This is the 2026 economy. Thirty-eight source-cited comparisons across
seven sections. Hover any chart for full source citations. Every
number traces to a named source: a federal agency, a journal article,
or a research or advocacy organization.
86%Non-disabled working-age SNAP households with earnings
99%U.S. counties where the maximum SNAP benefit per meal falls short of a modestly priced meal (2024)
46%Drop in U.S. child poverty during 2021 CTC + SNAP expansion
Summary
The Supplemental Nutrition Assistance Program, or SNAP, the federal food benefit once called food stamps, is sometimes described as “just supplemental.” I argue that the word fit the economy of 1975. It does not fit 2026. Fixed costs such as housing and health care ate the margin that once covered the rest of the food bill.
The 1975 federal minimum wage of $2.10 an hour equals $13.11 in 2026 dollars. The 2026 minimum is $7.25. It has not changed since July 2009. A median home cost 3.3 times median household income in 1975 and 5.0 times in 2025.
Of people on SNAP, 39 percent are children, 19 percent are 60 or older, and 9 percent are disabled adults. Among working-age SNAP households without a disability, 86 percent have earnings.
The benefit is about $6.16 per person a day in 2026, by the Center on Budget and Policy Priorities’ figure. In 99 percent of counties, the most SNAP pays per meal was less than the cost of a modestly priced meal in 2024. SNAP provides about nine meals for every one from Feeding America’s food banks.
In 2021 Congress briefly expanded the Child Tax Credit and raised SNAP benefits. Child poverty fell to 5.2 percent that year. The expanded credit ended with 2021. Child poverty went back to 12.4 percent in 2022.
Summary added 2 October 2026.
The argument in one paragraph
“Supplemental” was an accurate word in 1975.
In that economy, households had discretionary income to cover the
rest. Food was about a fifth of a household’s spending. A minimum-wage
worker had roughly twice the real purchasing power of today. A
median home cost 3.3× median income, not 5×. Healthcare
was 8% of GDP, not 18%. Childcare barely existed as a household
expense. Calling SNAP ‘just supplemental’ in 2026 doesn’t
describe a program. It describes an economy we no longer live in.
I. The macroeconomic baseline that made 'supplemental' accurate is gone
The word 'supplemental' described a household with discretionary income to cover the rest. That household existed in 1975. Wages, housing, college, tax structure, worker power, executive pay, childcare, and medical debt have all moved against that assumption.
The 1975 federal minimum wage was $2.10/hour, which is $13.11/hour in 2026 dollars. The actual 2026 minimum wage is $7.25/hour, frozen since July 2009. That is a 45% loss of real purchasing power. A 1975 minimum-wage worker could afford roughly twice what a 2026 minimum-wage worker can. This is the wage floor on which the word 'supplemental' was designed. That floor no longer exists.
A full-time working month is 173 hours. In 1975, a minimum-wage worker could cover a month's 1BR rent in 101 hours, about 2.5 weeks of full-time work. In 2026, the same worker needs 192 hours of minimum-wage labor to cover median 1BR rent. That exceeds a full-time month. Which means full-time work at minimum wage does not, mathematically, pay for a place to live. No amount of 'planning carefully' closes that gap.
In 1975, the median home cost 3.3 times median household income. In 2025, it costs 5.0 times, a 52% worse ratio. A median household today needs 1.7 additional years of total income, relative to 1975, to purchase a median home. That is the housing share of the 'discretionary income was supposed to cover food' assumption. It's gone.
The all-in cost of four-year public in-state college rose 133% in real terms, more than double, between 1975-76 and 2022-23. Total outstanding U.S. student loan debt now stands at $1.84 trillion. Student debt was not a mass household line item in 1975. It is now the second-largest category of consumer debt after mortgages. That is a cost that simply did not exist in the 'supplemental' design era.
Between 1979 and 2023, real wages for the bottom 90% of U.S. workers grew 44%. Real wages for the top 1% grew 182%. Productivity grew 80.9% over roughly the same period, while bottom-90% compensation grew only 29.4%, a 51-percentage-point gap between what workers produced and what they were paid. The 'supplemental' framework assumes that a working household's income grows with the economy. For most households, it has not.
In 1975, the top marginal individual income tax rate was 70% and the top corporate tax rate was 48%. In 2024, those rates are 37% and 21%. The economic baseline that made 'SNAP is just supplemental' accurate was one in which top earners and large corporations paid substantially more into the federal revenue pool that funded the rest of the safety net. We kept the 'supplemental' framing while cutting the taxation that made the rest of the household's income possible.
U.S. labor union density dropped from roughly 25.5% of wage and salary workers in 1975 to about 10% in 2023, and under 6% in the private sector. Union membership is one of the largest explanatory variables for wages tracking productivity. The 'supplemental' framework assumed workers had the collective bargaining leverage of 1975. They don't. 'Plan carefully' advice to a 2026 worker assumes a worker with 1975-level negotiating power. That worker is statistically unusual now.
In 1965 the CEO-to-worker pay ratio at large U.S. firms was 21 to 1. In 1975 it was 27 to 1. In 2023 it was 290 to 1. The economy that made 'supplemental' a reasonable description of SNAP was one in which executive pay was compressed against worker pay by more than an order of magnitude relative to today. The wage gap the 'just plan carefully' argument assumes is not the one that now exists. The one that exists is, instead, the largest executive-to-worker compensation gap in U.S. history.
Center-based childcare for a single child costs $10,900 to $17,000 per year in 2024 depending on state, with a national median around $13,950. In 1975, this expense category barely existed at scale: women's labor force participation was ~46% (vs ~58% in 2024), and most childcare happened inside the family or in unpaid, informal arrangements. A net-new multi-thousand-dollar annual household line item materialized between 1975 and now. The 'supplemental' framework's 30-percent-for-food math assumed a household with no childcare bill. That household is not the median household today.
In a survey of 910 people who filed for bankruptcy in the United States from 2013 to 2016, 66.5% said medical bills or illness-related loss of work contributed to their bankruptcy (Himmelstein et al., American Journal of Public Health, 2019). Medical bills were cited by 58.5%, and work lost to illness by 44.3%. Some cited both. The survey asked about contributing causes. That is not the same as the main cause. The response rate was 29.4%. A study that linked California hospital records to credit reports found that hospital stays triggered fewer than 5% of the bankruptcies in its sample (Dobkin et al., American Economic Review, 2018). Its authors call that likely a lower bound on all medical bankruptcies. It leaves out illness that never led to a hospital stay. In a KFF survey in early 2022, 41% of U.S. adults said they had debt from medical or dental bills. In 2022, KFF Health News and NPR put the total at more than 100 million people. Medical debt is one more claim on the 'discretionary income' that the 'supplemental' framework presumed households had.
✦
The economic baseline that made ‘supplemental’ an accurate word in 1975 no longer exists, in any of its components.
II. Federal priorities: the hypocrisy is in the budget
We DO have money. The argument is only about where we decide to send it.
The FY2026 defense budget is roughly $1.05 trillion. Combined FY2026 SNAP and WIC spending is roughly $115 billion. For every $1 the federal government spends on food assistance, it spends roughly $9 on defense. Calling the $115 billion bar 'supplemental' while the $1.05 trillion bar is not called 'supplemental' is a rhetorical choice, not a budget description.
1975 defense spending was $85.8 billion nominal, approximately $536 billion in 2026 dollars. 2026 defense spending is about $1.05 trillion. In real, inflation-adjusted dollars, the United States now spends roughly twice what it spent on the military in 1975. Over the same five decades, SNAP's real per-person purchasing power was frozen for 45 years (until the 2021 Thrifty Food Plan reset). 'We can't afford the safety net' does not hold when you look at what we chose to fund twice over.
The United States appropriated roughly $12.6 billion in military aid to Israel in FY24-FY25 and roughly $110.7 billion in defense-specific aid to Ukraine across FY22-FY24. SNAP reductions enacted in July 2025 will remove approximately $187 billion from domestic food assistance over ten years, by the Congressional Budget Office's estimate. Foreign military aid is treated as unquestionable budget arithmetic. A domestic program that feeds 42 million Americans is treated as 'supplemental' and therefore optional.
The United States is the single largest donor to the United Nations World Food Programme, contributing approximately $3.5 billion in 2023 to feed hungry people in other countries. At the same time, the SNAP reductions enacted in July 2025 will remove an average of roughly $18.7 billion per year from food assistance to Americans. It is possible to fund both. That has been our policy for decades. The argument that we cannot is a new political position, not a budget constraint.
Between 2009 and 2011, federal and state governments spent an average of $152.8 billion a year, in 2013 dollars, on four public assistance programs for working families (Jacobs, Perry and MacGillvary, UC Berkeley Labor Center, 2015, funded by the Service Employees International Union). The four are Medicaid with the Children's Health Insurance Program, the Earned Income Tax Credit, SNAP, and cash assistance under Temporary Assistance for Needy Families. The federal share was $127.8 billion, including $26.7 billion of SNAP. The states paid $25 billion. A working family here is any family with a member who works at least twenty-seven weeks a year, ten or more hours a week. The figure covers working families in general. It is not a count of any one employer's staff. An earlier study by the Labor Center and the University of Illinois found that working families accounted for 73% of enrollments in the same four programs over 2007 to 2011. Data on particular employers are thinner. The Government Accountability Office could get February 2020 records from eleven states. Walmart ranked among the top four employers of working adults on SNAP, or on Medicaid, on all fifteen state program lists. McDonald's and Amazon appeared on most of them. No single employer had more than 4% of working enrollees in any of those states. GAO also notes that a larger employer is more likely to make such a list than a smaller one, all else equal (GAO-21-45, October 2020). In the Labor Center's words, taxpayers bear “a significant portion of the hidden costs of low-wage work.” The federal government also supports farm producers: crop insurance premium subsidies for 2023 crops came to about $11.3 billion (USDA Risk Management Agency). In July 2025 Congress cut SNAP, one of those four programs.
Walmart captures roughly 25-30% of all SNAP dollars spent in the United States, approximately $13 billion per year, more than Kroger and Albertsons/Safeway combined. Walmart is also consistently among the top employers of workers who themselves rely on SNAP. The company receives billions of SNAP dollars through its grocery registers AND depends on SNAP to feed its own workforce. Cutting SNAP reduces Walmart's revenue AND forces Walmart to raise wages. Neither of those is happening. The program is the workaround.
The combined 2024 net income of the five largest U.S. health insurers (UnitedHealth, Cigna, Elevance, Humana, CVS/Aetna) is approximately $38.4 billion, roughly 38% of the entire annual SNAP program budget. Cigna alone earned approximately $3.4 billion in net income on $247 billion in revenue. The 'we cannot afford SNAP' argument sits uneasily next to the fact that a single insurance sector generates enough profit to fund more than a third of the program. The money is there. It is at the insurers.
✦
For every $1 of SNAP + WIC, the federal government spends roughly $9 on defense. The word for the smaller bar is ‘supplemental.’ The word for the larger bar is not.
III. Who SNAP and WIC actually serve
The public image of the 'SNAP recipient' bears almost no resemblance to the actual SNAP population.
39% of SNAP participants are children under 18. 19% are seniors 60 and over. 9% are disabled adults. Of the remaining working-age adults, 86% of non-disabled SNAP households have at least one person reporting earnings. The mental model of 'SNAP recipient' in public discourse, that of a non-working, non-elderly, non-disabled adult, is a minority of actual SNAP participants. 'Plan carefully' advice is being aimed at households that are, overwhelmingly, kids, seniors, and working parents.
Black households experience food insecurity at 22.4%, more than double the rate of white households (9.6%). Hispanic households are close behind at 19.3%. The 'supplemental' argument treats SNAP as a uniform program for a uniform population. The program lands on populations with already-unequal access to food, wealth, wages, and grocery-store density. 'Just plan carefully' is an argument that imagines a household at the white rate and ignores the household at the Black rate.
9.3% of Americans aged 60 and older, roughly 5.5 million seniors, experienced food insecurity in 2021. One in five low-income seniors report they have skipped food to pay for heating or medication. This is the population being told SNAP is 'just supplemental.' For many of these seniors, SNAP is the only thing standing between them and a forced choice between food, medication, and heat.
Half of new SNAP entrants leave the program within 12 months. Three-quarters are off within 24 months. The cultural image of SNAP, chronic and multi-decade dependency, is not what the data show. SNAP is used as a short-term bridge during job loss, illness, divorce, or disability onset, and households cycle off when circumstances improve. The 'just plan carefully until you get on your feet' implied criticism is describing what SNAP participants already do.
Peer-reviewed evidence from AHRQ's 2022 systematic review in Annals of Internal Medicine finds WIC participants have 14% lower infant mortality, 10% lower preterm birth rates, and up to 24% lower low-birth-weight rates than non-participants. Mathematica's evaluation for USDA FNS estimates every $1 invested in WIC saves $1.77 to $3.13 in Medicaid costs in the first 60 days after birth alone. Cutting WIC is not a cost-saving measure. It costs more downstream than it saves upstream, in dollars, and in children.
✦
Thirty-nine percent of SNAP participants are children. Nineteen percent are seniors. Nine percent are disabled adults. ‘Plan carefully’ is advice aimed at a demographic that is, statistically, not on SNAP.
IV. What the programs actually do
The efficiency, the evidence, and the one natural experiment that proved they work.
Feeding America, the entire U.S. food bank system, distributed approximately 6 billion meals in 2024. SNAP's meal-equivalent is approximately 54 billion per year. For every one meal Feeding America delivers, SNAP delivers nine. A 10% cut to SNAP would require the food-bank system to nearly double its national output to fill the gap. A larger cut would require tripling it. Neither looks achievable for a charity network of that size. 'Private charity can replace SNAP' is not a plan. It is arithmetic that does not work.
In 99% of U.S. counties, the maximum SNAP benefit per meal did not cover the cost of a modestly priced meal in 2024 (Urban Institute). These figures predate the cuts enacted in July 2025. The benefit is currently calibrated to a food plan whose real value was held flat from 1975 to 2021, adjusted each year only for inflation. 'Supplemental' implies the benefit covers something, and that a household covers the rest. The coverage math on the first part has already failed in nearly every county in the country.
Every federal dollar of SNAP benefits generates approximately $1.54 in GDP. It is one of the highest-multiplier federal programs because the money gets spent almost immediately, mostly at grocery stores in the recipient's own community. Every federal dollar of tax cuts to top earners generates approximately $0.39 in GDP, because high earners save a much larger share. On the pure math of economic stimulus, cutting SNAP and redirecting to tax cuts is one of the most efficient ways to shrink the economy.
SNAP trafficking, the share of benefits redirected for cash, runs at roughly 1.5 to 2%. Medicaid improper payments run at ~5.1%. Medicare fee-for-service improper payments run at ~7.7%. SSI improper payments run at ~10.6%. The Department of Defense has failed its annual audit seven consecutive times on $824 billion in assets. It cannot account for its own spending at the level SNAP can. The 'SNAP is full of fraud' argument is not supported by the comparative data.
In 2021, Congress briefly expanded the Child Tax Credit and increased SNAP benefits in response to COVID. The supplemental-poverty-measure child poverty rate fell from 9.7% in 2020 to 5.2% in 2021, a 46% reduction and the lowest rate on that measure's record, as the Census Bureau first published it. When the expanded CTC expired at the end of 2021, child poverty more than doubled, to 12.4% in 2022. (A corrected Census series released in August 2026 puts the three rates at 9.6%, 5.1% and 12.0%; the pattern is the same.) This is the empirical answer to 'do these programs work.' They work when we fund them. We know this because we tested it and then chose to end it.
The top five U.S. health insurers (UnitedHealth, Cigna, Elevance, Humana, CVS/Aetna) posted combined 2024 net income of approximately $38.4 billion, roughly 38% of the entire annual SNAP budget. Cigna alone posted approximately $3.4 billion in net income on $247 billion in revenue. The profit of a single sector of the healthcare economy in a single year would finance more than a third of SNAP. The framing that 'we cannot afford' SNAP ignores that the largest healthcare players in the country afford their own margins without difficulty.
✦
When Congress expanded the safety net in 2021, U.S. child poverty fell 46% in a single year. When the expansion expired, child poverty doubled back within twelve months. The programs work. The question is whether we fund them.
V. Structural cost explosion: where the 'discretionary income' went
Households are not less frugal than they were in 1975. Fixed costs ate the margin.
In 1970, U.S. health spending was 6.9% of GDP. In 2024, it is 18.0%, more than two and a half times larger as a share of the economy. Per-capita health spending, in constant 2024 dollars, rose from $2,208 to $15,474, a 7× real increase. 'Supplemental' was an accurate word in an economy where healthcare took less than one-fourteenth of each person's economic output. It is not accurate in an economy where healthcare takes nearly one-fifth. The margin to supplement has been absorbed.
Food fell from 19.3% of household spending in 1972-73 to 12.9% in 2024. Housing grew from 30.8% to 33.4%. Healthcare (household out-of-pocket) grew from 6.5% to 7.9%. And the true national-accounts healthcare share of household costs is far higher, because employer-paid premiums are hidden in wages. Pensions and personal insurance grew from 5.0% to 12.4%, as pensions disappeared and 401(k) savings became a worker responsibility. The benefit formula still expects a household to put 30 percent of its net income toward food, and that is not how U.S. households spend money anymore.
VI. Food insecurity as a clinical and public-health problem
The ground-level clinical reality that 'plan carefully' advice is issued against.
19 million Americans live in food deserts as defined by the U.S. Department of Agriculture: low-income census tracts in which the nearest supermarket is more than 1 mile away (urban) or more than 10 miles away (rural). 'Just plan carefully and shop at Aldi' is not an available strategy when Aldi is fifteen miles down the highway and the household's transportation options are limited. The advice presumes a choice architecture that does not exist for 19 million of the people it is aimed at.
Food swamps are areas where restaurants, fast food and convenience stores heavily outnumber the stores that sell fresh food. A food desert has no grocery store nearby. A food swamp may have one. It is outnumbered. A national count published in June 2026, led by researchers at the American Cancer Society, classed 88.5 percent of U.S. census tracts as food swamps in 2023, up from 80.2 percent in 2003 (Wiese and colleagues, American Journal of Public Health, 2026). That measure is broad. It counts every restaurant as a less healthy outlet, sit-down ones included. A tract counts as a food swamp when supermarkets, grocery stores and other healthy food retailers make up some, but fewer than one in three, of the food outlets in the tract and the half mile around it. Tracts with none count as food deserts. No cutoff is universally accepted, the authors note. An earlier study of more than 3,000 counties used 2009 Department of Agriculture data. It found that the presence of a food swamp predicted adult obesity rates better than the absence of full-service grocery stores (Cooksey-Stowers, Schwartz and Brownell, International Journal of Environmental Research and Public Health, 2017). Advice to 'plan carefully' and 'cook from scratch' does not account for the mix of stores around a household.
Peer-reviewed research finds nutrient-dense food costs approximately $1.45 per 100 calories, compared to $0.55 per 100 calories for ultra-processed food, a 2.6× price difference per calorie. SNAP benefits are calculated on calories, not nutrients. The math forces a rational household to buy the cheapest calories available. That household is then blamed for eating cheap calories. 'Plan carefully' and 'make healthier choices' does not change the price tag on a head of broccoli.
Among low-income adults aged nineteen to sixty-four, diabetes was about twice as common in households with very low food security as in households with high food security in 2019 to 2022: 9.6% against 4.5%, after adjusting for age, income, insurance and other factors (USDA Economic Research Service, 2024). Among Americans sixty-five and older, 16.8% of those in food-insecure families did not take their medication as prescribed to save money in 2021 and 2022, against 2.8% of those in food-secure families, about six times the rate (National Center for Health Statistics, 2024). Adults who were food insecure in 2011 had 1.47 times the rate of emergency department visits and 1.47 times the rate of hospitalizations of food-secure adults over 2012 and 2013, after adjusting for income, insurance and other factors (Berkowitz et al., American Journal of Managed Care, 2018). Berkowitz and colleagues also estimated about $52.9 billion in excess health care spending associated with food insecurity in 2016 (Berkowitz et al., Preventing Chronic Disease, 2019). Cutting SNAP may shift some costs to health care. In one national study, low-income adults who took part in SNAP had about $1,400 lower estimated health care spending a year in 2012 and 2013 than other low-income adults who did not, after adjustment (Berkowitz et al., JAMA Internal Medicine, 2017). The study shows an association, not cause and effect. The difference is also smaller than the benefit itself, which averaged about $133 per person a month in fiscal 2013, about $1,600 a year (USDA Food and Nutrition Service).
For cancer patients, food insecurity is not an abstract social issue. In several studies it is linked to whether patients keep up with treatment and to how long they live. In a national cohort of 5,603 U.S. adults aged forty and older with a cancer history, interviewed in 2011 and 2012 and followed through 2019, those who were food insecure died of any cause at 1.28 times the rate of those who were food secure. That figure accounts for age, sex, race and ethnicity, region, income, smoking, drinking, weight and other illnesses (Lin et al., JAMA Health Forum, 2025). Adjusted only for age and the survey year, the figure was 1.66. In a survey of 394 New Mexico survivors of breast, colorectal or prostate cancer, aged twenty-one to sixty-four and diagnosed between 2008 and 2016, some were food insecure in both the year before and the year after diagnosis. They had three times the odds of skipping, delaying or changing a prescription as those who stayed food secure (McDougall et al., JCO Oncology Practice, 2020). Not every study finds a link. Among 1,098 Latino and Black patients at New York City cancer clinics, needing help with food was not significantly linked to self-reported missed chemotherapy or radiation appointments (Costas-Muñiz et al., Ethnicity and Health, 2016). In a small, preliminary trial of 117 food-insecure patients at New York City safety-net cancer clinics, none of them receiving SNAP, some were given a monthly $230 food debit card on top of a clinic food pantry. The amount was based on the average SNAP benefit for one person when the study began. They completed treatment more often than those given the pantry alone, 94.6 percent against 77.5 percent (Gany et al., Journal of Clinical Oncology, 2022). SNAP and WIC are not 'nice-to-haves' in cancer care.
✦
Food insecurity is a clinical problem. Not a lifestyle difficulty. The roughly $53 billion in excess health care spending associated with food insecurity in 2016 belongs on the same ledger as any SNAP cut.
VII. Historical and international context
The long view: bipartisan history of the program, and how the U.S. compares to peer nations.
SNAP (previously Food Stamps) was expanded by Presidents Nixon, Carter, Reagan, G.W. Bush, and Biden, administrations of both parties across six decades. The program had a bipartisan political foundation until the late 2010s. The framing of SNAP as a controversial, ideologically contested program is recent, not a longstanding American debate. The architects of the program, Republican and Democrat, considered it core to both rural farm-economy stability and urban anti-hunger policy.
Among OECD peer nations, the United States has one of the highest child poverty rates, roughly 20.9% by relative-income measures, compared to 3.6% in Finland, 6.0% in Norway, and 11-12% in Germany, France, the UK, and Canada. The U.S. is not generous to its poorest households by peer-nation standards. It is, on this metric, one of the least generous wealthy nations in the world. The 'we can't afford it' argument is even harder to sustain when every economically comparable peer affords substantially more.
U.S. public social spending as a share of GDP is 18.7%, lower than France (31.0%), Finland (29.8%), Italy (30.0%), Germany (25.9%), and the UK (22.1%), and roughly comparable to Canada (17.3%). The United States is not a generous welfare state by any peer-nation standard. Every comparable wealthy country spends more of its economy on social protection than the U.S. does. The 'we are being overly generous' framing of SNAP critique is an assertion that cannot survive international comparison.
Correction, 2 October 2026. My own review of this piece found several errors, and I have corrected each one in place. Child poverty on the Census Bureau’s supplemental measure fell to 5.2 percent in 2021 from 9.7 percent in 2020, not from 12.5 percent. The 12.5 percent was the 2019 rate. The 46 percent drop is measured from 2020. The piece also called it the largest single-year drop in recorded history. The Bureau’s own claim is narrower: 2021 was the lowest rate on that measure’s record. A corrected Census series released in August 2026 gives 9.6, 5.1 and 12.0 percent for 2020 to 2022. The pattern is the same. Health spending was 6.9 percent of the economy in 1970 and 7.9 percent in 1975, not 6 percent (Centers for Medicare and Medicaid Services). Its share grew about two and a half times by 2024, not three. Food took about a fifth of household spending in 1972 and 1973 (19.3 percent, Bureau of Labor Statistics), not a third. The benefit formula does not assume a third either. It expects a household to put 30 percent of its net income toward food. The SNAP cuts of about $187 billion over ten years that I called proposed had already been enacted, on 4 July 2025, nine months before this piece ran (Congressional Budget Office estimate, 21 July 2025). The Urban Institute finding is that the maximum benefit per meal fell short of the cost of a modestly priced meal in 99 percent of counties in 2024. It does not mean SNAP cannot buy a single meal. Two of these corrections weaken supporting points. Health spending grew less than I wrote, and food was a smaller share of household spending in the 1970s. One makes the case more pressing. The cuts were already law when this was published. The central argument survives. Smaller changes: the ten-year figure for the cuts is now $187 billion and the yearly figure is the budget office average of $18.7 billion, the line about my sources and the wording on how the food plan was adjusted for inflation were corrected, a sentence about clinical experience was removed, the closing description of my practice’s services was updated, and a private person’s first name was removed from one sentence.
Later the same day, a check of the sources this piece cites found more figures I had described wrongly. They follow in the order they appear. The 66.5 percent comes from a survey of 910 people who filed for bankruptcy from 2013 to 2016 (Himmelstein and colleagues, American Journal of Public Health, 2019). It is the share who said medical bills or illness-related loss of work contributed to their bankruptcy. It is not the share of bankruptcies primarily caused by medical bills. Medical bills were cited by 58.5 percent. I also called the paper the most-cited peer-reviewed analysis. The journal classes it as an editorial, and I had no basis for “most-cited.” Both descriptions are gone. A study that used credit reports found a far lower figure for one cause, hospital stays. The piece now says so.
The $152.8 billion from the UC Berkeley Labor Center is what federal and state governments spent a year, on average, from 2009 to 2011, in 2013 dollars. It went to working families in general, through four programs: Medicaid with the Children’s Health Insurance Program, the Earned Income Tax Credit, SNAP and cash assistance. Housing aid was not among them. The figure was not limited to workers at large employers such as Walmart, McDonald’s or Amazon. And $25 billion of it was state money, not federal. The $17.7 billion I gave for farm commodity subsidies in 2023 is not a commodity subsidy total. It is close to what crop insurance paid in indemnities on 2023 crops. Payments for 2023 crops under the two main commodity programs, Agriculture Risk Coverage and Price Loss Coverage, were estimated at more than $447 million (Farm Service Agency). I have removed that figure. The charts for the bankruptcy and spending figures have been redrawn. These corrections weaken supporting points. The $152.8 billion is also older than the piece implied, and payments for 2023 crops under Agriculture Risk Coverage and Price Loss Coverage were far smaller than the figure I gave.
The piece said that about 17 percent of U.S. census tracts are food swamps, and credited the figure to Cooksey-Stowers, Schwartz and Brownell (2017). That study measured counties, not census tracts. The figure is not in it. I searched the cited paper, the national food-environment studies I could find, and CDC and USDA reports, and could not trace it. The piece also said food swamps have little or no access to a full-service grocery, and a chart label called them low-income areas. The first describes a food desert. Neither study cited here uses income to define a food swamp. A food swamp is an area where less healthy outlets heavily outnumber stores that sell fresh food. It may have a grocery store. A national count published in June 2026, led by researchers at the American Cancer Society (American Journal of Public Health), classed 88.5 percent of census tracts as food swamps in 2023. Its measure is broad. It counts every restaurant as a less healthy outlet. The chart and text now give that figure with its definition. The number is larger, but the point is weaker. By that measure most of the country qualifies, so it says little about low-income neighborhoods in particular.
Two figures in the section on food insecurity and health were wrong. The 47.9 and 10.5 percent for prescriptions came from a Canadian study (Men and colleagues, CMAJ Open, 2019). There they are the shares of people who were food insecure among those who did not take their medicines as prescribed because of cost (47.9 percent) and among those who did (10.5 percent). They are not how often food-insecure and food-secure Americans skip medicines. So the 4.56 times was wrong. The National Center for Health Statistics reports that in 2021 and 2022, 16.8 percent of Americans sixty-five and older in food-insecure families did not take their medication as prescribed to save money, against 2.8 percent in food-secure families. That is about six times the rate. For diabetes, I could not find a source that gives 16 and 9 percent as diabetes rates among food-insecure and food-secure Americans. The only match I found is a 2021 letter in Diabetes Care, where they are the shares of adults with and without diabetes who were food insecure. Either way, no source I found supports the 1.78 times, and none of these sources measured who develops diabetes. The Economic Research Service of the Department of Agriculture puts diabetes at 9.6 percent of low-income adults aged nineteen to sixty-four in households with very low food security, against 4.5 percent in households with high food security. That is for 2019 to 2022, after adjustment, and about twice as common. The chart and the text now use these figures. Both cover narrower groups than the text claimed, and the diabetes figure shows how common the disease is, not who develops it, so these two corrections narrow the point without reversing it.
The section on cancer patients was also wrong. It said food-insecure cancer patients had about 2.5 times the risk of interrupted treatment and missed chemotherapy appointments, citing Gany and colleagues in the Journal of Oncology Practice in 2014. I could find no such paper. I cannot trace where the figure came from. Francesca Gany’s 2014 paper measured how common food insecurity was among low-income cancer patients in New York City. A later study from her group, of Latino and Black patients there, found no significant link between needing help with food and missed chemotherapy or radiation appointments (Costas-Muñiz et al., Ethnicity and Health, 2016). I could not find a source for the claim that hospital readmissions ran 1.4 times higher. I have removed it. The figure of about 1.7 for deaths matches a 2025 national study’s estimates before income and health were taken into account (1.66 and 1.69). With income, health habits and other illnesses taken into account, the same study gives 1.28 (Lin et al., JAMA Health Forum). The section and its chart now report that study, a 2020 study of skipped and delayed prescriptions, the 2016 study, and what the 2022 trial by Gany and colleagues measured. This correction weakens another supporting point: the evidence on cancer patients is weaker and narrower than I wrote.
Smaller changes: the medical debt figure is now dated to its 2022 survey. Two unsourced comparisons, with other countries and with 1975, were removed from the medical bankruptcy passage and its chart. The Labor Center brief is now noted as funded by the Service Employees International Union. Crop insurance premium subsidies for 2023 crops are now given as about $11.3 billion (Risk Management Agency), not $11.5 billion. One more sentence that still described the July 2025 cuts as under debate was corrected. An estimate that 14,500 Walmart employees were in households on SNAP was removed, because I could not confirm its source. The UC Berkeley Labor Center was also taken off that chart’s source line. Its brief names no company. The emergency visit figures are now credited to Berkowitz and colleagues in the American Journal of Managed Care (2018), not the Journal of General Internal Medicine. The $52.9 billion estimate is now credited to a paper Berkowitz led in Preventing Chronic Disease (2019), not to Seligman. It is now described as spending associated with food insecurity in 2016, not a yearly cost attributable to it. A claim that cutting SNAP does not save money, and an unsourced claim that it shifts the cost to the clinic at a worse dollar ratio, were replaced with a more cautious sentence and a national study of SNAP and health care spending (Berkowitz and colleagues, JAMA Internal Medicine, 2017). The lower spending that study found was smaller than the average benefit. None of these later corrections strengthens the case. Several weaken it. The central argument survives.
Evidence over opinion. Always.
If this was useful, consider sharing the link with a clinician
or policymaker who is still working from the 1975 assumptions.
Nutrition services at Vitae Arete focus on GLP-1 and metabolic
health, built on the same evidence-first posture applied here.
