Inequality of Opportunity in developing economies A cross-country - - PowerPoint PPT Presentation

Inequality of Opportunity in developing economies

A cross-country analysis with LIS harmonised data Ana Suárez Álvarez

University of Oviedo Department of Applied Economics

LIS Research Visit, 2017

Table of contents

1

Introduction

2

Data and variables

3

Estimation of IOp

4

Bootstrapping methodology

5

Concludes

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 2 / 24

Introduction

Table of contents

1

Introduction

2

Data and variables

3

Estimation of IOp

4

Bootstrapping methodology

5

Concludes

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 3 / 24

Introduction

Background I: The concept of Inequality of Opportunity (IOp)

Inequality depends on

• Circumstances (C) Unfair Inequality
• Efforts (E) Fair Inequality

Inequality of Opportunity (IOp) Tries to estimate the part of overall inequality considered unfair since individuals cannot be held responsible for it.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 4 / 24

Introduction

Background I: The concept of Inequality of Opportunity (IOp)

Inequality depends on

• Circumstances (C) Unfair Inequality
• Efforts (E) Fair Inequality

Inequality of Opportunity (IOp) Tries to estimate the part of overall inequality considered unfair since individuals cannot be held responsible for it.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 4 / 24

Introduction

Background I: The concept of Inequality of Opportunity (IOp)

Inequality depends on

• Circumstances (C) Unfair Inequality
• Efforts (E) Fair Inequality

Inequality of Opportunity (IOp) Tries to estimate the part of overall inequality considered unfair since individuals cannot be held responsible for it.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 4 / 24

Introduction

Background II: Temporal framework

It is of great important to analyse the evolution of inequality and IOp indicators because we can: Assess the effect of different economic scenarios in Ineq and IOp Understand the relationship between Inequality and IOp indicators Know if the contribution of Circumstances to IOp and Inequality changes over time Improve the design of public policies to soften or eliminate Inequality and/or IOp For this purpose I analyse the significance of over time changes using the bootstrap methodology.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 5 / 24

Introduction

Background II: Temporal framework

It is of great important to analyse the evolution of inequality and IOp indicators because we can: Assess the effect of different economic scenarios in Ineq and IOp Understand the relationship between Inequality and IOp indicators Know if the contribution of Circumstances to IOp and Inequality changes over time Improve the design of public policies to soften or eliminate Inequality and/or IOp For this purpose I analyse the significance of over time changes using the bootstrap methodology.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 5 / 24

Introduction

Background II: Temporal framework

It is of great important to analyse the evolution of inequality and IOp indicators because we can: Assess the effect of different economic scenarios in Ineq and IOp Understand the relationship between Inequality and IOp indicators Know if the contribution of Circumstances to IOp and Inequality changes over time Improve the design of public policies to soften or eliminate Inequality and/or IOp For this purpose I analyse the significance of over time changes using the bootstrap methodology.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 5 / 24

Introduction

Background II: Temporal framework

It is of great important to analyse the evolution of inequality and IOp indicators because we can: Assess the effect of different economic scenarios in Ineq and IOp Understand the relationship between Inequality and IOp indicators Know if the contribution of Circumstances to IOp and Inequality changes over time Improve the design of public policies to soften or eliminate Inequality and/or IOp For this purpose I analyse the significance of over time changes using the bootstrap methodology.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 5 / 24

Introduction

Background II: Temporal framework

It is of great important to analyse the evolution of inequality and IOp indicators because we can: Assess the effect of different economic scenarios in Ineq and IOp Understand the relationship between Inequality and IOp indicators Know if the contribution of Circumstances to IOp and Inequality changes over time Improve the design of public policies to soften or eliminate Inequality and/or IOp For this purpose I analyse the significance of over time changes using the bootstrap methodology.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 5 / 24

Introduction

Background II: Temporal framework

It is of great important to analyse the evolution of inequality and IOp indicators because we can: Assess the effect of different economic scenarios in Ineq and IOp Understand the relationship between Inequality and IOp indicators Know if the contribution of Circumstances to IOp and Inequality changes over time Improve the design of public policies to soften or eliminate Inequality and/or IOp For this purpose I analyse the significance of over time changes using the bootstrap methodology.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 5 / 24

Introduction

Objectives

1 Estimate Inequality and Inequality of Opportunity 2 Estimate the contribution of each circumstance to IOp 3 Assess changes overtime on income inequality and inequality of

• pportunity

Stratified bootstrap estimation on Inequality and IOp indices

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 6 / 24

Introduction

Objectives

1 Estimate Inequality and Inequality of Opportunity 2 Estimate the contribution of each circumstance to IOp 3 Assess changes overtime on income inequality and inequality of

• pportunity

Stratified bootstrap estimation on Inequality and IOp indices

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 6 / 24

Introduction

Objectives

1 Estimate Inequality and Inequality of Opportunity 2 Estimate the contribution of each circumstance to IOp 3 Assess changes overtime on income inequality and inequality of

• pportunity

Stratified bootstrap estimation on Inequality and IOp indices

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 6 / 24

Data and variables

Table of contents

1

Introduction

2

Data and variables

3

Estimation of IOp

4

Bootstrapping methodology

5

Concludes

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 7 / 24

Data and variables

LIS Data

6 countries are analysed:

1 Brazil

2006, 2009, 2011 and 2013

2 Egypt

2012

3 Guatemala

2006, 2011 and 2014

4 India

2004 and 2011

5 Peru

2004, 2007, 2010 and 2013

6 South Africa

2008, 2010, 2012

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 8 / 24

Variables

Variable of Advantage Equivalised disposable income Variable for which we measure inequality and IOp Accurate approximation to the truly available income each individual benefits from OECD-Equivalence scale: e = 1 + 0.5(N14+ − 1) + 0.3N13− where N14+ is the number of HH members with 14 or more years old and N13− the number

• f HH members below 14 years.

Circumstances Variables used to approximate the true Inequality of Opportunity Gender ⇒ 2 categories Parental education ⇒ 3 categories Immigrant or not ⇒ 2 categories Age group ⇒ 5 categories Density ⇒ 2 categories Ethnic 2 categories for:

Brazil (white & others) Guatemala (non indig.&

• thers) India (high castes &
• thers) South Africa (african

& others)

Variables

Variable of Advantage Equivalised disposable income Variable for which we measure inequality and IOp Accurate approximation to the truly available income each individual benefits from OECD-Equivalence scale: e = 1 + 0.5(N14+ − 1) + 0.3N13− where N14+ is the number of HH members with 14 or more years old and N13− the number

• f HH members below 14 years.

Circumstances Variables used to approximate the true Inequality of Opportunity Gender ⇒ 2 categories Parental education ⇒ 3 categories Immigrant or not ⇒ 2 categories Age group ⇒ 5 categories Density ⇒ 2 categories Ethnic 2 categories for:

Brazil (white & others) Guatemala (non indig.&

• thers) India (high castes &
• thers) South Africa (african

& others)

Variables

Variable of Advantage Equivalised disposable income Variable for which we measure inequality and IOp Accurate approximation to the truly available income each individual benefits from OECD-Equivalence scale: e = 1 + 0.5(N14+ − 1) + 0.3N13− where N14+ is the number of HH members with 14 or more years old and N13− the number

• f HH members below 14 years.

Circumstances Variables used to approximate the true Inequality of Opportunity Gender ⇒ 2 categories Parental education ⇒ 3 categories Immigrant or not ⇒ 2 categories Age group ⇒ 5 categories Density ⇒ 2 categories Ethnic 2 categories for:

Brazil (white & others) Guatemala (non indig.&

• thers) India (high castes &
• thers) South Africa (african

& others)

Estimation of IOp

Table of contents

1

Introduction

2

Data and variables

3

Estimation of IOp

4

Bootstrapping methodology

5

Concludes

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 10 / 24

Estimation of IOp

Method and indicators

Parametric Procedure Estimation of lnyi = Ciβ + ui, reduced form of lnyi = Ciα + Eiδ + vi, where E(C, e). Absolute and relative IOp: IOpNP

A

= I({µt

i })

IOpNP

R

= I({µt

i })

I(y)

Assess the contribution of each C: Shapley value procedure Inequality indices used     

• GE(0) Path-Independent Decomposition

Foster and Shneyerov [2000]

• Gini Widely used and easy to interpret (Bounded)
• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 11 / 24

Estimation of IOp

Method and indicators

Parametric Procedure Estimation of lnyi = Ciβ + ui, reduced form of lnyi = Ciα + Eiδ + vi, where E(C, e). Absolute and relative IOp: IOpNP

A

= I({µt

i })

IOpNP

R

= I({µt

i })

I(y)

Assess the contribution of each C: Shapley value procedure Inequality indices used     

• GE(0) Path-Independent Decomposition

Foster and Shneyerov [2000]

• Gini Widely used and easy to interpret (Bounded)
• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 11 / 24

Estimation of IOp

Method and indicators

Parametric Procedure Estimation of lnyi = Ciβ + ui, reduced form of lnyi = Ciα + Eiδ + vi, where E(C, e). Absolute and relative IOp: IOpNP

A

= I({µt

i })

IOpNP

R

= I({µt

i })

I(y)

Assess the contribution of each C: Shapley value procedure Inequality indices used     

• GE(0) Path-Independent Decomposition

Foster and Shneyerov [2000]

• Gini Widely used and easy to interpret (Bounded)
• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 11 / 24

Estimation of IOp

Method and indicators

Parametric Procedure Estimation of lnyi = Ciβ + ui, reduced form of lnyi = Ciα + Eiδ + vi, where E(C, e). Absolute and relative IOp: IOpNP

A

= I({µt

i })

IOpNP

R

= I({µt

i })

I(y)

Assess the contribution of each C: Shapley value procedure Inequality indices used     

• GE(0) Path-Independent Decomposition

Foster and Shneyerov [2000]

• Gini Widely used and easy to interpret (Bounded)
• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 11 / 24

Estimation of IOp

Some graphs I: Income Inequality and Inequality of Opportunity

Figure: Income Inequality Figure: Relative Inequality of Opportunity

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 12 / 24

Estimation of IOp

Some graphs II: Contribution of circumstances to IOp

Figure: Contribution of C to IOp

Average contribution by year Gender Rural Educ Immigr Ethnic Age 2.91% 15.02% 56.04% 2.62% 21.35% 8.34%

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 13 / 24

Estimation of IOp

Some graphs II: Contribution of circumstances to IOp

Figure: Contribution of C to IOp

Average contribution by year Gender Rural Educ Immigr Ethnic Age 2.91% 15.02% 56.04% 2.62% 21.35% 8.34%

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 13 / 24

Bootstrapping methodology

Table of contents

1

Introduction

2

Data and variables

3

Estimation of IOp

4

Bootstrapping methodology

5

Concludes

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 14 / 24

Bootstrapping methodology

The bootstrap method:

Definition Being F the distribution function, we denote by ˆ F the distribution extracted through the Bootstrapping procedure which relies on B samples with replacement of size n. ˆ F is therefore equivalent to F since estimates are built from a random sample representative to the population. Why is the Bootstrap method useful? Estimates of Inequality and Inequality of Opportunity are non parametric Bootstrapping technique (Efron and Tibshirani [1986], Hall [1994]). Allow us not to impose any functional form to the data. Relies on random samples with replacement We can construct measures of accuracy: CI can be calculated to asses changes over time.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 15 / 24

Bootstrapping methodology

The bootstrap method:

Definition Being F the distribution function, we denote by ˆ F the distribution extracted through the Bootstrapping procedure which relies on B samples with replacement of size n. ˆ F is therefore equivalent to F since estimates are built from a random sample representative to the population. Why is the Bootstrap method useful? Estimates of Inequality and Inequality of Opportunity are non parametric Bootstrapping technique (Efron and Tibshirani [1986], Hall [1994]). Allow us not to impose any functional form to the data. Relies on random samples with replacement We can construct measures of accuracy: CI can be calculated to asses changes over time.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 15 / 24

Bootstrapping methodology

Stratified Bootstrap procedure

Previously tested and studied by Bickel and Freedman [1984] For each dataset I have a sample n taken from an unknown distribution function F. I conduct B = 1000 samples for each year of size n Involves taking into account the structure of the original sample by types of individuals. The procedure is drawn at type-level and the share of individuals within each type remains constant. Confidence Intervals: Percentile Interval method Easy to implement: involves using as lower and upper bounds the α − th percentiles of the ˆ F distribution. Hence an equal-tailed confidence interval of 1 − 2α from the original sample estimator ˆ θ is given by: (ˆ θB

α , ˆ

θB

1−α)

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 16 / 24

Bootstrapping methodology

Stratified Bootstrap procedure

Previously tested and studied by Bickel and Freedman [1984] For each dataset I have a sample n taken from an unknown distribution function F. I conduct B = 1000 samples for each year of size n Involves taking into account the structure of the original sample by types of individuals. The procedure is drawn at type-level and the share of individuals within each type remains constant. Confidence Intervals: Percentile Interval method Easy to implement: involves using as lower and upper bounds the α − th percentiles of the ˆ F distribution. Hence an equal-tailed confidence interval of 1 − 2α from the original sample estimator ˆ θ is given by: (ˆ θB

α , ˆ

θB

1−α)

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 16 / 24

Bootstrapping methodology

Summary of the Standard Bootstrap results

If CI do not overlap we can say changes over time are significant: Brazil: Significant decrease in both Income Inequality & IOp between 2006-2009 & 2009-2011. For the last two years 2011-2013 only the decrase in IOp is significant. Guatemala: Significant decrase in Income Inequality and IOp during the whole period analysed (2006,2011,2014) India: Significant increase between 2004-2011 for both Income inequality & IOp Peru: Significant decrase IOp & Income inequality in all periods (2004, 2007, 2010, 2013) South Africa: IOP & Ineq decreases are not significant only IOp measured by the Gini index

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 17 / 24

Bootstrapping methodology

Summary of the Standard Bootstrap results

If CI do not overlap we can say changes over time are significant: Brazil: Significant decrease in both Income Inequality & IOp between 2006-2009 & 2009-2011. For the last two years 2011-2013 only the decrase in IOp is significant. Guatemala: Significant decrase in Income Inequality and IOp during the whole period analysed (2006,2011,2014) India: Significant increase between 2004-2011 for both Income inequality & IOp Peru: Significant decrase IOp & Income inequality in all periods (2004, 2007, 2010, 2013) South Africa: IOP & Ineq decreases are not significant only IOp measured by the Gini index

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 17 / 24

Bootstrapping methodology

Summary of the Standard Bootstrap results

If CI do not overlap we can say changes over time are significant: Brazil: Significant decrease in both Income Inequality & IOp between 2006-2009 & 2009-2011. For the last two years 2011-2013 only the decrase in IOp is significant. Guatemala: Significant decrase in Income Inequality and IOp during the whole period analysed (2006,2011,2014) India: Significant increase between 2004-2011 for both Income inequality & IOp Peru: Significant decrase IOp & Income inequality in all periods (2004, 2007, 2010, 2013) South Africa: IOP & Ineq decreases are not significant only IOp measured by the Gini index

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 17 / 24

Bootstrapping methodology

Summary of the Standard Bootstrap results

If CI do not overlap we can say changes over time are significant: Brazil: Significant decrease in both Income Inequality & IOp between 2006-2009 & 2009-2011. For the last two years 2011-2013 only the decrase in IOp is significant. Guatemala: Significant decrase in Income Inequality and IOp during the whole period analysed (2006,2011,2014) India: Significant increase between 2004-2011 for both Income inequality & IOp Peru: Significant decrase IOp & Income inequality in all periods (2004, 2007, 2010, 2013) South Africa: IOP & Ineq decreases are not significant only IOp measured by the Gini index

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 17 / 24

Bootstrapping methodology

Summary of the Standard Bootstrap results

If CI do not overlap we can say changes over time are significant: Brazil: Significant decrease in both Income Inequality & IOp between 2006-2009 & 2009-2011. For the last two years 2011-2013 only the decrase in IOp is significant. Guatemala: Significant decrase in Income Inequality and IOp during the whole period analysed (2006,2011,2014) India: Significant increase between 2004-2011 for both Income inequality & IOp Peru: Significant decrase IOp & Income inequality in all periods (2004, 2007, 2010, 2013) South Africa: IOP & Ineq decreases are not significant only IOp measured by the Gini index

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 17 / 24

Concludes

Table of contents

1

Introduction

2

Data and variables

3

Estimation of IOp

4

Bootstrapping methodology

5

Concludes

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 18 / 24

Concludes

Concludes

The bootstrap procedure allow us to present a more robust evidence of changes in inequality indicators and the contribution of circumstances to IO Significant changes are observed for all the countries analysed. IOp increases in:

India

On the contrary, IOp decreases in

Brazil Guatemala Peru South Africa

Finally, it is important to highlight the role played by the circumstance Parental Education which is the most important for Inequality of Opportunity.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 19 / 24

Concludes

Concludes

The bootstrap procedure allow us to present a more robust evidence of changes in inequality indicators and the contribution of circumstances to IO Significant changes are observed for all the countries analysed. IOp increases in:

India

On the contrary, IOp decreases in

Brazil Guatemala Peru South Africa

Finally, it is important to highlight the role played by the circumstance Parental Education which is the most important for Inequality of Opportunity.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 19 / 24

Concludes

Concludes

The bootstrap procedure allow us to present a more robust evidence of changes in inequality indicators and the contribution of circumstances to IO Significant changes are observed for all the countries analysed. IOp increases in:

India

On the contrary, IOp decreases in

Brazil Guatemala Peru South Africa

Finally, it is important to highlight the role played by the circumstance Parental Education which is the most important for Inequality of Opportunity.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 19 / 24

Concludes

Concludes

The bootstrap procedure allow us to present a more robust evidence of changes in inequality indicators and the contribution of circumstances to IO Significant changes are observed for all the countries analysed. IOp increases in:

India

On the contrary, IOp decreases in

Brazil Guatemala Peru South Africa

Finally, it is important to highlight the role played by the circumstance Parental Education which is the most important for Inequality of Opportunity.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 19 / 24

Concludes

Concludes

The bootstrap procedure allow us to present a more robust evidence of changes in inequality indicators and the contribution of circumstances to IO Significant changes are observed for all the countries analysed. IOp increases in:

India

On the contrary, IOp decreases in

Brazil Guatemala Peru South Africa

Finally, it is important to highlight the role played by the circumstance Parental Education which is the most important for Inequality of Opportunity.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 19 / 24

Inequality of Opportunity in developing economies

A cross-country analysis with LIS harmonised data Ana Suárez Álvarez

University of Oviedo Department of Applied Economics

LIS Research Visit, 2017

Concludes

References I

• P. J. . Bickel and D. A. Freedman. Asymptotic normality and the bootstrap

in stratified sampling. Ann. Stat., 12(2):470–482, 1984.

• A. Björklund, M. Jäntti, and J. E. Roemer. Equality of opportunity and the

distribution of long-run income in Sweden. Soc. Choice Welfare, 39(2-3): 675–696, 2012.

• F. Bourguignon, F. H. G. Ferreira, and M. Menéndez. Inequality of

Opportunity in Brazil. Rev. Income Wealth, 53(4):585–618, dec 2007.

• D. Checchi and V. Peragine. Inequality of opportunity in Italy. J. Econ.

Inequal., 8(4):429–450, 2010.

• D. Checchi, V. Peragine, and L. Serlenga. Fair and unfair income

inequalities in Europe. IZA Discuss. Pap., 5025(174):1–29, 2010.

• D. Checchi, V. Peragine, and L. Serlenga. Income Inequality and

Opportunity Inequality in Europe: Recent Trends and Explaning Factors. 2015.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 21 / 24

Concludes

References II

• D. Cogneau and S. Mesplé-Somps. Inequality of opportunity for income in

five countries of Africa. Res. Econ. Inequal., 16(OCTOBER 2008): 99–128, 2008.

• B. Efron and R. Tibshirani. Bootstrap Methods for Standard

Errors,Confidence Intervals, and Other Measures of Statistical Accuracy.

• Stat. Sci., 1(1):54–77, 1986. ISSN 0883-4237, 2168-8745.
• F. H. G. Ferreira and J. Gignoux. The measurement of inequality of
• pportunity: Theory and an application to Latin America. Rev. Income

Wealth, 57(4):622–657, 2011.

• J. E. Foster and A. A. Shneyerov. Path independent inequality measures.
• J. Econ. Theory, 91(2):199–222, 2000.
• P. Hall. Methodology and Theory for the Bootstrap. In R. Engle and
• D. McFadden, editors, Handb. Econom., chapter Chapter 39, pages

2341–2381. Elsevier Science, 1994.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 22 / 24

Concludes

References III

• G. A. Marrero and J. G. Rodríguez. Inequality of Opportunity in Europe.
• Rev. Income Wealth, 58(4):597–621, 2012.
• S. Mussard and M. N. P. Alperin. Heterogeneous Groups and Income

Sources Measuring Significance of Inequalities with Heterogeneous Groups and Income Sources. CREDI Work. Pap., 06-13:1–16, 2013.

• P. Piraino. Intergenerational Earnings Mobility and Equality of Opportunity

in South Africa. World Dev., 67(November 2012):396–405, 2015.

• A. Singh. Inequality of opportunity in earnings and consumption

expenditure: The case of Indian men. Rev. Income Wealth, 58(1): 79–106, 2012.

• A. Suárez Álvarez

IOp in Developing economies LIS Research Visit 23 / 24

Inequality of Opportunity in developing economies

A cross-country analysis with LIS harmonised data Ana Suárez Álvarez

University of Oviedo Department of Applied Economics

LIS Research Visit, 2017