 1) = zj(t) + B(j)zj(t)θ1xj(t)θ2

• ≡g(x,z,j;t)
• x− goods invested in skills
• θ1 > 0 and 0 < θ2 < 1
• Complementarity between current ability and investments.

Decisions – Static Profit Maximization

• Managerial income for a manager with ability z at date t is

given by π(z, r, w, A, t) ≡ max

n,k {A(t)z1−γ

kαn1−αγ − w(t)n − (r(t) + δ)k}

• Factor demands and profits profits are linear in z. Follows that

π(z, r, w, A, t) = πz(r, w, A, t).z

Decisions – Managers

• Standard Euler equation for asset accumulation

1 c(t) = β(1 + r(t + 1)) 1 c(t + 1).

Decisions – Managers

• Standard Euler equation for asset accumulation

1 c(t) = β(1 + r(t + 1)) 1 c(t + 1).

• Skill investment

(1 + r(t + 1))

• marginal cost

= πz (t + 1) gx(t) + gx(t) gx(t + 1)[1 + gz(t + 1)]

• marginal benefit

.

• Given the functional forms, can be shown that
• Managers with higher abilities invest more
• Higher θ1 and θ2 amplify the differences between managers
• ver their life cycle.

Decisions

• Decisions for workers are standard and governed by Euler

equation for consumption.

Decisions

• Decisions for workers are standard and governed by Euler

equation for consumption.

• Focus on stationary equilibria.

At start of life, given their initial managerial ability, individuals choose to be either workers or managers.

Decisions

• Decisions for workers are standard and governed by Euler

equation for consumption.

• Focus on stationary equilibria.

At start of life, given their initial managerial ability, individuals choose to be either workers or managers. There is a threshold value such that those with z ≥ z∗ become managers, and those with z < z∗ become workers.

Decisions

• Decisions for workers are standard and governed by Euler

equation for consumption.

• Focus on stationary equilibria.

At start of life, given their initial managerial ability, individuals choose to be either workers or managers. There is a threshold value such that those with z ≥ z∗ become managers, and those with z < z∗ become workers. Details Balanced Growth Path

) (

1

W ) (

1 z

V

*

z z

managers workers

Benchmark Economy

• Model period: 5 years. Agents live for 12 model period (8 as

workers and 4 as retirees).

• Log-normal distribution of initial ability, (µz, σz)
• Calibrate model parameters to reproduce aggregate and

cross-sectional targets.

Model and Data

Statistic Data Model Mean Size 17.9 17.7 Capital Output Ratio 2.3 2.3 Fraction of Small (0-9 workers) establishments 0.725 0.719 Fraction of Large (100+ workers) establishments 0.026 0.030 Employment Share of Large establishments 0.462 0.444 Managerial Income (40-44/25-29) 1.18 1.20 Managerial Income (60-64/25-29) 1.25 1.25 Parameters

1.3

Age Earnings!Profile

1.25 1.2 1.15 1.1

Data

1.05

Model

1 25 29 30 34 35 39 40 44 45 49 50 54 55 60 60 64

Age

0.4 0.5 0.6 0.7 0.8

e Distribution

Data Model

0.1 0.2 0.3 less than 10 10-19 20-49 50-99 100-499 500+

Plant Size Classes

0.35

Employment Shares

0 25 0.3

Data Model

0.2 0.25 0.15 0.1 0.05 less than 10 10-19 20-49 50-99 100-499 500+

Plant Size Classes

Effects of Economy-Wide Productivity

Effects of Economy-Wide Productivity

60.0000 80.0000 100.0000 120.0000

Aggregate Effects of Lower A

0.0000 20.0000 40.0000 1.00 0.95 0.90 0.85 0.80 0.75 0.70

A

Output Mean Size

Effects of Economy-Wide Productivity

1.0000 1.0500 1.1000 1.1500 1.2000 1.2500 1.3000 25-29 30-34 35-39 40-44 45-49 50-54 55-

Ages

Figure 7: Aggregate TFP and Age-Earnings Profiles of Managers

• ✁

Effects of Economy-Wide Productivity

Australia Austria Belgium Canada Denmark Finland France Germany Iceland Ireland Israel Italy Luxembourg Netherlands Norway Spain Sweden Switzerland UK US_Census

−.2 .2 .4 Log (Income Growth, 50−54/25−29) −.4 −.2 .2 .4 Log (GDP per Worker Relative to the US) data model

(data and model) Figure 8: GDP per Worker and Life−Cycle Income Growth for Managers

Size-Dependent Distortions

• We model distortions as size-dependent output taxes.
• An establishment with output y faces an average tax rate

T(y) = 1 − λy −τ

Size-Dependent Distortions

• We model distortions as size-dependent output taxes.
• An establishment with output y faces an average tax rate

T(y) = 1 − λy −τ

• τ = 0, all establishments pay (1 − λ).

Size-Dependent Distortions

• We model distortions as size-dependent output taxes.
• An establishment with output y faces an average tax rate

T(y) = 1 − λy −τ

• τ = 0, all establishments pay (1 − λ).
• τ > 0, the distortions are size-dependent

Size-Dependent Distortions

• We model distortions as size-dependent output taxes.
• An establishment with output y faces an average tax rate

T(y) = 1 − λy −τ

• τ = 0, all establishments pay (1 − λ).
• τ > 0, the distortions are size-dependent
• size-dependent distortions distort input choices and reallocate

resources from large to small units – standard effect.

Size-Dependent Distortions

• We model distortions as size-dependent output taxes.
• An establishment with output y faces an average tax rate

T(y) = 1 − λy −τ

• τ = 0, all establishments pay (1 − λ).
• τ > 0, the distortions are size-dependent
• size-dependent distortions distort input choices and reallocate

resources from large to small units – standard effect.

• size-dependent distortions distort investment decisions of

managers – new effect.

Japan

Japan

• Lower output per worker than U.S. (76%).

Japan

• Lower output per worker than U.S. (76%).
• Documented distortions, and large size differences relative to

United States. What accounts for the differences between U.S. and Japan?

Japan

• Lower output per worker than U.S. (76%).
• Documented distortions, and large size differences relative to

United States. What accounts for the differences between U.S. and Japan?

• Exercise: Pick country-wide productivity and distortions to

match (i) output per-worker differences; (ii) fraction of small establishments, (iii) employment in large establishments.

0.8 0.9

S ize Dis tribution, J apan

0.7 0.5 0.6 0.4 0.5

Data Model

0.3 0.1 0.2 less than 10 10‐19 20‐49 50‐99 100+

Plant S ize C las s es

0.3

Employment Shar

0.25

Data Model

0.2 0.15 0.1 0.05 less than 10 10-19 20-49 50-99 100+

Japan

• Choose A, λ, and τ.

BM Japan Japan Japan (U.S.) Data Model Model (No Distortions) Economy-Wide Productivity 1

• 0.875

0.875 Mean tax rate on output (%)

• 6.1

Output 100 0.76 0.76 0.81 Fraction Small Plants (0-9) 0.72 0.79 0.79 0.72 Fraction Large Plants (100+) 0.03 0.01 0.01 0.03

• Empl. Large Plants (100+)

0.44 0.26 0.26 0.44 Mean Size 17.7 9.7 10.8 17.7 Investment in Skills 100

• 51.8

85.2 Average Managerial Ability 100

• 62.3

97.5

Japan

• Choose A, λ, and τ.

BM Japan Japan Japan (U.S.) Data Model Model (No Distortions) Economy-Wide Productivity 1

• 0.875

0.875 Mean tax rate on output (%)

• 6.1

Output 100 0.76 0.76 0.81 Fraction Small Plants (0-9) 0.72 0.79 0.79 0.72 Fraction Large Plants (100+) 0.03 0.01 0.01 0.03

• Empl. Large Plants (100+)

0.44 0.26 0.26 0.44 Mean Size 17.7 9.7 10.8 17.7 Investment in Skills 100

• 51.8

85.2 Average Managerial Ability 100

• 62.3

97.5 About 80% of output gap accounted for by economy-wide productivity

• differences. Distortions account for nearly all the differences in size.

Conclusions

• We document that earnings of managers grow faster with age

than earnings of non managers in richer countries.

Conclusions

• We document that earnings of managers grow faster with age

than earnings of non managers in richer countries.

• We use a life-cycle, span-of-control model along a balanced

growth path to interpret these facts.

Conclusions

• We document that earnings of managers grow faster with age

than earnings of non managers in richer countries.

• We use a life-cycle, span-of-control model along a balanced

growth path to interpret these facts.

• KEY: Managers invest in their managerial abilities.

Distribution of managerial abilities is endogenous.

Conclusions

• We document that earnings of managers grow faster with age

than earnings of non managers in richer countries.

• We use a life-cycle, span-of-control model along a balanced

growth path to interpret these facts.

• KEY: Managers invest in their managerial abilities.

Distribution of managerial abilities is endogenous.

• Differences in economy-wide productivity lead to lower

investments in managerial skills and less steep managerial income profiles in line with data.

Conclusions

• We document that earnings of managers grow faster with age

than earnings of non managers in richer countries.

• We use a life-cycle, span-of-control model along a balanced

growth path to interpret these facts.

• KEY: Managers invest in their managerial abilities.

Distribution of managerial abilities is endogenous.

• Differences in economy-wide productivity lead to lower

investments in managerial skills and less steep managerial income profiles in line with data.

• Distortions lead to significant effects on plant size and

managerial ability. Effects on output are of second-order.

EXTRA SLIDES

Australia AustriaBelgium Canada Denmark Finland France Germany Iceland Ireland Israel Italy Luxembourg Netherlands Norway Spain Sweden Switzerland UK US_Census

−.4 −.2 .2 .4 Log (Income Growth, 50−54/25−29) 11 11.2 11.4 11.6 Log (GDP per Worker)

Sample: without_self_employed_nonmanagers. Slope: 0.584. Corr: 0.454

GDP per Worker and Relative Income Growth

Australia Austria Belgium Canada Denmark Finland France Germany Iceland Ireland Israel Italy Luxembourg Netherlands Norway Spain Sweden Switzerland UK US_Census

−.2 .2 .4 .6 Log (Income Growth, 50−54/25−29) 11 11.2 11.4 11.6 Log (GDP per Worker)

Sample: without_self_employed. Slope: 0.662. Corr: 0.475

GDP per Worker and Relative Income Growth

Australia Austria Belgium Canada Denmark Finland France Germany Iceland Ireland Israel Italy Luxembourg Netherlands Norway Spain Sweden Switzerland UK US_Census

−.2 .2 .4 .6 Log (Income Growth, 50−54/25−29) 11 11.2 11.4 11.6 Log (GDP per Worker)

Sample: without_professionals. Slope: 0.584. Corr: 0.489

GDP per Worker and Relative Income Growth

Effects of Economy-Wide Productivity

Statistic A= 1 A= 0.9 A= 0.8 A= 0.7

GDP per worker 100 85.03 71.00 57.92 Mean Size 17.74 17.74 17.74 17.22 Number of Managers 100 100 100 102.89 Average Managerial Ability 100 97.99 96.08 91.90 Employment Share of Large estab. 0.444 0.440 0.435 0.429 Managerial Income (40-44/25-29) 1.198 1.170 1.144 1.118 Managerial Income (60-64/25-29) 1.253 1.217 1.182 1.149

Who is a Manager?

Australia

Senior officials and managers Corporate managers Managers of small enterprises

Canada

Senior management occupations, other management occupations

Who is a Manager?

Austria, Belgium, Finland, France, Germany, Iceland Ireland, Netherlands, Norway, Sweden, UK, Spain

Senior officials and managers Corporate managers Managers of small enterprises

US

Chief executives and public administrators, Financial managers, Human resources and labor relations managers, Managers and Specialists in marketing, advertising, and public relations, Man. in educ. and related fields, Man. of medicine and health occupations, Postmasters and mail superintendents, Managers of food services and lodging occupations, Managers of properties and real estate, Funeral directors, Managers of service organizations, Managers and administrators

Decisions – Life-Cycle

Life-Cycle

• The problem of the manager is

max

cj(t), aj(t),xj(t) J

∑

j=1

βj−1 log(cj),

• subject to

cj(t) + xj(t) + aj+1(t + 1) = π(z, r, w, A, t) + (1+ r(t))aj(t), j < J cj(t) + aj+1(t + 1) = (1 + r(t))aj(t), j ≥ JR, and zj+1(t + 1) = zj(t) + B(j)zj(t)θ1xj(t)θ2, with aJ+1(.) = 0.

Balanced Growth Path

Balanced Growth Path

• Let g be the growth rate of per capita output.
• Production technology requires that along a balanced growth

path 1 + g = (1 + gA)

1 1−αγ (1 + gz) 1−γ 1−αγ

• On the other hand, the skill investment technology implies
• 1

1 + gz θ1 1 1 + g θ2−1 = (1 + gA)

1 1−γ

• 1

1 + g γ(1−α)

1−γ

.

• We obtain

1 + g = (1 + gA)ψ, where ψ ψ ≡ 1 − θ1 γ(1 − α) + (1 − θ2)(1 − γ) − θ1(1 − αγ).

Parameter Values

Benchmark Economy Parameter values Population Growth Rate (n) 0.011 Productivity Growth Rate (g) 0.0255 Depreciation Rate (δ) 0.040 Importance of Capital (α) Returns to Scale (γ) 0.77 Mean Log-managerial Ability (µz) Dispersion in Log-managerial Ability (σz) 2.875 Discount Factor (β) 0.944 Skill accumulation technology (θ) 0.92 Skill accumulation technology (δθ) 0.048 Skill accumulation technology (θ1) 0.68 Skill accumulation technology (θ2) 0.49

Size-Dependent Distortions

Size-Dependent Distortions

• Evaluate effects of distortions that lead to same implicit tax

collections but...

Size-Dependent Distortions

• Evaluate effects of distortions that lead to same implicit tax

collections but...

• Different tax wedges between establishments large and small.

Effects of Size-Dependent Distortions

Tax Rate (% Output) 10 10 10 Level (λ) 1 0.9 1.038 1.165 Size Dependency (τ) 0.032 0.066 Wedge (1 − T(5y))/(1 − T(y)) 1 1 0.95 0.9 Output 100 94.48 91.96 87.66 Mean Size 17.74 17.74 11.78 8.23 Investment in Skills 100 74.28 55.45 42.21 Number of Managers 100 100.00 146.71 203.41 Average Managerial Ability 100 97.99 68.07 49.46

Effects of Size-Dependent Distortions

Tax Rate (% Output) 10 10 10 Level (λ) 1 0.9 1.038 1.165 Size Dependency (τ) 0.032 0.066 Wedge (1 − T(5y))/(1 − T(y)) 1 1 0.95 0.9 Employment Share (100+) estab. 0.44 0.44 0.29 0.16 Managerial Income (40-44/25-29) 1.20 1.17 1.11 1.07 Managerial Income (60-64/25-29) 1.25 1.22 1.14 1.14

Effects of Size-Dependent Distortions

60.00 70.00 80.00 90.00 100.00

Effect of Distortions on Output and Plant Size 10% Average Taxes

Output Size 20.00 30.00 40.00 50.00 1.00 0.95 0.90 0.85 Progressivity Wedge

Effects of Size-Dependent Distortions

• Reallocation effect
• distortions lower the wage rate and the fraction of managers

increase.

• distorted managers lower their output and factor demand.
• Skill accumulation effect
• distorted managers invest less.
• undistorted managers invest more.