According to the United Nations, income inequality is rising in many countries around the world. In addition, the COVID-19 pandemic has exacerbated global income disparities, with some countries facing greater economic losses than others.
Policymakers are increasingly focused on finding ways to reduce inequality and create a more just and equal society for all. When deciding how best to intervene, policymakers often rely on the Gini coefficient, a statistical measure of the distribution of resources, including wealth and income levels, among a population. The Gini coefficient measures zero for perfect equality and one for maximum inequality, with higher numbers indicating greater concentration of resources in the hands of a few.
This measure has long dominated our understanding of what inequality means, largely because it is used by governments around the world, published by statistical offices in multiple countries, and in the news media and policy Often discussed in discussions.
In our paper, recently published in natural human behaviorwe argue that researchers and policymakers are overly reliant on the Gini coefficient—by expanding our understanding of how inequality is measured, we can both uncover its effects and intervene to more effectively correct for it.
One size doesn’t fit all
In its simplest terms, inequality is the degree to which some people have more than others. To help compare different societies and measure inequality over time, researchers often use the Gini coefficient to capture the concentration of income in geographic locations. In many cases, the Gini coefficient is a useful metric to make such comparisons.
However, the Gini coefficient may not always reflect how resources are actually allocated. To illustrate why it’s important to consider other sizes, imagine what it’s like to buy pants. If it’s your usual size, you can go to the store and ask for a size M trousers. In many cases, these pants may fit well—because the medium conveys the most relevant message you want to convey. However, sometimes an all-encompassing medium isn’t enough to get the sizing and fit just right.
“The Gini coefficient — while usually a decent measure — is not the best way to capture the information contained in the distribution of U.S. incomes.”
Because people come in different body shapes and sizes, many retailers offer pants that fit two different sizes: waist and length. After all, two people who might be wearing a size M might actually need pants of different lengths and widths, and when choosing clothes that take these two extra sizes into account, they might end up choosing pants that fit better. For retailers, the goal is to design measures that capture changes in people’s preferences well, while minimizing the number of measurements needed to reduce the cost of producing and storing different types of pants.
Ortega provides a new answer
The same goes for measures of inequality: In our analysis of US data, we show that the Gini coefficient—although usually a good measure—is not the best way to capture the information contained in the distribution of income in the United States.
Instead, after comparing 17 different measures, we found that a measure consisting of two independent variables (called the Ortega parameter) reflects our dataset of more than 3,000 income distributions at the county level in the United States the best-fit model. The Ortega parameters together contain more information than Gini alone can capture. In fact, each of Ortega’s parameters focuses on a different aspect of income distribution. The first reflects how well income is distributed between low- and upper-middle-income earners, while the second reflects how ultra-high-income earners compare to the rest of the population. Aggregating the two Ortega parameters provides the Gini coefficient.
This nuance allows us to identify different types of inequality in society.For example, when measured by the Gini coefficient, two regions have very high levels of inequality: Teton County, Wyoming — home to the popular resort town of Jackson Hole and the Federal Reserve’s annual economic policy seminar last week Where it will be — and Monroe County, Alabama — home to Monroeville, home to Harper Lee to kill a robin.
But when two Ortega parameters are used, we gain a deeper understanding of the source of income disparity: inequality in Teton County is driven primarily by a minority of the ultra-rich, while inequality in Monroe County is driven by a minority Driven by the super rich. There are wider differences between low-income earners and upper-middle-income earners. We created an interactive map showing the overall degree of inequality (measured by the Gini coefficient) in different U.S. counties, and where this inequality is concentrated through more specific Ortega parameters, available here.
Are obesity and education linked to levels of inequality?
Not only can these measures of inequality yield more information about income distribution, but using these measures can also provide new insights into how inequality affects important social outcomes. We collected 100 policy-relevant indicators—including education, obesity, and others—and compared their associations with inequality, using either the Gini coefficient as done in previous studies or the Orr, which we propose here Special plus parameters to measure. In the vast majority of the 100 cases we examined, we found that the Ortega parameter bikini coefficient revealed more information about the relationship with social outcomes.
For example, consider the link between inequality and obesity. In examining our data, we found no significant association between inequality and obesity when using the Gini coefficient. However, both of Ortega’s parameters showed significant correlations in opposite directions: We found higher rates of obesity in areas where differences between low-income and upper-middle-income earners contributed to inequality. At the same time, we found less obesity in areas where differences between ultra-high earners and the rest of the population led to inequality.
We also examine the correlation between inequality and educational outcomes. Our data showed that the relationship between the Gini coefficient and the proportion of the population with a bachelor’s degree was not statistically significant, but the Ortega parameter again showed a statistically significant association. More specifically, where inequality is driven by differences between low- and upper-middle-income earners, we find that bachelor’s degrees are underrepresented in the population, whereas inequality is driven by differences between ultra-high-income earners area. Earners and other populations are associated with a larger percentage of bachelor’s degrees.
Using the two Ortega parameters, as these two examples show, we can pinpoint the causes of inequality in obesity and education. These parameters can be used in other ways to drive a host of other policy-relevant outcomes, including initiatives related to health, crime, and social mobility. The Ortega parameters reveal more detailed results that have important implications for both researchers and policymakers.
call for change
In recent years, inequality has received increasing attention in the academic, policy and public spheres, with many calling for a change in the status quo. In fact, a recent survey showed that a majority of Americans believe that economic inequality is too great. At the same time, public support for addressing inequality has been uneven.
Our research highlights that one way to understand differing beliefs about inequality and redistributive preferences may be to focus on the specific types of inequality that respondents are least satisfied with. If we made policies to address inequality based solely on the Gini coefficient, we would treat places like Teton and Monroe counties equally. But that might not be the right thing to do.
“Reducing inequality can be achieved by closing the gap between low- and upper-middle-income earners, and by closing the gap between ultra-high earners and the rest of the population.”
For example, reducing inequality can be achieved by reducing the gap between low- and upper-middle-income earners, such as by raising the minimum wage; and reducing the gap between ultra-high-income earners and the rest of the population can also reduce inequality, such as increasing the income tax. Our findings suggest that going beyond the overall concentration of inequality reflected by the Gini coefficient can be productive in identifying different types of inequality and determining how meaningful changes can be made to correct them.
A limitation of our study is that our results are limited to our US dataset, which may call into question whether these two Ortega parameters are also the best measures in other datasets, including other countries. The problem is that most datasets currently available to researchers do not contain enough information to perform the various analyses we show here. Therefore, if we want to look beyond the U.S. to study inequality, we need to make publicly available similarly high-quality data from other countries in order to follow the approach we suggest here and determine which measures are best suited to capture inequality.
Ultimately, we hope our work will encourage statistical offices around the world to publish more detailed inequality data for public review. After all, the stakes are high, and doing so will only help to better understand and address inequalities around the world.
Jon M. Jachimovic is an assistant professor in the Department of Organizational Behavior at Harvard Business School. Kristen Bresch PhD student at the University of Bremen. Oliver Hauser is Associate Professor of Economics at the University of Exeter Business School.
You might also like:
Feedback or ideas to share? Email the Working Knowledge Team at firstname.lastname@example.org.