I just completed my latest round of global inequality calculations several days ago. The data refer to the benchmark year 2011. This is the seventh benchmark year since I started working on global inequality. My first paper on the topic included only two years: 1988 and 1993. These were (as the title of the paper says) the first calculations of global inequality based on household survey data alone. Afterwards, in my book "Worlds Apart", I added the year 1998. And then it went on with years 2002 and 2005 (discussed here), and finally with a panel of country/deciles created specifically for the paper I wrote with Christoph Lakner. The quality of the data improved significantly. While in 1988, I had micro data for less than half of all surveys, in 2008 and 2011, all surveys (with the unfortunate exceptions of China in 2011 and Japan in both years) are based on micro data.
But on the other hand, the coverage of the world population has not improved much in the past twenty five years. I still have data for about 92-95 percent of world population and generally slightly more of world income. My own rule was that unless I have more than 90% of world population and dollar GDP covered, and at least 110 countries included, I cannot claim to have a global distribution. It is that yardstick that I reached last week, and that’s why my 2011 numbers are now ready. For 2011, the population coverage is 92 percent, income coverage (measured in current dollar terms) 90 percent, and the number of countries included is 112.
The missing populations are generally people in poor countries. These countries do not conduct regular household surveys and thus our global income inequality numbers are almost certainly underestimates. There are two additional reasons why they are underestimates but I will mention them below.
Is global inequality on the decline?
In a paper that discussed changes in global inequality between 1988 and 2008, Christoph Lakner and I find a slight, but unmistakable, decrease in global Gini, from 72 points in 1988 and 1993 to 70.5 Gini points in 2008. These were panel data. On a more detailed sample for the benchmark year 2008, I find an even lower Gini of 69. So, the decline in global inequality of about 2 to 3 Gini points was already present during the first decade of the 21 century.
The 2011 data show that the decline has continued. The 2011 global Gini is 67, some 2 Gini points lower than three years before. This is a remarkable change, driven, as before, by the fast growth rates of China and India, and between 2008 and 2011 also by the absence of growth in rich countries. The average real per capita income, calculated from Chinese household surveys, has increased by 45 percent between 2008 and 2011; in India, the increase was 11 percent. But now global inequality is decreasing also because the rich world is not growing. The US mean income in 2011 is lower than in 2008. That makes global convergence of people’s incomes easier.
What are the caveats?
There are three. In addition to our imperfect coverage of poor countries, global inequality is underestimated also by surveys’ non-inclusion of the very rich and, when they are included at all, by the underestimation of their incomes, especially of those derived from the ownership of capital. This has been a well-known issue for years but has become more serious because of globalization that allows people to move their assets more easily (we are speaking here of legally-owned assets, income from which owners tend to “forget” when they fill survey questionnaires). When Christoph Lakner and I made some rough adjustments for the underestimate of top 1% incomes, it turned out that almost all but ½ of a Gini point of global inequality decline dissipated. So, this is the second caveat: it could be that the global inequality decline is significantly less if we were to account better for the incomes of the very rich.
The third caveat is similar to the second. There are (we are now much more aware of this after the pioneering work by Gabriel Zucman) large financial assets that are not included either in fiscal nor in household survey data. This is money held in tax havens. Zucman estimates it at 8% of global financial assets. Information on income from these assets is extremely unlikely to be reported since even the legality of these assets (unlike those of the previous group) is questionable. Using some back-of-the envelope calculation, income from the assets “parked” in tax havens could amount to 0.7-1 percent of global GDP. Since it is very likely to be received by the global top 1%, we are thereby further underestimating global inequality.
The issue of the new 2011 PPPs
The Gini values I have provided so far are based on the Purchasing Power Parity (PPP) dollar exchange rates derived from the 2005 International Comparison Project (ICP). One of the reasons I chose to update my global inequality data for the year 2011 is that the new ICP was conducted in that year. When household surveys are conducted in the same year as ICP, domestic currency incomes are divided by PPP exchange rates which have also been calculated for that same year. One source of error, projection of PPPs to the year when the global ICP is not conducted, is thus avoided. The 2011 PPPs, as is by now well-known, have resulted in some significant reassessment of price levels in Asian countries (Indonesia, India, China). As their price levels were found to be lower than previously thought, PPP exchange rates were reduced, and real incomes accordingly increased. (If country's price level is lower then obviously with the same nominal income, you can buy more goods.) In global inequality calculations, higher real incomes of China, India and Indonesia (among the large countries) translate into reduced global inequality.
The effect of 2011 PPPs was essentially a level effect, meaning that the changes and the trend were left more or less unaffected while the level of global inequality was reduced throughout (that is, for all years) by about 3 Gini points. Thus, 2011 global inequality, measured in 2011 PPPs, is “only” 64 Gini points.
In 2011, the emergence of the global “middle class” continues. If we take one of the definitions of the middle class, made popular by the World Bank’s work on Latin America, as people with incomes above $PPP10 per day and less than $PPP50 per day, 29% of world's population belongs to that group and they control 43% of global income. In 1998, equally defined "middle class" included only 17% of world population. Unlike in 1988 and 1993, when global income distribution showed two peaks with high percentages of people among either the extremely poor or the very rich, today’s global income distribution is single-peaked with the “thickening” of the distribution around the middle. (Note however that the global median income of $PPP5.7 per day is still vastly below the minimum income that we use to define our “global middle class”).
The global top 1%
The global top 1% is still heavily dominated by the “old rich” OECD countries. There are just above 60 million people in the global top 1%. One-half of them are Americans: the richest 11% of Americans are the members of this “club.” (If we look at the global 2%, the entire top quintile of the US income distribution belongs there.) From other rich and relatively populous countries (Germany, France, Japan, Great Britain) 4-5% of the richest people belong to the global top 1%. Outside the “old OECD”, only Chile and Taiwan have more than 1 percent of their populations in the global top 1%. Brazil, South Africa and Russia each have their own top 1% also in the global top 1%, but not more than that.
National vices, global virtue?
There are two take-away messages. The first is that while many national income and wealth inequalities are increasing, or are thought to be unacceptably high, global income inequality is charting a modest decline. This is not surprising because the bulk of global inequality depends on differences in mean incomes between rich and poor countries, and these are now less than at any time since the end of World War II. But for this trend to continue, fast growth rates of China and India are crucial. Second take-away message is this: while many national middle classes in rich countries are being “hollowed out”, a global “middle class” is emerging. The shape of the world distribution has moved from being bi-polar, with one peak of the distribution among the very poor and another among the rich, to being single-peaked with a more sizable concentration of people into something that may be called “the global middle class.”