Inequality and Economic Performance

A Theoretical Perspective

Roland Bénabou

Princeton University

Columbia University, December 2014

Back to the Future

Large and active literature in 90�s

I e.g. Benabou (2000) �Unequal Societies: Income Distributionand the Social Contract� . Also work by Galor-Ziera, Durlauf,Fernandez-Rogerson, Ferreira,...

Imperfect capital markets (esp. for �nancing human capital, andespecially earl on) )

I Tomorrow�s distribution of income / wealth depends on today�s:inequality is endogenous

Redistribution, social insurance and other progressive policies(e.g., in education �nance) reduce future inequality + can increasesocial mobility

Back to the Future

Large and active literature in 90�s

I e.g. Benabou (2000) �Unequal Societies: Income Distributionand the Social Contract� . Also work by Galor-Ziera, Durlauf,Fernandez-Rogerson, Ferreira,...

Imperfect capital markets (esp. for �nancing human capital, andespecially earl on) )

I Tomorrow�s distribution of income / wealth depends on today�s:inequality is endogenous

Redistribution, social insurance and other progressive policies(e.g., in education �nance) reduce future inequality + can increasesocial mobility

Back to the Future

Large and active literature in 90�s

I e.g. Benabou (2000) �Unequal Societies: Income Distributionand the Social Contract� . Also work by Galor-Ziera, Durlauf,Fernandez-Rogerson, Ferreira,...

Imperfect capital markets (esp. for �nancing human capital, andespecially earl on) )

I Tomorrow�s distribution of income / wealth depends on today�s:inequality is endogenous

Redistribution, social insurance and other progressive policies(e.g., in education �nance) reduce future inequality + can increasesocial mobility

Back to the Future

The political economy of ex-ante e¢ cient / growth enhancingredistributions is radically di¤erent from that of �classical�, ine¢ cientredistributions:

I Classical case: equity-e¢ ciency tradeo¤, ine¢ cient / growth reducingredistributive policies imposed nonetheless by poor median voter.The more inequality, the more redistribution

I E¢ cient redistributions: universal support behind veil of ignorance,wide support in a relatively homogeneous society. As inequality rises,will be opposed by increasing fraction of the rich ) support declines(at least initially)

)The more inequality, the less redistribution (via wage compression,taxes, education �nance, residential and social integration).

I The more so, greater the elasticity of political in�uence to wealth, themore so. (Imperfect political systems, state capture)

Back to the Future

The political economy of ex-ante e¢ cient / growth enhancingredistributions is radically di¤erent from that of �classical�, ine¢ cientredistributions:

I Classical case: equity-e¢ ciency tradeo¤, ine¢ cient / growth reducingredistributive policies imposed nonetheless by poor median voter.The more inequality, the more redistribution

I E¢ cient redistributions: universal support behind veil of ignorance,wide support in a relatively homogeneous society. As inequality rises,will be opposed by increasing fraction of the rich ) support declines(at least initially)

)The more inequality, the less redistribution (via wage compression,taxes, education �nance, residential and social integration).

I The more so, greater the elasticity of political in�uence to wealth, themore so. (Imperfect political systems, state capture)

Back to the Future

The political economy of ex-ante e¢ cient / growth enhancingredistributions is radically di¤erent from that of �classical�, ine¢ cientredistributions:

I Classical case: equity-e¢ ciency tradeo¤, ine¢ cient / growth reducingredistributive policies imposed nonetheless by poor median voter.The more inequality, the more redistribution

I E¢ cient redistributions: universal support behind veil of ignorance,wide support in a relatively homogeneous society. As inequality rises,will be opposed by increasing fraction of the rich ) support declines(at least initially)

)The more inequality, the less redistribution (via wage compression,taxes, education �nance, residential and social integration).

I The more so, greater the elasticity of political in�uence to wealth, themore so. (Imperfect political systems, state capture)

Back to the Future

The political economy of ex-ante e¢ cient / growth enhancingredistributions is radically di¤erent from that of �classical�, ine¢ cientredistributions:

I Classical case: equity-e¢ ciency tradeo¤, ine¢ cient / growth reducingredistributive policies imposed nonetheless by poor median voter.The more inequality, the more redistribution

I E¢ cient redistributions: universal support behind veil of ignorance,wide support in a relatively homogeneous society. As inequality rises,will be opposed by increasing fraction of the rich ) support declines(at least initially)

)The more inequality, the less redistribution (via wage compression,taxes, education �nance, residential and social integration).

I The more so, greater the elasticity of political in�uence to wealth, themore so. (Imperfect political systems, state capture)

Back to the Future

Importance or peer and neighborhood e¤ects ) residential and socialstrati�cation has e¤ects similar to regressive redistribution (but ofnon-�nancial inputs into human and social capital).

Back to the Future (Implications / Predictions)

1 Redistribution, welfare state more likely to decline with inequalitythan to expand (�RH�e¤ect; U-shaped)

2 Emergence and stability of di¤erent social models (steady-states):

US / Europe, Latin America / East Asia, Scandinavia, etc.

I More equal ones (pretax) have more generous welfare state

I History matters: initial distribution of income, initial politicalinstitutions (s Engermann-Sokolo¤)

3 More unequal societies are less mobile (�GG curve�)

4 More spatially segregated cities are more unequal and have lowersocial mobility (Chetty et al.)

5 Inequality and growth can be negatively correlated (but need not bemonotonic)

Back to the Future (Implications / Predictions)

1 Redistribution, welfare state more likely to decline with inequalitythan to expand (�RH�e¤ect; U-shaped)

2 Emergence and stability of di¤erent social models (steady-states):

US / Europe, Latin America / East Asia, Scandinavia, etc.

I More equal ones (pretax) have more generous welfare state

I History matters: initial distribution of income, initial politicalinstitutions (s Engermann-Sokolo¤)

3 More unequal societies are less mobile (�GG curve�)

4 More spatially segregated cities are more unequal and have lowersocial mobility (Chetty et al.)

5 Inequality and growth can be negatively correlated (but need not bemonotonic)

Back to the Future (Implications / Predictions)

1 Redistribution, welfare state more likely to decline with inequalitythan to expand (�RH�e¤ect; U-shaped)

2 Emergence and stability of di¤erent social models (steady-states):

US / Europe, Latin America / East Asia, Scandinavia, etc.

I More equal ones (pretax) have more generous welfare state

I History matters: initial distribution of income, initial politicalinstitutions (s Engermann-Sokolo¤)

3 More unequal societies are less mobile (�GG curve�)

4 More spatially segregated cities are more unequal and have lowersocial mobility (Chetty et al.)

5 Inequality and growth can be negatively correlated (but need not bemonotonic)

Back to the Future (Implications / Predictions)

1 Redistribution, welfare state more likely to decline with inequalitythan to expand (�RH�e¤ect; U-shaped)

2 Emergence and stability of di¤erent social models (steady-states):

US / Europe, Latin America / East Asia, Scandinavia, etc.

I More equal ones (pretax) have more generous welfare state

I History matters: initial distribution of income, initial politicalinstitutions (s Engermann-Sokolo¤)

3 More unequal societies are less mobile (�GG curve�)

4 More spatially segregated cities are more unequal and have lowersocial mobility (Chetty et al.)

5 Inequality and growth can be negatively correlated (but need not bemonotonic)

Back to the Future (Implications / Predictions)

1 Redistribution, welfare state more likely to decline with inequalitythan to expand (�RH�e¤ect; U-shaped)

2 Emergence and stability of di¤erent social models (steady-states):

US / Europe, Latin America / East Asia, Scandinavia, etc.

I More equal ones (pretax) have more generous welfare state

I History matters: initial distribution of income, initial politicalinstitutions (s Engermann-Sokolo¤)

3 More unequal societies are less mobile (�GG curve�)

4 More spatially segregated cities are more unequal and have lowersocial mobility (Chetty et al.)

5 Inequality and growth can be negatively correlated (but need not bemonotonic)

Back to the Future (Implications / Predictions)

1 Redistribution, welfare state more likely to decline with inequalitythan to expand (�RH�e¤ect; U-shaped)

2 Emergence and stability of di¤erent social models (steady-states):

US / Europe, Latin America / East Asia, Scandinavia, etc.

I More equal ones (pretax) have more generous welfare state

I History matters: initial distribution of income, initial politicalinstitutions (s Engermann-Sokolo¤)

3 More unequal societies are less mobile (�GG curve�)

4 More spatially segregated cities are more unequal and have lowersocial mobility (Chetty et al.)

5 Inequality and growth can be negatively correlated (but need not bemonotonic)

Missing From Rational Political-Economy Account(s)Cannot attribute failure to implement e¢ ciency / growth enhancingpolicies to sole power of the rich over political institutions

Lack of support among much of the classes that would bene�t (estatetax, Obamacare, recent elections, etc.)

Key determinant of attitudes toward inequality and redistribution,both across individuals and across countries:

I Beliefs about e¤ort and willpower vs. luck and social circumstances askey determinant of life success, poverty, etc.

I These beliefs are often completely out of line with the reality of socialmobility

Sources of belief distortions / dissonance reduction / �falseconsciousness�:

I �Supply side� : propaganda by those opposed to such policiesI �Demand side�:�belief in a just world,��system justi�cation�.Brings in behavioral economics / psychology.

Missing From Rational Political-Economy Account(s)Cannot attribute failure to implement e¢ ciency / growth enhancingpolicies to sole power of the rich over political institutions

Lack of support among much of the classes that would bene�t (estatetax, Obamacare, recent elections, etc.)

Key determinant of attitudes toward inequality and redistribution,both across individuals and across countries:

I Beliefs about e¤ort and willpower vs. luck and social circumstances askey determinant of life success, poverty, etc.

I These beliefs are often completely out of line with the reality of socialmobility

Sources of belief distortions / dissonance reduction / �falseconsciousness�:

I �Supply side� : propaganda by those opposed to such policiesI �Demand side�:�belief in a just world,��system justi�cation�.Brings in behavioral economics / psychology.

Missing From Rational Political-Economy Account(s)Cannot attribute failure to implement e¢ ciency / growth enhancingpolicies to sole power of the rich over political institutions

Lack of support among much of the classes that would bene�t (estatetax, Obamacare, recent elections, etc.)

Key determinant of attitudes toward inequality and redistribution,both across individuals and across countries:

I Beliefs about e¤ort and willpower vs. luck and social circumstances askey determinant of life success, poverty, etc.

I These beliefs are often completely out of line with the reality of socialmobility

Sources of belief distortions / dissonance reduction / �falseconsciousness�:

I �Supply side� : propaganda by those opposed to such policiesI �Demand side�:�belief in a just world,��system justi�cation�.Brings in behavioral economics / psychology.

Missing From Rational Political-Economy Account(s)Cannot attribute failure to implement e¢ ciency / growth enhancingpolicies to sole power of the rich over political institutions

Lack of support among much of the classes that would bene�t (estatetax, Obamacare, recent elections, etc.)

Key determinant of attitudes toward inequality and redistribution,both across individuals and across countries:

I Beliefs about e¤ort and willpower vs. luck and social circumstances askey determinant of life success, poverty, etc.

I These beliefs are often completely out of line with the reality of socialmobility

Sources of belief distortions / dissonance reduction / �falseconsciousness�:

I �Supply side� : propaganda by those opposed to such policiesI �Demand side�:�belief in a just world,��system justi�cation�.Brings in behavioral economics / psychology.

Need to Bring in Behavioral Economics

Douglass North (1990), Institutions, Institutional Change andEconomic Performance.

Social scientists have incorporated the costliness ofinformation in their models, but have not come to grips with thesubjective mental constructs by which individuals processinformation and arrive at conclusions that shape their choices....

The subjective mental constructs of the participants willevolve an ideology that not only rationalizes the society�sstructure but accounts for its poor performance. As a result, theeconomy will evolve policies that reinforce the existing incentivesand organizations.

Need to Bring in Behavioral Economics

Douglass North (1990), Institutions, Institutional Change andEconomic Performance.

Social scientists have incorporated the costliness ofinformation in their models, but have not come to grips with thesubjective mental constructs by which individuals processinformation and arrive at conclusions that shape their choices....

The subjective mental constructs of the participants willevolve an ideology that not only rationalizes the society�sstructure but accounts for its poor performance. As a result, theeconomy will evolve policies that reinforce the existing incentivesand organizations.

Need to Bring in Behavioral Economics

Douglass North (1990), Institutions, Institutional Change andEconomic Performance.

Social scientists have incorporated the costliness ofinformation in their models, but have not come to grips with thesubjective mental constructs by which individuals processinformation and arrive at conclusions that shape their choices....

The subjective mental constructs of the participants willevolve an ideology that not only rationalizes the society�sstructure but accounts for its poor performance. As a result, theeconomy will evolve policies that reinforce the existing incentivesand organizations.

Evidence and Theory

Alesina et al. (2001)

Evidence and TheoryBenabou-Tirole (2006) �Belief in a Just World & RedistributivePolitics�

I Endogenous belief distortions / rationalizations concerning the originsof inequality constitute another general-equilbrium feedback force thattends to make (high-inequality, low redistribution) and (low inequality,high redistribution) steady-states / social models persistent.

I Di¤erent social models go along with (endogenously) di¤erent beliefdistortions / forms of reality denial: geologies

Need more work on the determinants of political attitudes, and inparticular how they vary with individual mobility experiences andaggregate �uctuations, e.g.,

I Giuliano et al. (2013) �Growing Up in a Recession: Beliefs and theMacroeconomy�

I Call to integrate such �belief�questions into panel-data studies.

Di¢ cult issues, settling facts / theories may take time.

Evidence and TheoryBenabou-Tirole (2006) �Belief in a Just World & RedistributivePolitics�

I Endogenous belief distortions / rationalizations concerning the originsof inequality constitute another general-equilbrium feedback force thattends to make (high-inequality, low redistribution) and (low inequality,high redistribution) steady-states / social models persistent.

I Di¤erent social models go along with (endogenously) di¤erent beliefdistortions / forms of reality denial: geologies

Need more work on the determinants of political attitudes, and inparticular how they vary with individual mobility experiences andaggregate �uctuations, e.g.,

I Giuliano et al. (2013) �Growing Up in a Recession: Beliefs and theMacroeconomy�

I Call to integrate such �belief�questions into panel-data studies.

Di¢ cult issues, settling facts / theories may take time.

High Pay: Rents, or Competitive Rewards to Talent?

Key dichotomy in, e.g., literature on CEO pay, �nancial sector.

Argue that the they are both!I Benabou- Tirole �Bonus Culture: Competitive Pay, Screening, andMultitasking� (2014, forthcoming)

A very large part of increase in earnings inequality has come throughrise of performance-based pay

I Bonuses, etc., but in fact all throughout wage structure, in US and UKI Evidence that increased competition for talent (and some SBTC) is amajor driver of such evolutions

So is all well and fair? No

In multitask setting (short vs. long term, expected return vs. hiddenrisks, cooperation vs. backstabbing, etc.), we know that givingtoo-high poewred incentives is harmful.

I Presumably, �rms know it too. So why do they still often give them?

High Pay: Rents, or Competitive Rewards to Talent?

Key dichotomy in, e.g., literature on CEO pay, �nancial sector.

Argue that the they are both!I Benabou- Tirole �Bonus Culture: Competitive Pay, Screening, andMultitasking� (2014, forthcoming)

A very large part of increase in earnings inequality has come throughrise of performance-based pay

I Bonuses, etc., but in fact all throughout wage structure, in US and UKI Evidence that increased competition for talent (and some SBTC) is amajor driver of such evolutions

So is all well and fair? No

In multitask setting (short vs. long term, expected return vs. hiddenrisks, cooperation vs. backstabbing, etc.), we know that givingtoo-high poewred incentives is harmful.

I Presumably, �rms know it too. So why do they still often give them?

High Pay: Rents, or Competitive Rewards to Talent?

Key dichotomy in, e.g., literature on CEO pay, �nancial sector.

Argue that the they are both!I Benabou- Tirole �Bonus Culture: Competitive Pay, Screening, andMultitasking� (2014, forthcoming)

A very large part of increase in earnings inequality has come throughrise of performance-based pay

I Bonuses, etc., but in fact all throughout wage structure, in US and UKI Evidence that increased competition for talent (and some SBTC) is amajor driver of such evolutions

So is all well and fair? No

In multitask setting (short vs. long term, expected return vs. hiddenrisks, cooperation vs. backstabbing, etc.), we know that givingtoo-high poewred incentives is harmful.

I Presumably, �rms know it too. So why do they still often give them?

High Pay: Rents, or Competitive Rewards to Talent?

Key dichotomy in, e.g., literature on CEO pay, �nancial sector.

Argue that the they are both!I Benabou- Tirole �Bonus Culture: Competitive Pay, Screening, andMultitasking� (2014, forthcoming)

A very large part of increase in earnings inequality has come throughrise of performance-based pay

I Bonuses, etc., but in fact all throughout wage structure, in US and UKI Evidence that increased competition for talent (and some SBTC) is amajor driver of such evolutions

So is all well and fair? No

In multitask setting (short vs. long term, expected return vs. hiddenrisks, cooperation vs. backstabbing, etc.), we know that givingtoo-high poewred incentives is harmful.

I Presumably, �rms know it too. So why do they still often give them?

Bonus Culture

When competiting to attract or retain �talent�, which is imperfectlyobservable, performance pay becomes a screening device, itsdetermination no longer in the hands of each single �rm

I Holmstrom-Milgrom + Rosthschild-Stiglitz + Hotelling

I At low levels of competition, (e.g., European labor markets), �rms seekto extract rents from talented workers ) contracts o¤er excessivelylow-powered incentives

I At high levels of competition (e.g., �nancial sector), �rms must leavemore and more rents to high-talent workers in order to attract / retainthem ) now o¤er excessively high-powered incentives, even from�rms�(shareholder�s) point of vew.

) Nefarious �bonus culture�

I A fortiori, from social point of view if short-termism, hidden risks, etc.,have negative social externalities.

Bonus Culture

When competiting to attract or retain �talent�, which is imperfectlyobservable, performance pay becomes a screening device, itsdetermination no longer in the hands of each single �rm

I Holmstrom-Milgrom + Rosthschild-Stiglitz + Hotelling

I At low levels of competition, (e.g., European labor markets), �rms seekto extract rents from talented workers ) contracts o¤er excessivelylow-powered incentives

I At high levels of competition (e.g., �nancial sector), �rms must leavemore and more rents to high-talent workers in order to attract / retainthem ) now o¤er excessively high-powered incentives, even from�rms�(shareholder�s) point of vew.

) Nefarious �bonus culture�

I A fortiori, from social point of view if short-termism, hidden risks, etc.,have negative social externalities.

Bonus Culture

When competiting to attract or retain �talent�, which is imperfectlyobservable, performance pay becomes a screening device, itsdetermination no longer in the hands of each single �rm

I Holmstrom-Milgrom + Rosthschild-Stiglitz + Hotelling

I At low levels of competition, (e.g., European labor markets), �rms seekto extract rents from talented workers ) contracts o¤er excessivelylow-powered incentives

I At high levels of competition (e.g., �nancial sector), �rms must leavemore and more rents to high-talent workers in order to attract / retainthem ) now o¤er excessively high-powered incentives, even from�rms�(shareholder�s) point of vew.

) Nefarious �bonus culture�

I A fortiori, from social point of view if short-termism, hidden risks, etc.,have negative social externalities.

Bonus Culture

When competiting to attract or retain �talent�, which is imperfectlyobservable, performance pay becomes a screening device, itsdetermination no longer in the hands of each single �rm

I Holmstrom-Milgrom + Rosthschild-Stiglitz + Hotelling

I At low levels of competition, (e.g., European labor markets), �rms seekto extract rents from talented workers ) contracts o¤er excessivelylow-powered incentives

I At high levels of competition (e.g., �nancial sector), �rms must leavemore and more rents to high-talent workers in order to attract / retainthem ) now o¤er excessively high-powered incentives, even from�rms�(shareholder�s) point of vew.

) Nefarious �bonus culture�

I A fortiori, from social point of view if short-termism, hidden risks, etc.,have negative social externalities.

Bonus Culture

E¢ ciency / social welfare is hump-shaped with respect to the degreeof labor market competition. A given sector or country may fall oneither side. Simple test.

When in region of nefarious �bonus culture�, room for caps on bonuspay (as fraction of total compensation). May, however, lead to otherdistortions. Role for progressive taxation.

Future research: externalities / competition among di¤erent types of�rms, / industries Some for which poerformance pay is e¤fcienecy(little unobserved component to e¤ort, e.g., hedge funds), otherswhere it is not (e.g. traditional banks) forced to follow.

Empirical side: need to looks at di¤erent forms of earningsdi¤erentials, e.g., �xed versus performance-based pay, short-term vs.deferred compensation.

Bonus Culture

E¢ ciency / social welfare is hump-shaped with respect to the degreeof labor market competition. A given sector or country may fall oneither side. Simple test.

When in region of nefarious �bonus culture�, room for caps on bonuspay (as fraction of total compensation). May, however, lead to otherdistortions. Role for progressive taxation.

Future research: externalities / competition among di¤erent types of�rms, / industries Some for which poerformance pay is e¤fcienecy(little unobserved component to e¤ort, e.g., hedge funds), otherswhere it is not (e.g. traditional banks) forced to follow.

Empirical side: need to looks at di¤erent forms of earningsdi¤erentials, e.g., �xed versus performance-based pay, short-term vs.deferred compensation.

Bonus Culture

E¢ ciency / social welfare is hump-shaped with respect to the degreeof labor market competition. A given sector or country may fall oneither side. Simple test.

When in region of nefarious �bonus culture�, room for caps on bonuspay (as fraction of total compensation). May, however, lead to otherdistortions. Role for progressive taxation.

Future research: externalities / competition among di¤erent types of�rms, / industries Some for which poerformance pay is e¤fcienecy(little unobserved component to e¤ort, e.g., hedge funds), otherswhere it is not (e.g. traditional banks) forced to follow.

Empirical side: need to looks at di¤erent forms of earningsdi¤erentials, e.g., �xed versus performance-based pay, short-term vs.deferred compensation.