1. World inequalities

2. Introduction Rawls (1970) and Buchanan (1985) veil of ignorance
Milanovic (2015) → only known fact is country of residence → How much of this person’s income will be determined by this factor, unrelated to individual effort or luck?
Assignment to country is fate for approximately 97% of the people in the world
Differences between mean country incomes are large →more than 2/3 of global inequality between individuals is due to national income → important for one’s life chances

3. Research question
Allocation of individual to a country implies 2 PG unalterable by individual effort → Roemer (1998) «circumstances»
average income of the country
inequality of income distribution
Income y of ith individual in jth country can be written as a function of:
country-specific circumstances 𝛼 𝑖𝑗 (e.g., average income of the country or its level of inequality);
individual specific circumstances 𝛾 𝑖𝑗 (e.g., parental income, gender, or race) whose effect also depends on country (hence subscripted by j);
Individual life effort Eij
a random shock (luck) uij
How much of yij can be explained by two characteristics of set 𝛼
mean income of country
Gini coefficient of country

4. Importance of research question
Innovation is studying global, as opposed to national, inequality of opportunities
At the global level «opportunities» are easier to proxy/define than at the national level
Circumstances offered by country of origin
Does effort pays off at the global level?
In a given country if one’s income is entirely determined by one’s parents’ income → unjust and no economic rationale for working
If income depend entirely on country of birth (→inequality within each country is 0) → individual has no incentive to work because she alone cannot influence her country’s growth rate → migration is only resource
Two questions raised:
Ethical question → Is it fair that income should be decided at birth?
Economic question → where should efforts of people in poor countries be directed: to work or to migrate?

5. World income distribution (Mladenovic, 2015)
WYD includes microdata from representative national household surveys
Last year is 2008 →includes 118 countries’ household surveys → 94% of the world’s population and 96% of world dollar income
Division in 100 income classes of equal size → percentiles
Each percentile=1% of total population
Allows making cross country comparisons → 23° percentile of NIG=75° percentile of CHN
Within each percentile Gini is 0.01 or 0.02 → within a percentile income differences across individuals are negligeable
Only in top 1% percentile Gini is in double digits (in some countries)

6. Income levels in the world by country and income class

7. Explanation of the diagram
Germany → rich country with moderate income inequality
Germans in country’s poorest percentile have per capita disposable income of about PPP $2,200 per year (horizontal broken line) → equal to 94° percentile of India
Richest German percentile has an income per capita of about PPP $104,000 → Ratio between the richest and the poorest percentile is less than 50:1
In China ratio between the top and bottom percentile is 66:1
Poorest percentile in China has annual per capita income just under PPP $300; richest percentile of almost PPP $20,000 (probably underestimate)
Only 40% of the Chinese population is richer than the poorest Germans
Brazil has a very unequal income distribution → covers almost the entire global spectrum (vertical axis)
Poorest people in Brazil at less than PPP $300; richest percentile at PPP $60,000 → ratio 200:1

8. Predicting income from country of residence
Income level of people belonging to percentile i in country j can be expressed as:
yij → annual average household per capita income in $PPP
mj → country’s GDP per capita in PPP terms
Gj → inequality in income distribution obtained from household surveys and measured by the Gini coefficient
Both variables on RHS are strictly exogenous to individual effort
Question: how much of global income variability can be accounted for by such an extremely parsimonious formulation
Two approaches about country’s population size:
IV («Individual Vision») → does not matter toindividual if his concern is how he would have fared were he born in a differentcountry
WAII («World As It Is»)→ population size does matter, given the actual structure of the economy and society

9. How income depends on circumstances

10. Illustration of results: IV approach
Dependent variable: log of household per capita income in $PPP for each country
Unweighted by country population
3 specifications
Mean country income
Education
Country dummies
With country income (regression 1) → elasticity of own income with respect to country’s GDP per capita is → “location premium”
Gini coefficient with negative sign → living in a more unequal country on average reduce one’s income → a 1 Gini point increase associated with a 1.5% decrease in own income → higher inequality numerically benefits fewer people than it harms
These 2 circumstances explain 2/3 of the variability of individual percentile incomes across the world → R2=0,66
With education → average number of years of schooling (regression 2) → increase of a country’s average educational level by one additional year of schooling associated with an increase of individual incomes of more than 30%
Explanatory power drops to 50% → R2=0,48
With country dummies (regression 3) → explain almost 3/4 of variability of individual percentile incomes across the world (R2=0,73)
Omitted country dummy is DR Congo → poorest country in the sample → coefficients on individual country dummies show locational premium that a person on average obtains by being a resident of a country other that DR Congo
U.S. locational premium is 355%, Sweden’s is 329%, Brazil’s is 164%, but Yemen’s is only 32%

11. Illustration of results: WAII approach
Regressions weighted by population show importance of circumstances as actually experienced by the people in the world → more populous countries matter more in estimates
Basically same results as IV (unweighted) estimates (regressions 4-6)
Elasticity of own income with respect to GDP per capita close to 1
Greater inequality (controlled for income) reduces more people’s incomes than it raises them
With education → location premium is higher than IV scenario → more populous countries seem to benefit more from a given increase in the average educational level
With country dummies the location premia are unchanged

12. Is the importance of country of residence for one’s income increasing over time?
To answer this question consider the “standard inequality of opportunity index” → people in the world differ only by country of residence
Within each country they are averaged out
→ Global inequality of opportunity equals the between country component of an inequality statistic (Gini, Theil 0)
Comparable data on global interpersonal inequality for 6 benchmark years between 1988 and 2008 → calculate the between-country component for each year
Population covered by the household surveys accounts for more than 90% of the world’s population (in all years except 1988)
All incomes are expressed in international PPP $.

13. Between country component of global interpersonal inequality over time

14. Illustration of results
Both Gini and Theil 0 indexes display more or less steady decline except for the period
Theil 0 almost 20% over the entire period, while Gini only about 4% →Gini more dependent on the mode of the distribution
Between-country component as a share of global interpersonal inequality has also gone down over the same period, from accounting for 81% of the total to 70%.
Global inequality of opportunity due to place of residence is huge but decreasing → decrease driven by rapid growth of relatively poor and populous countries, (China and India)
Whatever scenario or specification → at least 50% of the variability in real ($PPP) personal percentile incomes in the world can be attributed to two circumstances beyond individual control
The part that remains for effort and ‘‘episodic luck’’ is, within the worldwide context, relatively limited
True in 2008 despite a steady erosion of the importance of between-country components in global interpersonal inequality

15. Locational premium across income classes
So far location premium assumed to be equal across all percentiles of income distribution
→ the same for a given country regardless of person’s place in his country’s income distribution
Locational premium need not be uniform across the entire distribution
Look at income ventiles (each ventile contains 5% of population, ranked from the poorest to the richest)
Estimate same model as before with the person’s own income ventile held constant
One separate regression per ventile → regress ventile income on country’s GDP per capita and Gini coefficient
Only IV estimates, given the similarity with WAII vision

16. Estimates of location premium across classes

17. Illustration of results - 1
Country’s GDP per capita and Gini coefficient explain about 90% of the variability of income (R2 between 0,87 and 0,91)
if we take all people who are in a given ventile of their country’s income distributions → 90% of the variability of their incomes explained by GDP per capita and Gini coefficients of the country where they live
→ average income of the people in each ventile largely depend on the mean income of their country and on its distribution
Locational premium varies across ventiles
Relatively low for the bottom ventile (0.769)
Then it rises
At the maximum is about 0.88
→ locational premium holds for everyone → people in any ventile are better off if they live in a richer than in a poorer country
→ premium is less for those in the lowest ventiles of income distribution

18. Illustration of results - 2
Mean income and income inequality can also be substitutes
Given her income class, a person might gain more by being ‘‘allocated’’ into a more equal society even if its mean income is lower
Alternatively if a person is allocated to a top income class → gain from belonging to a more equal society will be negative
→ Trade-off between mean country income and inequality is not the same across all income classes
To assess the tradeoff at the bottom income class (regression 1) each Gini point increase is associated with a 5.75% loss of income → greater inequality, for a given mean income, will harm the poor
To exactly offset this, a person in the bottom ventile would have to be relocated in a country whose GDP per capita is about 7.5% higher (5.75 divided by the coefficient on mean income in regression 1, which is 0.769)
For the second lowest ventile, the GDP per capita increase needed to offset 1 Gini point higher inequality is 5.3%, and so on

19. Income/equality tradeoffs per ventile

20. Illustration of results - 3
Figure shows how much a 1 point increase in Gini is worth in terms of GDP per capita
The «1 Gini point equivalent» GDP per capita increases gradually decline as we move towards higher ventiles
It become close to 0 around the 16° ventile
People of the 17° ventile and richer benefit from increased inequality
To leave them with the same income, more unequal national distribution (from which they gain) has to be combined with lower GDP per capita (from which they, like everybody else, lose)
For the top ventile each Gini point change is offset by about a 3.2% change of GDP per capita in the opposite direction → For the nationally rich, the national distribution matters but less than for the poorest
The importance of change in national income distribution (represented by an increase or decrease of 1 Gini point) displays a U-shaped pattern with a substantially higher left end

21. Illustration of results - 4
Figure also shows that those for whom national inequalities are important are either:
those at the bottom → they gain from lower inequality
those at the very top → they gain from higher inequality
For the middle class (ventiles 13 to 18) → equality or inequality of national income distributions matters very little
→ Their income shares are about the same in both equal and unequal countries
For this people the mean income of the country where they live is of crucial importance
Three conclusions
All classes benefit from higher mean income → but that benefit is proportionately greater for the rich classes
Distributional change matters to the poor and to the rich (in the opposite directions) while it is of little importance to the middle class
For the income of the middle class what matters is whether the county is getting richer or poorer, not whether it is becoming more or less equal

22. What have we learned - 1
With only two circumstances, GDP per capita and income inequality of country of residence, or simply with country dummy variables, one can account for more than half of the variability in personal percentile incomes around the world
Other features (gender, race, or ethnicity) would increase the explanatory power of circumstances → role of place of residence is a lower bound to global inequality of opportunity
Locational premium is very large → compared to living in the poorest country in the world (DR Congo), a person gains more than 350% if she lives in the United States but only 32% if she lives in Yemen
Ability to predict a person’s income from only these two country characteristics holds also for each income class separately → given the income class of a person (her country and income ventile) knowledge of the country where that person lives explains about 90% of the global variability of incomes
Locational premium is positive for the entire spectrum of national income distributions
Given the person’s own income class, there is a trade-off between the GDP per capita of the country and its income distribution
A person in a low income class prefer to live in a more egalitarian country even if that country’s GDP per capita is lower
The opposite, of course, holds for a high income class person
For the middle class the national income distribution is relatively unimportant → income shares of the middle ventiles do not vary much across nations, whether the nations are equal or not
For the middle classes mean income of the country where they live is the key factor in determining their own income level

23. Global income inequalities and migration
An individual in any country must basically hope that her country will do well
If the country’s income increases relatively to other ones → the country will then move up the global income ladder and carry its entire population upwards
If the individual’s effort (a movement higher up along the national income distribution) is combined with an upward movement of the country itself (increase in national mean income) → he may substantially climb up in the global income distribution
If the country does not do well but the individual does, the two movements counteract each other, but the country’s effect is likely much larger
Alternatively the individual might try to move from a poorer country to a richer country → migration
Even if he does not end up at the high end of the new country’s income distribution → he might still gain significantly
Own efforts, hope that one’s country does well, and migration are three ways in which people can improve their global income position