The following summary was prepared by Kate Vyborny
Junior Fellow, Carnegie Endowment for International Peace

In his new book, Worlds Apart: Measuring International and Global Inequality, top World Bank economist Branko Milanovic analyzes income distribution worldwide using, for the first time, household survey data from more than 100 countries. Worlds Apart, offers a more accurate way of measuring inequality and discusses the relevant policies of first-world countries and NGOs.  The Carnegie Endowment's Trade Equity and Development Project recently hosted a discussion with Branko Milanovic, Francois Bourguignon, Chief Economist of the World Bank, and Thomas Pogge of Columbia University on Worlds Apart

Branko Milanovic, World Bank / Carnegie Endowment for International Peace

I will explore concepts of inequality, whether and why global inequality matters, and what can be done about it.  My book expands on three distinct conceptions of inequality on a world level:

1. Concept 1 inequality is inequality between nations, as measured by the differences in their Gross Domestic Income (GDI) per capita.  Essentially, this counts each nation-state equally, as a single entity, and disregards the size of the countries. 

2. Concept 2 inequality takes the size of countries into account, but still uses GDI per capita for the entire country.  So, it essentially assumes that each Chinese person has income equal to the GDI per capita of China – ignoring inequality within countries. 

3. Concept 3 inequality measures inequality on a global level, taking each person as an individual.

The three measures are useful for different purposes.  For example, if we want to look at why many countries which adhered to the same policies have experienced radically different outcomes, then Concept 1 inequality is a useful tool to examine this question.  Concept 2 is useful primarily as a proxy for Concept 3; it is useful because it is easier to measure than Concept 3.  Concept 3 inequality truly measures inequality between individuals in the world. 

From the 1950s to the present, Concept 1 inequality (between countries) has increased.  This is because the developed countries, China and India have grown, while Africa and the middle income countries (Latin America, Eastern Europe and former UUSR) experienced relative decline.  Some even declined in absolute terms; on average, 33% of middle income countries, and 43% of LLDCs (Least Developed Countries) have experienced negative growth rates in a given year. 

We arbitrarily classify countries into four “worlds”: to define the First World, we include the West and its offshoots.  Then we take the income per capita of the poorest First World Country (Portugal) and divide remaining countries into three additional groups according to their income relative to Portugal.  The Second World includes countries with a per capita GDP 1/3 less than Portugal’s, the Third World includes those from 1/3 to 2/3 below Portugal’s per capita GDP, and the Fourth World includes those more than 2/3 below Portugal.  Looking at these groups over time, we see several trends: a shrinking of the Second World, China’s movement from the Fourth to the Third world, a decline of Latin America in relative terms, and the decline of Africa.  To put this last trend into perspective, 350 million Africans live in countries which are poorer in 2000 than in 1980 (per capita GDP is lower), during the Carter administration; 180 million Africans live in countries which were poorer in 2000 than in 1963, under the Kennedy administration.  In that same period, while those countries became poorer, US GDP per capita doubled. 

While concept 1 inequality increased from the 1950s to the present, Concept 2 inequality has decreased, largely due to the great increases in per capita GDP of India and China, which are heavily weighed due to their large populations.  Excluding India and China shows a roughly flat trend in Concept 2 inequality over this period. 

If, however, we use alternative data, by breaking larger countries into their component states or rural/urban areas, using alternative GDI per capita data for China, and expanding the sample size to include “failed” countries, we see no decline in Concept 2 inequality. 

Looking at a longer time span, from 1870 to 2000, we can say there has been a significant increase in Concept 1 and Concept 2 inequality because of the industrial revolution.  Concept 3 inequality increased but also changed in composition: in 1870, intra-country inequality (inequality due to class structure) was greater; now, inter-country inequality (inequality due to location) is greater. 

Now let us examine Concept 3 inequality today, that is, inequality between world citizens today.  The data is more difficult to gather because household-level data for each country is necessary to measure Concept 3.  Concept 3 inequality has been influenced by a number of contradictory changes, including greater inequality within nations, greater differences between countries’ mean incomes, and the “catching up” of large, poor countries like India and China. 

Today, we see a roughly 90-10 split: in other words, if we adjust for differences in price levels (Purchasing Power Parity or PPP income), the bottom 90% of the world’s population has half of world income, and the top 10% has the other half.  In simple dollar terms (not adjusted for price levels), the top 10% has two-thirds of the world’s income. 

If we look at the income distribution in individual countries, we can see that in many cases rich countries have first order dominance over poorer countries.  In other words, not even the richest people in rural India have as much as the poorest people in France.  But this is not true for Brazil, for example: about a third of Brazil’s population is better off than the poorest 10% of France’s population.  This has important implications for policy.

So, we know that Concept 3 inequality is very high; but, because of the lack of data, we don’t have a clear indication of its direction of change. 

So, why does global inequality matter?  There is no global government, no global taxation authority, and arguably no one can do much about it.  However, in an era of globalization, people have a higher level of awareness of their relative standard of living, leading to migration.  Also, we know that inequality at a country level is linked with conflict; at a world level, it is likely to lead to conflict as well.

Which is more important for people (from which do they derive more utility): their absolute level of income, or their relative level of income?  If relative income is more important, then with globalization, as people have more information about living standards in other places, the poor understand that they are relatively poor.  This decreases their well-being.  Analysts disagree as to the importance of absolute versus relative level of income. 

What can be done about global inequality?  To improve the situation of poor countries, changes could be made in global trade rules, including the removal of agricultural subsidies, free trade in textiles, steel, and other sensitive products, and changes in WTO rules to emphasize intellectual property rights and financial liberalization less.  But what about global transfers, like a global safety net as we have within most countries to some extent? 

What kind of rules should such transfers follow?  They should be progressive, that is flowing from the rich to the poor. 

• “Progressivity 1”: flowing from a rich to a poor country – this is relatively simple.

• But they should also be globally progressive, that is flowing from a richer individual to a poorer individual; and

• Nationally progressive, or improving the inequality within countries – in the rich country granting the transfer, the taxpayers should be rich, and in the poor country recipient, the beneficiaries of the transfer should be the poor.

In order to ensure these conditions are met, we should reexamine the issue of national distributions of income.  If we make a transfer between two countries that do not overlap in their distribution, it is assured to be globally progressive (but not nationally progressive) no matter which individuals are taxpayers and recipients of the transfer.  For this reason, we should treat more favorably poor countries with low inequality as transfer recipients, since this makes globally progressive transfers more likely. 

This represents the beginning of a paradigm shift; instead of thinking about transfers from state to state, or from inter-state organizations to states, but rather from a global authority to individual citizens; a natural complement to any global tax authority is a relationship with poor citizens as individuals, rather than poor states. 


Francois Bourguignon, World Bank

The idea of Concept 3 inequality is a useful tool, and for many purposes it makes sense to move towards using Concept 3 as a measure rather than concept 1.  I have a few footnotes to add to these useful ideas.

First, what are the implications of the different concepts of inequality?  If we ask people who have polarized views what has been the result of globalization, they will use different concepts of inequality to support their different views – this distinction between the three concepts clarifies the seemingly contradictory evidence. 

I would also point out that the three concepts still give a static view of distribution, whereas if we look at the dynamic story, we should examine individual mobility within the world distribution – in other words, whether people move within the distribution, stay in place, or actually deteriorate within the distribution of income over time. 

If we use the idea of Pareto improvement, meaning that a change is an improvement only if everyone is the same or better off as before, then we realize that redistribution requires a value judgment of the relative importance of people at different points of the distribution than others.  For instance, if we make transfers from rich to poor, we are placing value on those at the bottom of the distribution.  We can either place an emphasis on the situation of those at the bottom of the distribution, or we can look at the whole picture of society. 

My second point is regarding the relationship between inequality and globalization.  In the first part of the 20th century, we saw a huge increase in Concept 1 inequality; then for the last twenty, years we have seen a decrease in Concept 1, but an increase in inequality within countries - because of increased globalization of capital and trade flows.  The logical limit of this is that in a fully globalized world, with completely free trade and capital flows, there would be no more inequality across countries, but only inequality within countries. 

We do see, though, that a level of inequality that would be unacceptable at the country level has become acceptable at the world level; and the redistribution through Official Development Assistance is miniscule compared to the larger levels of transfers within countries. 

Third, income is only one dimension of welfare.  Equity in fact is a broader concept than simply income distribution, but rather includes access to public services, as well as the actual outcomes for people – their level of health, education, lifespan, etc.  And we have seen over time a convergence of some of these outcomes, such as lifespan.  (The effect of AIDS negates this, but netting out AIDS, lifespans have converged.) 

In conclusion, I would outline the historical trend that world inequality has increased since the 19th century to the middle of the 20th century.  However, after the Second World War, development became a world issue and concern.  For that reason, inequality, particularly if we look at it holistically in terms of people’s general welfare, is not a lost cause, and is a worthwhile objective. 


Thomas Pogge, Columbia University

As we can see in a recent Economist article “Catching Up,” (August 21, 2003), the authors show trends in Concept 1 and Concept 2 inequality, but do not examine Concept 3.  If we look at global inequality from a moral point of view, Concept 3 is clearly the most accurate way to measure inequality. 

If we look at a few statistics, the gravity of the situation becomes clear; the ratio in income between the top 10% of the world’s population and the bottom 10% is 320 to 1.  Everyone in this room is in that top decile; if each of us took just 1/320 of their income and gave to someone in the bottom decile, their income would actually double.  Meanwhile, one third of human deaths today have poverty related causes.  While some may argue that poverty is important, while inequality is not, in fact we have to consider not just inequality in the abstract, but this inequality, the inequality that does exist, which is strongly linked to poverty.  Studies show that people’s relative poverty impacts people’s happiness with their situation as well as their health.  Because of improved communications, people know more about their situation relative to others; this can help to breed dissatisfaction and resentment.  In addition, we know that inequality actually works against growth. 

In considering how societies develop, we must also take into account the political competition between actors over the economic rules of the game.  The actors which are economically and politically powerful work to design legislation in their own favor – we see this in the United States through the process of corporate lobbying.  Thus societies can stabilize at various levels of inequality depending on how those powerful actors influence the rules of the game.  This process occurs on an international level as well.  For this reason, we must be wary of accepting great levels of inequality in the short run as a byproduct of rapid growth in the hope that inequality can be solved later, as is the case now in China; societies can become path dependent, since once economic and political power is concentrated to a few actors, those actors may redesign the rules in their favor, ensuring that the high level of inequality persists.  Lula has run up against this problem in Brazil. 

We tend to look at information on inequality in the context of deeming globalization a good or bad phenomenon.  But, in reality, this is not a binary choice; we should talk about what kind of globalization we want; we should examine inequality in the context of what can or should we do to improve the situation. 


Question and Answer Session

Q: Could you clarify what you mean by “class” versus “location” in saying that the composition of Concept 3 inequality has changed from 1870 to the present?  What do you mean by class?

A: (Milanovic) This does not represent a complex idea of class; rather, I used “class” and “location” to represent intra-society inequality (class), and inter-society inequality (location). 

Q: What are the implications of this study for World Bank Operations?  Also, what kind of publicity will the World Bank give this work, as opposed to research which shows that globalization is positive, which is much better publicized? 

A: (Bourguignon) Of course give maximum publicity to our own products, and this work will be well advertised.  As you may know, the most recent World Development Report had an emphasis on Equity and Development; there is a chapter in it which is drawn principally from Branko’s work. 

(Milanovic) The operational significance of this work is limited to a great extent.

(Bourguignon) However, it can play a significant part of the Bank’s advocacy role, rather than its day-to-day operations.  The results will be used in documents which push for poverty reduction in the least developed countries. 

Q: What are the policy implications you, Branko, draw from this?

A: (Milanovic) I draw two sets of policy conclusions from this work:

1. Rules for global redistribution, which should be progressive in the three ways (outlined above):
a. Progressivity 1 (transfers go only from richer to poorer countries),
b. Globally Progressive (transfers go only from richer to poorer individuals), and
c. Nationally Progressive (improves equality within both donor and recipient countries). 

2.  The need in global governance for global agencies rather than international agencies which deal with individuals, rather than nation-states, and represent their interests rather than those of shareholders, as do the International Financial Institutions. 

(Bourguignon) Aid is clearly progressive in the first way (Progressivity 1), but it is not clear whether it fits the conditions of global and national progressivity.  Trade protection is regressive rather than progressive, and much more than offsets the amount of aid given.  This is particularly the case as regards agreements on intellectual property rights, for instance. 

Q: I would argue that absolute poverty is at least as important as is inequality.  Also, it is essential to understand the reason for the inequality.  Is it justified or explained?  An increase in inequality related to globalization would be an enigma for classical economists, as it runs counter to the Heckscher-Ohlin theorem, which states that trade equalizes incomes even if capital and labor are not mobile. 

A: We should not be surprised because the real world does not conform perfectly to theory.  Rather, it is shaped in part by the rules of the game; for instance, the flight of capital from capital-poor developing countries to Swiss banks does not fit economic theory, but it is widespread, often due to embezzlement.  The amount of this capital flight dwarfs Official Development Assistance.  The same is true for bribes: often interest groups bribe officials to spend public money in ways that are bad for their country. 

If we consider absolute poverty as the main issue, we have to ask ourselves whether there will ever be a point at which we can say there is no more absolute poverty in the world.  At that point we are likely to redefine the norm or benchmark of what we consider absolute poverty, so in effect it is all relative to some extent. 

Q: Are the politics and policies of domestic and international regimes to some extent also shaped by middle income groups, particularly where those are growing in numbers? 

A: (Pogge) Yes.  One way of looking at this is to set a constant equal to the income of the poorest person.  Then, subtract that constant from each person’s income to approximate their level of political influence.  However, there is also the issue that these middle income people are more dispersed and less coordinated than the few powerful actors at the top.