Professional Documents
Culture Documents
Inequality and Globalization: A Review Essay: Martin Ravallion
Inequality and Globalization: A Review Essay: Martin Ravallion
https://doi.org/10.1257/jel.20171419
620
Ravallion: Inequality and Globalization: A Review Essay 621
countries) and microeconomics (why some cesses, with today’s rich world taking off
households are better able to participate in economically from the early nineteenth cen-
that growth), as well as from history and the tury (though with some late starters such as
history of thought. Both knowledgeably span Japan). Average inequality within countries
the world, from poor to rich countries—as was stagnant or even falling over much of
is appropriate for books on global inequal- this period, most notably over the middle
ity. Bourguignon draws more completely on half of the twentieth century—known as the
the theories and evidence found in modern Great Leveling in the rich world.
economic writings on the subject. Milanovic This pattern changed dramatically toward
takes a somewhat more idiosyncratic the end of the twentieth century, with an
approach, grounded in his own interpreta- overall pattern of falling inequality between
tions of the evidence he assembles. (I expect countries alongside rising average inequal-
that Milanovic’s style sells more copies.) They ity within countries. This new pattern in the
broadly agree on the evidence. Bourguignon evolution of global inequality is the main
probes more deeply in trying to understand focus of both books. Figure 1 shows the
that evidence as an economist. Both authors series of global inequality measures provided
sprinkle their text with their own opinions, by Bourguignon, using a Theil index.2 We
though Bourguignon provides a clearer idea see the fall in global inequality, markedly so
of what they are based on. in the new millennium. This has been driven
This essay begins with an overview of what by a decline in inequality between countries,
these books tell us about the trends in global which accounts for the bulk of total inequal-
inequality. It then critically examines what ity. Average inequality within countries
they say about the causative factors and pol- (
population-weighted) has edged upwards
icy responses. Finally, comments are offered since 2000.3
on some broader concerns, applicable to There are numerous data issues under-
much of the literature on global inequality. lying figure 1, related to household sur-
veys, price indices, census data, and the
role played by national accounts. Neither
2. The Evidence on Global Inequality
book goes into much detail on these issues,
The historical patterns identified in though chapter 1 of each book provides a
both books can be summarized as follows. brief summary of how their estimates for
Looking back over 200 years, one finds that
global inequality—defined as the relative
inequality of incomes among all peoples of
2 The country data underlying these aggregate statis-
the world ignoring where they live—was on
tics come from national household surveys that included
a rising trend from 1820 to about 1990.1 This income or consumption. The data for developing countries
long period of rising inequality was driven, come mainly from the World Bank’s PovcalNet site. The
in the main, by the divergent growth pro- data for developed countries come from the OECD and
the Luxembourg Income Study. Anand and Segal (2008)
provide a good overview of the data and measurement
issues, and of the (often inconsistent) estimates of global
1 The series on inequality back to 1820 draws on an inequality in past literature, mostly prior to the work pre-
important paper by Bourguignon and Morrisson (2002), sented in the two books under review here. Other evi-
which combined Maddison’s (1995) estimates of national dence on the new pattern of changes in global inequality
income with historical distributional data. Naturally, the can be found in Nino-Zarazua, Roope, and Tarp (2014) and
reliability of all these data sources becomes more ques- Ravallion (2014, 2016b, chapter 5). All the sources I know
tionable the further back one goes. The Bourguignon and are consistent with the trends discussed here.
Morrisson series ended in 1992, prior to the recent change 3 For the Theil index, unlike the more popular Gini
in the trajectory for global inequality (discussed below). index, the two add up exactly to total inequality.
622 Journal of Economic Literature, Vol. LVI (June 2018)
1.0
Total global inequality
0.8
Inequality between countries
0.6
Theil index
0.4
0.2
0.0
88
90
92
94
96
98
00
02
04
06
08
10
12
19
19
19
19
19
19
20
20
20
20
20
20
20
Figure 1. Global Inequality and its Between- and Within-Country Components
global inequality were derived.4 Nor will about u nderreporting of incomes, especially
this essay focus on data issues. But I flag income from capital. Estimates using income
only one issue that should be kept in mind. tax records have indicated larger “high-end”
Along with both authors, I suspect that the incomes than found in surveys (Atkinson,
within-country component in figure 1 is
Piketty, and Saez 2011). It is likely that the
being underestimated. There are a number true level of within-country inequality is
of reasons. Selective compliance in surveys higher than currently measured. It is less
is a concern almost everywhere; in particu- clear how much these measurement errors
lar, it is plausible that the rich are less likely matter to the trend, but my own expectation
to participate in household surveys. The bias is that inequality within countries is rising
could well be large: Korinek, Mistiaen, and more than the data in figure 1 suggest, on
Ravallion (2006) estimate that correcting the presumption that many newly affluent
for such selective compliance adds about respondents are reticent to fully reveal their
five percentage points to the Gini index for gains or even to participate in surveys.
the United States. There are also concerns The aggregate summary statistics in fig-
ure 1 do not reveal much about how income
4 Fuller discussions of these issues can be found in
changes were distributed across the pop-
Anand and Segal (2008) and Ravallion (2016b, chapters ulation. Milanovic starts out with a more
3–5). informative tool for describing the evolution
Ravallion: Inequality and Globalization: A Review Essay 623
80
70
Real income change 1988–2008 (in percent)
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
of income distribution in the world, using a the GIC in figure 2 is explained in Lakner
graph from Lakner and Milanovic (2016a) and Milanovic (2016a).
that plots the proportionate gain in income The Lakner–Milanovic graph has been
over 1988–2008 against fractiles of the dubbed the “elephant chart,” since it traces
income distribution, as reproduced in fig- the shape of an elephant’s head with its trunk
ure 2. This is a version of a “growth incidence held high. Readers who are used to hear-
curve” (GIC), as defined in formal terms for ing about rising inequality in the rich world
continuous distributions by Ravallion and will see that feature in the graph; strikingly,
Chen (2003), who discuss the curve’s prop- between the eightieth percentile (from the
erties.5 The methodology used to construct bottom) and the top 1 percent globally we
see a steeply positive curve (the elephant’s
5 Note that the version of the GIC in Lakner and raised trunk), rising from near zero growth to
Milanovic (2016a) gives growth rates for ventiles (with over a 60 percent gain for the top percentile.
the top 1 percent separated out) rather than percentiles. But readers also see something as striking—
This smooths their curve. The percentile version can be
found in Corlett (2016). This shows negative growth rates the marked proportionate rise in incomes for
among the poorest and in a neighborhood of the eighti-
eth percentile. These have been averaged out in the
Lakner and Milanovic version, as also used in Milanovic not held fixed. This is consistent with the fact that the
(2016). The negative values at the bottom probably reflect “
quasi-non-anonymous” GIC in Lakner and Milanovic
compositional effects, given that the set of countries is (2016, figure 5) does not show any negative growth rates.
624 Journal of Economic Literature, Vol. LVI (June 2018)
100
1988
2008
90
80
Cumulative share of global income (percent)
70
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
those near the middle of the global distribu- this came with a marked inward shift of the
tion (the elephant’s massive and expanding Lorenz curve between the thirtieth and
head). This came with considerably slower eightieth percentiles, but an outward shift
growth for the poorest. among the top decile and a declining share
As Lakner and Milanovic (2016a) note, for the poorest 5 percent. The L akner–
the Lorenz curves intersect internally, as can Milanovic estimates imply that the share of
be seen in figure 3.While the overall Gini the world’s top 1 percent rose from 12 per-
index fell (from 72 percent to 71 percent), cent to 15 percent between 1988 and 2008.
Ravallion: Inequality and Globalization: A Review Essay 625
So rather than suggesting a decline in good news from the point of view of abso-
global inequality, this is actually quite an lute poverty. For example, Ravallion and
ambiguous picture of distributional change. Chen (2013) find that the proportion of the
Given that there is not Lorenz dominance, world’s population living in absolute poverty
the claim that global inequality is falling is fell from 36 percent in 1990 to 19 percent in
not robust to the choice of index; some valid 2008. Over this period, Chen and Ravallion
inequality measures (such as the Gini index (2013) find fi
rst-order dominance up to the
and the Theil index, as in figure 1) can show US poverty line, implying declining poverty
a decrease while other equally valid mea- measures for the developing world over all
sures do not.6 With sufficiently strong aver- reasonable poverty lines and a broad class
sion to inequality, one would declare that of poverty measures.9 However, there has
global inequality has in fact risen. Consider, been much less progress in reducing relative
for example, the Atkinson (1970) index, poverty, as judged by poverty lines typical of
which has a parameter ε reflecting the aver- each country given its average income (Chen
sion to inequality; a higher value of ε implies and Ravallion 2013). And there has been
that one is willing to incur a greater loss much less progress for the world’s poorest,
when transferring money from the rich to who can reasonably be said to have been “left
the poor (i.e., a lower share actually reach- behind” (Ravallion 2016a).
ing the poor) and yet still judge that social
welfare has increased.7 I calculate that the
3. Interpreting the Evidence
Atkinson index of global inequality has fallen
over 1988–2008 for ε ≤ 4, but that inequality Two aspects of the evolution of global
has risen for ε = 5or 6.8 The upshot of these inequality (as summarized above) attract the
observations is that with sufficiently strong bulk of the attention in these books: rising
aversion to inequality, one will judge that inequality within some countries of the rich
global inequality has risen over this period. world, and the falling between-country com-
Nonetheless, the fact that growth is positive ponent of global inequality. The books differ
for such a large segment of the population is in how they go about explaining these fea-
tures. Bourguignon’s approach is grounded
6 This statement applies a w
in neoclassical economics, though with
ell-known result from
Atkinson (1970); the key property for a valid measure plenty of r eal-world features such as market
is the P igou–Dalton transfer principle. Lakner and failures and poverty traps. This allows him to
Milanovic (2016a) note that the Lorenz curves for 1988 provide a quite comprehensive accounting of
and 2008 intersect, but they do not note the implication
for n
on-robustness of the ordinal inequality comparison. the forces likely to be at work, though often
7 The Atkinson index can be written as pointing to the uncertainties and ambiguities
bold, all of which allows him to put forward (p. 10).11 To his eyes, the graph shows how
some intriguing hypotheses about why we the rich world’s lower-middle class has seen
have seen these patterns; we do not always little or no gain from globalization—these
come away confident that he is right, but his are the people (he claims) living around the
views are definitely interesting. eightieth percentile of figure 2, with near zero
gain over the period.12 By contrast, the poor
3.1 Globalization
and middle class of the developing world have
In this context, “globalization” is primarily seen substantial gains. The largest percent-
about greater economic integration across age gain in the elephant graph is close to the
countries, which mainly means greater open- global median.13 In Milanovic’s interpretation,
ness to external trade and greater mobility of the emerging middle class in the developing
financial capital.10 In both books, the pres- world have been the big gainers from global-
ent period of globalization is essentially seen ization, while the losers were the (relatively)
as the joint cause of both falling inequal- poor and middle class within the rich world.
ity between countries and rising inequal- The elephant chart attracted consider-
ity within countries. The two (opposing) able popular attention in traditional and
effects are thus linked. So far, at least, the social media.14 It has been seen by some as
inequality-decreasing, between-country, an explanation for both the Brexit vote in the
effect of globalization has dominated. Much United Kingdom and the outcome of the US
of the rest of this article will try to assess the presidential election in 2016. One observer
case for agreeing with this thesis on the link described it as “the most powerful chart of
between globalization and inequality. the last decade.”15 The link with globalization
In arguing that global economic integra- has seemed obvious, though Kawa (2016)
tion has been the major force in the evo- drove the point home by calling figure 2 the
lution of inequality between and within
countries, both authors are in agreement 11 In response to the criticism by Corlett (2016), Lakner
with much popular opinion today, though and Milanovic (2016b, p. 3) say that they do not subscribe
one hears very different views on whether to the “monocausal explanation” of their elephant graph as
it is a good thing or not. R ich-world critics being due to globalization. Lakner and Milanovic (2016a)
actually say very little about globalization. However, that is
of globalization claim that it has destroyed not true of Milanovic (2016), in which the headlines and
jobs at home and led to stagnant or falling the bulk of the interpretation of the elephant graph is very
living standards for all except the wealthy, close to monocausal.
12 With reference to those near the eightieth percen-
who have the (financial and human) capital tile, Milanovic (2016, p. 19) claims that “almost all (are)
to benefit. Supporters point to, among other from the rich economies of the OECD.” In his book talk
things, the gains to the developing world’s he goes further to label the low point around the eightieth
percentile the “US lower middle class.” (See https://piie.
poor, including from the jobs created. com/system/files/documents/milanovic20160509ppt.pdf.)
The elephant graph suggests that there 13 This illustrates the limitations of focusing on the
might be truth on both sides. Milanovic median as the key distributional statistic, as advocated by
Birdsall and Meyer (2015). The growth rate of the global
introduces this graph at the beginning of his median is hardly representative of the results in figure 2; it
chapter 1 to illustrate that “The gains from is especially deceptive about how incomes of the poorest
globalization are not evenly distributed” have evolved. The literature on poverty measurement has
taken a different approach in measuring economic growth,
accounting for distribution (Ravallion and Chen 2003).
14 See, for example, this article in The Economist
10 The word “globalization” has taken on some other magazine: https://www.economist.com/news/finance-
means, including (for example) cultural globalization, such and-economics/21707219-charting-globalisations-
as the global diffusion of American entertainment and discontents-shooting-elephant
cuisine. 15 This was a tweet by Toby Nangle in April 2016.
Ravallion: Inequality and Globalization: A Review Essay 627
16 Lakner and Milanovic (2016b) reply to Corlett 3.2 Inequality between Countries
(2016), though they do not take issue with the claims by
Corlett referred to here. Also see the comments in Freund
(2016), who also questions Milanovic’s interpretation of the Both authors point to the importance of
elephant graph, drawing in large part on Corlett’s work. economic growth in initially poor countries
628 Journal of Economic Literature, Vol. LVI (June 2018)
to the pattern of falling inequality between lished studies, but not always, depending on
countries. The higher growth rates seen in (among other things) what other predictors
the developing world have clearly been an are included.19 Of course, assessments using
important driving force in the changes we cross-country regressions are not typically
see in figures 1–3. China was, of course, a population weighted, while that is the case
major contributor to the size of the ele- for the w ithin-country component of global
phant’s head in the Lakner–Milanovic graph inequality (such as plotted in figure 1).
(figure 2).17 But since 2000, all regions of External trade has clearly played an import-
the developing world have seen most of ant role in China’s growth, though here too
their economies growing at appreciably there are other factors in play, including
higher rates. Bourguignon’s book will be many domestic policy reforms in the transi-
more helpful to readers keen to learn how tion to a market economy such as the promo-
this has been achieved and how it might tion of internal trade.
continue, including on the role for external
3.3 Inequality within Countries
aid, trade restrictions by rich countries, and
liberalization within poor countries. Looking When we see similar things happening
ahead, Bourguignon sees good prospects for across multiple countries, it is very tempting
the middle-income countries but sees more to look for a global answer. But similar things
uncertainty facing low-income countries, are not happening everywhere, as both
now concentrated in sub-Saharan Africa, authors know. There is heterogeneity across
given their dependence on primary com- countries and over time in the changes in
modity exports. the aggregate statistics for within-country
My reading of the literature on the empir- inequality in figure 1. Inequality has been
ical determinants of economic growth at rising in a majority of countries in the rich
country level does not give me confidence world, but not everywhere. Bourguignon
that trade openness has been as an import- points to the example of France, which
ant driving force as the authors suggest. A has avoided the marked upward trend in
reasonable summary of the evidence would inequality seen elsewhere since the 1980s.
probably be that trade has helped promote France is not alone among the OECD coun-
growth and poverty reduction in the devel- tries; depending on the time period (and
oping world as a whole, but that is only that can matter a lot), one also finds falling
one of a number of relevant factors, which inequality in (for example) Belgium, Greece,
include aspects of the initial distribution of Hungary, and Spain.20 There appears to
income and human development.18 “Trade have been even more heterogeneity within
openness” (usually measured by exports the developing world. Indeed, during peri-
plus imports as a share of GDP) is often a ods of economic growth we have seen fall-
significant predictor of growth rates in pub- ing inequality within countries about as
17 See Rosnick (2016), who removes China from the 19 Trade openness does not emerge as one of the
GIC of Lakner and Milanovic (2016a); the elephant is still robust predictors of growth in the m
eta-study of growth
there, but with some 30 percentage points knocked off its regressions by
Sala-i-Martin, Doppelhofer, and Miller
head! (2004).
18 Contributions include Dollar and Kraay (2004), 20 Evidence on the changes over time in inequality in
Lundberg and Squire (2003) and Berg, Ostry, and the OECD countries can be found in OECD (2011). Also
Zettelmeyer (2012). For overviews of the arguments and see Morelli, Smeeding, and Thompson (2014) and Freund
evidence, focusing on developing countries, see Ravallion (2016). Note that inequality comparisons over time can
(2006, 2016a, chapter 9). depend on the measures used and specific time periods.
Ravallion: Inequality and Globalization: A Review Essay 629
often as we have seen rising inequality assumes that growing capitalist economies
(Ravallion 2014). Granted, the developing tend to (more or less) automatically see ris-
countries with a trend increase in inequal- ing inequality (echoing Piketty 2014). Once
ity over the last twenty years or so include the Industrial Revolution had delivered sus-
the two most populous, China and India, tained growth, inequality rose steadily, reach-
which are clearly putting upward pressure ing very high levels by the early twentieth
on the (population-weighted) within-country century. The wave broke with the First World
component of global inequality, such as War, which is seen by Milanovic as the out-
in figure 1—as well as bringing down the come of a struggle among capitalists for mar-
between-country component. But any idea kets, rather than some exogenous “inequality
of a common global force of economic inte- shock.” By the early twentieth century, citi-
gration driving up inequality everywhere can zens were demanding action against high
be readily dismissed. Inequality appears to inequality and their governments were ready
fall in some countries when they are opened to take action, and that happened over much
to trade and increases in others. And there of the twentieth century. When inequality
are clearly many other forces in play. There gets very high, there is pressure on govern-
is more to the story. ments to take actions to lower it, but not
One possible place to look for clues is the when it is low. Thus, to Milanovic’s eyes, we
literature on growth and inequality in devel- see waves in which inequality rises then falls.
oping countries. Here much attention has By implication, the present period of rising
been given to the “inverted U” hypothesis of inequality in many rich countries will come to
Kuznets (1955), which has long been influ- an end at some time. Milanovic points to fac-
ential in thinking about development policy. tors that may push the wave to break, though
Kuznets argued that inequality would first one does not get the impression he thinks it
increase within poor countries as their econ- will happen anytime soon.
omies grew through urbanization, but after The Kuznets wave idea is more an inter-
some point inequality would fall.21 I continue pretation of history than an economic model.
to be surprised at how much attention the Nor is it the only way one might interpret the
Kuznets hypothesis still gets as it has found historical data. Rather than being the trough
rather little support in subsequent empirical of a repeated wave, maybe the Great Leveling
work; over time, very few developing countries was a unique period historically—a large,
have followed the predictions of the Kuznets sustained, but unusual, “inequality shock”
hypothesis, as shown by Bruno, Ravallion, and in a rising upward trajectory of inequal-
Squire (1998) and Fields (2001). ity under capitalism, as argued by Piketty
Milanovic introduces the idea of what (2014). Of course, given that a standard
he calls “Kuznets waves.” This is a bold yet inequality index is bounded above and below,
simple idea that allows him to provide, in a pattern of fluctuations around a stationary
chapter 2, a short coherent economic his- value is a plausible characterization of the
tory of the long-run evolution of inequality dynamics; inequality cannot either increase
within countries. In a nutshell, Milanovic (or decrease) indefinitely. The real challenge
must then be in explaining why it changes its
direction when it does, what long-run value
21 Kuznets did not provide a formal theoretical argu- it takes, and why the country-specific waves
ment but this was provided in subsequent literature, appear to be synchronized across many coun-
notably Anand and Kanbur (1993), which identifies the
assumptions required for the Kuznets inverted-U in a dual- tries. Readers will not find much that can
istic developing economy. help explain any of this in Milanovic’s book.
630 Journal of Economic Literature, Vol. LVI (June 2018)
Whether capitalism is inequality increasing While trade openness has often (though
or not depends on initial conditions (including not always) been a significant predictor of
the distribution of endowments and how growth in the published studies, the picture
competitive markets are) and policies. At is less clear for inequality. The cross-country
any one time, we see income inequality evidence to date makes it hard to general-
increasing in some countries and decreas- ize. Dollar and Kraay (2004) find that an
ing in others. One clue to the patterns in the expanding volume of trade neither increases
data is found in neoclassical growth theory, nor decreases inequality on average, although
which implies inequality convergence—that Lundberg and Squire (2003) find a small
inequality measures tend to fall when it is inequality-increasing effect of trade expan-
high and rise when it is low (Bénabou 1996; sion. The OECD (2011) argues that neither
Ravallion 2003). (Essentially all moments of trade openness nor financial integration have
the distribution, when they exist, should con- been important factors empirically in explain-
verge under the neoclassical growth process, ing rising inequality. There is also evidence
not just the mean.) The signs we have seen that trade liberalizations (in Latin America)
over the last thirty years or so of inequality have been inequality decreasing (Ferreira,
convergence could well stem from the same Leite, and Wai-Poi 2007). Much popular
economic forces that have generated mean attention in the United States has been given
convergence. Distinguishing inequality con- to the implications for American workers of
vergence from waves in the s teady-state val- import competition with China (in particular),
ues is likely to be difficult empirically. and there is indeed evidence of significant
Inequality convergence can also be labor-market adjustment costs that have been
explained by how economic policy con- borne especially by low-wage workers (Autor
vergence in the world has interacted with et al. 2014). Growing industrial interdepen-
pre-reform differences in the extent of
dence across countries due to the offshoring
inequality. To see why, suppose that reform- of production appears to have come at a cost
ing developing countries fall into two cate- to workers in Europe and the United States,
gories. First, there are those countries/dates especially those who perform more routine
for which p re-reform controls on the econ- tasks, who tend to be less well educated.
omy benefited the rich, keeping inequality However, here too, the effects that have been
artificially high. Arguably this was the case identified empirically appear to be quite mod-
in much of Latin America up to the 1980s, est (Parteka and Wolszczak-Derlacz 2016).
but we have seen falling inequality in that There are often subtle and ambiguous
region since the 1990s.22 Second, there are ways in which each of the multiple factors
countries/periods in which the controls had involved, including trade openness, impact
the opposite effect, keeping inequality low; inequality. A cross-country regression coef-
arguably this was the case in China (prior ficient averages out many things that mat-
to the 1980s), and Vietnam and the former ter, sometimes going in opposite directions.
Soviet Union (prior to the 1990s). Then lib- Among equally poor people, for example,
eralizing economic policy reforms can entail some gain and some lose from trade open-
sizable redistribution between the poor and ness, given their heterogeneity in net trad-
the rich, but in opposite directions in the two ing positions in relevant markets. Thus,
groups of countries. while the regression coefficients may be
nearly zero, it would be deceptive to con-
22 For recent evidence on the evolution of inequality in clude that trade does not matter. Similarly,
this region, see Rodríguez-Castelán et al. (2016). while in a unified labor market, labor-market
Ravallion: Inequality and Globalization: A Review Essay 631
in-work poverty. Minimum-wage rates have Greater public awareness in rich countries of
naturally played less of a role in poor coun- the overall net benefits of liberalizing migra-
tries, where a high degree of informality tion (including benefits in those countries)
limits enforcement. There are exceptions would no doubt also help.
such as Brazil, where rising minimum-wage Bourguignon gives somewhat more atten-
rates appear to have played a role in reduc- tion than Milanovic to redistributive policies
ing income inequality (Brito, Foguel, and using taxes and transfers. These policies have
Kerstenetzky 2016). India’s National Rural come to play a significant redistributive role
Employment Guarantee Scheme (NREGS) in rich countries, and have helped some-
can be interpreted as a policy instrument what in mitigating the rise in the inequal-
for enforcing a minimum-wage rate in a set- ity of primary incomes (as discussed by
ting in which compliance with the statutory Bourguignon). Numerous m icrosimulation
minimum is weak, although how well this is and decomposition exercises have suggested
achieved in practice can be questioned based that the tax-benefit systems found in many
on the evidence on program performance rich countries have helped in attenuating
(Dutta et al. 2014). high inequality.26 The Great Leveling in the
So far, globalization has brought more inte- twentieth century stemmed in part from such
gration in the markets for goods and services, policies (though communism also played a
and in capital markets, than for labor mar- role in reducing inequality, notably within
kets. Huge intercountry differences in mar- the Soviet block and China). The differences
ginal products of labor remain. This suggests we see across countries in the inequality of
that greater integration of global labor mar- disposable incomes reflect in part the differ-
kets through migration would reduce global ences in the redistributive efforts of states.
inequality and poverty, and be growth pro- Such policies have been less prominent
moting globally. Both authors clearly favor historically in developing countries, but
fewer restrictions on international migration. this too has been changing in the new mil-
Critics of efforts to liberalize international lennium, with rapidly expanding coverage
migration argue that inequality-increasing of ostensibly redistributive policies and evi-
forces will emerge within the destination dence that some developing countries have
countries. This does not appear to be borne been successful against inequality using taxes
out clearly by the evidence, some of which and transfers.27 It appears to be that when
also points to longer-term gains in aver- the political will is sufficiently strong, and
age incomes of the destination countries, backed up by administrative capacity, any
notably through enhanced productivity.25 country can make significant inroads against
Nonetheless, these are average impacts. high inequality.
Here, too, there will probably be losers (and
gainers) at any given income level. Resistance
26 The EUROMOD
emerges among those bearing the costs, or microsimulation model of the
direct tax and transfer systems of European countries
in fear of doing so. Both authors are sensi- has been an important contribution; for an overview, see
tive to the likelihood of political resistance, Figari, Paulus, and Sutherland (2015). For an example of
and modify their policy proposals accord- the use of such models in studying the impacts on inequal-
ity of policy changes using decomposition methods see
ingly; for example, Milanovic makes a plea Bargain and Callan (2010).
for greater use of temporary work permits. 27 On the surge in the use of direct interventions in
developing countries and their heterogeneity in reaching
poor people, see Ravallion (2016b, chapter 10). On the
25 Recent evidence on this point can be found in impacts of this class of fiscal policies on inequality in devel-
Jaumotte, Koloskova, and Saxena (2016). oping countries, also see Lustig (2016).
Ravallion: Inequality and Globalization: A Review Essay 633
While not a major theme for either author, omy.30 The informational requirements of
both note the potential for social policies to a basic income on its own are minimal, and
help protect workers and their families from the incentive effects (mainly reduced labor
down-side risks associated with globalization. supply through a positive income effect on
Many rich countries now have safety net poli- demand for leisure) and administrative costs
cies that act as more or less automatic income would probably be modest, although a full
stabilizers. (For example, as has long been assessment must also consider the method
recognized, progressive income taxes pro- of financing. (When financed by progressive
vide a degree of insurance in the presence of income taxes, the policy package is formally
income fluctuations.) However, while system- identical to the negative income tax, as advo-
atic evidence does not appear to be available, cated by Friedman 1962.) A revenue-neutral
the experience of many targeted transfer pro- switch from existing welfare programs to
grams in developing countries does not sug- a basic income could well have greater
gest they typically have the flexibility that is impact on poverty; see, for example, Murgai,
needed in adjusting to (positive or negative) Ravallion, and van de Walle (2016) on India’s
income changes at household level.28 NREGS. In 2017, Finland introduced a
The incentive effects of transfers to poor trial basic income for unemployed workers,
people have been a concern, as Bourguignon replacing the existing unemployment benefits
points out, but we need to be careful in and other allowances. Many people will be
properly assessing that concern. There can watching Finland’s basic-income experiment
be little doubt that finely targeted policies with interest.
that impose very high marginal tax rates Globalization has been seen as a threat to
(even close to 100 percent) on poor people social protection—the welfare state—includ-
create adverse incentives for work and sav- ing by both authors (especially Milanovic).
ing. However, the existing evidence across One concern is that the international mobil-
countries at all stages of development does ity of capital and of skilled workers erodes
not suggest that incentive effects are likely to the domestic tax base, leading to lower lev-
be a major concern in practice, to the point els of provision for protection in rich coun-
of outweighing the social welfare gains from tries—a “race to the bottom.” One response
greater equity, at least as long as very high has been to call for a universal (global) basic
marginal tax rates are avoided. income; see, for example, Van Parijs and
Neither book has much to say about the idea Vanderborght (2015). However, moving to
of a basic income—an unconditional transfer a new country will never be costless, even
payment at a common level for all persons, when impediments to integration have been
whether living in a poor household or not.29 To removed (which is still a long way off) and
many observers, this idea has a strong appeal there are country-specific non-income fac-
in allowing guaranteed protection from pov- tors relevant to welfare and (hence) migra-
erty, while retaining the economic advantages tion choices. And the ability to move freely
of an open and competitive market econ- across borders can also be an important
source of insurance.
Another concern one hears is that global integration. Globalization may well
(to paraphrase) “globalization drives up be getting too much credit, and being
inequality, which undermines the welfare blamed for too much.
state since the rich want to opt out.” This,
too, is simplistic. As I have said, it is not clear
4. Measures of Inequality and Their
that globalization is (or has to be) i nequality
Limitations
increasing. Nor is it clear that social pro-
tection is undermined by inequality; mod- Stepping back, at the core of the literature
ern welfare states emerged in today’s rich on global inequality as a whole (reflected
world when it was as unequal as today, or in these two books) one finds a measure-
even more so. Inequality can also gener- ment concept of a world without countries;
ate political pressures for redistribution. all incomes are pooled as if one is measur-
The concerns one hears that globalization ing inequality in just one country. This is
undermines the scope for social protection essentially the same way global poverty is
are surely exaggerated. If the political will measured (as in Chen and Ravallion 2010).
is present, then effective and sustained pro- Of course, there is an important difference,
poor redistribution is possible, as a number in that global poverty (or some more inclu-
of countries have demonstrated.31 sive measure of aggregate social welfare
such as mean log income) depends on both
3.5 In Summary
the overall mean and the relative distribu-
The two key features of how global tion (“inequality” roughly speaking). We
inequality has been changing in the last few can debate what trade-off is to be allowed
decades are the falling between-country between the mean and inequality, but we
component alongside a rising w ithin-country clearly value both, implying that we cannot
component. While there can be little doubt decide the matter looking at only one. My
that trade openness and capital mobility impression is that both authors agree.32 But
have had distributional impacts, both verti- then their books must be judged as rather
cal and horizontal, the jury is still out on the partial in perspective, though less so for
thesis of both books (in keeping with widely Bourguignon. A partial perspective is possi-
held views today) that globalization has bly easier to excuse, given that so many of
been the main force jointly creating both the books written about economic growth
features. There has been considerable vari- have ignored inequality. But is that how we
ance across countries in both their growth should restore balance?
rates and the changes in inequality, and As noted already, there are both
trade openness does not seem to stand out between-country and w
ithin-country com-
as the major generalizable causative factor ponents to this concept of global inequality.
that these books, and many other observers, Both authors make much use of this decom-
assume. Technological change in unequal position; indeed, it defines the very structure
settings could well be a much stronger force of Milanovic’s book. While pedagogically
than expanding trade. Policies have mat- useful, it is a mechanical decomposition.
tered to both growing poor economies and
redressing inequality within countries. And
these policies can coexist with considerable 32 This is obvious for Bourguignon. Tellingly, Milanovic
(p. 5) writes that “wars led to declines in inequality but
also, unfortunately and more importantly, to declines in
31 See, for example, the analysis in Huber et al (2006) mean incomes” (though one might have expected him to
for Latin America. Also see the discussion in Green (2016). say loss of life).
Ravallion: Inequality and Globalization: A Review Essay 635
“Countries” only have salience as arbitrary and poverty measures. But it is not intui-
groupings of people. Remarkably, there is tively obvious nor widely held, and it does
no concept of nationality here. Yet, it is clear need a stronger defense. Without that, one
that people care a lot more about inequality worries that the measures being used in
within their country of citizenship or resi- this literature do not properly capture the
dence (or maybe even their neighborhood, or salience that the idea of nationality clearly
some other reference group) than g lobally.33 has to almost everyone.
Certain between-group inequalities have a The issue can be thought about in terms
salience that is belittled by these mechan- of individual welfare. Inequality and pov-
ical decompositions. Milanovic rightly erty measures can be thought of as summary
points to the dangers of focusing solely on statistics of a distribution of money metrics
between-group inequalities—“existential of welfare. The type of global inequality
inequalities” as he calls them (in chapter 5). measure found in this literature implicitly
But the point here is that the importance of characterizes individual welfare in a rather
inequality between countries, versus within, narrow way, as solely a function of individ-
need not be weighed properly in this prevail- ual consumption or income. The judgments
ing concept of global inequality. made about how much “inequality” exists are
There are strongly competing views today strangely divorced from plausible welfare
on the relevance of national borders to how judgments. Yet it is surely clear that there
we think about inequality and what we think are real welfare effects at the individual level
should be done about it. It would not be of social factors, such as relative deprivation
defensible to only point to the (obvious) fact and social exclusion, and also that there can
that the bulk of political discourse about be external welfare gains from living in a
inequality is at the national level as justifi- richer country. How should these factors fig-
cation for ignoring global inequality. One ure in our thinking about “global inequality?”
naturally looks for a deeper rationale for the Neither of these books offers guidance to the
weight given to national versus international answer. In the context of global poverty mea-
inequalities. The intellectual high ground of surement, it can be argued that a person is
moral philosophy provides support for both not poor globally if they are neither poor by
sides of the issue. For example, Rawls (1999) a common international standard nor poor
argues that rich countries have no moral obli- by the standard of the country in which they
gation to help poor countries as long as the live.34 Something similar is needed for global
latter are reasonably well governed. Other inequality.35 Ravallion (2017) proposes a
philosophers, such as Singer (2010), argue tractable approach to m easuring global
instead that national borders, or distance, are
not morally relevant to the case for helping 34 Atkinson and Bourguignon (2001) provided such a
disadvantaged people whom we can help. concept for global poverty measurement and a generalized
One might respond that, on normative version with a global empirical implementation has been
developed by Ravallion and Chen (2011).
grounds, one should care about everyone, 35 Brandolini and Carta (2016) propose one possible
no matter where they live—what Brandolini approach in which the social welfare function for citizens
and Carta (2016) dub the “cosmopolitan of a given country depends on the welfare levels of those in
other countries, but with a (nonnegative) weight that can
view.” This is essentially the ethical posi- be less than unity, thus embracing both the “cosmopolitan”
tion that justifies standard global inequality and “nationalist” views as limiting cases. Brandolini and
Carta also refer to an unpublished conference paper by
Bourguignon that allows the weight on residents of another
33 Kanbur (2006) makes this point on “between-group” country to depend (negatively) on the spatial distance from
inequality, such as that based on gender or ethnicity. the country of own residence.
636 Journal of Economic Literature, Vol. LVI (June 2018)
i nequality when it is recognized that national Even if relative inequality does not change,
income can matter (either positively or neg- given existing inequality, the absolute income
atively) to individual welfare at given own gains to the rich will obviously be greater
income. Following this approach, one can than the gains to the poor. Figure 4 gives the
get either falling or rising global inequality absolute gains corresponding to figure 2.38
measures since 1990, depending on the The elephant’s head has shrunk greatly rel-
assumptions one makes. ative to the trunk. Over this 20 year period,
The value judgments made in measuring the absolute gain in mean daily income of
inequality carry considerable weight for the the poorest 5 percent was 7 cents per per-
position one takes in the ongoing debates son, while for the richest 1 percent it was to
on globalization. An important distinction almost $70 (and the latter number could well
in this context is between relative and abso- be an under-estimate). In absolute terms, the
lute inequality (Ravallion 2004). Both books developing world’s middle class and (espe-
focus more on relative inequality—whereby cially) its poor must surely be judged to have
the measure is unchanged when all incomes gained rather little; it is only because they
are multiplied by a constant. As we have started off so poor that the elephant’s head
seen, some measures (such as the popular is so large in figure 2. While the relative Gini
Gini index) show an overall decline in global index fell slightly (from 0.722 to 0.705), the
relative inequality, although it was also noted absolute Gini index rose appreciably (from
that there are other measures embodying a 0.722 to 0.898).39
strong aversion to inequality that show an A specific way in which inequality is ris-
increase. ing in the developing world is the rising
However, relative inequality is not the only absolute gap between the poorest and the
valid concept that can be considered here. mean standard of living. As I have tried to
As both authors note (albeit briefly), there demonstrate elsewhere, there has been only
is another concept of inequality, namely modest growth in the lower bound of the dis-
“absolute inequality,” which depends instead tribution of permanent consumption in the
on the absolute differences.36 This is not a world, which is still barely above a survival
purely academic distinction. Perceptions on level (Ravallion 2016a). At the same time,
the ground that “inequality is rising” appear there has been considerable progress against
often to refer to the absolute concept. A absolute poverty, using standard measures.
number of experiments (all with university In other words, the nature of that progress
students, to my knowledge) have found that is that there are fewer people living near the
40–60 percent of participants (in the United floor, not that the floor has risen. Yet social
Kingdom, Israel, Germany, and the United policy discussions put emphasis on the need
States) think about inequality in absolute to raise the floor—for example, the desire to
rather than relative terms.37 “leave no one behind” is prominent in the
36 While a relative index is invariant to multiplying all at Georgetown University. Of the 400 students I have sur-
incomes by a constant, an absolute index is invariant to veyed at Georgetown, a litle over half think about inequal-
adding a constant. Early theoretical writing on inequal- ity in absolute terms.
ity measurement noted this distinction, including Dalton 38 Milanovic also gives the absolute gains by fractiles,
(1920) and Kolm (1976), and it was explored further the- though here I have presented it in a comparable form to
oretically by Blackorby and Donaldson (1980), but it has the elephant graph.
dropped off the radar screen in almost all applied work. 39 The absolute Gini index is simply the relative index
37 The literature has closely followed Amiel and Cowell not normalized by the mean. For the purpose of this cal-
(1999). Ravallion (2014) discusses this further and points culation, I have normalized the indices such that the two
to other evidence, including my own surveys of students are equal in 1988.
Ravallion: Inequality and Globalization: A Review Essay 637
70
Absolute real gain 1988–2008 ($/person/day)
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
Source: Author’s calculation using the estimates made by Lakner and Milanovic (2016a)
United Nations’ new development goals.40 In has long been recognized as a policy concern
the period from the middle of the n ineteenth in economics; for example, Pigou’s (1949,
century to the middle of the twentieth cen- pp. 14–15) “principle of equity” in taxation
tury, during which time today’s rich world demands horizontal equity on the grounds
virtually eliminated extreme absolute pov- that anything else created “… a sense of
erty, more progress appears to have been being unfairly treated … in itself an evil.” In
made in raising the consumption floor than the context of trade openness, there are many
we are seeing in the developing world today. sources of horizontal inequality. Geographic
The horizontal impacts of trade openness disparities in access to human and physical
are also notable, but hidden from view in the infrastructure (such as between urban and
types of inequality measures used in these rural areas) affect prospects for participating
books and the literature more broadly. There in the opportunities created by greater open-
can be gainers and losers at all levels of living, ness to external trade among ex ante equally
even when a standard measure of inequality well-off people. Differences between devel-
or poverty is unchanged. Horizontal inequity oping countries in the degree of inequality
in cultivatable land holding imply different
40 See http://www.un.org/sustainabledevelopment/ distributions of the gains from agricultural
sustainable-development-goals/. growth (as might stem from trade openness).
638 Journal of Economic Literature, Vol. LVI (June 2018)
5. Should We Care, and Why? they happen to have been born, this arguably
points to a concern about an appropriate
There is a fundamental question left concept of global poverty rather than global
begging: why should we care about global inequality. So, too, does a desire for greater
inequality, as defined in these two books, in global equity in opportunities; it is surely the
keeping with the literature? I understand lack of opportunities of the less advantaged
why most citizens of the world, no matter that is the real concern, not the inequality of
where they live, care about global poverty, as opportunities per se. We can certainly care
a natural implication of human empathy and that high inequality (of outcomes or oppor-
compassion for deprivations and lost oppor- tunities) in a country can impede economic
tunities experienced by others. But the case growth, poverty reduction, and human devel-
is less obvious for global inequality. Unlike opment, and foster social ills such as crime,
poverty, less of which is surely always bet- political paralysis, or excessive political influ-
ter, inequality can also be too low from the ence of a rich elite. (Both authors point to
point of view of our valued social objectives. such costs of high inequality.) But the con-
Bourguignon clearly recognizes this point, cerns here are about inequality within coun-
and makes some references to goals such as tries. And each of these arguments point
poverty reduction and human development. to other higher goals, not inequality per se.
However, neither book provides a convinc- There are other, more international, argu-
ing case for the intrinsic importance of global ments that point to external costs to rich
inequality as a concern that warrants action folk from global inequality (such as through
in its own right. pandemics) or point to a case for compensa-
“Inequality” is a big word, and in my expe- tion for actions (or inactions) by rich coun-
rience the idea rarely gets serious attention tries that impose costs on poor ones (such as
from policy makers until one unpacks it, to global warming, trade restrictions, or implicit
identify the specific types of inequality that support for money laundering). But we can
matter. Unequal opportunities in life—as probe each of these arguments and find that
one aspect of “inequality”—are objection- the real concern is not inequality, which is
able to almost anyone if they abstract from instrumentally relevant and potentially very
their own position. So, too, are inequalities damaging at high levels, but not intrinsically
that reflect material economic deprivations so. Recognizing this point helps clarify the
for poor people. By the same token, if rising arguments for public action.
inequality brings sufficient benefits to the Both books point to how the new growth
least advantaged, then it may be considered trajectories seen in poor countries since 2000
justified.44 have helped bring down global inequality
While a case for supporting greater global despite the upward pressure coming from
equity might be grounded on empathy for within many countries. But surely the truly
the plight of those less fortunate, wherever good news is that these growth processes
have also come with massive absolute pov-
erty reduction. Whether that progress can be
found in Lambert, Ravallion, and van de Walle (2010).
44 In moral philosophy, this idea was famously cap-
maintained is another matter.
tured in a strong form by the “difference principle” of
Rawls (1971). Prioritarianism allows a conceptually simi- References
lar result—that social welfare can be enhanced by gains to
the poor even if overall inequality rises—without requiring Amiel, Y., and F. A. Cowell. 1999. Thinking about
that we only care about the least advantaged person; see, Inequality: Personal Judgment and Income Dis-
for example, Arneson (2000). tributions. Cambridge and New York: Cambridge
640 Journal of Economic Literature, Vol. LVI (June 2018)