Global income distribution: From the fall of the Berlin Wall to the Great Recession

Christoph Lakner , Branko Milanovic, 27 May 2014



The period between the fall of the Berlin Wall and the Great Recession saw probably the most profound reshuffle of individual incomes on the global scale since the Industrial Revolution. This was driven by high growth rates of populous and formerly poor or very poor countries like China, Indonesia, and India; and, on the other hand, by the stagnation or decline of incomes in sub-Saharan Africa and post-communist countries as well as among poorer segments of the population in rich countries.

Anand and Segal (2008) offer a detailed review of the work on global income inequality. In Lakner and Milanovic (2013), we address some of the limitations of these earlier studies and present new results from detailed work on household survey data from about 120 countries over the period 1988–2008. Each country’s distribution is divided into ten deciles (each decile consists of 10% of the national population) according to their per capita disposable income (or consumption). In order to make incomes comparable across countries and time, they are corrected both for domestic inflation and differences in price levels between countries. It is then possible to observe not only how the position of different countries changes over time – as we usually do – but also how the position of various deciles within each country changes. For example, Japan’s top decile remained at the 99th (2nd highest from the top) world percentile, but Japan’s median decile dropped from the 91st to the 88th global percentile. Or, to take another example, the top Chinese urban decile moved from being in the 68th global percentile in 1988 to being in the 83rd global percentile in 2008, thus leapfrogging in the process some 15% of the world population – equivalent to almost a billion people.

When we line up all individuals in the world, from the poorest to the richest (going from left to right on the horizontal axis in Figure 1), and display on the vertical axis the percentage increase in the real income of the equivalent group over the period 1988–2008, we generate a global growth incidence curve – the first of its kind ever, because such data at the global level were not available before. The curve has an unusual supine S shape, indicating that the largest gains were realised by the groups around the global median (50th percentile) and among the global top 1%. But after the global median, the gains rapidly decrease, becoming almost negligible around the 85th–90th global percentiles and then shooting up for the global top 1%. As a result, growth in the income of the top ventile (top 5%) accounted for 44% of the increase in global income between 1988 and 2008.

Figure 1. Anonymous global growth incidence curve: Real income change at various percentiles of the global income distribution between 1988 and 2008 (%)

Fortunes of income deciles in different countries over time

The curve in Figure 1 is drawn using a simple comparison of real income levels at given percentiles of the global income distribution in 1988 and 2008. It is ‘anonymous’ because it does not tell us what happened to the actual people who were at given global income percentiles in the initial year, 1988. In fact, the regional composition of the different global income groups changed radically over time because growth was uneven across regions. A ‘quasi non-anonymous’ growth incidence curve in Figure 2 adjusts for this – the growth rates are calculated for all individual country/deciles at the positions they held in the initial year (1988). The growth rate on the vertical axis (calculated from a non-parametric fit) thus shows how the country/deciles that were poor, middle-class, rich, etc. in 1988 performed over the next 20 years. The supine S shape still remains, although it is now slightly less dramatic.

People around the median almost doubled their real incomes. Not surprisingly, 9 out of 10 such ‘winners’ were from the ‘resurgent Asia’. For example, a person around the middle of the Chinese urban income distribution saw his or her 1988 real income multiplied by a factor of almost 3; someone in the middle of the Indonesian or Thai income distribution by a factor of 2, Indian by a factor of 1.4, etc.

It is perhaps less expected that people who gained the least were almost entirely from the ‘mature economies’ – OECD members that include also a number of former communist countries. But even when the latter are excluded, the overwhelming majority in that group of ‘losers’ are from the ‘old, conventional’ rich world. But not just anyone from the rich world. Rather, the ‘losers’ were predominantly the people who in their countries belong to the lower halves of national income distributions. Those around the median of the German income distribution have gained only 7% in real terms over 20 years; those in the US, 26%. Those in Japan lost out in real terms.

Figure 2 Quasi non-anonymous global growth incidence curve: Real income change between 1988 and 2008 across 1988 percentiles of the global income distribution


The particular supine S-shaped growth incidence curve (Figure 1) does not allow us to immediately tell whether global inequality might have gone up or down because the gains around the median (which tend to reduce inequality) may be offset by the gains of the global top 1% (which tend to increase inequality). On balance, however, it turns out that the first element dominates, and that global inequality – as measured by most conventional indicators – went down. The global Gini coefficient fell by almost 2 Gini points (from 72.2 to 70.5) during the past 20 years of globalisation. Was it then all for the better?

Probably yes, but not so simply. The striking association of large gains around the median of the global income distribution – received mostly by the Asian populations – and the stagnation of incomes among the poor or lower middle classes in rich countries, naturally opens the question of whether the two are associated. Does the growth of China and India take place on the back of the middle class in rich countries? There are many studies that, for particular types of workers, discuss the substitutability between rich countries’ low-skilled labour and Asian labour embodied in traded goods and services or outsourcing. Global income data do not allow us to establish or reject the causality. But they are quite suggestive that the two phenomena may be related.

Figure 3 Real per capita income of the 2nd income decile in the US and the 8th urban income decile in China between 1988 and 2011

A dramatic way to see the change brought by globalisation is to compare the evolution over time of the 2nd US income decile with (say) the Chinese urban 8th decile (Figure 3). Indeed we are comparing relatively poor people in the US with relatively rich people in China, but given the income differences between the two countries, and that the two groups may be thought to be in some kind of global competition, the comparison makes sense. Here we extend the analysis to 2011, using more recent and preliminary data. While the real income of the US 2nd decile has increased by some 20% in a quarter century, the income of China’s 8th decile has been multiplied by a factor of 6.5. The absolute income gap, still significant five years ago, before the onset of the Great Recession, has narrowed substantially.

Political implications

And even if the causality cannot be established because of many technical difficulties and an inability to define credible counterfactuals, the association between the two cannot pass unnoticed. What, then, are its implications? First, will the bottom incomes of the rich countries continue to stagnate as the rest of China, or later Indonesia, Nigeria, India, etc. follow the upward movement of Chinese workers through the ranks of the global income distribution? Does this imply that the developments that are indeed profoundly positive from the global point of view may prove to be destabilising for individual rich countries?

Second, if we take a simplistic, but effective, view that democracy is correlated with a large and vibrant middle class, its continued hollowing-out in the rich world would, combined with growth of incomes at the top, imply a movement away from democracy and towards forms of plutocracy. Could then the developing countries, with their rising middle classes, become more democratic and the US, with its shrinking middle class, less?

Third, and probably the most difficult: What would such movements, if they continue for a couple of decades, imply for global stability? The formation of a global middle class, or the already perceptible ‘homogenisation’ of the global top 1%, regardless of their nationality, may be both deemed good for world stability and interdependency, and socially bad for individual countries as the rich get ‘delinked’ from their fellow citizens.


In a nutshell, the movements that we witness do not only lead to an economic rebalancing of the East and West – in which both may end up with global output shares close to what they had before the Industrial Revolution – but to a contradiction between the current world order, where political power is concentrated at the level of the nation-state, and the economic forces of globalisation which have gone far beyond it.


Anand, Sudhir and Paul Segal (2008), “What Do We Know about Global Income Inequality?”, Journal of Economic Literature, 46(1): 57–94.

Lakner, Christoph and Branko Milanovic (2013), “Global income distribution: from the fall of the Berlin Wall to the Great Recession”, World Bank Working Paper No. 6719, December.

Topics: Global economy, Politics and economics, Poverty and income inequality
Tags: democracy, globalisation, income inequality, Inequality

Extended Term Consultant, Development Research Group (Poverty & Inequality team), World Bank
Visiting Presidential Professor, Graduate Center University of New York; Senior Scholar, Luxembourg Income Center