global inequality · global inequality: relatively lower, absolutely higher 1 miguel nino-zaraz~ ua...

Global inequality:Relatively lower, absolutely higher1

Miguel Nino-Zarazua2, Laurence Roope3 and Finn Tarp4

ABSTRACT

This paper measures trends in global interpersonal inequality during 1975-2010 using data

from UNU-WIDER’s World Income Inequality Database. The picture that emerges of the evo-

lution of inequality during the past 35 years using ‘absolute,’ and even a ‘centrist’ measure of

inequality, is found to be irreconcilable with the results obtained using standard ‘relative’ inequal-

ity measures such as the Gini or Mean Log Deviation. Relative global inequality has declined

steadily and substantially over the decades, driven by declining inequality between countries,

in large part resulting from the emergence of India and China. In contrast, ‘absolute’ inequal-

ity, as captured by the Variance, has increased dramatically and unabated. Like the Variance,

our ‘centrist’ inequality indicator, the Krtscha measure, also registers a pronounced increase in

inequality over the decades. Unlike the Variance, but in line with our ‘relative’ inequality mea-

sures, the Krtscha registers a notable decline in global inequality during 2005 to 2010. A critical

question posed by our findings is whether it is plausible that ‘absolute’ and ‘centrist’ measures

of inequality would ever signify declining inequality in the face of global economic growth of the

scale experienced during recent decades. Using counterfactual analyses we conclude that it is not

feasible for ‘absolute’ measures to decline in such circumstances but possible, if very demanding,

for ‘centrist’ measures like the Krtscha to do so.

This version: 10th February, 2015.

Keywords: global interpersonal inequality, inequality, inequality measurement

JEL Classifications: D31, D63, E01, O15,

1We are grateful to participants at seminars at the Universities of Helsinki, Oxford, Bielefeld, andBeijing Normal University, and at the September 2014 UNU-WIDER Conference on “Inequality - mea-surement, trends, impact and policy” in Helsinki, for their helpful comments on earlier versions of thispaper. Naturally, any remaining errors are ours.

2UNU-WIDER, Finland. Corresponding author email: [email protected] of Oxford. Email: [email protected]. Email: [email protected]

1 Introduction

Since the turn of the century, inequality has become one of the most prominent political

issues of our time. The World Economic Forum’s Global Risks 2013 report identified

‘global income disparity’ as the global risk most likely to manifest itself over the next

ten years. Issues of taxation and redistribution were central to the debate in the 2012

US presidential election and in a number of recent general elections in Europe. In 2014,

Thomas Piketty (2013)’s 700 page epic treatise on wealth and inequality reached number

one on the New York Times Best Sellers List for best selling hardcover nonfiction.

There has recently been significant interest in the economic literature in the level of,

and trends in, various concepts of global inequality. The earliest of these papers were

predominantly focused on either ‘within-country’ inequality, as in Cornia and Kiiski

(2001), or ‘between-country’ inequality (see, for example, Boltho and Toniolo (1999);

Firebaugh (1999, 2003); Melchior, Telle and Wiig (2000)). Much of the impetus for these

studies came from concerns as to what impact the recent era of globalization may have

had on inequality (see, for example, Richardson (1995); Wood (1995) and Williamson

(1999), and also UNDP (1999), which explicitly called for policies to mitigate rising

inequality caused by economic globalization).

To quote Milanovic (2002:52), a direct implication of globalization is “that national

borders are becoming less important, and that every individual may, in theory, be re-

garded simply as a citizen of the world.” The literature on global inequality trends began

to focus on estimating levels of ‘global interpersonal inequality.’ In this approach, the

global distribution of income of all the citizens of the world is constructed from national

accounts and/or survey data.1 Inequality is subsequently measured, based on this global

interpersonal distribution of income. Notable contributions in this area have been made

by Korzeniewicz and Moran (1997); Chotikapanich, Valenzuela and Rao (1997); Schultz

(1998); Milanovic (2002, 2005, 2012); Bourguignon and Morrisson (2002); Dickhavoc

and Ward (2002); Bhalla (2002); Dowrick and Akmal (2005); Sala-i-Martin (2006); and

Atkinson and Brandolini (2010). See also Anand and Segal (2008) for a critical review

of this strand of literature.

This study adds to the body of literature on trends in global interpersonal inequality

which, for convenience, we will sometimes refer to hereafter simply as ‘global inequality.’

Using the most recent version of UNU-WIDER’s World Income Inequality Database

(WIID), we estimate global inequality levels and their trends during the period from

1Actually some studies have focused on income and others on consumption. We use the term ‘income’loosely for now but will discuss some of the issues arising from the important distinction between thetwo in Section 3.

2

1975 to 2010. Most of the aforementioned studies consider trends in global inequality

only up to the mid 1990s. To the best of our knowledge, this is the first comprehensive

study on global inequality which spans the pre- and post-financial crisis periods. It is

also one of the first studies to analyse global inequality using not only ‘relative’ inequality

measures, but also ‘absolute’ and ‘centrist’ measures, with their very different properties

and normative underpinnings.2

It has long been accepted in the literature on social choice and welfare economics

that arguments for ‘relative’ inequality measures vis-a-vis ‘absolute’ or ‘centrist’ mea-

sures are firmly in the domain of normative economics. Amiel and Cowell (1992, 1999),

among others, have demonstrated in experimental work that people have a range of

views regarding how distributions should be ranked with respect to inequality. Yet there

has been a paucity of empirical studies within Economics that have employed ‘absolute’

or ‘centrist’ measures. There are signs that this is beginning to change. A number of

prominent studies have recently emphasised the importance of avoiding unnecessarily

restricting the discourse on inequality to relative inequality alone (e.g. Ravallion (2003),

Subramanian and Jayaraj (2014) and, with particular reference to global inequality,

Atkinson and Brandolini (2010) and Bosmans et al. (2014)).

In this paper we find that the picture that emerges of the evolution of inequality

during the past 35 years using absolute, and even a centrist measure of inequality, is

irreconcilable with the results obtained using standard relative inequality measures such

as the Gini or Mean Log Deviation (MLD). Our headline findings are that relative global

inequality has declined steadily and substantially over the decades, driven primarily by

declining inequality between countries, in large part resulting from the emergence of India

and China. In contrast, absolute inequality, as captured by the Variance, has increased

dramatically and unabated throughout the period analysed, and has only been dampened

very modestly indeed by India and China catching up to become middle income countries.

Like the Variance, our centrist inequality indicator, the Krtscha measure, also registers a

very pronounced increase in inequality over the decades. Unlike the Variance though, the

Krtscha measure regards the emergence of India and China as an important dampening

force on inequality. Also unlike the Variance, but in line with our relative inequality

measures, the Krtscha registers a notable decline in global inequality during 2005 to

2010. While our analysis is purely descriptive, this might be interpreted as suggesting

2‘Relative’ inequality measures are those which are invariant under equiproportional increases in allincomes; ‘absolute’ inequality measures are those which register no change when the same absoluteamount of income is added to all incomes; ‘centrist’ inequality measures are those which register anincrease in inequality if all incomes increase equiproportionally, and a decrease if the same absoluteamount of income is added to all incomes.

3

that one impact of the global financial crisis is a decline in global inequality, due to its

relatively larger adverse impact on high income developed economies such as the US

and Europe than on populous lower and middle income countries, including India and

China, which continued to close the gap.

The divergent nature of the trends in inequality obtained using relative and abso-

lute inequality measures pose some important questions and challenges for policymakers.

One such critical question is whether it is plausible that absolute and centrist measures

of inequality would ever register a decrease in the face of global economic growth of the

scale experienced during recent decades. We analyse this question by constructing coun-

terfactual income distributions, in which all countries in the world grew as they actually

did during 1975 to 2010, but where the domestic income distributions in all countries in

2010 were at Nordic levels, among the most equal in the world. When global inequality

is re-estimated for such counterfactual distributions, it turns out that absolute measures

continue to register a very large increase in inequality, suggesting that declining absolute

inequality is probably incompatible with substantial global growth. Analogous counter-

factual analysis using the Krtscha measure suggests that high sustained growth and

declining inequality, according to this measure, is possible, though highly demanding.

The rest of the paper is organized as follows. In Section 2 we discuss the importance

of measuring inequality in general. In Section 3 we discuss concepts and challenges in-

volved in measuring global inequality. In particular, we discuss the relationship between

global inequality and within-country and between-country inequality. In Section 4 we

discuss some theoretical aspects of inequality measurement, with a particular focus on

the different normative underpinnings of ‘relative,’ ‘absolute’ and ‘centrist’ inequality

measures. In Section 5 we describe the data and discuss some of the empirical chal-

lenges and techniques. We also formally describe a number of counterfactual analyses

performed in the study. In Section 6 we provide all our estimates of trends in global

inequality, for our different types of measures. We also provide the results for our various

counterfactual analyses. In Section 7 we do some sensitivity analysis on our main results

and discuss the robustness of our estimates. We also discuss how our results compare

to other studies in the literature. A discussion of our main findings, and concluding

remarks on the implications of our study, are offered in Section 8.

2 The importance of inequality measurement

There are many reasons why one might have a concern for inequality and wish to measure

it. Perhaps the most obvious reason is that high levels of inequality may be deemed to be

4

socially unfair. Since the time of ancient societies, scholars have been concerned about

the possible negative effects of inequality on peace and prosperity. In his dialogue with

Adeimantus, and reproduced in Plato’s Republic (1901:422), Socrates was already aware

of the pervasive effects of indiscriminate wealth in deteriorating peace and order. Also,

under the influence of Plato, Aristotle (1954:1379a) saw in inequity a source of conflict

and anger. In that context, the state was seen as fundamental to ensuring peace and

prosperity through the procurement of justice and social equality (Plato 1901).

Classical economists, from Adam Smith and David Ricardo to Karl Marx, were con-

cerned about the effects of unfair distribution of income on factors of production, and

social classes. These were, of course, discussions in the domain of normative principles.

Others have also argued for the significance of inequality of opportunity as an obstacle

for progress and development. Dworkin (1981a, 1981b), for example, argued that egal-

itarians should seek to equalize resources rather than outcomes. Roemer (1993, 1998)

introduced a model which separated the determinants of the welfare outcomes a person

experiences into circumstances and effort. He argued that individuals should only be

held responsible for the latter. In contrast to effort, a person has no choice with re-

spect to the circumstances of the environment they are born into. In Roemer’s (1993,

1998) framework, an equal-opportunity policy is an intervention which ‘levels the playing

field’ by ensuring that equal outcomes in achievement accrue to individuals who have

expended the same amount of effort.3 4 Whilst inequality of opportunity is beyond the

scope of this paper, its importance for the analysis of inequality is undeniable.

With the rise of development economics theory, the concerns of inequality were linked

to the development process, giving an emphasis to the trend of increasing inequality

as countries transited from agrarian to industrial societies; see Lewis (1954); Kuznets

(1955).

More recently, and following the neoclassical paradigm of the Solow growth model

(Solow 1956), there has been a particular focus on the relationship between inequality

and economic growth. The first empirical studies of the relationship between growth

and inequality found an unambiguous detrimental impact of inequality on growth. For

3For further philiosophical and normative discussion of equality of opportunity and related issues seeArneson (1989); Cohen (1989); Fleurbaey (2008); and Rawls (1971).

4There have also been some recent empirical attempts to measure inequality of opportunity. Forexample, drawing on Roemer’s (1993, 1998) distinction between circumstances and effort, Bourguignon,Ferreira and Menendez (2007) have decomposed earnings inequality in Brazil into a component due tounequal opportunities and a residual term. They found that around a quarter of total inequality wasdue to differences in observable circumstances. In another recent study, Checchi and Peragine (2010)proposed a methodology for decomposing total inequality into “ethically acceptable” and “ethicallyoffensive” components and, in an application to data from Italy, found that inequality of opportunityaccounts for around 20 per cent of total inequality.

5

example, Alesina and Perotti (1996) found that income inequality, by fuelling social

discontent, increases sociopolitical instability. They found, furthermore, that this in-

stability leads to reduced investment, by creating uncertainty in the politico-economic

environment. As a consequence, income inequality and investment, an important engine

of growth, are inversely related.5 A review by Benabou (1996:13) finds that “These

regressions, run over a variety of data-sets and periods with many different measures

of income distribution, deliver a consistent message: initial inequality is detrimental to

long-run growth.”

In contrast, Forbes (2000) found that in the short and medium term, an increase in

a country’s level of income inequality has a significant positive relationship with subse-

quent economic growth. Still other studies have found more nuanced relations between

inequality and growth.6 Barro (2000), for example, found that higher inequality tends

to retard growth in poor countries but promote growth in richer countries. Banerjee and

Duflo (2003), on the other hand, found the growth rate to be an inverted U-shaped func-

tion of net changes in inequality, in which any changes in inequality are associated with

reduced growth in the subsequent period. The lack of consensus and the importance of

the topic will ensure that the debate rages on.

In our introduction we drew attention to possible links between high levels of inequal-

ity and financial crises. High levels of inequality have also been blamed for, among many

other things, political instability (Alesina and Perotti, 1996), crime (Kelly, 2000), cor-

ruption (You and Khagram, 2005), and poor health (Wilkinson and Pickett, 2006). Such

issues are typically complex and multi-faceted, with possible reverse causality. Again, a

fuller consideration of these issues is beyond the scope of this paper, but it is clear that

good measurement of inequality is essential for any such empirical analyses.7

Some of the discussion above may seem more obviously applicable to domestic,

‘within-country,’ inequality. However, as the world becomes increasingly inter-connected,

it is natural that relations between global inequality and global levels of growth, health,

corruption, political stability, crime and so on will increasingly be of interest. Both do-

mestic and global inequality are important in these regards and this paper is concerned

5For a discussion on the effects of redistributive and fiscal policies on growth see Alesina and Rodrik(1994) and Perotti (1996).

6It was in that context that the empirical work of Caselli et al. (1996) and Islam (1995) introduced theGeneral Method of Moments proposed by Holtz-Eakin, Newey and Rosen (1988) and Arellano and Bond(1991) to tackle issues of inconsistency from individual effects and endogeneity found in cross-countrystudies of the relationship between inequality and growth.

7There might also be good reasons to measure other concepts of income inequality, not directly relatedto the present study. For example, one might wish to conduct an empirical analysis of ‘convergence’theory, which predicts that per capita incomes across countries should converge over time. But notethat, as explained in the next section, this is not the same thing as ‘between-country’ inequality.

6

with each of them. In order to clarify precisely what we are, and are not, analysing in

this paper, various concepts of inequality are discussed in the next section.

3 Concepts and challenges in global interpersonal inequal-

ity measurement

3.1 Within-country, between-country, and global inequality

Milanovic (2005) provided a useful framework, subsequently extended by Anand and

Segal (2008), for distinguishing between different concepts of inequality.8 Keeping to

this framework we can define four concepts of inequality. ‘Concept zero’ is inequality

among countries, where countries are ranked according to their total income, and every

country receives an equal weight. We are not concerned with this concept of inequality

in the present study but it is the concept which would be most appropriate if one wished

to analyse, for example, relations between countries’ total incomes and their power on

the world stage.

‘Concept one’ is inequality among countries, where countries are ranked according

to their average per capita income, and every country receives an equal weight. This

is not the focus of this study either, but is the most suitable concept for analyzing

certain economic questions, such as whether ‘convergence’ theory appears to stand up

empirically.

‘Concept two’ is what we refer to throughout this paper as between-country inequal-

ity. This is what the inequality among all the individuals in the world would be if each

person received the average per capita income for his country.

‘Concept three’ inequality is global interpersonal inequality (or global inequality),

the inequality inherent in the actual global distribution of income, of all the citizens of

the world.

In this study we focus on ‘Concept two’ and ‘Concept three’ inequality. We also

consider the ‘within-country’ component of global inequality but stop short of defining

it as a new concept. This refers to the level of global inequality which is not attributable

to between-country inequality. This is a more involved concept than it might appear at

first glance and, as discussed in the next section, can only be appropriately measured

using a very specific class of inequality measures.

8This subsection closely follows the discussion in Anand and Segal (2008).

7

3.2 Income inequality and consumption inequality

Thus far we have used the term ‘income’ rather loosely. It is important to distinguish

between ‘income’ and ‘consumption’ inequality. It is well-known that, in general, income

inequality is likely to be considerably higher than consumption inequality. The reason

is quite straightforward. The lowest quantiles of a distribution based on consumption

typically take a greater share of the ‘consumption pie’ than the corresponding quantiles

of income do. Conversely, the highest quantiles usually get a higher share of the ‘income

pie’ than they do of consumption. Based on their analysis, Deininger and Squire (1996)

have suggested adding 6.6 per cent to Gini coefficients based on expenditure to make

them more comparable with income Ginis. In this study we focus on income inequality.9

3.3 Inequality at market exchange rates and at PPP

The measurement of global inequality requires an appropriate set of exchange rates to

convert the various national currencies into a common numeraire. The natural choice,

and that adopted in most of the literature and in this paper, is to convert national

currencies into purchasing power parity (PPP). In this paper, for example, all currencies

are converted into 2005 US$ at PPP. In theory, one dollar at 2005 PPP enables one to

purchase the same quantity of goods and services in any country and is an equivalent

amount to that which US$1 would have purchased in the US in 2005.

The same does not apply when incomes are converted to a common numeraire,

such as US$, using market exchange rates. This is because market exchange rates are

determined only by the relative prices of traded goods across countries. The relative price

of untraded goods - such as housing, transport, and education - is typically considerably

lower in developing countries. Evaluating incomes in developing countries at market

prices therefore tends to understate incomes in terms of purchasing power and would be

expected to lead to exaggerated measures of both global inequality and between-country

inequality.

Constructing consistent PPP conversion factors is a considerable undertaking. It re-

quires finding comparable ‘baskets of goods’ to compare purchasing power across coun-

tries. Yet purchasing habits and patterns differ across countries, as do the kinds of

goods that are available. The two most commonly used methods are the Geary-Khamis

(GK) and the Elteto-Koves-Szulc (EKS) methods.10 Anand and Segal (2008) provide

9In order to increase our sample of country-year observations, we do resort to using expenditure datain places, but make adjustments. This procedure is described in Section 5.

10Another method, known as the ‘Afriat’ method was developed by Dowrick and Quiggin (1997)though is less widely used.

8

an excellent discussion of the relative merits of the different methods, in the context of

global inequality measurement, and come down in favour of the EKS method. The 2005

US$ PPP conversion factors used in this study are those estimated by, and obtainable

from, the World Bank. Their more recent PPP estimates use the EKS method and the

full details of their construction are described in World Bank (2008).

4 Inequality measures

Of central importance to any study on inequality is the selection of the index used to

measure it. The choice of the index embodies fundamental normative judgements that

are important to be aware of when interpreting any results.

The most widely used measure of inequality is the Gini coefficient. It is often defined

graphically, with respect to the Lorenz curve, which depicts the cumulative share of, e.g.,

income or consumption expenditure, corresponding to the cumulative population share.

In a uniform, completely equal, income distribution the corresponding Lorenz curve is

a 45 degree line, known as the line of equality. The Gini coefficient is the area which

lies between the line of equality and the actual Lorenz curve, divided by the total area

under the line of equality.

More formally, let x = (x1, . . . , xn) ∈ Rn+ represent the distribution of incomes within

a society of n individuals which, without loss of generality, we assume to be rank-ordered

so that xi≤ xi+1for all i ∈ {1, . . . , n− 1}. Then the Gini coefficient can be expressed as

follows:

Gini =2∑n

i=1 ixin∑n

i=1 xi− n + 1

n. (1)

The popularity of the Gini index is largely due to its attractive intuitive geometric

interpretation, taking values betwen 0 and 1, with 0 reflecting perfect equality and

1, perfect inequality. The Gini also has some normatively appealing characteristics. It

satisfies the Pigou-Dalton transfer principle whereby, ceteris paribus, a transfer of income

from a better-off individual to a less well-off individual is deemed to lead to a reduction

in inequality. It also satisfies a population invariance principle, which enables consistent

comparison of populations of different sizes.

In the present context, one important drawback of the Gini coefficient is that it is

not decomposable into within-country and between-country inequality components.11

11For a discussion on decomposable income inequality measures, see Bourguignon (1979).

9

In contrast, the Theil L measure, which belongs to the family of generalized entropy

measures, is additively decomposable, with population share weights. It is also known

as the mean log deviation (hereafter, MLD), because it gives the average deviation of

logged income from its mean. Letting x = 1n

∑ni=1 xi,the MLD can then be expressed

as:

MLD =1

n

n∑i=1

ln(x

xi). (2)

As Anand and Segal (2008:85) point out, of the various inequality indices which have

been use to measure global inequality in the literature, the MLD is the only measure

which “has a consistent interpretation of its between- and within-group components.”12

In this study we use both the Gini coefficient (mainly on account of its popularity

and for the sake of comparability with other studies) and the MLD (mainly due to its

decomposability).

As noted above, use of any inequality measure embodies certain normative judge-

ments. The Gini and the MLD each implicitly adopt one particular judgement that not

everyone may support. They satisfy a ‘scale invariance’ property, in which a propor-

tional increase in all incomes is considered to leave inequality unchanged. That is, they

belong to the class of ‘relative’ inequality measures. Relative inequality measures have

been by far the most widely used in empirical studies but a strong case can also be made

for attaching some importance to absolute differences in income. Indeed Kolm (1976)

went as far as describing the relative inequality approach as “rightest.” Conversely, the

‘absolute’ inequality approach, of regarding inequality as being unaffected by an increase

of the same absolute amount to all incomes, was described by Kolm (1976) as “leftist.”

Experimental evidence (see for example Amiel and Cowell 1992, 1999) has shown support

for diverse views as to how distributions should be ranked with respect to inequality.

To accommodate a diversity of judgements about what consistutes inequality, we

therefore also use both an absolute measure, the Variance, and Krtscha, a centrist mea-

sure. These are defined as follows:

V ariance =1

n

n∑i=1

(xi − x)2 . (3)

12The Theil T index is another popular decomposable inequality measure. However, it is decomposablewith income-share, rather than population-share weights. In our view, decomposition by population shareis preferable in the context of global inequality; it is a country’s population-share, not its income-share,which should determine the contribution it is allowed to make to global inequality.

10

Krtscha =1

nx

n∑i=1

(xi − x)2 . (4)

‘Centrist’ inequality measures have the property that inequality is deemed to increase

when all incomes increase proportionally, and to decrease when all incomes increase by

the same absolute amount. As Bosmans et al. (2014) discuss, centrist measures can

range from being very close to being relative measures, to being very close to being

absolute measures. We choose to use the Krtscha measure primarily because it lies very

close to the middle of this range and might be described as being one of the most centrist

of centrist measures (see, for example, Bosman et al. (2014)’s Table 1 and footnote 7).

The Variance and Krtscha measures also have the advantage of being unit-consistent.13

5 Data and empirical issues and techniques

5.1 Data compilation

The analysis in this paper is conducted using the latest version of UNU-WIDER’s

World Income Inequality Database (UNU-WIDER, 2014), which contains repeated cross-

country information on Gini coefficients and income (or consumption) quantiles for 174

countries, spanning the period 1970-2013.14 It is the most comprehensive and complete

database of worldwide distributional data currently available. [Say a bit more about

the data, perhaps quoting Jenkins (2014)?]

As is to be expected with a secondary database of this scale, the data originate from

many different sources. The various household surveys and other sources from which

the WIID is compiled differ in many important respects. Some of the differences are

conceptual. For example, some surveys are based on income and others on consumption.

Among those data which refer to income, some are before tax and some are after tax. The

surveys also differ in coverage; some are national, others are focused solely on rural or

on urban areas and others still exclude certain groups, such as the self-employed. Some

surveys take the household as the unit of analysis and others the individual. Importantly,

the data also differ with respect to their quality and reliability. In its latest incarnation,

many (though not all) country-year observations are assigned a quality rating ranging

from 1 to 4, where 1 denotes the highest quality and 4 the lowest. A score of 1, for

13As Zheng (1997) shows, the Variance and the Krtscha are, in fact, respectively, the only knownabsolute and centrist inequality measures which are both unit-consistent and decomposable.

14The WIID contains further data, dating as far back as 1867, but the focus in this study is on the1970s onwards.

11

example, means that the underlying concepts are known and the survey is judged as

sufficient according to a number of criteria.15

The focus in this study centres on four specific years - 1975, 1985, 1995, 2000, 2005

and 2010. The gaps between the latter chosen years are shorter, partly due to increased

data availability and partly due to interest in what happened immediately before and

immediately after the global financial crisis of 2007-2008. In each of the years analysed,

there is an inevitable trade-off between using data as close as possible to the desired

years, while maintaining as high a coverage as possible of the global population at

those times. The compromise adopted was to choose these six years and to include

observations within a maximum of five years of each data point - with a preference,

naturally, for observations as close to each of these years as possible.16 So, for example,

the observations for 1985 actually come from 1980-1990, but are concentrated around

1985 as much as possible.

As well as favouring data close to the six specified years, all other things being equal,

we had a number of other preferences. Our inequality estimates are ultimately built up

from quantile share data. In order to obtain more precise estimates, we had a preference

for data based on deciles or, better still, the lower nine deciles plus the top two vingtiles,

rather than quintile shares. Since it is a study on global interpersonal inequality, we

also had a preference for those data in which the person, rather than the household,

was the unit of analysis. Naturally, we preferred data based on surveys with a more

representative coverage of the entire population and those in which the quality of the

data is deemed to be highest.

We had one final important preference. As highlighted in Section 3.2, our focus is

on global income (rather than expenditure) inequality. Naturally, all other things being

equal, we used income data rather than expenditure data. Nevertheless, ignoring the

consumption based data completely would have dramatically reduced the coverage of the

desired countries and years. Where no suitable income-based data were available but we

had data on expenditure, we used the expenditure data and adjusted it as described in

the following subsection.

A number of additional adjustments to the data could certainly be entertained to

account for the various other conceptual and coverage issues discussed above. However,

15For further information on the WIID database, including details of the quality criteria, see theWIID User Guide and Data Sources, downloadable from [http://www.wider.unu.edu/research/WIID3-0B/en GB/WIID-documentation/].

16We regarded five years as an absolute cut-off in this respect. If there were only observations morethan five years from the desired country-year, these were not used. For the latter three years analysed,observations more than three years from the desired country-year were not used.

12

no more were made and, in this regard, we plead a similar defence to Milanovic (2002:56)

who, faced with comparable issues, wrote that, “...no adjustments to the surveys were

made first, because information on sources of the bias survey-by-survey is unavailable,

and second, even if we had information regarding omission of certain population cat-

egories, it is simply beyond the scope of knowledge of any single researcher to make

meaningful corrections for such a great and varied number of surveys.”

Before turning to the adjustment procedure, we note that, finally, where there was

seemingly nothing to choose between more than one source for a given country-year,

we took an average of the quantile shares from these different sources. At the end of

the process we were left with 55, 86, 122, 119, 135 and 107 country-year observations

in 1975, 1985, 1995, 2000, 2005 and 2010 respectively. This provided us with a sample

which covers 77% of the world’s population in 1975, 85% in 1985, 93% in 1995, 87% in

2000, 94% in 2005 and 83% in 2010. The full list of country-year observations for each

of the respective years, divided into regional categories, and detailing regional coverage,

is outlined in Tables A.1-A.7 in Appendix A.

5.2 Converting consumption quantile shares into income quantile shares

Deininger and Squire (1996), in the context of their dataset, suggest adding 6.6 Gini

points to Gini coefficients based on consumption to obtain the corresponding income

Gini coefficients. In this study, as described in the following subsection, all our inequality

estimates are made directly using quantile share data. This clearly requires a different

approach to that of Deininger and Squire (1996), but it might be regarded as being in a

similar spirit.

We began, starting with the full WIID database, by comparing the average quantile

shares for income with the corresponding quantile shares for consumption. However, in

order to ensure that we were comparing like with like as far as possible, we focused only

on those country-years for which there are income and consumption data in exactly the

same year. Where there was a choice of sources for a given country-year’s income or

consumption data, we had a preference for instances where the sources for the income

and consumption data where the same. This was in order to minimize differences due

to other factors, such as different survey designs. The average shares per decile for

consumption and for income, and the average differences between them, are displayed

in Table 1.

As expected, the lowest deciles for consumption have a higher share than the corre-

sponding deciles for income, while the highest decile for consumption has a lower share

13

Table 1: Converting decile quantile shares to income quantile shares

Decile 1 2 3 4 5 6 7 8 9 10

Mean consumption share (%) 2.45 3.67 4.68 5.60 6.69 7.89 9.49 11.67 15.69 32.17

Mean income share (%) 1.58 2.77 3.76 4.67 5.78 7.01 8.73 11.08 15.65 38.89

Adjustment (% points) -0.87 -0.90 -0.92 -0.93 -0.91 -0.88 -0.76 -0.59 -0.04 6.72

than the highest decile for income. Where we had consumption-based decile data for a

given country-year, the shares were adjusted by the amounts indicated in Table 1.17

5.3 Estimating global inequality indices from country quantile data

Thus far we have discussed the collation of income-based quantile share data, at the

country level, for each of the countries and years indicated. Estimating global inequality

requires constructing a global distribution of income, using these country-level quantile

data. To do this, we need to consider both the number of individuals and the income

per capita within each country. The number of individuals per country were obtained

from population data from a number of sources.18 The income levels per capita were

then calculated using GDP data. GDP for the various country-years, in 2005 US$ at

PPP, were obtained from the World Bank’s databank.19

A common approach in previous studies has been to make the simplifying assump-

tion that all individuals in the same country-quantile-year have the same income. As

17In a few exceptional cases, where the adjustment took some of the bottom quantiles’ shares belowzero, these were instead simply taken to be zero and an equivalent subtraction taken from the topquantile.

18The main sources were: (1) United Nations Population Division. World Population Prospects, (2)United Nations Statistical Division. Population and Vital Statistics Report (various years), (3) Censusreports and other statistical publications from national statistical offices, (4) Eurostat: DemographicStatistics, (5) Secretariat of the Pacific Community: Statistics and Demography Programme, and (6)U.S. Census Bureau: International Database.

19In most cases we were able to obtain GDP values, in 2005 US$ PPP, directly from the databank.There are a few exceptions. GDP values in 2005 US$ PPP were unavailable for 1975, and so wereestimated by assuming the same average growth rate during 1975 to 1980 as in 2000 US$ PPP (whichwere available). A similar approach was used to estimate the Maldives 2000 value from its known 2001value. For Belarus 1985 we used the 1990 value as an estimate. For Czechoslovakia in 1985 we addedthe 1985 value for the Slovak Republic to the 1990 value of the Czech Republic to produce our estimate.For Kazakhstan 1985 we used the 1990 value as an estimate. For Kyrgyz Republic 1985 we used the1986 value as an estimate. For Lithuania 1985 we used the 1990 value as an estimate. For Poland 1985we used the 1990 value as an estimate. For the Russian Federation 1985 we used 1989 as a (possiblyvery poor) estimate. For Slovenia 1985 we used the 1990 value as an estimate. For Turkmenistan 1985we used the 1987 value as an estimate. For Ukraine 1985 we used the 1987 value as an estimate. ForUzbekistan 1985 we used the 1987 value as an estimate. For the West Bank and Gaza in 2010 we usedits 2005 value as an estimate. The Jamaican values for 1975, 1985 and 1995 are based on data no longeravailable on the World Bank’s website: http://data.worldbank.org/

14

Milanovic (2002) has discussed, there are some reasonable grounds for taking this ap-

proach. Nevertheless, as is well recognized, the method should be expected to bias

the resulting inequality estimates downwards.20 Like Bhalla (2002) and Sala-i-Martin

(2006), though using a different method, we constructed smooth within-country distri-

butions, and based our global inequality figures on these estimates. We used a technique

developed by Shorrocks and Wan (2009), which constructs a synthetic sample of obser-

vations which conform exactly with the known quantile shares.21 22 The approach was

found by Shorrocks and Wan (2009) to be capable of reproducing individual data from

grouped statistics with a high degree of accuracy.

In the first stage of the method, a lognormal distribution is fitted to the reported

quantile data and an equal-weighted synthetic sample of 1,000 observations is gener-

ated.23 The resulting sample is approximately consistent with the known quantile shares.

The second stage then adjusts the values of the observations within each quantile until

the quantile shares for the synthetic sample exactly match the actual quantile shares.

Shorrocks and Wan (2009) find that while some other functional forms tend to provide

a better initial fit to income distributions, particularly in the upper tail, the second

stage procedure improves the accuracy of the lognormal based sample so much that the

outcome is as good as, or better then, leading alternatives.24

For each of our years of analysis, this smoothing method was applied separately to

each country. This provided us with a sample of 1,000 synthetic individual income-share

observations for each country, which were consistent with the reported income shares (or

those estimated based on expenditure shares). These shares were then scaled up by the

country’s mean GDP (in 2005 US$ at PPP), weighted by the country’s population size,

and merged into a single synthetic global income distribution, upon which our global

inequality estimates are based.

Formally, the synthetic global income distribution in year t can be denoted as follows.

Each country c ∈ {1, . . . , C} in year t ∈ {t1, . . . tT } has an income distribution xct ∈ Rnct+ ,

where nct∈ N denotes its population at that time. Let Nt = ΣCc=1nctdenote the global

population size at time t. The global income distribution is then given by a concatenation

20As Anand and Segal (2008) discuss, such estimates should be interpreted as being lower bounds.21This technique has recently been used by Davies et al. (2011) in the closely related context of

estimating the global distribution of household wealth.22The program we used to implement the technique is freely available as part of the Distributive

Analysis Stata Package (DASP) (Araar and Duclos, 2007).23Shorrocks and Wan (2009) find that any benefits from using more than 1,000 observations are likely

to be modest. Davies et al. (2011) also use a sample of 1,000 observations in their study.24The alternative forms considered were the General quadratic Lorenz curve (Villasenor and Arnold,

1989), Beta Lorenz curve (Kakwani, 1980), Generalized Beta (McDonald, 1984) and Singh-Maddala(Singh and Maddala, 1976).

15

of all domestic distributions at this time, as follows:

xwt ∈ RNt+ =

x1t

x2t

...

xCt

5.4 Estimating counterfactual global inequality indices

As mentioned in our introduction, in this paper we explore a number of counterfactual

scenarios. Without loss of generality, it will be helpful in what follows to refer to China

as country 1, India as country 2, Sweden as country 3, and Iceland as country 4. Since

we focus on analysing six particular years, we also have that T = 6; t1 = 1975, t2 = 1985,

t3 = 1995, t4 = 2000, t5 = 2005 and t6 = 2010.

Counterfactual Scenario 1:

In our first counterfactual scenario, we assume that India’s and China’s incomes

per capita, and distribution of incomes (i.e. quantile shares), remained at 1975 levels,

and estimate what global inequality levels would have emerged in such circumstances.25

Formally, this amounts to estimating the inequality of the counterfactual distribution

given by

xwt6 ∈ RNt6+ =

x1t6

x2t6

x3t6...

xCt6

where, for c ∈ {1, 2} , xct6 ∈ Rnct6

+ is a counterfactual income distribution with the

same quantile shares and mean as xct1.


In our second counterfactual scenario, we suppose instead that India’s and China’s

incomes per capita grew at the same rates as they actually did over 1975-2010, but that

they maintained the same domestic quantile shares as in 1975. Formally, this amounts

25Their populations are assumed to have grown as they actually did over this period.

16

to estimating the inequality of the counterfactual distribution given by

xwt6 ∈ RNt6+ =

x1t6

x2t6

x3t6...

xCt6

where, for c ∈ {1, 2} , xct6 ∈ Rnct6

+ is a counterfactual income distribution with the

same same mean as xct6 but the same quantile shares as xct1 .


In our third counterfactual scenario, all countries are assumed to have their actual

incomes per capita and population sizes in 2010. However, we suppose that instead of

their actual domestic distributions of income, all countries have the same quantile shares

as those of Sweden in 2010. Formally, this amounts to estimating the inequality of the

counterfactual distribution given by

xwt6 ∈ RNt6+ =

x1t6

x2t6...

xCt6

where, for c ∈ {1, . . . , C} , xct6 ∈ Rnct6

+ is a counterfactual income distribution with

the same same mean as xct6 but the same quantile shares as x3t6 .


This counterfactual scenario is essentially the same as Counterfactual 3, except that

all countries are assumed to have the same quantile shares as those of Iceland in 2010,

rather than those of Sweden.

6 Results and Analysis

In this section we provide our global inequality estimates for 1975 to 2010, using our

chosen relative, absolute and centrist measures. We also describe the results of our

counterfactual analyses, and briefly discuss how our results compare to others in the

literature.

17

6.1 Global interpersonal inequality: relative measures

We begin by providing our relative global inequality estimates. The overall findings are

summarized in Table 2.

Table 2: Global Inequality: Relative Estimates

Inequality Measure 1975 1985 1995 2000 2005 2010

Gini 0.739 0.708 0.705 0.697 0.680 0.631MLD 1.349 1.121 1.072 1.003 0.985 0.805MLD within-country component 0.262 0.229 0.324 0.317 0.333 0.297MLD between-country component 1.087 0.893 0.748 0.687 0.653 0.507

Source: WIID / Authors’ estimates

The overall headline is that relative global inequality has fallen during 1975 to 2010,

from 0.739 to 0.631 according to the Gini coefficient, and from 1.349 to 0.805 according

to the MLD index. As is easily seen from Figure 1, this decline has, for the most part,

occurred fairly steadily over this time period.

Figure 1: Relative Global Inequality 1975-2010


Interestingly, the decline in relative global inequality appears to have accelerated

since the global financial crisis of 2007-08. A strong clue as to what has driven the

changes in inequality is apparent from an examination of the within-country and between-

country components of the MLD. Like overall relative global inequality, the between-

18

country component has decreased dramatically during 1975-2010, from 1.087 to 0.507.

Moreover, it is apparent from Figure 1 that the trend in between-country inequality

quite closely mirrors that of overal global inequality; both declined steeply between 1975

and 1985, more modestly during 1995 to 2005, and very steeply between 2005 and 2010.

In contrast, within-country inequality has increased over the period of analysis, though

the trend is less consistent, and the size of the net change much less dramatic, than in

the case of between-country inequality. It had increased notably from 0.262 in 1975 to

0.324 in 1995 (after a dip in 1985). It then remained fairly stable until 2005, with a

slight subsequent decline to 0.297 by 2010. Ceteris paribus, this would, have led to a

modest increase in overall global inequality during 1975 to 2010 but the dynamic has

been much more than offset by the dramatic reduction in between-country inequality.

As well as causing global inequality to fall, the changes that have occurred in within-

country and between-country inequality have also led to a considerable change in its

composition. It can be inferred from the results in Table 2 that in 1975 the within-

country component of global inequality was just 19%. By 2010, this had almost doubled,

to 37%.

We now turn to our results for the counterfactual scenario in which we assume that, in

2010, India’s and China’s incomes per capita, and distribution of incomes, had remained

at 1975 levels. These results are presented in Table 3.

Table 3: Counterfactual Scenario 1: Relative global inequality levels had India’s and China’s

income per capita and distribution remained at 1975 levels

Inequality Measure 1975 2010

Gini 0.739 0.757MLD 1.349 1.493MLD within-country component 0.262 0.261MLD between-country component 1.087 1.232


As Table 3 indicates, in this counterfactual scenario, global inequality would have

instead increased during 1975-2010, from 0.739 to 0.757 according the Gini coefficient,

and from 1.349 to 1.493 according to the MLD. An examination of the decomposition of

the MLD index reveals that the increase that would have occurred in this scenario would

have been entirely driven by increases in between-country inequality. China and India

were low-income countries in 1975. If their incomes per capita had remained unchanged

during the subsequent 30 years, a period in which mean world incomes soared, an increase

in between-country inequality would, in light of their large populations, have been very

19

much expected. Instead, China and India grew much faster than the global mean and,

rather than a large increase in between-country inequality, we observe instead a large

decrease.

Strikingly, the within-country component of global inequality would have remained

virtually unchanged in this counterfactual scenario. This suggests that one way of in-

terpreting the actual observed increase in within-country inequality over this period is

to view it as entirely a result of the changes in India’s and China’s income distributions.

However, this masks substantial increases and decreases in within-country inequality in

different countries and regions, as we discuss in Section 6.4.

Combining the results from Table 2 and Table 3, it is clear that the changes that

took place between 1975 and 2010 in India’s and China’s income distributions have had

a very pronounced dampening effect on relative global inequality. Paradoxically, this

is despite the fact that there were increases in both India’s and (especially) China’s

domestic inequality over this same period. China’s inequality increased from 0.287 in

1975 to 0.398 in 2010, according to the Gini coefficient, and from 0.139 to 0.288 according

to the MLD index. India’s Gini coefficient also increased, though only very marginally,

from 0.412 in 1975 to 0.417 in 2010 and, similarly, its MLD index increased from 0.280

to 0.287.26 The explanation for this dichotomy lies, of course, in the dramatic growth

which occurred in both countries over this timeframe, taking them from poor countries

to middle-income countries.

The picture so far then is that the changes to China’s and India’s domestic in-

come distributions which have caused dramatic growth, but also increased domestic

inequality (in China at least), have caused a large decline in relative global inequality.

So the changes that have taken place in these two countries have been simultaneously

‘inequality-increasing’ and ‘inequality-reducing,’ depending on which income distribu-

tion one feels is the more relevant one.

In our second ‘counterfactual’ scenario we consider what would have happened, with

respect to global inequality, if India and China had been able to grow their per capita

incomes at the same rates as they actually did over 1975-2010, while maintaining the

same domestic relative inequality levels as in 1975. The results of this analysis are

presented in Table 4.

Our results suggest that, in this scenario, global inequality would have fallen even

further by 2010; to 0.621 according to the Gini coefficient, and to 0.769 according to the

MLD index. As expected, and as confirmed by the decomposition of the MLD index in

26It is quite possible that these estimates understate the increases in inequality that have occurred inIndia and China. We consider the implications this would likely have for our results in Section 7.

20

Table 4: Counterfactual Scenario 2: Relative global inequality levels assuming actual growth in

per capita income for India and China but domestic inequality at 1975 levels




Table 4, this further fall in inequality, relative to what actually happened, would have

been driven exclusively by there being a smaller level of within-country inequality than

that which actually emerged. This serves to emphasize the point that the increases in

domestic inequality in India and China have not been ‘good’ for global inequality per

se. They have only been ‘good’ to whatever extent that they have been an unavoidable

side effect of high economic growth that has led to a marked reduction in between-

country inequality. Relative global inequality would have been reduced even further if

the same growth could have been achieved in these countries without associated increases

in domestic inequality.

6.2 Global interpersonal inequality: absolute and centrist measures

We now turn to our absolute and centrist global inequality estimates. The overall findings

are summarized in Table 5 and the trends displayed graphically in Figure 2.

Table 5: Global Inequality: Absolute and Centrist Estimates


Variance 1.0387E+08 1.321E+08 1.914E+08 2.432E+08 2.876E+08 3.069E+08

Krtscha 19,342 20,977 26,337 31,281 32,036 28,902

Source: Authors’ estimates

It is immediately apparent that the Variance and Krtscha, our respective absolute

and centrist measures of inequality, paint a very different picture of the overall trend in

global inequality during 1975 to 2010.27 Far from there having been a dramatic reduction

in inequality over these 35 years, both measures indicate a dramatic increase in global

inequality. The Variance suggests that global inequality has increased steadily, with a

substantial increase in inequality from each time period to the next, albeit at a slightly

27Another absolute measure of inequality, the Absolute Gini was also computed, and found to tell avery similar story to the Variance.

21

reduced rate between 2005 and 2010. The Krtscha measure considers inequality to have

increased in each period analysed up until 2005, but, having increased at a rather slower

rate between 2000 and 2005, to have then decreased markedly between 2005 and 2010.

Nevertheless, the Krtscha still regards global inequality in 2010 as being at a higher level

than at any period prior to 2000.

Figure 2: Global Inequality 1975-2010: Absolute and Centrist Measures


We now revisit, with the Variance and Krtscha, our first counterfactual scenario,

in which we assumed that India’s and China’s incomes per capita and distribution of

incomes remained at 1975 levels. These results are presented in Table 6.

Table 6: Counterfactual Scenario 1: Global (absolute and centrist) inequality levels had India’s

and China’s income levels and distribution remained at 1975 levelsInequality Measure 1975 2010

Variance 1.037E+08 3.230E+08Krtscha 19,342 37,339


Both the Variance and the Krtscha measures agree with the relative inequality es-

timates in Table 3 that in this counterfactual scenario, global inequality would have

increased since 1975. However, the results are much more emphatic. In fact, inequal-

ity would have increased so much that our relative and absolute measures also agree

22

that actual inequality in 2010 was lower than it would have been without the dramatic

changes that took place in China and India over this period.

Returning to our second ‘counterfactual’ scenario, we now consider what would have

happened to global inequality, as captured by the Variance and Krtscha, if India and

China had been able to grow their per capita incomes at the same rate as they actually

did over 1975-2010, while maintaining the same domestic income quantile shares as in

1975.28 The results of this analysis are presented in Table 7.

Table 7: Counterfactual Scenario 2: Global (absolute and centrist) inequality levels assuming

actual growth for India and China but domestic quantile shares at 1975 levels


Variance 1.037E+08 3.038E+08Krtscha 19,342 28,615


In marked contrast to the judgement of our relative inequality measures, both the

Variance and the Krtscha register a large increase in inequality during 1975 to 2010

under this hypothetical situation. Inequality here is lower in 2010 than in the preceding

counterfactual scenario, where there were no changes in India and China while the rest

of the world changed. It is also lower, though only marginally, than its actual 2010

levels. For these measures, the reductions in inequality due to India and China catching

up with global mean income have been massively outweighed by the increased absolute

differences in income throughout the global income distribution.

6.3 Is it feasible for absolute or centrist measures of inequality to fall

as global incomes rise?

The results in the previous subsection raise the question as to whether increased global

income per capita of the magnitude that took place between 1975 and 2010 is ever

conceivably compatible with declining absolute and centrist notions of inequality. The

Nordic countries stand out as being high income countries with comparatively low levels

of relative inequality. It may very well be possible for some countries to become more

affluent, and more equal, than the Nordic countries. Nevertheless, for most countries,

maintaining current income per capita levels, or growing further, while emulating Nordic

28While maintaining the same domestic income shares as mean income grows equates to relativeinequality measures remaining constant, this is not true for absolute and centrist measures, which registeran increase in such circumstances. It is of interest, nevertheless, to see what would happen to the Varianceand Krtscha in such circumstances.

23

levels of relative inequality would be applauded by most economists and policymakers

as a considerable achievement. Thus, if it is ever likely to be possible for absolute /

centrist measures of global inequality to fall in the face of large increases in incomes,

we might expect a necessary condition along the following lines. Consider our third

counterfactual scenario, in which all countries have their actual incomes per capita in

2010, but their quantile shares (and therefore domestic relative inequality levels) are the

same as those of Sweden in 2010.29 If an inequality measure considers global inequality

to have increased during 1975-2010 in such a scenario, the prospects for that measure

ever registering a decrease alongside substantial global economic growth seems, in our

judgement, to be very slim.

The results of this counterfactual scenario are displayed in Table 8, which also in-

cludes, for comparison, the corresponding relative inequality estimates.

Table 8: Counterfactual Scenario 3: Relative, absolute and centrist global inequality levels

assuming actual growth in per capita income globally but domestic income shares at 2010 Swedish

levelsInequality Measure 1975 2010

Absolute & Centrist measuresVariance 1.037E+08 1.932E+08Krtscha 19,342 18,191Relative measuresGini 0.739 0.569MLD 1.349 0.609


As expected, both relative inequality measures register a very substantial decline in

inequality in such a scenario, with inequality far below actual 2010 levels. The situation

is, however, very different for the Variance, which almost doubles. This suggests that

there is a considerable tension between absolute inequality and growth in mean incomes;

it is virtually impossible to imagine circumstances in which the global growth levels

actually experienced could have taken place without increasing absolute inequality. In

contrast, the Krtscha registers a decrease, suggesting that any tension between growth

and inequality is much more modest for this measure; this is not entirely unexpected

given the fall observed in the Krtscha between 2005 and 2010.30

29Sweden had the fourth lowest Gini coefficient in our sample of countries in both 1975 and 2010 andwas virtually unchanged, from 0.239 in 1975 to 0.241 in 2010.

30These findings are also consistent with Roope (2015), who demonstrates that, ceteris paribus, incre-mental increases in an individual’s income can never lead to a rise in inequality according to the Krtschaif their income is below the mean, and that their income typically needs to be far above the mean.

24

Yet it turns out that substantial growth, and a falling Krtscha measure, is still an

extremely demanding requirement. In 2010, Iceland had a Gini coefficient of 0.256,

the eighth lowest in the world according to our estimates. In our fourth counterfactual

analysis, analogous to that displayed in Table 8, but assuming Icelandic, rather than

Swedish quantile shares, levels of inequality in 2010 according to Krtscha were higher

than they were in 1975.

Table 9: Counterfactual Scenario 4: Relative, absolute and centrist global inequality levels as-

suming actual growth in per capita income globally but domestic income shares at 2010 Icelandic

levelsInequality Measure 1975 2010

Absolute & Centrist measuresVariance 1.037E+08 2.098E+08Krtscha 19,342 19,755Relative measuresGini 0.739 0.574MLD 1.349 0.617


These results, presented in Table 9, suggest that given actual levels of GDP per capita

in 2010, Icelandic relative inequality levels are a kind of upper bound to the domestic

relative inequality levels that would be required globally to reduce the Krtscha to 1975

levels.

6.4 Global interpersonal inequality: regional analysis

We have considered inequality within and between countries. While not the main focus

of this paper, it is also of interest, in passing, to investigate what trends in inequality

within and between regions are apparent during 1975 to 2010.31 A decomposition of the

MLD into within-region and between-region components is presented in Table 10 and

displayed graphically in Figure 3.

Table 10: Global Inequality: MLD Within and Between Regions


MLD 1.349 1.121 1.072 1.003 0.985 0.805MLD within-region component 0.696 0.387 0.602 0.460 0.529 0.421MLD between-region component 0.653 0.734 0.470 0.543 0.457 0.383


31To conserve space, our regional analysis of inequality is restricted to relative inequality.

25

With the notable exception of the mid 1980s, for most of the period of analysis,

within-country and between-country components of inequality have been of a similar

magnitude. Notwithstanding a number of aberrations between periods, it is clear that

the long-term decreases in the two components have been roughly equally important in

explaining the pronounced long-term decline in the MLD.

Figure 3: Decomposition of Relative Global Inequality Within and Between Regions 1975-2010


While relative inequality within regions has fallen overall, as Table 11 indicates,

there have been both increases and decreases in inequality within specific regions during

1975-2010.32

Despite issues with data coverage, Sub-Saharan Africa stands out as having extremely

high levels of both within-country and between-country inequality. There has been a

very substantial reduction in relative inequality in East Asia & Pacific, driven by a stag-

gering reduction in between-country inequality, and despite an increase in within-country

inequality. This result, naturally, is largely driven by China. There is evidence of an

inverted U-shape in inequality in Europe & Central Asia, with a peak around 1995-

2000, and which broadly reflects trends in both within-country and between-country

inequality. Inequality in Latin America & the Caribbean is almost all within-country

32Note that the population coverage in our sample is typically rather poor for the Middle East andNorth Africa, and for Sub-Saharan Africa. There are also some issues of comparability over time in theseregions in terms of the countries which compose the regional sample changing over time. See TablesA.1-A.7 in Appendix A for full details of regional composition and population coverage.

26

Table 11: Regional Relative Inequality Levels: Within and Between Countries


East Asia & PacificMLD 1.186 0.536 0.884 0.449 0.605 0.477MLD within-country component 0.190 0.184 0.334 0.295 0.281 0.280MLD between-country component 0.996 0.352 0.550 0.154 0.324 0.196

Europe & Central AsiaMLD 0.183 0.272 0.528 0.514 0.422 0.331MLD within-country component 0.175 0.154 0.254 0.221 0.206 0.185MLD between-country component 0.008 0.118 0.274 0.292 0.216 0.147

Latin America & CaribbeanMLD 0.624 0.546 0.604 0.650 0.555 0.483MLD within-country component 0.593 0.501 0.566 0.579 0.509 0.456MLD between-country component 0.031 0.045 0.038 0.071 0.046 0.027

Middle East & North AfricaMLD 0.848 0.488 0.514 0.383 0.373 0.235MLD within-country component 0.501 0.381 0.391 0.240 0.223 0.215MLD between-country component 0.346 0.107 0.123 0.144 0.151 0.021

North AmericaMLD 0.200 0.238 0.268 0.274 0.286 0.320MLD within-country component 0.199 0.237 0.265 0.272 0.284 0.318MLD between-country component 0.001 0.001 0.002 0.002 0.002 0.002

South AsiaMLD 0.274 0.201 0.207 0.312 0.465 0.311MLD within-country component 0.264 0.185 0.188 0.296 0.437 0.284MLD between-country component 0.009 0.016 0.019 0.016 0.028 0.027

Sub-Saharan AfricaMLD 0.535 0.565 0.948 0.885 0.745 0.735MLD within-country component 0.500 0.447 0.567 0.449 0.346 0.413MLD between-country component 0.036 0.119 0.380 0.435 0.399 0.322


27

inequality, and a pronounced decline is apparent between 2000 and 2010. Finally, in-

equality in North America (which for these purposes does not include Mexico) is almost

entirely within-country inequality, and it has increased steadily over time.

Finally, within regions, there is typically considerable variation with respect to levels

of, and changes in, domestic inequality over the period of analysis. For example, in

Europe, the UK’s Gini has increased by 38%, while France’s has fallen by 16%; in Latin

America, Argentina’s Gini has increased by 25%, while Brazil’s has fallen by 10%; in

South Asia, Bangladesh’s Gini has increased by 60%, while Nepal’s has decreased by

38% and in East Asia, China’s Gini has increased by 39%, while the Republic of Korea’s

has decreased by 14%.33 The latter comparison is an interesting one. China and Korea

have both been extraordinarily successful with respect to growth in mean incomes over

the period studied, yet have had very different experiences with respect to changes in

domestic inequality.

6.5 How our results compare to those of other studies in the literature

The overwhelming majority of previous studies on global inequality have investigated

only relative inequality. Our overall relative global inequality estimates lie broadly in the

same ball park as previous studies. For example, Dowrick and Akmal (2005) reported

Ginis of 0.698 in 1980 and 0.711 in 1993, Sala-i-Martin (2006) found Ginis of 0.660 in

1980 and 0.637 in 2000, Bhalla (2002) reported Ginis of 0.686 in 1980 and 0.651 in 2000,

Bourguignon and Morrisson (2002) found a Gini of 0.657 in both 1980 and 1992 and

Milanovic (2005) reported Ginis of 0.622 in 1988 and 0.641 in 1998.

Our estimates lie towards the upper end of the existing literature, closest to Dowrick

and Akmal (2005). This is likely to be partially related to the manner in which we

have adjusted for data which are reported as consumption quantiles rather than income

quantiles. As we have discussed, this imparts an (intentional) upward impact on our

global inequality measurements.

Our analysis of the impact of China on global inequality is consistent with that of

Sala-i-Martin (2006), who found that without China, global inequality would have risen

from 0.620 to 0.648 from 1970 to 2000, whereas in fact it actually decreased.

Our analysis of trends in within-country and between-country inequality and of the

impact of India and China seem at least partially consistent with that of Bourguignon

and Morrisson (2002). They found that between 1970-92, global income inequality, as

measured by the MLD index, increased slightly from 0.823 to 0.827. Whilst we found

33Quantile data were unavailable for the Republic of Korea for 2010. However the WIID does havethe relevant Gini data, ultimately sourced from OECD statistics.

28

that global income inequality decreased between 1975 and 1995, our within-country and

between-country components moved in the same directions as theirs. (Their within-

country component increased from 0.304 to 0.332, while their between-country compo-

nent decreased from 0.518 to 0.495).

Bourguignon and Morrisson (2002:738) also found that “...the relatively poor growth

performances of China and India until late in the 20th century” was one of the main

“disequalizing forces” while “China’s outstanding growth performance in the last decade

or two of the period” was one of the main “equalizing forces.” While we consider a much

shorter period of analysis (which starts much later but also finishes later), our results as

to the impact of China and India are consistent with theirs.

There are also some notable points of divergence in our results. For example, Mi-

lanovic (2002) found that inequality, as measured by the Gini index, increased from 0.63

in 1988 to 0.66 in 1993. Moreover, this increase was driven more by differences in mean

incomes between countries than by inequalities within countries. Although we do not

consider these exact years, these results do seem somewhat at odds with our findings

for 1985-95, especially with respect to the trends in within-country and between-country

inequality; as discussed, we find that within-country inequality increased considerably

over this period, while between-country inequality decreased substantially.

In one of the very few studies which has employed relative, absolute and centrist

measures, Bosmans et al. (2014) studied trends in ‘between-country’ inequality during

1980 to 2009. Our results are largely consistent with theirs in that they found that

‘between-country’ inequality had increased over their period of analysis according to a

comprehensive range of both absolute and centrist measures of inequality. While they

also found that ‘between-country’ inequality had increased according to some relative

inequality measures, they found it to have decreased according to those used in this

study (Gini and MLD).

7 Sensitivity analysis and robustness of estimates

As in any study on global inequality, there are a multitude of potential sources of error in

our estimates, including sampling error, measurement error and, as discussed by Anand

and Segal (2008), assuming a single PPP price level for each country. We discuss the

general implications of these issues for the robustness of our estimates in some detail

in Section 7.2. First though, recognising the very considerable impact on our overall

inequality estimates of our quantile data for India and China, we consider the sensitivity

of our estimates to significant measurement error in those particular data.

29

7.1 Sensitivity Analysis

In this paper, through our counterfactual analyses, we have seen that overall levels of

global inequality can be very substantially affected by changes in income distribution in

India and China. This fact also serves to emphasise just how sensitive any estimate of

global inequality must be to measurement error in the underlying quantile data for these

countries. In this subsection we consider the extent to which our main results would

likely be affected if our quantile data are badly wrong.

With respect to relative inequality, our key findings are that global inequality has

fallen steadily and substantially during 1975-2010, driven primarily by falling between-

country inequality, while within-country inequality has increased modestly. In tandem,

our absolute and centrist measures of inequality have increased dramatically. As the

analysis in Section 6.3 indicates, the latter findings would be most unlikely to change

using almost any conceivable quantile data for India and China; we therefore focus on

the former results. Perhaps the most obvious threat to these findings comes from the

possibility that relative inequality was much lower in India and China in 1975 than our

data indicate, and much higher in 2010 than our data suggest. In our sensitivity analysis

we therefore estimate the Gini coefficient and MLD for a counterfactual situation where,

in 1975, India and China had the same quantile shares as Hungary, which had the lowest

MLD in our sample in 1975, and the same quantile shares in 2010 as South Africa, which

had the highest MLD in our sample at that time. All other countries are assumed to

have the same quantile shares as in our main analysis. The results are presented in Table

11.

Table 12: Sensitivity Analysis: Relative global inequality levels assuming extreme measurement

error in India’s and China’s quantile data




As expected, a comparison of the Gini and MLD estimates to those in Table 2

indicates lower inequality in 1975, and higher inequality in 2010, than in our main

analysis. However, the direction of change is is not affected; relative inequality has still

fallen according to both the Gini and MLD (from 0.735 to 0.691, and from 1.294 to

1.118, respectively). Substantively, the big difference here to our main analysis is in the

30

different picture with regard to what has happened to the within-country component

of inequality. Rather than a modest increase from 0.262 to 0.297, in this scenario the

within-country component of the MLD increases dramatically from 0.207 to 0.611. To

put the sensitivity analysis in context, the scenario just computed is tantamount to

assuming measurement error so high that China’s and India’s Gini coefficients in 2010

were incorrectly measured as 0.398 and 0.417, respectively, when they should actually

have been 0.696. Furthermore, they were incorrectly measured in 1975 as 0.287 and

0.412, respectively, when they should actually have been 0.226. Overall, the sensitivity

analysis strongly suggests that the main findings regarding inequality trends are robust

to imprecise measurement of China and India’s income distributions, with the important

caveat of uncertainty over the robustness of the direction of change of inequality’s within-

country component.

7.2 Robustness of Estimates

Anand and Segal (2008:90) have argued that sensitivity analysis should be undertaken

to assess the possible impact of the many sources of uncertainty in global inequality esti-

mates, even if there is insufficient information to estimate statistical confidence intervals.

They also illustrate in some detail why the standard errors estimated in the literature

do not address these concerns. We broadly agree with Anand and Segal (2008). With

respect to the difficulties involved in estimating statistical confidence intervals, we would

perhaps even go as far as to argue that such a task is impossible. There is simply too

much uncertainty regarding the sizes and directions of some of the biases. We will discuss

some of the issues in turn and consider their possible impact.

7.2.1 Sampling error and bias

There are well-established techniques for estimating the standard errors arising from a

sample not being representative of the population under study. The difficulty in the

present context, as pointed out by Anand and Segal (2008), is that we are not dealing

with a random sample of the global population. Rather we are dealing with a constructed

sample, based on many different national surveys, each with their own sampling variance.

Estimating the sampling variance of the resulting distribution is a difficult problem and

a solution is not easily apparent.

A related issue is the extent to which our results are likely to suffer from sample

selection bias. Given the nature of household surveys, one might expect this to impart

a downward bias on our estimates, since it is rare for very high earners, or the very

31

poorest, to be surveyed. However, we make this conjecture somewhat tentatively and

without a method for estimating the extent of this potential bias. Moreover, any such

bias needs to be considered in tandem with possible biases arising from measurement

error, as discussed next.

7.2.2 Measurement error

Measurement error is likely to arise from a number of sources, notably the quantile

share data from the various surveys and the GDP data measured in US$ PPP. Indeed

measurement error in the latter might itself arise from either the raw GDP calculations

or the estimation of the PPP conversion rates. Measurement error in the population

data is a further possibility.

Anand and Segal (2008) argue that since the responses from those who are surveyed

are likely to be noisy, this would bias the variance of the responses upwards which might

cause measured inequality to be overstated. This is a reasonable argument. However,

as Anand and Segal (2008) also point out, surveys are also likely to suffer from underre-

porting of the incomes of the rich. This dynamic would be expected to bias inequality

in the opposite direction.

Now, if the GDP data are biased upwards (respectively, downwards) for low-income

countries, or downwards (respectively, upwards) for high-income countries, measured

global inequality will be biased downwards (respectively, upwards), due to the resulting

impact on the between-country component of inequality.

If population sizes are biased upwards (respectively, downwards) in either very low-

or very high-income countries, this will tend to bias the between-country component

of global inequality upwards (respectively, downwards). If population sizes are biased

upwards (respectively, downwards) in countries with high domestic inequality, this will

tend to bias the within-country component of global inequality upwards (respectively,

downwards).

The direction and size of biases resulting from measurement error in the GDP data

and of country population sizes are very difficult to predict. Overall, given the multitude

of surveys and the various uncertainties involved, it is unfortunately not at all clear what

even the direction of the net bias due to measurement error should be expected to be,

let alone its size. This is certainly an area for future research enquiry.

32

7.2.3 Assuming a single PPP price level for each country

Further biases are likely due to the necessary evil of assuming a single PPP price level for

each country-year. This issue has been discussed by Aten and Heston (2010) and Anand

and Segal (2008). In short, if prices faced by individuals are positively correlated with

incomes, as is often the case, the assumption of a single PPP price level will tend to bias

measured inequality upwards.34 On the other hand, economies of scale tend to favour the

better off, who are more likely to be able to buy in bulk. This may result in a negative

correlation between prices and incomes and a downward bias on measured inequality.

The overall direction of the net bias even at a given country-level is difficult to predict;

estimating the overall effect of such biases globally appears an almost insurmountable

task.

8 Discussion and Concluding remarks

Using the most up-to-date and complete database of worldwide distributional data

presently available, we have estimated global interpersonal inequality levels and their

trends during the period from 1975 to 2010. To the best of our knowledge, it is the first

comprehensive study on global inequality which spans the pre- and post-financial crisis

periods.

In contrast to the majority of previous studies on global inequality, we employed a

number of inequality measures with wildly varied normative underpinnings; ‘relative,’

‘absolute’ and ‘centrist.’ Taken together, the results in the paper echo Atkinson and

Brandolini (2010) in emphasising just how central the choice of measure is to any dis-

cussion of what has happened to global inequality levels during recent decades. Con-

structing a global distribution of income upon which to base inequality estimates is a

considerable undertaking, with scope for many kinds of errors, as discussed in the pre-

vious section. Yet whatever errors may remain in constructed distributions such as ours

and those of other authors, the resulting error in, say the Gini coefficient, is as nothing

compared to the differing conclusions reached about what has happened to inequality in

moving even to a centrist measure such as the Krtscha, let alone an absolute one.

According to our results, relative global inequality, as captured by the Gini and MLD,

while still staggeringly high, has fallen steadily and quite substantially over the decades.35

This has been driven by a dramatic decline in inequality between countries, largely

34It is well known that incomes are often higher in expensive regions.35Global relative inequality in 2010 remains much higher than domestic levels in even the most unequal

countries of Latin America and sub-Saharan Africa.

33

arising from India’s and China’s outstanding performances. Had the dramatic changes

which took place in those populous countries not occurred, global relative inequality

would instead have increased during 1975 to 2010. In stark contrast, the Variance, an

absolute inequality measure, finds that not only has global inequality rocketed during

1975 to 2010, but the changes in India and China which have taken place have had

only a very modest dampening impact on the increased inequality. The Krtscha, one

of the most centrist of the centrist measures agrees with absolute measures that global

inequality was considerably higher in 2010 than in 1975, but that it peaked around 2005,

and was substantially dampened by the above changes in India and China.

What should policymakers with a concern for global inequality take from these appar-

ently wildly disparate findings? The evidence from questionnaire studies is that people

do have diverse views about what exactly consitutes inequality. Perceptions of inequal-

ity are very likely related to absolute differences in income, not only to the relative

differences which have been so focused on in empirical studies in the literature. Yet our

analysis in Section 6.3 suggests a serious difficulty with using absolute inequality mea-

sures to inform policy; it appears most implausible that the global economy can grow

substantially without large increases in absolute global inequality. During the timeframe

analysed, millions of people around the world have been lifted out of absolute notions

of poverty, and, in our judgement, it seems unfeasible that this could have occurred

without absolute inequality indicators registering very considerable increases. Whether

the same could have been achieved without centrist measures registering an increase is

a more moot point. In the specific case of the Krtscha, the requirement, Icelandic levels

of relative inequality, is certainly extremely demanding, but perhaps possible. Other

centrist measures, closer to the relative end of the spectrum but which still penalise

absolute differences in income to an extent, are likely to present less demanding, more

easily achievable, targets.

In the end, our results can perhaps be interpreted as suggesting that, for better or

for worse, a degree of increased inequality, at least in its absolute form, is an inevitable

by-product of economic ‘progress.’ It would take a foolhardy policymaker, and would

be a huge mistake, to attempt to halt or reverse economic growth in order to appease

an absolute inequality indicator. On the other hand, there is increasing evidence that

lowering levels of relative inequality may not only be desirable in its own right but

may also be growth promoting. A recent study by Berg and Ostry (2014) finds relative

inequality to be negatively related to growth, even after controlling for redistribution, the

possible distortionary effects of which are sometimes proposed as a plausible mechanism

via which inequality might damage growth.

34

Developing policies aimed at reducing relative inequality may have multiple benefits.

In the short-run, redistributive policies are an obvious channel to achieve this. In the

longer run, devoting greater resources to the promotion of human capital, especially, as

Heckman and others have recently convincingly argued, in early childhood, and especially

among disadvantaged children, is perhaps an even more potent tool. This brings us to

the point on which we wish to conclude. All too often people on lower incomes suffer

from lack of attainment across multiple domains. Progress, with or without increased

income inequality, is not real progress if improvements are not also apparent in non-

income domains, be they people’s health, life satisfaction, or, as Sen (1979, 1985) has

forcefully argued, the freedom to live the kind of life one has reason to value. Monetary

outcomes are only one facet of wellbeing. In our view, a key challenge for policymakers

is to employ a dashboard of indicators to assess progress and, to whatever extent income

inequality, relative or absolute, is unavoidable, policymakers should do all they can to

minimise its replication in other domains.

References

[1] Alesina, A. and Perotti, R. (1996), “Income distribution, political instability, and

investment,” European Economic Review 40, 1203-1228.

[2] Amiel, Y. and Cowell, F. (1992), “Measurement of Income Inequality: Experimental

Test by Questionnaire,” Journal of Public Economics 47, 3-26.

[3] Amiel, Y. and Cowell, F. (1999), “Income transformations and income inequality,”

In D. Slottje (ed.), Advances in Econometrics, Income Distribution and Scientific

Methodology. Heidelberg: Physica Verlag, pp. 209–32.

[4] Anand, S.and Segal, P. (2008), “What do we know about global income inequality?”

Journal of Economic Literature 46, 57-94.

[5] Araar, A. and Duclos, J-Y. (2007), DASP: Distributive Analysis Stata Package,

PEP, World Bank, UNDP and Universite Laval.

[6] Arneson, R. (1989), “Equality of opportunity for welfare,” Philos. Stud. 56, 77–93.

[7] Aristotle (1954) Rhetoric. Book II, Chapter 2, 1379a. Translation of Rhys Roberts.

Compiled by Lee Honeycult, web edition at www.public.iastate.edu

35

[8] Arellano, M. and Bond, S. (1991), “Some Tests of Specification for Panel Data:

Montecarlo Evidence and an Application to Employment Equations,” Review of

Economic Studies 58, 277-297.

[9] Aten, B and Heston, A. (2010), “Use of Country Purchasing Power Parities for

International Comparisons of Poverty Levels: Potential and Limitations,” in De-

bates in the Measurement of Global Poverty, ed. S. Anand, P. Segal and J. Stiglitz.

Oxford and New York: Oxford University Press.

[10] Atkinson and Brandolini (2010), “On analysing the world distribution of income,”

The World Bank Economic Review 24, 1-37.

[11] Banerjee, A. and Duflo, E. (2003), “Inequality And Growth: What Can The Data

Say?,” Journal of Economic Growth 8, 267-299.

[12] Barro, R. (2000), “Inequality and Growth in a Panel of Countries,” Journal of

Economic Growth 5, 5-32.

[13] Benabou, R. (1996), “Inequality and Growth,” in Ben S. Bernanke and Julio J.

Rotemberg, eds., NBER macroeconomics annual 1996. Cambridge, MA: MIT Press,

pp. 11-74.

[14] Berg, A. and Ostry, J. (2014), “Redistribution, inequality and growth,” Paper pre-

sented at UNU-WIDER conference ‘Inequality – measurement, trends, impacts and

policies,’ Helsinki, September 2014.

[15] Bhalla, S. (2002), Imagine there’s no country: Poverty, Inequality and Growth in

the Era of Globalization. Washington, D.C.: Institute for International Economics.

[16] Boltho, A. and Toniolo, G. (1999), “The Assessment: The Twentieth Century-

Achievements, Failures, Lessons,” Oxford Review of Economic Policy 15, 1-17.

[17] Bosmans, K., Decancq, K. and Decoster, A. (2014), “The Relativity of Decreasing

Inequality Between Countries,” Economica 81, 276–292.

[18] Bourguignon, F. (1979) “Decomposable Income Inequality Measures,” Economet-

rica, Vol. 47, No. 4, pp. 901-920

[19] Bourguignon, F. and Morrisson, C. (2002), “Inequality among World Citizens: 1820-

1992,” American Economic Review 92, 727-744.

36

[20] Bourguignon, F., Ferreira, F. and Menendez, M. (2007), “Inequality of opportunity

in Brazil,” Review of Income and Wealth 53, 585-618.

[21] Caselli, F., Esquivel, G. and Lefort, F. (1996), “Reopening the Convergence Debate:

A New Look at Cross-Country Growth Empirics,” Journal of Economic Growth 1,

363-389.

[22] Chotikapanich, D., Valenzuela, R. and Prasado Rao, D. (1997), “Global and Re-

gional Inequality in the Distribution of Income: Estimation with Limited and In-

complete Data,” Empirical Economics 22, 533-546.

[23] Cohen, G.A. (1989), “On the currency of egalitarian justice,” Ethics 99, 906-944.

[24] Cornia, G. A. and Kiiski, S. (2001), “Trends in Income Distribution in the Post-

World War II Period: Evidence and Interpretation,” UNU-WIDER Discussion Pa-

per No. 2001/89.

[25] Davies, J., Sandstrom, S., Shorrocks, A. and Wolff, E. (2011), “The Level and

Distribution of Global Household Wealth,” The Economic Journal 121, 551, 223-

254.

[26] Deininger, K. and Squire, L. (1996), “A New Data Set Measuring Income Inequal-

ity,” World Bank Economic Review 10, 565-591.

[27] Dowrick, S. and Akmal, M. (2005), “Contradictory Trends in Global Income In-

equality: A Tale of Two Biases,” Review of Income and Wealth 51, 201-229.

[28] Dowrick, S. and Quiggin, J. (1997), “True Measures of GDP and Convergence,”

American Economic Review 87, 41-64.

[29] Dworkin, R. (1981a), “What is equality? Part one: Equality of welfare,” Philosophy

& Public Affairs 10, 185-246.

[30] Dworkin, R. (1981b), “What is equality? Part two: Equality of resources,” Philos-

ophy & Public Affairs 10, 283-345.

[31] Firebaugh, G. (1999), “Empirics of World Income Inequality,” American Journal of

Sociology 104, 1597-1630.

[32] Firebaugh, G. (2003), The New Geography of Global Income Inequality. Cambridge

and London: Harvard University Press.

37

[33] Fleurbaey, M. (2008), Fairness, Responsibility, and Welfare. Oxford University

Press, London.

[34] Forbes, K. (2000), “A Reassessment of the Relationship between Inequality and

Growth,” American Economic Review 90, 869-887.

[35] Holtz-Eakin, D., Newey, W., & Rosen, H. S. (1988), “Estimating vector autoregres-

sions with panel data,” Econometrica 56, 1371-1395.

[36] Islam, N. (1995), “Growth Empirics: A Panel Data Approach,” Quarterly Journal

of Economics 110, 1127-70.

[37] Kakwani, N. (1980), “On a class of poverty measures,” Econometrica 48, 437-446.

[38] Kelly, M. (2000), “Inequality and Crime,” Review of Economics and Statistics 82,

530-539.

[39] Kolm, S. (1976). “Unequal Inequalities I and II,” Journal of Economic Theory, 12,

416-442, and 13, 82-111.

[40] Korzeniewicz, R. and Moran, T. (1997), “World-Economic Trends in the Distribu-

tion of Income, 1965-1992,” American Journal of Sociology 102, 1000-1039.

[41] Kuznets, S (1955), “Economic Growth and Income Inequality,” American Economic

Review, 45(1), pp. 1-28.

[42] Lewis, W. A. (1954), “Economic Development with Unlimited Supplies of Labour,”

The Manchester School, 22: 139-191.

[43] McDonald, J. (1984), “Some generalized functions for the size distribution of in-

come,” Econometrica 52, 647-663.

[44] Melchior, A., Telle, K. and Wiig, H. (2000), “Globalization and Inequality: World

Income Distribution and Living Standards, 1960-1998,” Royal Norwegian Ministry

of Foreign Affairs, Studies on Foreign Policy Issues Report, No. 6B.

[45] Milanovic, B. (2002), “True World Income Distribution, 1998 and 1993: First Cal-

culation Based on Household Surveys Alone,” Economic Journal 112, 51-92.

[46] Milanovic, B. (2005), Worlds Apart: Measuring International and Global Inequality.

Princeton and Oxford: Princeton University Press.

38

[47] Milanovic, B. (2012), “Global inequality recalculated and updated: the effect of new

PPP estimates on global inequality and 2005 estimates,” The Journal of Economic

Inequality 10, 1-18.

[48] Piketty, T. (2014), Capital in the 21st Century. Cambridge: Harvard Uni (2014).

[49] Plato (1901) The Republic. Chapter 4 on wealth, poverty and virtue, No. 422.

Translation of Benjamin Jowett, web edition at www.mdx.ac.uk, Stephanus page

numbers used.

[50] Ravallion, M. (2003), “The debate on globalization, poverty and inequality: why

measurement matters,” International Affairs, 79, 739–53.

[51] Rawls, J. A. (1971), Theory of Justice. Harvard University Press, Cambridge.

[52] Richardson, D., (1995), “Income inequality and trade: how to think, what to con-

clude,” Journal of Economic Perspectives 9, 33-56.

[53] Roemer, J.E. (1993), “A pragmatic theory of responsibility for the egalitarian plan-

ner,” Philosphy & Public Affairs 22, 146-166.

[54] Roemer, J. E. (1998), Equality of Opportunity. Cambridge, MA: Harvard University

Press.

[55] Roope, L. (2015), “Inequality and growth: a simple structural relationship,” pa-

per prepared for the CSAE Conference 2015: Economic Development in Africa,

University of Oxford.

[56] Sala-i-Martin, X. (2006), “The World Distribution of Income: Falling Poverty

and...Convergence, Period,” Quarterly Journal of Economics 121, 351-397.

[57] Schultz, P. (1998), “Inequality in the Distribution of Personal Income in the World:

How it is Changing and Why,” Journal of Population Economics 11, 307-344.

[58] Singh, S. and Maddala, G. (1976), “A function for the size distribution of incomes,”

Econometrica 44, 963-970.

[59] Solow, R. (1956), “A Contribution to the Theory of Economic Growth,” The Quar-

terly Journal of Economics, 70, 65-94.

[60] Shorrocks, A. and Wan, G. (2009), “Ungrouping income distributions: synthesising

samples for inequality and poverty analysis,” (in K. Basu and R. Kanbur, eds)

39

Arguments for a Better World: Essays in Honor of Amartya Sen, Vol. I: Ethics,

Welfare and Measurement, pp. 414-434, Oxford: Oxford University Press.

[61] Subramanian, S. and Jayaraj, D. (2014), “Growth and Inequality in the Distribution

of India’s Consumption Expenditure: 1983—2009-10,” Paper presented at UNU-

WIDER Conference on Inequality—Measurement, trends, impacts, and policies,

Helsinki, 5-6 September, 2014.

[62] UNDP (1999), Human Development Report, 1999. New York and Oxford: Oxford

University Press.

[63] UNU-WIDER (2014), “World Income Inequality Database (WIID3.0b),” September

2014.

[64] Villasenor, J. and Arnold, B. (1989), “Elliptical Lorenz curves,” Journal of Econo-

metrics 40, 327-338.

[65] Wilkinson, R. and Pickett, K. (2006), “Income inequality and population health: A

review and explanation of the evidence,” Social Science & Medicine 62, 1768-1784.

[66] Williamson, J. (1999), “Globalization and inequality then and now: the late 19th

and late 20th centuries compared,” National Bureau of Economic Research Working

Paper no. 5491, Cambridge, Mass.

[67] Wood, A. (1995). “How trade hurt unskilled workers,” Journal of Economic Per-

spectives 9, 57-80.

[68] World Bank (2008), “Global Purchasing Power Parities and Real Ex-

penditures: 2005 International Comparison Program,” available online at

[http://siteresources.worldbank.org/ICPINT/Resources/icp-final.pdf], accessed 24

February 2013.

[69] You, J-S and Khagram, S. (2005), “A Comparative Study of Inequality and Cor-

ruption,” American Sociological Review 70, 136-157.

[70] Zheng, B. (1997), “Unit-Consistent Decomposable Inequality Measures,” Econom-

ica 74, 97-111.

40

A Appendix

Table A.1: East Asia & Pacific Sample Composition1975 1985 1995 2000 2005 2010

China Australia Australia Cambodia Australia Australia

Fiji China Cambodia China Cambodia Cambodia

Hong Kong Indonesia China Hong Kong China China

Indonesia Korea, Rep. Fiji Indonesia Fiji Fiji

Japan Malaysia Hong Kong Korea, Rep. Hong Konga Hong Kong

Korea, Rep. New Zealand Indonesia Lao PDR Indonesia Japan

Malaysia Philippines Japan Micronesia Japan Lao PDR

New Zealand Thailand Korea, Rep. Mongolia Korea, Rep. Malaysia

Philippines Lao PDR Philippines Malaysia Mongolia

Thailand Malaysia Singapore Micronesia New Zealand

Mongolia Thailand Philippines Philippines

New Zealand Timor-Leste Singapore Singapore

Papua New Guinea Vietnam Thailand Thailand

Philippines Timor-Leste Vietnam

Thailand Vietnam

Vietnam

Coverage 90% 83% 95% 87% 95% 82%

41

Table A.2: Europe & Central Asia Sample Composition1975 1985 1995 2000 2005 2010

Denmark Austria Albania Albania Albania Albania

Finland Belarus Armenia Armenia Armenia Armenia

France Belgium Austria Austria Austria Austria

Germany Bulgaria Azerbaijan Azerbaijan Belarus Azerbaijan

Hungary Czechoslovakia Belarus Belarus Belgium Belarus

Ireland Denmark Belgium Belgium Bosnia & Herz. Belgium

Italy Finland Bulgaria Bosnia & Herz. Bulgaria Bulgaria

Netherlands France Czech Republic Bulgaria Croatia Croatia

Norway Germany Denmark Croatia Cyprus Cyprus

Spain Hungary Estonia Czech Republic Czech Republic Czech Republic

Sweden Ireland Finland Denmark Denmark Denmark

United Kingdom Italy France Estonia Estonia Estonia

Kazakhstan Georgia Finland Finland Finland

Kyrgyz Republic Germany France France France

Latvia Greece Georgia Georgia Georgia

Lithuania Hungary Germany Germany Germany

Luxembourg Ireland Greece Greece Greece

Moldova Italy Hungary Hungary Hungary

Netherlands Kazakhstan Ireland Iceland Iceland

Norway Kyrgyz Republic Italy Ireland Ireland

Poland Latvia Kazakhstan Italy Italy

Portugal Lithuania Kyrgyz Republic Kazakhstan Kazakhstan

Romania Luxembourg Latvia Kyrgyz Republic Kyrgyz Republic

Russian Federation Macedonia Lithuania Latvia Latvia

Slovak Republic Moldova Luxembourg Lithuania Lithuania

Slovenia Netherlands Macedonia Luxembourg Luxembourg

Spain Norway Moldova Macedonia Macedonia

Sweden Poland Netherlands Moldova Moldova

Switzerland Portugal Norway Montenegro Montenegro

Turkey Romania Poland Netherlands Netherlands

Turkmenistan Russian Federation Portugal Norway Norway

Ukraine Yugoslavia Romania Poland Poland

United Kingdom Slovak Republic Russian Federation Portugal Portugal

Uzbekistan Slovenia Yugoslavia Romania Romania

Spain Slovak Republic Russian Federation Russian Federation

Sweden Slovenia Serbia Serbia

Switzerland Spain Slovak Republic Slovak Republic

Turkey Sweden Slovenia Slovenia

Turkmenistan Switzerland Spain Spain

Ukraine Tajikistan Sweden Sweden

United Kingdom Turkey Switzerland Switzerland

Uzbekistan Turkmenistan Tajikistan Tajikistan

Ukraine Turkey Turkey

United Kingdom Ukraine Ukraine

Uzbekistan United Kingdom United Kingdom

Uzbekistan

Coverage 43% 94% 98% 100% 98% 96%

42

Table A.3: Latin America & Caribbean Sample Composition1975 1985 1995 2000 2005 2010

Argentina Argentina Ecuador Suriname Argentina Argentina

Bahamas, The Bolivia Argentina Argentina Bolivia Barbados

Barbados Brazil Belize Belize Brazil Bolivia

Brazil Chile Bolivia Bolivia Chile Brazil

Colombia Colombia Brazil Brazil Colombia Chile

Costa Rica Costa Rica Chile Chile Costa Rica Colombia

Ecuador Dominican Rep. Colombia Colombia Dominican Rep. Costa Rica

Jamaica Ecuador Costa Rica Costa Rica Ecuador Dominican Rep.

Mexico Guatemala Dominican Rep. Dominican Rep. El Salvador Ecuador

Panama Honduras El Salvador Ecuador Guatemala El Salvador

Peru Jamaica Guyana El Salvador Honduras Honduras

Trinidad and Tobago Mexico Honduras Guatemala Jamaica Mexico

Uruguay Panama Jamaica Guyana Mexico Panama

Venezuela, RB Paraguay Mexico Haiti Nicaragua Paraguay

Peru Nicaragua Honduras Panama Peru

Trinidad and Tobago Panama Mexico Paraguay Uruguay

Uruguay Paraguay Nicaragua Peru Venezuela, RB

Venezuela, RB Peru Panama Uruguay

St. Lucia Paraguay Venezuela, RB

Trinidad and Tobago Peru

Uruguay Uruguay

Venezuela, RB Venezuela, RB

Coverage 82% 92% 93% 96% 95% 91%

Table A.4: Middle East & North Africa Sample Composition1975 1985 1995 2000 2005 2010

Egypt, Arab Rep. Algeria Algeria Djibouti Egypt, Arab Rep. Egypt, Arab Rep.

Iran, Islamic Rep. Iran, Islamic Rep. Djibouti Egypt, Arab Rep. Iran, Islamic Rep. Jordan

Israel Israel Egypt, Arab Rep. Iran, Islamic Rep. Iraq Malta

Jordan Jordan Iran, Islamic Rep. Israel Israel West Bank and Gaza

Morocco Morocco Israel Morocco Jordan

Tunisia Tunisia Jordan Tunisia Malta

Morocco Yemen, Rep. Morocco

Tunisia Syrian Arab Rep.

Yemen, Rep. Tunisia

West Bank and Gaza

Yemen, Rep.

Coverage 64% 48% 75% 63% 77% 24%

43

Table A.5: North America Sample Composition1975 1985 1995 2000 2005 2010

Canada Canada Canada Canada Canada Canada

United States United States United States United States United States United States

Coverage 100% 100% 100% 100% 100% 100%

Table A.6: South Asia Sample Composition1975 1985 1995 2000 2005 2010

Bangladesh Bangladesh Bangladesh Bangladesh Afghanistan Bangladesh

India India India India Bangladesh Bhutan

Nepal Nepal Nepal Maldives Bhutan India

Pakistan Pakistan Pakistan Pakistan India Maldives

Sri Lanka Sri Lanka Sri Lanka Sri Lanka Maldives Nepal

Nepal Pakistan

Pakistan

Coverage 98% 98% 98% 96% 99% 97%

44

Table A.7: Sub-Saharan Africa Sample Composition1975 1985 1995 2000 2005 2010

Botswana Botswana Botswana Angola Benin Angola

Cote d’Ivoire Cote d’Ivoire Burkina Faso Burkina Faso Burkina Faso Burkina Faso

Kenya Ethiopia Burundi Burundi Burundi Central African Rep.

Malawi Ghana Cameroon Cameroon Cameroon Cote d’Ivoire

Nigeria Lesotho Central African Rep. Cabo Verde Central African Rep. Ethiopia

Zambia Madagascar Cote d’Ivoire Cote d’Ivoire Chad Madagascar

Malawi Ethiopia Ethiopia Comoros Malawi

Mali Gambia, The Gambia, The Congo, Rep. Mali

Mauritania Ghana Ghana Congo, Dem. Rep. Mauritania

Nigeria Guinea Guinea-Bissau Ethiopia Mozambique

Rwanda Guinea-Bissau Kenya Gabon Namibia

Sierra Leone Kenya Madagascar Gambia, The Niger

Uganda Lesotho Malawi Ghana Nigeria

Madagascar Mali Guinea Rwanda

Mali Mauritania Kenya South Africa

Mauritania Rwanda Lesotho Sudan

Mozambique Sao Tome and Principe Liberia Swaziland

Namibia Senegal Madagascar Senegal

Niger Seychelles Malawi Uganda

Nigeria South Africa Mali

Senegal Swaziland Mauritania

South Africa Tanzania Mauritius

Swaziland Uganda Mozambique

Tanzania Zambia Namibia

Uganda Niger

Zambia Nigeria

Rwanda

Senegal

Seychelles

Sierra Leone

South Africa

Tanzania

Togo

Uganda

Zambia

Coverage 29% 46% 74% 53% 86% 63%

45

global inequality · global inequality: relatively lower, absolutely higher 1 miguel nino-zaraz~ ua...

Documents