Confronting Inequality: How Societies Can Choose Inclusive Growth

Much of the discussion in this book is about the inequality of incomes within countries. Other authors, notably Milanovic (2005), have studied global inequality, that is, the difference between the incomes of the rich and the poor without regard to nationality; this measure of inequality has declined over time, in large part due to rising incomes in populous countries such as China and India. Yet, as Milanovic notes, inequality appears to be on the rise within many countries around the world.

How should we measure the gap between the rich and the poor within countries? This chapter begins by describing the three measures that we use. The first is the Gini coefficient, which is the most widely used—but not the most easily understood—measure of inequality. The second is the share of income accruing to the richest segments of the population, for example the top 1, 5, or 10 percent of the population. The third is the share of income going to labor: the share made up by wages and salaries as opposed to other forms of income such as profits. The chapter concludes with a primer on the drivers of inequality.

Before we get to inequality, however, we have to understand the distinction between market income and net (or disposable) income, which plays an important role in this book. Most societies levy taxes on income and also provide some assistance to people in the form of transfers. Net income is market income adjusted for those taxes and transfers; that is, we subtract from market income the taxes paid on that income and add in the transfers received.

To illustrate the distinction, figure 2.1 shows annual market and net incomes for the average person in the United States and France from 1970 to 2014. For both countries, net incomes are below market incomes. This means that the average person pays more in taxes than he or she receives in transfers. People much poorer than the average are likely to be receiving more in transfers than they pay in taxes; hence, their net incomes would be higher than market incomes, and the opposite would be true for people much richer than the average.

Net incomes are derived from market incomes by subtracting taxes paid to the government and adding in transfers received from the government. The average person pays more in taxes than he or she receives in transfers. (Figures are in 2011 U.S. dollars.)

Now let’s move to comparing inequality in incomes in the two countries. Figure 2.2 shows how market and net incomes in the two countries are distributed across deciles, from the poorest tenth of the population (decile 1) to the richest (decile 10). The height of each bar shows the percent share of income going to that decile. If market incomes were equal across the population, each of the bars shown in the top panel would be of the same height and at 10 percent (and indeed the concept of a decile would be redundant). Even if market incomes were not equal, societies could choose, through taxes and transfers, to make net incomes equal, so that the net income bars in the bottom panel would be of the same height at 10 percent.

Of course, in both the United States and France, market incomes are far from equal across the population and so are net incomes, despite redistribution from rich to poor. For example, the share of market income going to the richest decile is about 35 percent in France and about 45 percent in the United States. Redistribution does lower these shares considerably: in both countries, the share of net income going to the richest decile falls to below 30 percent, but it is still well above the 10 percent share of a perfectly equal society.

Is market income inequality lower in France than in the United States? And how does redistribution affect the relative inequality in the two countries? The data in figure 2.2 can inform our answers to these two questions but cannot resolve them fully. Take the question about market inequality: It’s true that the share of incomes going to the very poor deciles is higher in France than in the United States, but even the middle-income deciles in France have a higher share than their U.S. counterparts. So, while the gap between the very poor and the very rich seems smaller in France, the gap between upper-middle incomes and the very poor could well be larger.

The second question is also difficult to answer purely from a visual inspection of the data. Redistribution does lower the share of incomes going to the very top, as we already noticed, but it moves it to both the very poor and the middle class, again making the impact on inequality uncertain.

If we were only interested in, say, the gap between the poorest and the richest deciles, we could make statements about the relative inequality in the two countries. But in general, we would like to be able to compare overall inequality in the two countries, not just at particular deciles. Here’s where the Gini coefficient comes to the rescue. Technically, the Gini measures the average difference in income between any two randomly chosen people in the population. It thus provides a summary measure of inequality within a country. The index is scaled so that it varies from 0 to 100: 0 means that everyone in society receives the same income and 100 means that one person gets all the income.

Figure 2.3 explains the concept of the Gini coefficient, continuing with the example of France and the United States and using net incomes. The vertical axis cumulates the share of income going to each decile. If income within a country was completely equally distributed, the points would all lie along the 45-degree line and the Gini coefficient would be 0. Now consider what France’s data look like compared to the line of complete equality. The poorest 10 percent of the population account for about 4 percent of the income. The next decile gets 5 percent of the income, bringing the cumulative total to 9 percent as shown. We proceed in this manner, adding deciles and showing the cumulative totals at each step. The figure also shows similar numbers for the United States.

Note: The numbers shown are the cumulative percent shares of net income received by people up to that decile. For instance, “30” and “25” at P5 indicate that the poorer half of the population receives 30 percent of the income in France and 25 percent of the income in the United States.

It is now evident that no matter which decile we consider, the cumulative share of income going to individuals with incomes up to that decile is higher in France than in the United States: there is indeed less inequality, overall, in France than in the United States. The mathematical formula used to compute the Gini summarizes this difference between the income distributions of the two countries by giving France a lower coefficient (29) than the United States (37).

Armed with this knowledge about the Gini coefficient, let’s return to the discussion of U.S. versus French market and net incomes and see how inequality has evolved over time in the two countries. Figure 2.4 shows the Gini index, computed using both market and net incomes. The top panel gives the striking result that the inequality of market incomes has been quite similar in the United States and France since 1995. What is different in the two countries is the inequality in net incomes, which since 1985 has been much lower in France.

Inequality of market incomes is similar in the United States and France. Inequality of net incomes in the United States is much higher than in France and it has also increased over time.

We can compute market and net Ginis for a large group of countries across the world. The results for 2018 are shown in figure 2.5. This figure shows market inequality on the horizontal axis and net inequality on the vertical axis. Each point represents a particular country. A country on the 45-degree line would have identical market and net inequality, implying that no redistribution was being carried out. A country far below this line would have much lower net than market inequality, meaning that a lot of redistribution was taking place.

Almost all countries lie below the line, implying some degree of redistribution. Some, such as China, do little of it; others, such as Sweden, do a lot. And on average, the distance from the line grows with the amount of market inequality, showing that countries with more unequal market incomes tend to redistribute more. We can verify from the figure that the market income Ginis for the United States and France are virtually the same, whereas the U.S. net income Gini is much higher than that of France.

Table 2.1 shows the countries with the highest and lowest Gini coefficients in 2010 for market and net incomes. It also shows the countries with the largest increases and decreases in inequality between 1990 and 2010.

Most inequality data, such as those on Ginis that we have been studying, come originally from surveys that typically ask households about income from various sources, taxes, and consumption. Surveys, which are expensive and complex, generally are undertaken only every few years at best. And there is no guarantee that the surveys are representative. The very rich may be less likely to participate or reluctant to reveal the true extent of their incomes in a survey (Ostry and Berg 2014).

Because of these limitations of surveys, economists began looking to tax records as a source of income distribution data (Atkinson, Piketty, and Saez 2011). These data are available for all taxpayers, so the rich are better represented and it is possible to look at small segments, such as the top 0.1 percent. Moreover, the data tend to be available annually and often as far back as the early twentieth century.

But there are also important disadvantages to using tax data. First, the many poor and even middle-class people who do not pay income taxes are excluded. Second, there is generally little information on actual taxes paid and transfers received that allow the calculation of net income. Third, data are available only for advanced economies and a handful of emerging markets. And fourth, tax-based data have their own measurement problems related to misreporting and the use of tax-avoidance strategies, many of which are perhaps particularly available to the richest.

Still, the share of income going to the very top provides a useful complement to the Gini measures. Figure 2.6 shows the share of income going to the richest 1 percent of households in the United States and France. In the 1970s, the share was about 10 percent in both countries. Since then, however, the share has increased in the United States so that the richest 1 percent has about 20 percent of total income; in France the share has remained stable.

Another way of looking at the distribution of income is to compute the share that goes to labor compared to capital. As Washington Post columnist Robert Samuelson notes, calculating “labor’s share is straightforward. It covers workers’ wages, salaries and fringe benefits. Capital is more complicated. It includes corporate profits, the income of small businesses and professional partnerships, rents from real estate and net interest on bank deposits, bonds and loans” (Samuelson 2013). Over the past few decades, labor’s share of income—the part that is relatively easy to compute—has fallen in many countries. “The shift to capital is worldwide,” Samuelson (2013) observes, due to forces such as “globalization, new technologies, and weaker unions. All tend to beat down wages through intensified competition, the substitution of machines for people and the loss of bargaining power.”

Figure 2.7 shows labor shares for the United States and France. The U.S. labor income share is much lower today than it was in 1970, about 60 percent compared to 66 percent. France too has seen its labor share decline, though much of it happened during the 1980s.

We conclude this chapter with an exploration into the drivers of inequality. We have carried out a comprehensive investigation of what drives inequality—as measured by the Gini coefficient—using data for ninety countries from 1970 to 2015.

Nobel laureate Simon Kuznets conjectured that inequality would initially increase as a country goes from being poor to middle-income—as it first opens up to the forces of competition—but then decline past a certain threshold level of income (Kuznets 1955). This “Kuznets curve” does indeed explain some of the evolution of inequality across countries and over time. But we find that it also leaves a lot of inequality unexplained. As emphasized by Milanovic (2016), a multitude of other factors account for changes in inequality. In our investigation, we group these into four categories.

Structural factors: As we discussed in chapter 1, inequality of incomes could be due to inequality of opportunity, but the latter is difficult to measure directly. The mortality rate provides a proxy; for instance, higher mortality rates could be an indication of lack of access to health care. Hence, we include the mortality rate as a control for such difficult-to-measure factors that influence inequality of opportunity. The share of industry in gross domestic product (GDP) is used as an indicator for the structural transformation of the economy. Inequality tends to be very high in countries still dominated by agriculture and declines as resources move out of that sector into industry and services.

Global trends: Trends are measured by technology and by openness to trade, both of which have tended to raise inequality in recent decades. Technological change has conferred an advantage on those adept at working with computers and information technology. Global supply chains have moved low-skill tasks out of advanced economies. Thus the demand for highly skilled workers in advanced economies has increased, raising their incomes relative to those less skilled. And often, the people who benefit from these developments in the developing economies tend to be the higher-skilled people in those economies.

Economic policies: We include here (1) the degree of financial openness—the extent to which countries are open to foreign flows of capital, (2) domestic financial reforms, and (3) the size of government—share of government spending in GDP. As noted earlier, we expect that opening up to foreign capital flows raises inequality because the poor do not gain access to foreign capital and lose out disproportionately if opening up is followed by a crisis. Some of the same reasons, for instance lack of access to financial services, could also lead to higher inequality in the aftermath of domestic financial deregulation. Based on our findings on the link between fiscal consolidation and inequality, we expect that shrinking the size of government will raise inequality.

Other factors: Crises are often mentioned as leading to inequality by having a greater impact on the poor than the rich—hence an indicator of currency crises is included. The last variable is a proxy for dictatorial governments (whether the head of government is a military officer), which tend to raise inequality by allowing greater expropriation of wealth by elites.

Each bar in figure 2.8 shows the impact on the Gini from an increase in each of these drivers; in each case the increase considered is an increase in the value of the indicator from the fiftieth percentile to the seventy-fifth percentile of its distribution. The first thing to note is that each set of factors plays a role in contributing to inequality. For example, a country that is more open to trade has a higher Gini coefficient, as does a country that is more open to modern technology. The impact of each of the other factors conforms to our expectations, as discussed above.

For the purposes of this book, the key finding is that inequality is not driven solely by global trends or deeper structural factors: economic policies within the control of governments also play an important role. The three indicators of economic policies we have included, and which will be the subject of chapters that follow, all contribute significantly to raising inequality even after accounting for the influence of several other factors.

It is worth noting that policies toward trade liberalization are also to some extent within the control of governments, though small countries may have little choice but to integrate with others. The impact of currency crises is also to a large extent dependent on policy choices. In short, the extent of inequality that we have attributed to economic policies may be an underestimate.

High Market Gini (2010)		Low Market Gini (2010)		High Net Gini (2010)		Low Net Gini (2010)
Country	Gini	Country	Gini	Country	Gini	Country	Gini
South Africa	68	Iceland	39	South Africa	59	Czech Republic	26
Zambia	60	New Zealand	39	Zimbabwe	52	Netherlands	26
Zimbabwe	60	Kazakhstan	38	China	51	Ukraine	26
Latvia	57	Tanzania	38	Honduras	51	Belarus	25
Ireland	56	Mali	37	Zambia	51	Belgium	25
Lithuania	56	Fiji	36	Rwanda	50	Denmark	25
Rwanda	56	Ukraine	35	Guatemala	49	Iceland	25
Brazil	54	Taiwan	33	Chile	48	Slovenia	25
Honduras	54	Belarus	32	Colombia	48	Sweden	25
United Kingdom	54	Korea, Rep. of	32	India	48	Norway	24
Largest Increase in Market Gini (1990–2010)		Largest Decrease in Market Gini (1990–2010)		Largest Increase in Net Gini (1990–2010)		Largest Decrease in Net Gini (1990–2010)
Country	Change	Country	Change	Country	Change	Country	Change
Latvia	+33	Sierra Leone	−28	Georgia	+19	Sierra Leone	−24
Lithuania	+31	Malawi	−17	Rwanda	+19	Malawi	−15
Georgia	+22	Senegal	−9	China	+16	Iran	−7
Rwanda	+21	Iran	−8	Latvia	+14	Senegal	−7
Cyprus	+20	Fiji	−6	Lithuania	+14	Brazil	−6
Macedonia	+20	Tanzania	−6	Macedonia	+14	Fiji	−6
Russia	+19	Turkey	−6	Armenia	+12	Tanzania	−6
Estonia	+18	Egypt	−5	Bulgaria	+12	Egypt	−5
Bulgaria	+17	Mali	−5	Cyprus	+12	Mali	−5
China	+16	Peru	−5	Slovakia	+12	Turkey	−5