David Johnson is the executive director of the International Association for Research in Income and Wealth. Prior to that, he served as a study director for the National Academies of Sciences, Engineering and medicine, for a report called, “Creating an Integrated System of Data and Statistics on Household Income, Consumption, and Wealth” Johnson also served for 25 years in the Federal Statistical system, where he was the only senior executive to have leadership roles at the Bureau of Labor Statistics, the Bureau of Economic Analysis and the US Census Bureau. At the Census, he led the implementation of the supplemental poverty measure and the reengineering of the Survey of Income and Program Participation.
Episode Description
According to the World Bank, some 3.5 billion people live on less than $7 a day. That's more than 40% of the global population. Almost 700 million of those individuals live in extreme poverty, getting by on less than $2.15 a day. In the US in 2024, almost 40 million Americans were living in poverty, according to the U.S. Census. But what do all these numbers mean? How do the people researching income inequality measure poverty, and how reliable are those measurements? That's the focus of this episode of Stats and Stories with guest David Johnson.
Timestamps
Defining poverty (2:14)
What is the measure of poverty? (9:30)
How’s this information used? (12:11)
An Updated Measure of Poverty: Redrawing the Line talk (16:10)
What has made this such a success? (21:06)
Transcript
Rosemary Pennington
According to the World Bank, some 3.5 billion people live on less than $7 a day. That’s more than 40% of the global population. Almost 700 million of those individuals live in extreme poverty, getting by on less than $2.15 a day. In the U.S., in 2024, almost 40 million Americans were living in poverty, according to the Census. But what do all these numbers mean? How do the people researching income inequality measure poverty, and how reliable are those measurements?
That’s the focus of this episode of Stats and Stories, where we explore the statistics behind the stories and the stories behind the statistics. I’m Rosemary Pennington. Stats and Stories is a production of Miami University’s Departments of Statistics and Media, Journalism, and Film, as well as the American Statistical Association.
Joining me is regular panelist John Bailer, emeritus professor of statistics at Miami University. Our guest today is David Johnson, the executive director of the International Association for Research in Income and Wealth. Prior to that, he served as a study director for the National Academies of Sciences, Engineering, and Medicine for a report called Creating an Integrated System of Data and Statistics on Household Income, Consumption, and Wealth.
Johnson also served for 25 years in the federal statistical system, where he was the only senior executive to have leadership roles at the Bureau of Labor Statistics, the Bureau of Economic Analysis, and the U.S. Census Bureau. At the Census Bureau, he led the implementation of the Supplemental Poverty Measure and the reengineering of the Survey of Income and Program Participation.
David, thank you so much for joining us here on Stats and Stories.
David Johnson
Thank you.
John Bailer
David, it’s great to have you here. Poverty is a word we use in everyday conversation, and we may have a sense of what it means—although that sense may differ from person to person. It’s often something related to insufficient resources to meet basic needs. But that kind of colloquial understanding doesn’t really serve us well. So, what do you need in order to measure and track something like poverty? What kind of formal definition is required?
David Johnson
Yeah, so I think our own conception of poverty is important in developing a measure. When I start presentations, I ask people: What is poverty? What do you need? How do you build up a definition? Most people would say you need food, a house, clothing, maybe a car. Nowadays, you probably need a telephone and internet access. But do you need health insurance? Do you need childcare?
This is the critical issue when we think about how to measure poverty. Historically, the numbers that were just mentioned referred to the Supplemental Poverty Measure for the number of people who are poor. The official poverty measure, which is slightly different, says about 36 million people in 2024 were in poverty. But that official measure is outdated.
Back in the 1960s, President Johnson wanted to fight a war on poverty, but he couldn’t do it without a measure. You can’t fight poverty without defining it. So, they developed a measure based on what people spent on food—what they needed for food—multiplied by three, because food accounted for about a third of a typical household budget. Since then, that measure has basically been updated only for inflation.
At the time, about $3,100 a year would put you in poverty. Now it’s about $31,000.
Rosemary Pennington
I was actually going to ask you about the official poverty measure and the Supplemental Poverty Measure. I was doing some work to prepare for this and digging through the Census Bureau’s reporting on both. You’ve explained what the official poverty measure is, and I know you led the implementation of the Supplemental Poverty Measure. Could you describe what that measure is—how it was developed and why it was needed?
David Johnson
So let’s start with why the official measure has problems. The official measure is basically the cost of the Thrifty Food Plan multiplied by three, right? But today, people spend about one-sixth of their income on food. So should we still be multiplying by three?
Back then, there also weren’t many government programs. To me, one of the main reasons to measure poverty is to track people’s well-being, the hardships they face, and how government spending helps. Let’s say we increase food stamps—or, alternatively, next month we decrease them. What happens to people in poverty?
Well, the official poverty measure would show no change because food stamps aren’t included as income in that measure. Let’s say your health care costs go up. What happens? Nothing changes in the official measure, because those costs aren’t accounted for.
The Supplemental Poverty Measure came out of 25 to 30 years of research at the National Academies, starting with a 1995 report aimed at developing a better measure of poverty. The Supplemental Poverty Measure counts resources after taxes and transfers. So, the Earned Income Tax Credit is included. Near-cash benefits are included—not just Social Security, but also food stamps. It also accounts for costs like childcare. If you need childcare to go to work, the Supplemental Poverty Measure includes that expense.
Health care, however, was not included in the threshold itself. The idea was to determine how much money you need to meet basic needs for food, clothing, shelter, and utilities—plus a small additional amount. Child support payments are also handled differently. If you receive child support, it counts as income. But if you pay for child support, that amount is subtracted when determining whether you are poor.
There were many changes made to the resource measure that most people agree with, but there were also significant changes to the threshold measure. Instead of basing it only on food, the Supplemental Poverty Measure uses average spending on food, clothing, shelter, and utilities. It also adjusts for housing status—people who own their homes outright, without a mortgage, have lower thresholds because they have lower expenses.
The measure also accounts for geographic differences in cost of living. The cost of living in Manhattan is different from that in Mississippi, so the thresholds vary across the country. Taxes are included as well.
Another key difference—and this is what confuses people when they look at the numbers—is that the threshold moves as spending patterns change. If spending on food, clothing, shelter, and utilities increases, the threshold can go up—sometimes by more than inflation—because it reflects actual expenditures. The official poverty measure, by contrast, is adjusted only for inflation.
Last September, the Census Bureau released poverty figures. The official poverty rate was 10.6%, but the Supplemental Poverty Measure was 12.9%. That’s because the Supplemental Poverty Measure subtracts taxes, includes additional resources, and uses higher, more responsive thresholds.
Between 2023 and 2024, the official poverty measure showed a decline of about half a percentage point. The Supplemental Poverty Measure, however, showed no change. The threshold increased more than inflation because spending increased.
This is why poverty measurement can be confusing. It’s complicated, and you need to understand what each measure includes. When headlines say, “Poverty went down,” that depends on which measure you’re using. Under one measure, it may have declined. Under another, it may have stayed the same.
John Bailer
This makes it a really hard story to tell. It’s difficult for the public, for journalists, and for lawmakers—especially those who don’t think about how challenging it is to measure poverty when there are different definitions.
As you were describing this, you mentioned that the official poverty measure—around $31,000, based on that calculation—has a certain appeal. It may be flawed, but it’s simple. You can compare one number to a threshold and count how many people fall below it.
The Supplemental Poverty Measure, by contrast, sounds much more layered—almost individualized. It seems like you have to determine poverty status case by case. Is that correct?
David Johnson
Yes, that’s right. The Supplemental Poverty Measure has around 45,000 different thresholds because it changes based on housing status—whether you rent, own with a mortgage, or own outright. It changes depending on the number of children and adults in the family. It also varies by geographic location. So, it adjusts for many different factors.
In my view, if we go back to the beginning and think about poverty, it’s actually quite individual. I sometimes wonder whether we should have a different poverty threshold for every person. This idea is somewhat related to President Johnson’s thinking. He grew up in poverty and understood the challenges people faced. His goal in launching the War on Poverty was to reduce hardship for individuals—not just in the aggregate.
Of course, it’s nearly impossible to design policies that address every individual differently. But I don’t think having more detailed thresholds is a bad thing. After all, we don’t calculate taxes using a single percentage. People use tax tables with many different categories and brackets. Poverty measurement can be thought of in a similar way.
John Bailer
I was thinking about your earlier comment about starting with three times the cost of food as the poverty threshold. Even if we updated that logic and moved to six times the cost of food, that would tell a very different story. If you’re talking about $60,000 a year as a threshold, that’s dramatically different from what we’re using now.
So how do officials use this information? Is the Supplemental Poverty Measure the one being used and promoted now, or does the Official Poverty Measure still have legs?
David Johnson
The Supplemental Poverty Measure was not designed to be used for program eligibility or allocation. Why? Because it includes benefits like food stamps in the resource calculation. You can’t use a measure that counts benefits to then determine eligibility for those same benefits.
So the Official Poverty Measure—and its thresholds—are still used for several purposes. One use is allocating funding. For example, Title I education funding is targeted to school districts based on their poverty rates, which are calculated using the official measure. That process is extremely complicated because the Census Bureau has to estimate poverty at the school district level. That’s why surveys like the American Community Survey are so important—they provide income and poverty estimates at smaller geographic levels.
You can also use poverty rates to assess how society is doing overall. In that context, the Supplemental Poverty Measure is often more informative. When poverty numbers are released, people ask: Are there more or fewer people in poverty? What impact did programs like food stamps, Social Security, or the child tax credit have?
For example, during the COVID-19 pandemic, the official measure showed little change because stimulus payments and expanded credits weren’t counted as income. But the Supplemental Poverty Measure showed a large decline in poverty—especially child poverty—when those programs were in place, and then an increase once they ended.
The third major use of poverty thresholds is determining eligibility. The federal government uses the official threshold—around $31,000 for a family of four—and the Department of Health and Human Services creates poverty guidelines from it. These guidelines determine eligibility for programs like SNAP, WIC, and roughly 40 other programs.
Most programs don’t use the threshold directly; instead, they use a percentage of it. For example, SNAP eligibility might be set at 135% or 185% of poverty. Medicaid eligibility can go up to 400% of poverty. That suggests the official threshold may not fully reflect economic need.
One of the most important uses of poverty measurement is evaluating government programs. Since many programs are designed to assist low-income households, we need a measure that can show whether those programs are working. That’s why the Supplemental Poverty Measure is often better—it reflects taxes, transfers, and policy changes more accurately.
Rosemary Pennington
You’re listening to Stats and Stories, and we’re talking about measuring poverty with David Johnson.
David, I know you were involved in a report a few years ago titled An Updated Measure of Poverty: Redrawing the Line. I may be oversimplifying, but it seems to argue that the Supplemental Poverty Measure should evolve into the primary measure we use. Could you give us a broad overview of the report’s argument and its main recommendations?
David Johnson
Sure. I had been involved with the Supplemental Poverty Measure for a long time and thought it was the best measure available. However, it had never really been formally evaluated. It was developed by committee under the Department of Commerce and the Office of Management and Budget, but there were issues it didn’t fully address.
One major issue was health care. Under the Supplemental Poverty Measure, health care costs are subtracted from resources, but health care isn’t included in the threshold itself. The assumption was that everyone needs health care, so costs were removed from income. The problem is that uninsured people often have very low out-of-pocket costs—because they don’t receive care. That can make them appear less poor than they actually are.
The report proposed what it calls a “Principal Poverty Measure.” Under this approach, health care costs would be included in the threshold—what people need to meet basic living standards. At the same time, the value of health insurance would be counted as income. For someone with insurance, it wouldn’t change much. But for an uninsured person, it would reflect both the need and the lack of resources.
Housing was another issue. The Supplemental Poverty Measure adjusts thresholds for renters versus homeowners, but it doesn’t fully account for differences in housing costs. The report suggests using fair market rents—similar to those used in housing programs—to set thresholds. For homeowners who own their homes outright, the implicit rental value of living in their own home would count as income.
Childcare was also treated too narrowly. The Supplemental Poverty Measure accounts for childcare costs mainly when both parents are working. But people may need childcare if they are disabled, attending school, or in other circumstances. The report recommends including childcare more comprehensively in the threshold and counting childcare benefits as income.
Finally, the report recommends redefining the poverty unit. The Supplemental Poverty Measure defines units based on family relationships, but many households contain multiple families. For example, two sisters living together with their children would be treated as separate poverty units—even though they share housing costs. The Principal Poverty Measure suggests using the household as a unit instead.
These changes affect who are counted as poor. Under the Principal Poverty Measure, more uninsured people would be classified as poor, fewer homeowners would be considered poor, and more young families would be recognized as poor due to childcare costs.
The report ultimately recommends adopting this Principal Poverty Measure as the official measure, arguing that it better reflects modern living conditions and the true composition of poverty in the United States.
John Bailer
This really feels like a lesson in the importance of continuously wrestling with how we define something like poverty. Maybe “struggle” is the right word. You’re constantly asking: What are the essential ingredients? What truly defines poverty?
Ultimately, the goal is to act on valid and accurate information. One part of the reports you’ve worked on highlights the collaboration among federal statistical agencies responsible for these measures. Given the recent cuts to some of these organizations, I worry about the potential loss of important work.
Can you talk about why this has been a success story in official statistics—and what the impact of staffing or budget cuts could mean for the future of poverty measurement?
David Johnson
I do think the Supplemental Poverty Measure is a real success story. As I mentioned, it was developed following a National Academies report and implemented in 2010. Under the Obama administration, the Census Bureau received $5 million to produce it. They moved quickly and have been publishing it since 2011.
Interestingly, the more recent National Academies report evaluating the Supplemental Poverty Measure was funded under the Trump administration. Even though some conservative groups, such as the Heritage Foundation, have criticized the Supplemental Poverty Measure, the administration still approved funding for the evaluation. And despite periodic attempts to cut it, funding for the measure remains in the Census Bureau’s budget.
To me, that’s a success story. When a measure is criticized by people on both sides of the political aisle, it often means it’s doing something meaningful. Conservatives may criticize how the thresholds move, while liberals may argue that childcare costs or the true cost of raising children aren’t fully captured. That kind of debate is healthy.
But beyond definitions, measurement itself is critical. The Census Bureau produces the poverty estimates each year based on the Current Population Survey (CPS), the same survey used to measure unemployment. Poverty data come from an income supplement fielded in February, March, and April. Similarly, food security data—recently defunded—come from a December supplement. So, there is always a risk that supplements could be cut.
The Supplemental Poverty Measure also relies on the Consumer Expenditure Survey from the Bureau of Labor Statistics to set thresholds. There is ongoing collaboration among the Census Bureau, BLS, HUD, Social Security, and others to evaluate and refine the measure. They made significant changes in 2021 and are now considering recommendations from the National Academies report—addressing housing, health insurance, childcare, and whether the household should replace the family as the poverty unit.
Staffing is critical to this work. It’s not just leadership; it’s career staff—GS-12s, GS-13s, GS-14s—who handle the data, run the calculations, and present findings. Right now, it can be difficult for staff to present their work publicly and receive feedback. There can also be pressure when results are politically sensitive.
A memorable example comes from The West Wing, which featured a clip about revising the poverty measure. In the show, officials decided against changing it because it would make poverty appear to increase. That reflects real tension. Changing the measure often depends on political will.
When I worked at the Census Bureau during the Obama administration, there was an instance where leadership wanted to highlight that 100,000 fewer people were in poverty. But that change was not statistically significant—it was within the margin of error. Since the estimate is survey-based, 100,000 people might represent responses from roughly 100 families that were weighed up. We reported that the poverty rate had not changed, but there was pressure to frame it differently.
Political spin happens under every administration. The real concern is whether policymakers attempt to fundamentally alter the measure for political reasons. So far, I haven’t seen that happen with poverty measurement, which is encouraging.
Rosemary Pennington
You’ve talked about the development of the Supplemental Poverty Measure and the proposed Principal Poverty Measure. What do you see as the next big horizon in measuring poverty?
David Johnson
The next big horizon is blending survey data with administrative data to produce better measures.
In the Current Population Survey, many people underreport benefits. For example, when comparing survey responses to administrative records, only about 60% to 70% of people who receive SNAP report it in the survey. Some elderly individuals don’t fully understand their pension or retirement benefits and underreport them. Medicaid participation is underreported. Social Security and wages are generally well reported, but not perfectly.
The Census Bureau is now matching survey responses with administrative records—IRS data, Social Security records, SNAP records from many states, WIC records, housing subsidy data, and more. They’ve created blended data sets combining these sources.
When they do this, poverty rates change. For example, in 2020, using blended data, the Supplemental Poverty Measure was about 2.5 percentage points lower than the published estimate. For the elderly, it was about 4.5 percentage points lower. That suggests many older Americans are not actually in poverty once all benefits are properly counted.
This isn’t primarily a response rate issue. The challenge is that people often cannot accurately report all of their benefits on a survey. It’s not necessarily stigma—SNAP benefits are delivered via electronic cards, and many recipients may not fully understand eligibility rules. Retirement benefits are also complicated.
Improving poverty measurement means improving how we measure resources. That could change both the level and the trend in poverty over time. The challenge is timeliness. Survey data collected early in the year are released by September for the previous calendar year. Administrative data matching can take another year or more. Some blended estimates are only available through 2021 so far.
But I believe blended data are the future—not just for poverty, but for income, consumption, price indexes, and even GDP. Combining survey and administrative data is essential for improving official statistics going forward.