3. Who Got What, Then and Now? A Fifty Years Overview from the Global Consumption and Income Project . 51
3.5. Aggregate ‘Social Welfare’
Given the uneven nature of growth as well as the changes in inequality, how might we go about assessing whether the world taken as a whole has experienced economic welfare gains during the period in consideration and if so to what extent? A useful tool in this regard is the Generalized Lorenz Curve (GLC) which allows us effectively to rank distributions in terms of welfare.
Shorrocks (1983) showed that for any welfare function that is Schur Concave (i.e. responds negatively to regressive Pigou-Dalton Transfers and is therefore inequality averse) and that is positively responsive to income, a given distribution of income would provide more welfare than another distribution if its GLC were everywhere higher. Moreover, all such income-focused
75
welfare functions would agree on the welfare ranking of two situations if and only if such
‘dominance’ is established. Figure 3.12 depicts the generalized Lorenz curve for the world distribution of income in four periods of time. Clearly, by any measure, compared to 1960, the welfare of the world had increased by 1990. Perhaps somewhat surprisingly however, there was no unambiguous welfare improvement between the 1990 and 2000 distributions of income, since the generalized Lorenz curves lie almost on top of each other. However, following that period, we see that by 2013, by any measure, global welfare had again increased. Assessment of the actual change in welfare in quantitative terms is also possible but requires the choice of a specific welfare function (or class of functions).
While the GLC provides a framework for welfare comparisons, using a growth incidence curve provides a more detailed depiction of the beneficiaries of growth across this period. Growth has been broadly ‘inclusive’ in the limited sense that it has taken place across percentiles of the world distribution, but it has been rather uneven across the different percentiles, and the temporal pattern of increases has also varied across percentiles, as shown in Figures 3.13 and 3.14.
Between 1960 and 2010, the poorest experienced a greater share of their cumulative growth, in particular of consumption, early in the period. However, we observe an interesting hump-shape (which was first described as an elephant shape by Lakner and Milanovic (2015), arising during the interval of the greatest dynamism (2000 to 2013 and 1990 to 2013)15. The middle-income groups – those between the 40th and 60th percentiles - saw their incomes rise rapidly in this period, while those in a rather higher income bracket (80th -95th percentiles) saw their incomes grow much more slowly, mainly because the richer countries in which they disproportionately lived experienced lower growth. Of course, problems of estimation of top-incomes in household surveys provide an essential qualification to any such conclusion. Lakner and Milanovic (2015) plot GICs for the period 1988-2008 and show a larger uptick in growth after the 95th global percentile than observed in the period 1990-2013 in our graphs. For the same period 1988-2008 we also see higher growth above the 95th percentile using GCIP data, but this effect diminishes if the period is extended to 2013. GICs for the longer time frame of 1960-2013 are far flatter than for the more recent period and also consumption GICs show a far more pronounced pattern than the income GICs.
15 The literature that has been spawned by the so-called elephant graph is already sizable. For a critique and response see Corlett (2016) and Lakner and Milanovic (2016)
76 Figure 3.12: Global Generalized Lorenz Curve (2005 PPP)
The importance of the Chinese experience in this transformation can be seen to be critical since the Chinese population makes up a disproportionate share of those in the fortieth to seventieth percentile range in 2013. Another way to arrive at this conclusion is to look at the relative
77
Figure 3.13: Global Consumption Growth Incidence Curve (2005 PPP)
Figure 3.14: Global Income Growth Incidence Curve (2005 PPP)
78
position of the populations of several large countries over time. This is done in Figures 3.15 and 3.16. In 1990, all of the US population enjoyed incomes that would place them in the top quintile of the world income distribution. The Chinese percentile distribution was in the group of bottom countries depicted. By 2013, however, the Chinese percentile distribution dominated those of several of the other countries and had caught up with Brazil. China is now truly the ‘Middle Kingdom’ of the world as most of its population lie between the fortieth and seventieth percentile of the world distribution. Nigeria is now dominated in relative terms by the other percentile distributions depicted here (although it was not previously). The Chinese relative position looks better when we plot income distributions but is slightly weaker when we plot the consumption distributions of different countries due to the relatively high savings rate of Chinese population. In income distributions,
China and Brazil overlap, but China is substantially lower than Brazil when one plots consumption distribution. Also, even though GDP per capita for Brazil in 2013 was about a third higher than China the survey means are almost the same. This is due to differences in the disparity between national accounts and surveys in the two countries. In
Figure 3.15: Relative Position of Select Countries in 1990 based on Income (2005 PPP)
79
China national accounts and survey means track each other closely whereas they don’t to the same extent in Brazil. These differences in bilateral comparisons due to the measures chosen shows that the broader picture too might be dependent on the particular indicator used for comparisons and that systematic exploration of such dependence is necessary to arrive with greater confidence at overall results.
Figure 3.16: Relative Position of Countries in 2013 based on Income (2005 PPP)