Measuring Intra-household Inequality KCP Project: Intra-Household - - PowerPoint PPT Presentation

Measuring Intra-household Inequality

KCP Project: Intra-Household Allocation of and Gender Di¤erences in Consumption Poverty - Dissemination Workshop Pierre-André Chiappori

Columbia University

World Bank HQ, Nov. 2019

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 1 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Pragmatic (most of the empirical literature)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Pragmatic (most of the empirical literature) Samuelson (Fixed welfare index: W h Ua, Ubi )

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Pragmatic (most of the empirical literature) Samuelson (Fixed welfare index: W h Ua, Ubi ) Becker: Rotten Kid Theorem ! start from a criticism of Samuelson; but requires strong assumptions (in particular, TU: Bergstrom 1989)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Pragmatic (most of the empirical literature) Samuelson (Fixed welfare index: W h Ua, Ubi ) Becker: Rotten Kid Theorem ! start from a criticism of Samuelson; but requires strong assumptions (in particular, TU: Bergstrom 1989)

Upsides of the unitary model: based on known foundations

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Pragmatic (most of the empirical literature) Samuelson (Fixed welfare index: W h Ua, Ubi ) Becker: Rotten Kid Theorem ! start from a criticism of Samuelson; but requires strong assumptions (in particular, TU: Bergstrom 1989)

Upsides of the unitary model: based on known foundations

Testability (Slutsky)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Pragmatic (most of the empirical literature) Samuelson (Fixed welfare index: W h Ua, Ubi ) Becker: Rotten Kid Theorem ! start from a criticism of Samuelson; but requires strong assumptions (in particular, TU: Bergstrom 1989)

Upsides of the unitary model: based on known foundations

Testability (Slutsky) Identi…ability (integrability)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Pragmatic (most of the empirical literature) Samuelson (Fixed welfare index: W h Ua, Ubi ) Becker: Rotten Kid Theorem ! start from a criticism of Samuelson; but requires strong assumptions (in particular, TU: Bergstrom 1989)

Upsides of the unitary model: based on known foundations

Testability (Slutsky) Identi…ability (integrability)

Weaknesses of the unitary model

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Pragmatic (most of the empirical literature) Samuelson (Fixed welfare index: W h Ua, Ubi ) Becker: Rotten Kid Theorem ! start from a criticism of Samuelson; but requires strong assumptions (in particular, TU: Bergstrom 1989)

Upsides of the unitary model: based on known foundations

Testability (Slutsky) Identi…ability (integrability)

Weaknesses of the unitary model

Theory: violates methodological individualism; disregards ‘power’

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Pragmatic (most of the empirical literature) Samuelson (Fixed welfare index: W h Ua, Ubi ) Becker: Rotten Kid Theorem ! start from a criticism of Samuelson; but requires strong assumptions (in particular, TU: Bergstrom 1989)

Upsides of the unitary model: based on known foundations

Testability (Slutsky) Identi…ability (integrability)

Weaknesses of the unitary model

Theory: violates methodological individualism; disregards ‘power’ Welfare: neglects intrahousehold inequality

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

The Twilight of the Unitary Model

Early works: the unitary model is paramount ! various justi…cations:

Pragmatic (most of the empirical literature) Samuelson (Fixed welfare index: W h Ua, Ubi ) Becker: Rotten Kid Theorem ! start from a criticism of Samuelson; but requires strong assumptions (in particular, TU: Bergstrom 1989)

Upsides of the unitary model: based on known foundations

Testability (Slutsky) Identi…ability (integrability)

Weaknesses of the unitary model

Theory: violates methodological individualism; disregards ‘power’ Welfare: neglects intrahousehold inequality Empirics: several empirical falsi…cations: Slutsky (Browning Chiappori 1998, Kapan 2009, Vermeulen 2005), Income Pooling (Thomas 1990, Lundberg et al 1997, Du‡o 2003, Attanasio Lechene 2014,...)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 2 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters In general, µ may be a function of:

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters In general, µ may be a function of:

Incomes (and wages, prices,...)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters In general, µ may be a function of:

Incomes (and wages, prices,...)

explains violations of income pooling

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters In general, µ may be a function of:

Incomes (and wages, prices,...)

explains violations of income pooling income shocks: renegotiation versus insurance

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters In general, µ may be a function of:

Incomes (and wages, prices,...)

explains violations of income pooling income shocks: renegotiation versus insurance

‘Distribution factors’, including:

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters In general, µ may be a function of:

Incomes (and wages, prices,...)

explains violations of income pooling income shocks: renegotiation versus insurance

‘Distribution factors’, including:

household-speci…c characteristics (e.g. individual incomes, wealth, networks, etc.) (although exogeneity may be an issue)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters In general, µ may be a function of:

Incomes (and wages, prices,...)

explains violations of income pooling income shocks: renegotiation versus insurance

‘Distribution factors’, including:

household-speci…c characteristics (e.g. individual incomes, wealth, networks, etc.) (although exogeneity may be an issue) the environment (e.g., divorce laws: CFL 2002, Voena 2015; the marriage market; global policies;...)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters In general, µ may be a function of:

Incomes (and wages, prices,...)

explains violations of income pooling income shocks: renegotiation versus insurance

‘Distribution factors’, including:

household-speci…c characteristics (e.g. individual incomes, wealth, networks, etc.) (although exogeneity may be an issue) the environment (e.g., divorce laws: CFL 2002, Voena 2015; the marriage market; global policies;...)

In particular, recovering µ

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters In general, µ may be a function of:

Incomes (and wages, prices,...)

explains violations of income pooling income shocks: renegotiation versus insurance

‘Distribution factors’, including:

household-speci…c characteristics (e.g. individual incomes, wealth, networks, etc.) (although exogeneity may be an issue) the environment (e.g., divorce laws: CFL 2002, Voena 2015; the marriage market; global policies;...)

In particular, recovering µ

directly addresses intra-household inequality

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

Cooperative Approach: the Collective Model

Basic assumption: e¢ciency Translation: max Ua + µUb, where µ Pareto weight ! µ as summarizing power : Power matters In general, µ may be a function of:

Incomes (and wages, prices,...)

explains violations of income pooling income shocks: renegotiation versus insurance

‘Distribution factors’, including:

household-speci…c characteristics (e.g. individual incomes, wealth, networks, etc.) (although exogeneity may be an issue) the environment (e.g., divorce laws: CFL 2002, Voena 2015; the marriage market; global policies;...)

In particular, recovering µ

directly addresses intra-household inequality leads to policies (targeting)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 3 / 17

The Collective Model with private goods

Simple translation: e¢ciency equivalent to the existence of a ’sharing rule’ ! two stage process

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 4 / 17

The Collective Model with private goods

Simple translation: e¢ciency equivalent to the existence of a ’sharing rule’ ! two stage process

Stage 1: individuals (cooperatively) allocate income (! sharing rule)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 4 / 17

The Collective Model with private goods

Simple translation: e¢ciency equivalent to the existence of a ’sharing rule’ ! two stage process

Stage 1: individuals (cooperatively) allocate income (! sharing rule) Stage 2: individuals each choose their consumption vector under BC

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 4 / 17

The Collective Model with private goods

Simple translation: e¢ciency equivalent to the existence of a ’sharing rule’ ! two stage process

Stage 1: individuals (cooperatively) allocate income (! sharing rule) Stage 2: individuals each choose their consumption vector under BC

Note that this is simply the …rst and second welfare theorems

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 4 / 17

The Collective Model with private goods

Simple translation: e¢ciency equivalent to the existence of a ’sharing rule’ ! two stage process

Stage 1: individuals (cooperatively) allocate income (! sharing rule) Stage 2: individuals each choose their consumption vector under BC

Note that this is simply the …rst and second welfare theorems Sharing rule: ‘in dollars’ ! easier interpretation than a Pareto weight (which is an ordinal concept) ! but price to pay: no public good (although conditional demands)!

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 4 / 17

The Collective Model with private goods

Simple translation: e¢ciency equivalent to the existence of a ’sharing rule’ ! two stage process

Stage 1: individuals (cooperatively) allocate income (! sharing rule) Stage 2: individuals each choose their consumption vector under BC

Note that this is simply the …rst and second welfare theorems Sharing rule: ‘in dollars’ ! easier interpretation than a Pareto weight (which is an ordinal concept) ! but price to pay: no public good (although conditional demands)! Identifying intra-household inequality amounts to recovering the sharing rule!

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 4 / 17

The Collective Model with private goods

Simple translation: e¢ciency equivalent to the existence of a ’sharing rule’ ! two stage process

Stage 1: individuals (cooperatively) allocate income (! sharing rule) Stage 2: individuals each choose their consumption vector under BC

Note that this is simply the …rst and second welfare theorems Sharing rule: ‘in dollars’ ! easier interpretation than a Pareto weight (which is an ordinal concept) ! but price to pay: no public good (although conditional demands)! Identifying intra-household inequality amounts to recovering the sharing rule! ! Basic question: what do we need to identify the sharing rule?

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 4 / 17

The Collective Model with private goods

Simple translation: e¢ciency equivalent to the existence of a ’sharing rule’ ! two stage process

Stage 1: individuals (cooperatively) allocate income (! sharing rule) Stage 2: individuals each choose their consumption vector under BC

Note that this is simply the …rst and second welfare theorems Sharing rule: ‘in dollars’ ! easier interpretation than a Pareto weight (which is an ordinal concept) ! but price to pay: no public good (although conditional demands)! Identifying intra-household inequality amounts to recovering the sharing rule! ! Basic question: what do we need to identify the sharing rule?

Obviously identi…ed if all individual, private consumptions are observed

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 4 / 17

The Collective Model with private goods

Simple translation: e¢ciency equivalent to the existence of a ’sharing rule’ ! two stage process

Stage 1: individuals (cooperatively) allocate income (! sharing rule) Stage 2: individuals each choose their consumption vector under BC

Note that this is simply the …rst and second welfare theorems Sharing rule: ‘in dollars’ ! easier interpretation than a Pareto weight (which is an ordinal concept) ! but price to pay: no public good (although conditional demands)! Identifying intra-household inequality amounts to recovering the sharing rule! ! Basic question: what do we need to identify the sharing rule?

Obviously identi…ed if all individual, private consumptions are observed Theory shows that one observed private consumption per person is su¢cient

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 4 / 17

The Collective Model with private goods

Simple translation: e¢ciency equivalent to the existence of a ’sharing rule’ ! two stage process

Stage 1: individuals (cooperatively) allocate income (! sharing rule) Stage 2: individuals each choose their consumption vector under BC

Note that this is simply the …rst and second welfare theorems Sharing rule: ‘in dollars’ ! easier interpretation than a Pareto weight (which is an ordinal concept) ! but price to pay: no public good (although conditional demands)! Identifying intra-household inequality amounts to recovering the sharing rule! ! Basic question: what do we need to identify the sharing rule?

Obviously identi…ed if all individual, private consumptions are observed Theory shows that one observed private consumption per person is su¢cient Pragmatic view: ‘the more the better’

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 4 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

‘Di¤erentiable’ approach: general tests

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

‘Di¤erentiable’ approach: general tests

Demand function ! Slutsky matrix: SNR1 (Browning Chiappori 98)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

‘Di¤erentiable’ approach: general tests

Demand function ! Slutsky matrix: SNR1 (Browning Chiappori 98) Distribution factors proportionality

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

‘Di¤erentiable’ approach: general tests

Demand function ! Slutsky matrix: SNR1 (Browning Chiappori 98) Distribution factors proportionality Links between the two

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

‘Di¤erentiable’ approach: general tests

Demand function ! Slutsky matrix: SNR1 (Browning Chiappori 98) Distribution factors proportionality Links between the two

Revealed preferences: general tests (L3+ complete works)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

‘Di¤erentiable’ approach: general tests

Demand function ! Slutsky matrix: SNR1 (Browning Chiappori 98) Distribution factors proportionality Links between the two

Revealed preferences: general tests (L3+ complete works)

Engel curves (no price variation):

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

‘Di¤erentiable’ approach: general tests

Demand function ! Slutsky matrix: SNR1 (Browning Chiappori 98) Distribution factors proportionality Links between the two

Revealed preferences: general tests (L3+ complete works)

Engel curves (no price variation):

Need 2 DF at least

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

‘Di¤erentiable’ approach: general tests

Demand function ! Slutsky matrix: SNR1 (Browning Chiappori 98) Distribution factors proportionality Links between the two

Revealed preferences: general tests (L3+ complete works)

Engel curves (no price variation):

Need 2 DF at least Then DF proportionality (or z-conditional demands)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

‘Di¤erentiable’ approach: general tests

Demand function ! Slutsky matrix: SNR1 (Browning Chiappori 98) Distribution factors proportionality Links between the two

Revealed preferences: general tests (L3+ complete works)

Engel curves (no price variation):

Need 2 DF at least Then DF proportionality (or z-conditional demands) With private consumptions only: additional conditions

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: Testability

Main message: the collective model is testable With price variations: two main approaches

‘Di¤erentiable’ approach: general tests

Demand function ! Slutsky matrix: SNR1 (Browning Chiappori 98) Distribution factors proportionality Links between the two

Revealed preferences: general tests (L3+ complete works)

Engel curves (no price variation):

Need 2 DF at least Then DF proportionality (or z-conditional demands) With private consumptions only: additional conditions

Labor supply (in particular when leisure is private) ! strong restrictions, with or without DF

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 5 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles!

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

‘Power matters’ ...

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

‘Power matters’ ... ... and powers are a¤ected by the economic environment

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

‘Power matters’ ... ... and powers are a¤ected by the economic environment ! policy may aim at ‘empowering’ speci…c subpopulations ...

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

‘Power matters’ ... ... and powers are a¤ected by the economic environment ! policy may aim at ‘empowering’ speci…c subpopulations ... ... and at reducing intra-household inequality

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

‘Power matters’ ... ... and powers are a¤ected by the economic environment ! policy may aim at ‘empowering’ speci…c subpopulations ... ... and at reducing intra-household inequality

Collective model often performs well

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

‘Power matters’ ... ... and powers are a¤ected by the economic environment ! policy may aim at ‘empowering’ speci…c subpopulations ... ... and at reducing intra-household inequality

Collective model often performs well

for consumption (BBCL , BC , AL ,...)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

‘Power matters’ ... ... and powers are a¤ected by the economic environment ! policy may aim at ‘empowering’ speci…c subpopulations ... ... and at reducing intra-household inequality

Collective model often performs well

for consumption (BBCL , BC , AL ,...) for production: Udry (1996), agricultural production in Burkina Faso ! social norms may harm e¢ciency; but Rangel & Thomas (2019): not clear after controling for (unobservable) land quality

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

‘Power matters’ ... ... and powers are a¤ected by the economic environment ! policy may aim at ‘empowering’ speci…c subpopulations ... ... and at reducing intra-household inequality

Collective model often performs well

for consumption (BBCL , BC , AL ,...) for production: Udry (1996), agricultural production in Burkina Faso ! social norms may harm e¢ciency; but Rangel & Thomas (2019): not clear after controling for (unobservable) land quality

Rough estimates of ‘sharing rules’, broadly consistent across studies; therefore of intra-household elasticities, etc.

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

‘Power matters’ ... ... and powers are a¤ected by the economic environment ! policy may aim at ‘empowering’ speci…c subpopulations ... ... and at reducing intra-household inequality

Collective model often performs well

for consumption (BBCL , BC , AL ,...) for production: Udry (1996), agricultural production in Burkina Faso ! social norms may harm e¢ciency; but Rangel & Thomas (2019): not clear after controling for (unobservable) land quality

Rough estimates of ‘sharing rules’, broadly consistent across studies; therefore of intra-household elasticities, etc. Strong policy implications

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: empirical testing

Main conclusions: Unitary model generally rejected for couples, but not for singles! Interpretation:

‘Power matters’ ... ... and powers are a¤ected by the economic environment ! policy may aim at ‘empowering’ speci…c subpopulations ... ... and at reducing intra-household inequality

Collective model often performs well

for consumption (BBCL , BC , AL ,...) for production: Udry (1996), agricultural production in Burkina Faso ! social norms may harm e¢ciency; but Rangel & Thomas (2019): not clear after controling for (unobservable) land quality

Rough estimates of ‘sharing rules’, broadly consistent across studies; therefore of intra-household elasticities, etc. Strong policy implications In all cases, opens to a ‘metatheory’ of Pareto weights determination

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 6 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

! generic identi…ability if one exclusion restriction per person

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

! generic identi…ability if one exclusion restriction per person more robust if DF

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

! generic identi…ability if one exclusion restriction per person more robust if DF But: identi…cation ‘up to’

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

! generic identi…ability if one exclusion restriction per person more robust if DF But: identi…cation ‘up to’ Revealed Preferences: bounds, which can be narrow

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

! generic identi…ability if one exclusion restriction per person more robust if DF But: identi…cation ‘up to’ Revealed Preferences: bounds, which can be narrow

Cross-sectional data: demand as function of income (or total expenditures) and a DF

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

! generic identi…ability if one exclusion restriction per person more robust if DF But: identi…cation ‘up to’ Revealed Preferences: bounds, which can be narrow

Cross-sectional data: demand as function of income (or total expenditures) and a DF

If one exclusive (or assignable) good per person: robust identi…cation

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

! generic identi…ability if one exclusion restriction per person more robust if DF But: identi…cation ‘up to’ Revealed Preferences: bounds, which can be narrow

Cross-sectional data: demand as function of income (or total expenditures) and a DF

If one exclusive (or assignable) good per person: robust identi…cation Identi…cation obtains even without assignable goods (less robust)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

! generic identi…ability if one exclusion restriction per person more robust if DF But: identi…cation ‘up to’ Revealed Preferences: bounds, which can be narrow

Cross-sectional data: demand as function of income (or total expenditures) and a DF

If one exclusive (or assignable) good per person: robust identi…cation Identi…cation obtains even without assignable goods (less robust) In all cases: sharing rule identi…ed up to an additive constant

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

! generic identi…ability if one exclusion restriction per person more robust if DF But: identi…cation ‘up to’ Revealed Preferences: bounds, which can be narrow

Cross-sectional data: demand as function of income (or total expenditures) and a DF

If one exclusive (or assignable) good per person: robust identi…cation Identi…cation obtains even without assignable goods (less robust) In all cases: sharing rule identi…ed up to an additive constant Note that the additive constant is welfare irrelevant (but matters for inequality) ! ‘welfare paradox’

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability from couples’ behavior only

Di¤erent approaches: With price variations (CE 09)

! generic identi…ability if one exclusion restriction per person more robust if DF But: identi…cation ‘up to’ Revealed Preferences: bounds, which can be narrow

Cross-sectional data: demand as function of income (or total expenditures) and a DF

If one exclusive (or assignable) good per person: robust identi…cation Identi…cation obtains even without assignable goods (less robust) In all cases: sharing rule identi…ed up to an additive constant Note that the additive constant is welfare irrelevant (but matters for inequality) ! ‘welfare paradox’ ... and could be pinned down by ‘out of sample’ global restrictions (e.g., y = 0 ) ρ = 0)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 7 / 17

Collective model: identi…ability with additional assumptions

Using singles

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 8 / 17

Collective model: identi…ability with additional assumptions

Using singles

Old idea (e.g. Laisney et all 2006, Myck et al. 2006): recovering ‘part

• f’ individual utilities from singles’ behavior

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 8 / 17

Collective model: identi…ability with additional assumptions

Using singles

Old idea (e.g. Laisney et all 2006, Myck et al. 2006): recovering ‘part

• f’ individual utilities from singles’ behavior

Requires assumptions:

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 8 / 17

Collective model: identi…ability with additional assumptions

Using singles

Old idea (e.g. Laisney et all 2006, Myck et al. 2006): recovering ‘part

• f’ individual utilities from singles’ behavior

Requires assumptions:

selection

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 8 / 17

Collective model: identi…ability with additional assumptions

Using singles

Old idea (e.g. Laisney et all 2006, Myck et al. 2006): recovering ‘part

• f’ individual utilities from singles’ behavior

Requires assumptions:

selection changes in preferences after marriage, ...

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 8 / 17

Collective model: identi…ability with additional assumptions

Using singles

Old idea (e.g. Laisney et all 2006, Myck et al. 2006): recovering ‘part

• f’ individual utilities from singles’ behavior

Requires assumptions:

selection changes in preferences after marriage, ...

More generally: comparing families with di¤erent compositions ! notion of ‘household technology’

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 8 / 17

Collective model: identi…ability with additional assumptions

Household technology (Becker): distinction between

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 9 / 17

Collective model: identi…ability with additional assumptions

Household technology (Becker): distinction between

commodities consumed ! vector q

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 9 / 17

Collective model: identi…ability with additional assumptions

Household technology (Becker): distinction between

commodities consumed ! vector q commodities purchased ! vector r

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 9 / 17

Collective model: identi…ability with additional assumptions

Household technology (Becker): distinction between

commodities consumed ! vector q commodities purchased ! vector r production technology: r = A (q1 + q2) with ∑

i

piri = y

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 9 / 17

Collective model: identi…ability with additional assumptions

Household technology (Becker): distinction between

commodities consumed ! vector q commodities purchased ! vector r production technology: r = A (q1 + q2) with ∑

i

piri = y

Main intuition: preferences do not change with marriage (or with family composition); only the technology changes

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 9 / 17

Collective model: identi…ability with additional assumptions

Household technology (Becker): distinction between

commodities consumed ! vector q commodities purchased ! vector r production technology: r = A (q1 + q2) with ∑

i

piri = y

Main intuition: preferences do not change with marriage (or with family composition); only the technology changes Consumption goods only and A diagonal ! Barten scales: ri = Aii

• qi

1 + qi 2

• (and Aii < 1 for economies of scale)...

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 9 / 17

Collective model: identi…ability with additional assumptions

Household technology (Becker): distinction between

commodities consumed ! vector q commodities purchased ! vector r production technology: r = A (q1 + q2) with ∑

i

piri = y

Main intuition: preferences do not change with marriage (or with family composition); only the technology changes Consumption goods only and A diagonal ! Barten scales: ri = Aii

• qi

1 + qi 2

• (and Aii < 1 for economies of scale)...

... but with the addition of intrahousehold allocation (and inequality) ! Indi¤erence scales

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 9 / 17

Indi¤erence scales

Old problem: compare income across families with di¤erent compositions; e.g., couple with income Y with a single ! which income y would the single need?

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 10 / 17

Indi¤erence scales

Old problem: compare income across families with di¤erent compositions; e.g., couple with income Y with a single ! which income y would the single need? Standard solution: equivalence scale

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 10 / 17

Indi¤erence scales

Old problem: compare income across families with di¤erent compositions; e.g., couple with income Y with a single ! which income y would the single need? Standard solution: equivalence scale De…nition: ‘y is such that the couple, endowed with Y , is as well o¤ as the single with y’

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 10 / 17

Indi¤erence scales

Old problem: compare income across families with di¤erent compositions; e.g., couple with income Y with a single ! which income y would the single need? Standard solution: equivalence scale De…nition: ‘y is such that the couple, endowed with Y , is as well o¤ as the single with y’ Problems with that de…nition:

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 10 / 17

Indi¤erence scales

Old problem: compare income across families with di¤erent compositions; e.g., couple with income Y with a single ! which income y would the single need? Standard solution: equivalence scale De…nition: ‘y is such that the couple, endowed with Y , is as well o¤ as the single with y’ Problems with that de…nition:

requires interpersonal comparison of utility

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 10 / 17

Indi¤erence scales

Old problem: compare income across families with di¤erent compositions; e.g., couple with income Y with a single ! which income y would the single need? Standard solution: equivalence scale De…nition: ‘y is such that the couple, endowed with Y , is as well o¤ as the single with y’ Problems with that de…nition:

requires interpersonal comparison of utility (worse) requires comparing utility of single with that of couple

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 10 / 17

Indi¤erence scales

Old problem: compare income across families with di¤erent compositions; e.g., couple with income Y with a single ! which income y would the single need? Standard solution: equivalence scale De…nition: ‘y is such that the couple, endowed with Y , is as well o¤ as the single with y’ Problems with that de…nition:

requires interpersonal comparison of utility (worse) requires comparing utility of single with that of couple (even worse) requires the notion of ‘utility of a couple’

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 10 / 17

Indi¤erence scales

Old problem: compare income across families with di¤erent compositions; e.g., couple with income Y with a single ! which income y would the single need? Standard solution: equivalence scale De…nition: ‘y is such that the couple, endowed with Y , is as well o¤ as the single with y’ Problems with that de…nition:

requires interpersonal comparison of utility (worse) requires comparing utility of single with that of couple (even worse) requires the notion of ‘utility of a couple’ (worst of all) not identi…ed in a deep sense (Blundell - Lewbel)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 10 / 17

Indi¤erence scales (cont.)

New notion: indi¤erence scale ! question: ‘y is such that the person, when single, is as well o¤ as she was when married’ Note that: No interpersonal comparison (compare utility of the same person in di¤erent situations) ... ... although requires utility to be comparable across marital status No ’utility of the couple’ - instead, utility of each spouse Moreover, (non parametrically) identi…ed ! revival of household technology!

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 11 / 17

Indi¤erence scales (cont.)

In practice: Couple: new prices (Barten!) π = A0p and sharing rule (ρ1, ρ2); i solves max Ui (qi) π0qi = ρi ! indirect utility V i (π, ρi) Single: max Ui (qi) p0qi = yi ! indirect utility V i (p, yi), estimated from singles’ demand IS based on ¯ yi such that V i (π, ρi) = V i (p, ¯ yi)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 12 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation But requires:

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation But requires:

singles’s demand observable (! problematic for children!)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation But requires:

singles’s demand observable (! problematic for children!) price variations + complex estimation techniques

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation But requires:

singles’s demand observable (! problematic for children!) price variations + complex estimation techniques

Simpler version: Dunbar-Lewbel-Pendakur (2011), Bargain-Donni (2012)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation But requires:

singles’s demand observable (! problematic for children!) price variations + complex estimation techniques

Simpler version: Dunbar-Lewbel-Pendakur (2011), Bargain-Donni (2012)

Basic assumption: ‘scale invariance’ (sharing rule proportional to income) ...

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation But requires:

singles’s demand observable (! problematic for children!) price variations + complex estimation techniques

Simpler version: Dunbar-Lewbel-Pendakur (2011), Bargain-Donni (2012)

Basic assumption: ‘scale invariance’ (sharing rule proportional to income) ... ... and a milder restrictions (parents’ preferences independent of the #

• f children)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation But requires:

singles’s demand observable (! problematic for children!) price variations + complex estimation techniques

Simpler version: Dunbar-Lewbel-Pendakur (2011), Bargain-Donni (2012)

Basic assumption: ‘scale invariance’ (sharing rule proportional to income) ... ... and a milder restrictions (parents’ preferences independent of the #

• f children)

Then simple and robust identi…cation (from Engel curves) ...

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation But requires:

singles’s demand observable (! problematic for children!) price variations + complex estimation techniques

Simpler version: Dunbar-Lewbel-Pendakur (2011), Bargain-Donni (2012)

Basic assumption: ‘scale invariance’ (sharing rule proportional to income) ... ... and a milder restrictions (parents’ preferences independent of the #

• f children)

Then simple and robust identi…cation (from Engel curves) ... ... even for children, despite no observable demand

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation But requires:

singles’s demand observable (! problematic for children!) price variations + complex estimation techniques

Simpler version: Dunbar-Lewbel-Pendakur (2011), Bargain-Donni (2012)

Basic assumption: ‘scale invariance’ (sharing rule proportional to income) ... ... and a milder restrictions (parents’ preferences independent of the #

• f children)

Then simple and robust identi…cation (from Engel curves) ... ... even for children, despite no observable demand Assumptions can be (somewhat) relaxed if DF

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Indi¤erence scales: identi…cation

Browning-Chiappori-Lewbel (2013) ! Main result: exact identi…cation But requires:

singles’s demand observable (! problematic for children!) price variations + complex estimation techniques

Simpler version: Dunbar-Lewbel-Pendakur (2011), Bargain-Donni (2012)

Basic assumption: ‘scale invariance’ (sharing rule proportional to income) ... ... and a milder restrictions (parents’ preferences independent of the #

• f children)

Then simple and robust identi…cation (from Engel curves) ... ... even for children, despite no observable demand Assumptions can be (somewhat) relaxed if DF Linearized version: Lechene-Pendakur-Wolf 2019

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 13 / 17

Empirical applications: DLP 2011

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 14 / 17

Empirical applications: Calvi 2019

Anderson and Ray (2010) estimate that in India, 1.7 million woman

• ver age 45 “are missing:” die at younger than expected ages.

! The number missing increases with age from 45 to 70. ! A puzzle - poverty kills, but household poverty rates do not correlate with women’s age.

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 15 / 17

Empirical applications: Calvi 2019

Anderson and Ray (2010) estimate that in India, 1.7 million woman

• ver age 45 “are missing:” die at younger than expected ages.

! The number missing increases with age from 45 to 70. ! A puzzle - poverty kills, but household poverty rates do not correlate with women’s age. Calvi (2019): Estimates DLP collective model for India couples ! Finds women’s resource shares decrease with age starting at 45. ! Calculates separate men’s and women’s poverty rates by age.

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 15 / 17

Empirical applications: Calvi 2019

Anderson and Ray (2010) estimate that in India, 1.7 million woman

• ver age 45 “are missing:” die at younger than expected ages.

! The number missing increases with age from 45 to 70. ! A puzzle - poverty kills, but household poverty rates do not correlate with women’s age. Calvi (2019): Estimates DLP collective model for India couples ! Finds women’s resource shares decrease with age starting at 45. ! Calculates separate men’s and women’s poverty rates by age. Finds that women’s poverty rate, after correcting for within household allocations by age, has .96 correlation with Anderson and Ray’s rate

• f missing women estimates by age.

! Measuring poverty at the individual level virtually completely explains the missing women puzzle!

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 15 / 17

Empirical applications: Calvi 2019

Anderson and Ray (2010) estimate that in India, 1.7 million woman

• ver age 45 “are missing:” die at younger than expected ages.

! The number missing increases with age from 45 to 70. ! A puzzle - poverty kills, but household poverty rates do not correlate with women’s age. Calvi (2019): Estimates DLP collective model for India couples ! Finds women’s resource shares decrease with age starting at 45. ! Calculates separate men’s and women’s poverty rates by age. Finds that women’s poverty rate, after correcting for within household allocations by age, has .96 correlation with Anderson and Ray’s rate

• f missing women estimates by age.

! Measuring poverty at the individual level virtually completely explains the missing women puzzle! More generally: individual poverty needs not coincide with household poverty!

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 15 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

standard household characteristics + expenditures (food and non-food)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

standard household characteristics + expenditures (food and non-food) fully individualized expenditure (both food and nonfood)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

standard household characteristics + expenditures (food and non-food) fully individualized expenditure (both food and nonfood)

Bargain-Lacroix-Tiberti (2019):

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

standard household characteristics + expenditures (food and non-food) fully individualized expenditure (both food and nonfood)

Bargain-Lacroix-Tiberti (2019):

Test identifying assumptions in DLP ! SAP mostly rejected, SAT and R-SAT not rejected (except for large families); independence of base not rejected

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

standard household characteristics + expenditures (food and non-food) fully individualized expenditure (both food and nonfood)

Bargain-Lacroix-Tiberti (2019):

Test identifying assumptions in DLP ! SAP mostly rejected, SAT and R-SAT not rejected (except for large families); independence of base not rejected Compare DLP-based predicted shares with actual ! works pretty well!

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

standard household characteristics + expenditures (food and non-food) fully individualized expenditure (both food and nonfood)

Bargain-Lacroix-Tiberti (2019):

Test identifying assumptions in DLP ! SAP mostly rejected, SAT and R-SAT not rejected (except for large families); independence of base not rejected Compare DLP-based predicted shares with actual ! works pretty well!

…rst kid: observed = 23.8% vs estimated = 23.5%; two kids: observed = 18.2% per kid vs estimated = 18.7% per kid

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

standard household characteristics + expenditures (food and non-food) fully individualized expenditure (both food and nonfood)

Bargain-Lacroix-Tiberti (2019):

Test identifying assumptions in DLP ! SAP mostly rejected, SAT and R-SAT not rejected (except for large families); independence of base not rejected Compare DLP-based predicted shares with actual ! works pretty well!

…rst kid: observed = 23.8% vs estimated = 23.5%; two kids: observed = 18.2% per kid vs estimated = 18.7% per kid Reasonable comparison of distributions of shares (Andrews test: not rejected, except for large families) + limited reranking

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

standard household characteristics + expenditures (food and non-food) fully individualized expenditure (both food and nonfood)

Bargain-Lacroix-Tiberti (2019):

Test identifying assumptions in DLP ! SAP mostly rejected, SAT and R-SAT not rejected (except for large families); independence of base not rejected Compare DLP-based predicted shares with actual ! works pretty well!

…rst kid: observed = 23.8% vs estimated = 23.5%; two kids: observed = 18.2% per kid vs estimated = 18.7% per kid Reasonable comparison of distributions of shares (Andrews test: not rejected, except for large families) + limited reranking

Results: Poverty assessed on a per capita basis ! overall poverty rate 36%; for children, 57% (obs.), 51% (est.)

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

standard household characteristics + expenditures (food and non-food) fully individualized expenditure (both food and nonfood)

Bargain-Lacroix-Tiberti (2019):

Test identifying assumptions in DLP ! SAP mostly rejected, SAT and R-SAT not rejected (except for large families); independence of base not rejected Compare DLP-based predicted shares with actual ! works pretty well!

…rst kid: observed = 23.8% vs estimated = 23.5%; two kids: observed = 18.2% per kid vs estimated = 18.7% per kid Reasonable comparison of distributions of shares (Andrews test: not rejected, except for large families) + limited reranking

Results: Poverty assessed on a per capita basis ! overall poverty rate 36%; for children, 57% (obs.), 51% (est.) GKLW 2019: compare various assignable goods

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Empirical applications: Bangladesh

Data: HIES, 2004 ! special component ! 1,039 households, randomly drawn from 33 districts (704 from rural areas).

standard household characteristics + expenditures (food and non-food) fully individualized expenditure (both food and nonfood)

Bargain-Lacroix-Tiberti (2019):

Test identifying assumptions in DLP ! SAP mostly rejected, SAT and R-SAT not rejected (except for large families); independence of base not rejected Compare DLP-based predicted shares with actual ! works pretty well!

…rst kid: observed = 23.8% vs estimated = 23.5%; two kids: observed = 18.2% per kid vs estimated = 18.7% per kid Reasonable comparison of distributions of shares (Andrews test: not rejected, except for large families) + limited reranking

Results: Poverty assessed on a per capita basis ! overall poverty rate 36%; for children, 57% (obs.), 51% (est.) GKLW 2019: compare various assignable goods Main message: more data of this type (with assignable goods)!

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 16 / 17

Conclusions: a booming literature, many directions

Intrahousehold allocation and inequality: ‘opening the black box’

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 17 / 17

Conclusions: a booming literature, many directions

Intrahousehold allocation and inequality: ‘opening the black box’

Standard view: only inequality across households considered; equivalence scales ! implicitly assumes equal (or fair) allocation ! largely counterfactual

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 17 / 17

Conclusions: a booming literature, many directions

Intrahousehold allocation and inequality: ‘opening the black box’

Standard view: only inequality across households considered; equivalence scales ! implicitly assumes equal (or fair) allocation ! largely counterfactual Indi¤erence scales: simultaneous estimation of domestic production patterns (economies of scale, etc.) and intrahousehold allocations

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 17 / 17

Conclusions: a booming literature, many directions

Intrahousehold allocation and inequality: ‘opening the black box’

Standard view: only inequality across households considered; equivalence scales ! implicitly assumes equal (or fair) allocation ! largely counterfactual Indi¤erence scales: simultaneous estimation of domestic production patterns (economies of scale, etc.) and intrahousehold allocations Main message: the black box can be opened ! Can evaluate intrahousehold inequality with (most) available data sets

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 17 / 17

Conclusions: a booming literature, many directions

Intrahousehold allocation and inequality: ‘opening the black box’

Standard view: only inequality across households considered; equivalence scales ! implicitly assumes equal (or fair) allocation ! largely counterfactual Indi¤erence scales: simultaneous estimation of domestic production patterns (economies of scale, etc.) and intrahousehold allocations Main message: the black box can be opened ! Can evaluate intrahousehold inequality with (most) available data sets Policy consequences: Donni (2009), Lise Seitz (2011), Bargain et al (2011), CSW (2017),...

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 17 / 17

Conclusions: a booming literature, many directions

Intrahousehold allocation and inequality: ‘opening the black box’

Standard view: only inequality across households considered; equivalence scales ! implicitly assumes equal (or fair) allocation ! largely counterfactual Indi¤erence scales: simultaneous estimation of domestic production patterns (economies of scale, etc.) and intrahousehold allocations Main message: the black box can be opened ! Can evaluate intrahousehold inequality with (most) available data sets Policy consequences: Donni (2009), Lise Seitz (2011), Bargain et al (2011), CSW (2017),...

! General conclusions:

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 17 / 17

Conclusions: a booming literature, many directions

Intrahousehold allocation and inequality: ‘opening the black box’

Standard view: only inequality across households considered; equivalence scales ! implicitly assumes equal (or fair) allocation ! largely counterfactual Indi¤erence scales: simultaneous estimation of domestic production patterns (economies of scale, etc.) and intrahousehold allocations Main message: the black box can be opened ! Can evaluate intrahousehold inequality with (most) available data sets Policy consequences: Donni (2009), Lise Seitz (2011), Bargain et al (2011), CSW (2017),...

! General conclusions:

Further improvements on these methods are continuing to be developed...

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 17 / 17

Conclusions: a booming literature, many directions

Intrahousehold allocation and inequality: ‘opening the black box’

Standard view: only inequality across households considered; equivalence scales ! implicitly assumes equal (or fair) allocation ! largely counterfactual Indi¤erence scales: simultaneous estimation of domestic production patterns (economies of scale, etc.) and intrahousehold allocations Main message: the black box can be opened ! Can evaluate intrahousehold inequality with (most) available data sets Policy consequences: Donni (2009), Lise Seitz (2011), Bargain et al (2011), CSW (2017),...

! General conclusions:

Further improvements on these methods are continuing to be developed... ... but systematic investigation of real data (including the construction

• f better data sets) indispensible

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 17 / 17

Conclusions: a booming literature, many directions

Intrahousehold allocation and inequality: ‘opening the black box’

Standard view: only inequality across households considered; equivalence scales ! implicitly assumes equal (or fair) allocation ! largely counterfactual Indi¤erence scales: simultaneous estimation of domestic production patterns (economies of scale, etc.) and intrahousehold allocations Main message: the black box can be opened ! Can evaluate intrahousehold inequality with (most) available data sets Policy consequences: Donni (2009), Lise Seitz (2011), Bargain et al (2011), CSW (2017),...

! General conclusions:

Further improvements on these methods are continuing to be developed... ... but systematic investigation of real data (including the construction

• f better data sets) indispensible

Don’t forget Distribution Factors!

Chiappori (Columbia University) Measuring Intra-household Inequality World Bank HQ, Nov. 2019 17 / 17