World Inequality Report (2022): Regularities in Inequalities

The World Inequality Report (2022) (WIR-2022) is out and has already gained much attention. The report includes chapters on global income and wealth inequality, environmental inequalities and inequalities through women's perspectives, and international tax policies.

They were nice enough to share some aggregates they have calculated, together with the descriptions of their methodologies. I felt like exploring the data a little bit.

An Unequal World

I feel responsible for starting with emphasizing a first observation WIR-2022 made: "We live in a data-abundant world, and yet we lack basic information about inequality." Not only do we live in a data-abundant world, but also with an abundance of people trained in statistics, able to use software and ask interesting questions, yet we can't utilize neither effectively. Highly restricted access to data (and publications for the same reason) constitute a massive obstacle in front of our field's development.

According to WIR-2022, inequalities we see globally are extreme. To give some numbers: "The global 50% captures 8% of total income measured at Purchasing Power Parity (PPP). The global bottom 50% owns 2% of wealth (at Purchasing Power Parity). The global top 10% owns 76% of total Household wealth and captures 52% of total income in 2021", noting, top-wealth households and top-income households are not necessarily overlapping perfectly.

One comment that seems to have gained attention is, "Contemporary global inequalities are close to early 20th century levels, at the peak of Western imperialism". Hauner, Milanovic, and Naidu, in their 2017 article "Inequality, Foreign Investment, and Imperialism," provided a detailed analysis of the relationship between inequality and global conflicts. I see WIR-2022's reference as no less relevant; global inequalities indeed are complemented with international conflicts (I tried to share some thoughts on this topic before here: Clusters (goingtomeettheman.blogspot.com).

Wealth inequality is observed between developed vs. developing worlds and within these regions. The report finds, "The Top 10% in Latin America captures 77% of total household wealth, versus 22% for the Middle 40% and 1% for the Bottom 50%. In Europe, the Top 10% owns 58% of total wealth, versus 38% for the Middle 40% and 4% for the Bottom 50%." Among South-East Asia, East Asia, North America, Sub-Saharan Africa, Russia & Central Asia, and MENA regions, there is no single place where the top 10% holds less than 60% of the total wealth. Similarly, the report finds until the 1980s, between-country income inequalities were rising, but since the 1980s, within-country income inequalities have been on the rise. Lastly, the report finds that the top 10% of emitters are responsible for nearly 50% of all emissions, while the bottom 50% produce 12% of the total.

The report points out the average income of a country may not be a good indicator of how the income is distributed in that country or where that country stands in the global distribution of income. It is suggested to use the Top 10%'s income ratio to the Bottom 50%'s income. This ratio answers the question "How many times more do the rich earn than the poorest half?" and reveals MENA as the most unequal part of the world, followed by Sub-Saharan Africa and Latin America.

Regularities in Income and Wealth

I wondered if these trends are visible in a more aggregate vision and what patterns they support. From the WIR-2022 methodology section, I used their calculations for the Top 10% and the Bottom 50% wealth. I found the data for years between 1995-2021, and the countries in this dataset do not correspond one-to-one for income inequality measures. Matching the available countries in two datasets, I tried to see if some form of a scaling law alike relationship between income and wealth inequality.

Realize, wealth inequality can get much more significant than income inequality. Thus the figure above is obtained on a logarithmic scale for the x-axis. The wealth and income inequality suggest a positive relationship when looked at on an aggregate level. Similarly, the country-level relationship rarely depicts a downwards sloping curve. Yet, it is hard to say the inequalities documented here suggest a scaling law-like relationship. Consistent with WIR-2022 findings, the inequalities are somewhat clustered, meaning country-level effects play a role in the relationship between wealth and income inequality. Although the aggregate relationship that is observed here can be seen as a piece of evidence to "higher wealth inequality -> higher income inequality" argument, first, within these countries, the top wealth owners are not necessarily the top income groups; second, the slope differs considerably in each country, suggesting the importance of country-specific effects. As a next step, I tried to build rank-size plots to explore income inequality, noting that the countries and years documented now will be different than the first experiment:

1980-1989

1990-1999

2000-2009

2010-2021

I'm not suggesting a statistical model for the patterns depicted above. However, two things stand out: first, starting from the late 1980s, maximum values p10/p50 ratios can get seem to be getting bigger; second, more and more countries each year are "jumping into extreme inequality wagon." The tail of the rank-size plot started by suggesting a mixture of two states of the world where the early and late 1980s were not presenting a common pattern, yet, that deviation seems to have disappeared with a significant number of countries becoming more unequal. Regarding statistical modeling, visual inspection suggests a mixture for pre-1990s and a Pareto for post-1990s.

Regularities in Emissions

As before, I will start with some observations made by WIR-2022: "In 2021, humans released nearly 50 billion tonnes of CO2 into the atmosphere, reversing most of the decline observed during the 2020 Covid pandemic. Of these 50 billion tonnes, about three quarters were produced in burning fossil fuels for energy purposes, 12% by the agricultural sector, 9% by industry (in cement production, among other things),  and 4% came from waste. On average, each individual emits just over 6.5 tonnes of CO2 per annum. These averages mask considerable disparities between countries and within them, as we discuss below."

I intend to explore the available datasets further, but I used their 2019, country-level emissions data as a first approximation. The main idea is, in 2019, in a selected country, what are the ratios of top to bottom emissions? So the emissions created by the top 10% emitters, divided by, let's say, the bottom 50% emitters. The following figure is the rank-size plot for that idea:

Unlike previous rank-size plots, I preferred adding loess and linear fits to emphasize the "mixture" point. Maybe not a very surprising finding: no matter what ratio you observe, there seems to be an "extreme inequality in emissions" group that diverges from the overall behavior depicted by other countries. What is maybe a little surprising is that while the income inequality is getting extreme for most countries and they join each other in being extreme inequality countries, the emissions inequality suggests little empirical evidence for a convergence of that sort.

What Regularities?

One has to appreciate the spirit in the reports and works like WIR-2022. Ultimately all types and forms of inequalities listed in the report are seen as results of policy, and the report's faith in beating them has to be celebrated. Of course, this small blog post is just a blog post that I've prepared for fun (and if there are any errors in data work or anything, they are my errors), which provides some support to this line of thinking. First, there doesn't seem to be a law-like dynamic governing the relationship between wealth and income inequality (and if there is, it is relatively weak). Second, extreme income inequality has become the defining feature of our times. Third, the inequality in emissions seems to reflect a mixture of behaviors: a variety of emissions that we describe with a functional form and a group that deviates from the tails of that functional form. Based on these three observations, one can make many suggestions, but that should probably be another blog post. Yet, and I can't believe I'm saying this, there seem to be reasons to believe/claim there's room for policy to reduce inequalities.