Mike Groll/AP Photo
The Commerce Department’s Bureau of Economic Analysis collects and calculates data to explain how America’s aggregate income is distributed.
“GDP has become something of a talisman. When GDP is growing, it gives people and companies hope, and when it declines, governments pull out all the policy stops to reverse the trend.” —World Economic Forum
For almost 90 years, the U.S. government has produced what many experts consider our economy’s single most important number: the gross domestic product, or GDP.
To do this, the Commerce Department’s Bureau of Economic Analysis collects, calculates, and processes thousands of data points that measure the economy’s performance from hundreds of sources. Last year, America’s current-dollar GDP grew 6.6 percent to over $26,000,000,000—one quarter of the entire world’s total output, produced by just 6 percent of the world’s population.
And so when the latest GDP number is announced, it’s front-page news in newspapers and magazines everywhere, a leading story for TV and radio, and the preoccupation of an almost uncountable number of online sites. But you may have noticed something strange about GDP: Because it’s a measure of the total economy’s output, it’s silent about how that output is divided among Americans. We’ve been living—and are living right now—in a nation with ever-increasing inequality, which makes the question of who gets how much of the GDP as important as GDP itself. And that makes it time for the government to start measuring and reporting more than GDP’s sum of our aggregate production. We need to know how GDP growth is distributed.
Let’s for the moment call such a measure the GDD—for “gross domestic distribution.” (I’ll explain how it would work in a moment.)
Most Americans already know that America has been growing ever more unequal—but beyond that, they know few details. Government tells them that GDP is still growing—by 5.2 percent in the last quarter—but not that nearly all of that growth is captured by the wealthiest 10 percent, and especially the top 1 percent.
Yet what most Americans know intuitively is that life for the Great American Middle Class has gotten much harder—something the government’s own data has been recording since the 1970s. That data shows that, to begin with, the broad middle class’s share of total household income has fallen from over 60 percent in the 1970s to barely 40 percent. Among young Americans—the millennials—far fewer are earning more than their parents at the same age, and the gap is growing wider, especially for the majority who lack college degrees.
Yet because that’s been a steady, slow-rolling decline, officials haven’t talked about what’s happening the way they do for a significant “event” like the Great Depression—or our recent Great Recession and COVID lockdown. Like the frog in the pot that gets hotter slowly, the temperature keeps rising—but the frog doesn’t jump.
But most of us aren’t like the frog. We know the water is getting hotter—and we want out of the pot. Nearly 80 percent of Americans now tell pollsters that reducing inequality should be “a top or important priority” issue. Among Democrats, it’s become the urgent focal point for President Biden’s (unexpected but welcome) break with the party’s neoliberals; for Republicans, it’s right up there alongside all the “culture war” issues fueling the angry right-wing populism driving their dream to make Trump president again.
What Would a GDD Do?
Here’s how creation of what I’m calling the GDD—a gross domestic distribution index—could help focus the debate about causes and cures of America’s steadily rising inequality.
If you visit the website of the Commerce Department’s Bureau of Economic Analysis, you’ll see on their landing page that, in addition to GDP, they collect and calculate—in quite detailed reports, backed by literally dozens of tables and charts—how to explain in some very particular ways how America’s aggregate income, the sum of all incomes, is distributed.
You’ll find the total personal income of all Americans broken down, for example, by types of income, such as wages, dividends, interest, proprietors’ income, and rents; by income sorted by states, counties, and metro areas; by income adjusted for taxes, for inflation, and both.
But you’ll also see a problem: The data isn’t available in a user-friendly form, and the details can overwhelm the unwary. For example, you can learn from the BEA that “in 2021, personal income increased in 3,075 counties, decreased in 36, and was unchanged in 3” or that “the $138.7 billion increase in [personal consumption] in September [2023] reflected an increase of $96.2 billion in spending for services and a $42.5 billion increase in spending for goods,” plus a great deal more.
Here’s a much bigger problem, however—one that having a GDD could solve. What the BEA doesn’t do is tell you how Americans’ income can be broken out and arranged by the size of their incomes—ranked by the top 10 percent, say, or the middle fifth, or the bottom half. Thus, you won’t learn from the BEA where income has been flowing over the years—how much to the one percent or the poor or the middle class. That is, needless to say, elementary information for addressing income inequality—a serious topic to, let me remind you, 80 percent of Americans nowadays.
Government tells us that GDP is still growing, but not that nearly all of that growth is captured by the wealthiest 10 percent, and especially the top 1 percent.
Yet here’s a surprise: The government has that information—in fact, it has breathtakingly detailed income information, all calculated, charted, and graphed. That data comes from multiple sources including its own extensive multiyear surveys, as well as the research done by individual branches of government such as the Treasury, the Bureau of Labor Statistics, the Department of Agriculture, the Federal Reserve, the IRS, and the multiagency Federal Statistical Research Data Centers. And, for those so inclined, it offers data-organizing tools and chart-building software, including multiple alternative metrics, to analyze and assess all the data.
The U.S. Census Bureau, which happens to be the BEA’s neighbor in a giant office complex just outside Washington, is the primary collector of this data. Over the past decade, the agency has spent millions of dollars upgrading the online organization and accessibility of its enormous data pool.
At the U.S. Census Bureau website, you’ll thus now find Americans’ income sorted by household size; by marital status; by the household head’s sex, race, and age; by region of the country, subdivided by urban/suburban/rural; by native or foreign-born; and by level of education. And for each of these, there’s a summary of median income as well as year-to-year changes. And that’s packed into just a few of the dozens of tables you’ll find.
Another Census table sorts the size of income groups by tenths from the poorest to richest, as well as the top 5 percent, and then calculates income ratios between the bottom, middle, and top tenth of income earners—and moreover has all this organized annually back to the mid-1960s. Yet another table lets you do this kind of sorting yourself by state, by race, by types of income, and even in some cases, by ZIP code.
In other words, the government has a vast amount of detailed, publicly available information about the distribution of Americans’ income. It’s just that apparently the public doesn’t know it.
Instead, what they do know comes largely from personal experience, shaped by some back-of-the-envelope categories we all know. Most Americans, for example, assume there are the rich, poor, middle class, upper middle class, lower middle class, and working class. But where the boundary lines fall between these classes, and how much income (and wealth) each has, turns out to be fuzzy at best.
Here, for example, is a figure from a study done at Harvard Business School several years ago, a poll on the subject of wealth (not income) distribution. The study asked interviewees what they think U.S. wealth distribution looks like—and what they think that distribution should be—compared to actual distribution.
What you immediately see are two things: first, that Americans underestimate the concentration of wealth at the top and overestimate it for everyone else—and second, that they think a fair distribution of wealth would be a good deal more equal than what they (mistakenly) think it is.
Let me quickly stop you before you say, like a great many people, “Well, yes, but what Americans think wealth distribution is and should be is skewed by party and ideology—Republicans and Democrats will give you very different answers.” Take a close look at the second figure below from the same Harvard study; this one asks the same questions about the “imagined” and “ideal” versus the “actual” distribution of wealth—but sorts out respondents by their own income level, party affiliation, and gender. What’s most interesting is the degree of uniformity of views across the various groupings.
The Gini Index: Could It Be the New GDD?
One of the most important metrics economists use to measure income inequality is the so-called “Gini index.” First modeled shortly before World War I by Corrado Gini, an Italian statistician and demographer, the Gini index arranges the distribution of individual incomes along a line from the smallest to the largest. Using a scale from 0 to 1, the 0 end represents a (purely hypothesized) economy in which all income is equally distributed—everyone earns, say, the same $10,000 or $100,000. The 1 end signifies an equally hypothetical world in which just one person has all the income—a sort of Musk World on AI-prescribed steroids.
The two ends are theoretical points, obviously—but by mapping them in this form, they allow graphing of a curved-line distribution (a “Lorenz curve” to economists) between 0 and 1 that in turn yields a single-number result, the so-called “Gini number” (economists prefer to say “Gini coefficient”). The lower the Gini number, the more equal the distribution of income; the higher the number, the more unequal.
These days, we have quite a lot of information about the Gini number for quite a lot of countries. We know, for example, that some of the smaller European countries—Slovenia, the Czech Republic, the Netherlands, and the Nordics—are collectively the most egalitarian, with Gini numbers around 0.25 (remember, this is a scale from 0 to 1). By contrast, some sub-Saharan African (South Africa and Namibia), Latin American (Belize and Brazil), and Middle Eastern countries (UAE) rank at nearly 0.6, much closer to 1.
What about the United States? The U.S., at 0.434, was ranked by the OECD three years ago as the most unequal of the G7, the most advanced economies in the world.
Another OECD study showed moreover that since 1985 the U.S.—which already by then ranked as the most unequal among this larger pool of advanced economies—has been steadily growing more unequal.
Our inequality, after four decades of decline beginning during the New Deal, started rising in the 1970s amidst stagflation, two massive oil price shocks, and the offshoring of U.S. manufacturing. That abrupt turn toward inequality was then amplified by public policies. Starting in 1980, then-Fed chairman Paul Volcker deliberately initiated two back-to-back recessions designed to crush inflation, and the Reagan administration followed Volcker with massive supply-side tax cuts on corporations and the wealthy. Neoliberal Democrats kept the process going with Wall Street’s deregulation, trade treaties that hollowed out unions and regions of the country, and a failure to reverse Republican tax policies.
If a GDD Can Be Added to the GDP, What Is to Be Done?
We know that the Census Bureau has nearly 60 years’ worth of Gini data available to transfer and fold into the BEA’s historical tables of GDP growth—and with the two agencies literally next-door neighbors, sharing data and methods hardly seems to represent an enormous problem.
There are a few technical questions, though, that need to be addressed.
The first concerns frequency of reporting: The Census Bureau generally releases its latest Gini number on an annual basis, whereas the BEA releases GDP both quarterly and annually. Because of time lags, however, in the collection of income distribution data, it isn’t practical to release a Gini index each quarter.
The second is formatting: The Gini relies on a 0-to-1 scale, with 0 as ultimate equality. Americans love sports whose winners have the higher score (golf being a rare exception). I’d thus invert the scale, with greater equality earning a higher, not lower, score. I’d then recalibrate 0-to-1 as 0-to-100—here again, because the general public will more easily recognize the scale.
Finally, there’s the matter of providing more details. Both the GDP and GDD are single summary numbers, and details of both aggregate growth—sectors, industries, regions, etc.—and distribution by age, sex, race, education, etc. are vitally important. Given the flexibility of online formatting, the Census Bureau could easily construct backup tables for GDD, so that with a click or two, you could find out what share of GDP growth was captured by, say, the top 5 percent and the bottom 90 percent. Click again, and you could also find distribution data by race, age, education level, and gender. Another click, and another table would show you the distribution before and after taxes and transfers, or by individual and by household.
Thus, when the annual GDP number is released alongside the new GDD, click-through supporting pages would offer granular data about who’s getting what share of the growth.
The Problem With Bureaucracy—and Funding
Curious to see whether the BEA would be open to adding GDD data from the Census Bureau to its GDP reports, I called their Washington office to ask that question—and got quite an interesting answer. First, I was told (rather archly) that I was mistaken: The BEA has its own income distribution data organized by size on its website, and I was pointed to it.
But what the BEA has, I found, is nothing like the Census Bureau’s enormous pool of information. Instead, filed away under the heading “special topics,” there are a set of “prototype” tables the BEA apparently began constructing only three years ago, and has only intermittently updated, most recently in October 2023. Along with the tables, there are a couple of research memos and a blog that collects miscellaneous research articles related to income distribution.
The prototype tables are, it turned out, nothing more than basic Excel spreadsheets that, going back from 2021 to 2000, show median and mean income for each year, and divide aggregate income by bottom-to-top tenths, then add the estimated income share of the top 5 percent and 1 percent as well as a Gini number for the year. Some newer refinements include tables that are state-level Excel sheet versions of the national data, plus what the BEA calls an “international comparisons table” that doesn’t actually compare the U.S. distribution to any other country’s. It is instead an Excel spreadsheet that follows the OECD’s somewhat different definitional guidelines for reporting U.S. income. Overall, the section is exactly what the BEA calls it: a “prototype,” a work in progress—one that’s quite frankly in no way as thorough as the data tables, charts, and tools available through the U.S. Census Bureau.
Don’t get me wrong: What’s there is not useless. A lengthy 2022 report by the nonprofit Washington Center for Equitable Growth, for example, used the BEA’s data tables to compare how inequitably income gains had been distributed after the Great Recession and after the dot-com bust of the early 2000s. The results were hardly surprising:
More importantly, the center’s report then went on to highlight several major deficiencies in the BEA’s prototype tables: Capital gains are not included, which means the BEA underreports a major source of income for the top 5 percent and 1 percent; the data BEA has started releasing comes with a two-year lag—when by comparison, its GDP data is released quarterly as well as annually, and includes routine follow-up revisions; the data doesn’t look back before 2000 (as Census data does), which severely limits the ability to track longer-term trends in terms of income distribution. It concluded that
The federal statistical system needs to be resourced to expand and continue reporting on inequality … Four decades of rising inequality calls for a more robust policy response to ensure broad-based growth in the U.S. economy. An important first step is to develop the data infrastructure to track growth in inequality over time, so that policymakers can monitor and respond to the problem, and voters can hold them accountable to producing strong growth for all U.S. households.
My suggestion here is that the BEA and Census Bureau not only need to talk to one another but should also jointly convene a group of economists, sociologists, demographers, and statisticians to hammer out some of these questions. I would add, however, the warning that the perfect can be an imperfect ally of the good. Getting to a usable, public report of how GDP is shared by Americans—and presenting it alongside GDP itself to ensure maximum public attention—shouldn’t be made to wait until all the technicalities are smoothed out.
Back in the 1930s, when the first national income accounts were being defined and constructed, their authors faced the same “technical difficulties” problems—and openly cautioned about the limits of what we today call GDP. Since the 1970s, new criticisms of GDP—for its mismeasurement of environmental costs, of women’s non-wage contributions to the economy, and a host of other faults—have even prompted creation of “alternative GDP” indexes—which just as easily and usefully could be done for GDD.
But the core fact is this: GDP—the measurement of aggregate economic growth, summarized in a single number—isn’t going away. Making it more useful by counting things that matter would seem worth trying. As the historian Theodore Porter once remarked about the ascendency of data and statistics in modern life, “Numbers … create new things and transform the meaning of old ones.”
