Skip to content

Investment vs. spending · November 1, 2026

What it looks like now vs. what it looks like instead

What it looks like now is federal spending on Social Security, Medicare, Medicaid, and other means-tested programs since fiscal 1946. What it looks like instead is the same dollars in the S&P 500, dividends reinvested, valued on November 1, 2026. Figures are nominal and in September 2026 dollars, the latest CPI-U month. An illustration of opportunity cost, not a portfolio the government held.

What it looks like now · spending

Nominal

$66.52 trillion

$195,635 per person

September 2026 dollars

$107.94 trillion

$317,472 per person

FY2026 outlays $3.99 trillion. OMB estimate. That year is already in today’s prices, so the inflation-adjusted amount matches.

What it looks like instead · invested

$1.17 quadrillion

$3,440,197 per person on November 1, 2026. Nominal and September 2026 dollars are the same figure: the portfolio is already priced in today’s market. Compounding the inflation-adjusted outlays at real total returns lands on this same value.

17.6× the nominal spending · 10.8× the inflation-adjusted spending

Fiscal position

The hypothetical stock next to the debt

Debt held by the public on 2026-11-04 was $32.77 trillion. Gross debt was $40.55 trillion. The invested value is about 36 times debt held by the public and about 29 times gross debt, in either nominal or September 2026 dollars, because the portfolio is already a today’s-dollar price. The spending beside it is $66.52 trillion nominal and $107.94 trillion in September 2026 dollars. The multiple is 17.6× nominal and 10.8× after inflation. The sections below follow the dollar toggle. Debt is a current-dollar stock either way.

Spent · September 2026

$107.94 trillion

$317,472 per person

Invested · Nov 1, 2026

$1.17 quadrillion

$3,440,197 per person · 10.8× spending

Debt held by the public

$32.77 trillion

2026-11-04 · $96,368 per person

OASDI 75-year unfunded obligation

$26.6 trillion

September 2026 dollars. Present value was 2025-01-01. Not a debt stock.

On this comparison, the same program dollars would be a larger stock than today’s debt. That is not a finding that the Treasury could have bought the S&P 500 in that size, and it does not subtract the benefits, care, and income people received when the money was spent. The $25.1 trillion figure is the present value of future OASDI shortfalls under the 2025 Trustees’ intermediate assumptions. It is not added to today’s gross debt here, and it is not subtracted from the hypothetical stock. Sources are linked at the bottom.

Four programs

Spending, the published outcome, and the invested value

One card each. The total adds the four OMB lines and does not count any dollar twice. Medicaid is its own line, not part of other means-tested entitlements. Federal grants to nonprofits are on the next card and are not in this total. Per person uses 340,000,000 people, the population used on the rest of this site.

OASDI

Social Security

Old-age, survivors, and disability insurance. The paid-in benefit is an annuity. The trust-fund gap is a financing fact, published by the Trustees.

FY2026
$1.67T
Estimate
Cumulative
$50.67T
$149,018 / person
Nominal $29.14T
Invested now
$704.59T
$2,072,324 / person
Nominal $704.59T
  • Combined trust-fund reserves

    Under the 2025 Trustees’ intermediate assumptions, combined OASI and DI reserves are projected to deplete in 2034. Continuing income would then cover about 81% of scheduled benefits. Considered alone, OASI reserves deplete in 2033.

    Social Security Trustees

  • 75-year unfunded obligation

    $25.1 trillion. That is the open-group unfunded obligation for OASDI over 2025–2099, in present-value dollars discounted to January 1, 2025. It is not a cash debt stock.

    Unfunded obligation

  • Poverty at 65 and older, 2024

    Official poverty rate 9.9%. Supplemental Poverty Measure 15.0%. Social Security moved 28.7 million people out of SPM poverty in 2024, the largest program effect Census measured.

    Census 2024

Hospital insurance and supplementary medical insurance

Medicare

Federal Medicare outlays from OMB. National health spending and life expectancy are the published outcome context, and they are not the same thing as federal outlays.

FY2026
$1.06T
Estimate
Cumulative
$25.16T
$73,991 / person
Nominal $16.45T
Invested now
$193.99T
$570,552 / person
Nominal $193.99T
  • HI trust fund

    The 2025 Medicare Trustees project Hospital Insurance trust-fund depletion in 2033, three years earlier than in the 2024 report. At depletion, continuing income would cover 89% of HI costs.

    Medicare Trustees

  • National health spending, 2024

    CMS national health expenditures were $5.3 trillion, $15,474 per person, and 18.0% of GDP. Medicare spending in those accounts, which is not the same as federal outlays, was $1,118.0 billion, 21% of national health spending.

    CMS

  • Life expectancy at birth

    78.8 years in 2019, 78.4 in 2023, and 79.0 in 2024 (CDC). The 2024 figure is above 2019. This is the whole population, not Medicare beneficiaries alone.

    CDC, 2019 · CDC, 2024

Grants to states

Medicaid

Federal Medicaid outlays. National Medicaid spending in the health accounts also includes state funds. The uninsured rate is a coverage outcome for the whole population, not a Medicaid-only score.

FY2026
$714.90B
Estimate
Cumulative
$15.18T
$44,661 / person
Nominal $10.29T
Invested now
$101.85T
$299,545 / person
Nominal $101.85T
  • National Medicaid spending, 2024

    CMS counts Medicaid spending, federal and state together, at $931.7 billion in 2024, 18% of national health expenditures. The spending line on this card is federal outlays only.

    CMS

  • Uninsured, all year

    In 2024, 8.0% of people (27.1 million) had no health insurance at any point in the year, Census CPS ASEC. That rate was not statistically different from 2023, when it was also 8.0% (26.4 million).

    Census, coverage

SNAP, SSI, EITC, and the rest of the OMB list

Other means-tested programs

Table 8.1’s other means-tested entitlements, not including Medicaid. The list in the Historical Tables introduction includes SNAP, child nutrition, Pell grants, CHIP, TANF, SSI, the refundable EITC and child credit, and other income-tested accounts.

FY2026
$544.80B
Estimate
Cumulative
$16.93T
$49,803 / person
Nominal $10.64T
Invested now
$169.24T
$497,776 / person
Nominal $169.24T
  • Official poverty rate

    10.5% in 2019, 11.5% in 2022, 11.1% in 2023, and 10.6% in 2024 (35.9 million people in 2024). Census official poverty uses pre-tax money income. It does not count SNAP, refundable tax credits, or housing assistance.

    Census 2019 · Census 2022 · Census 2023 · Census 2024

  • Supplemental Poverty Measure

    11.8% in 2019, 12.4% in 2022, 12.9% in 2023, and 12.9% in 2024. The SPM counts noncash benefits and refundable credits, and subtracts taxes, work expenses, and medical out-of-pocket costs. Those transfers lower measured SPM poverty relative to a measure that ignores them.

    Census SPM · Census 2023 · Census 2024

Four programs together

No double counting

FY2026 outlays
$3.99 trillion
OMB estimate
Cumulative since FY1946
$107.94 trillion
$317,472 per person
Invested · Nov 1, 2026
$1.17 quadrillion
$3,440,197 per person · 10.8×

FY2025, the last actual year in Table 8.1, was $3.75 trillion across the four lines ($11,026 per person). FY2026 in that table is still an estimate of $3.99 trillion.

Federal dollars only

Grants to nonprofits

USAspending obligations for block grants, formula grants, project grants, and cooperative agreements whose recipient type is nonprofit. Not private donations. Not total nonprofit assets. Not added to the four-program total, because some of these awards are already inside those outlays and because subawards are undercounted. The series returned by the API starts in FY2008. Earlier years are not estimated.

FY2025 obligations
$42,702,598,178
Latest complete year in this pull
Cumulative FY2008–FY2025
$798.63B
$2,349 per person
Invested · Nov 1, 2026
$2.40 trillion
$7,052 per person
Multiple vs. its spending
3.0×
September 2026 dollars
Federal grant and cooperative-agreement obligations to nonprofit recipients by fiscal year
Fiscal yearNominal obligations
FY2008$12,085,496,744
FY2009$25,494,495,521
FY2010$32,291,704,515
FY2011$26,119,763,868
FY2012$28,316,525,682
FY2013$27,065,432,584
FY2014$29,990,839,495
FY2015$32,377,732,828
FY2016$34,791,603,246
FY2017$29,686,143,163
FY2018$29,553,570,935
FY2019$30,127,059,946
FY2020$35,507,822,511
FY2021$41,124,467,729
FY2022$40,409,543,021
FY2023$45,908,590,853
FY2024$73,168,907,462
FY2025$42,702,598,178

On November 6, 2026, USAspending showed about $7.17 billion of FY2026 obligations for this filter, well below a finished year. The year is omitted rather than treated as complete.

Estimate

Government-funded share of 501(c)(3) assets

IRS SOI Table 1, tax year 2022, Form 990 filers that are 501(c)(3)s. Government grants (Form 990 Part VIII line 1e) were $346.33B, which is 10.07% of $3.438 trillion in total revenue. Total assets were $7.010 trillion. Net assets were $4.503 trillion. Applying the revenue share as a proxy gives about $706.23B of assets and $453.65B of net assets. Amounts are scaled from December 2022 prices into September 2026 dollars. The 10.07% share does not change.

Line 1e includes state and local governments, so the share is not a federal share. USAspending’s federal series above is much smaller than $305 billion because it misses pass-through subawards and uses a narrower award filter. This proxy is not a market value of a federal claim on nonprofit balance sheets.

November 1 record

Baseline, then every November 1 from 2020 through 2026

November 1, 1945 is the baseline, before FY1946 outlays. Each later row’s “that year” is the fiscal year that ended on September 30 just before that November 1. Cumulative spending runs from FY1946, including the 1976 transition quarter. The change column is since the previous row, so the 2020 change is the whole road from 1945, and each row after that is one November to the next.

The table and chart are in September 2026 dollars.

November 1 snapshots of program spending and the hypothetical S&P 500 value, 1945 and 2020 through 2026
November 1That year’s spendingCumulative spentInvested valueSpent per personInvested per personInvested, since prior row
1945—$0.00T$0.00T$0$0—
2020$3.83T$85.01T$585.68T$250,024$1,722,574$585.68T
2021$4.18T$89.19T$790.79T$262,312$2,325,862$205.12T
2022$3.62T$92.80T$630.88T$272,946$1,855,521−$159.92T
2023$3.63T$96.43T$675.69T$283,612$1,987,337$44.82T
2024$3.63T$100.06T$911.43T$294,293$2,680,669$235.73T
2025$3.89T$103.95T$1,078.83T$305,731$3,173,033$167.40T
2026$3.99TEstimate$107.94T$1,169.67T$317,472$3,440,197$90.84T

Cumulative spent vs. invested value

Every November 1 from 1946 through 2026, in September 2026 dollars. On a linear scale the spending line sits near the axis because the invested line is about 10.8× as large. The log scale keeps both lines readable.

Loading chart…

Start somewhere

Divert 1% to 5% of one year’s program outlays

FY2025 outlays for the four programs were $3.75 trillion. This projection puts 1% of that amount, $37,487,000,000 a year ($110 per person), into Freedom Accounts at the end of each year. One column compounds at the nominal geometric average, 10.02%. The other compounds at the real geometric average, 6.78%. A projection, not a promise.

10.02% nominal

Future nominal dollars

10 years

$597.97B

$1,759 per person

20 years

$2.15 trillion

$6,328 per person

30 years

$6.19 trillion

$18,199 per person

6.78% real

Future purchasing power, in September 2026 dollars

10 years

$512.52B

$1,507 per person

20 years

$1.50 trillion

$4,412 per person

30 years

$3.40 trillion

$10,007 per person

Contributions are added at each year-end, the same convention as the calculator on this site, so the last contribution has not yet earned a return. 10.02% is the geometric average of nominal S&P 500 total returns, 1928–2025. 6.78% is the geometric average of those returns after CPI-U December-to-December inflation. Neither is a guarantee, a fee-adjusted return, or advice to buy or sell anything.

Sources and methods

Where every figure comes from

Nominal figures add historical dollars as they were spent and compound them at nominal S&P 500 total returns. Inflation-adjusted figures are constant September 2026 dollars, the latest CPI-U month published by November 6, 2026 (BLS CUUR0000SA0, not seasonally adjusted; retrieved as FRED CPIAUCNS). Each fiscal year's outlays are multiplied by CPI in September 2026 divided by CPI in the month the outlays are invested (June through FY1976, September after that). October 2025 CPI-U was not published because of the 2025 lapse in appropriations; the November 1, 2025 row uses September 2025. October 2026 CPI-U was not yet published, so the November 1, 2026 valuation uses September 2026 and the portfolio's nominal value is already in those dollars. Compounding the inflation-adjusted outlays at real total returns (nominal return divided by CPI change) matches that result. The real multiple is the portfolio divided by inflation-adjusted spending. Per person divides by 340 million, the population used across this site. It is a present-day denominator, including for earlier November 1 rows.

The real geometric average is 6.78%, from 1928 through 2025. On November 1, 2026 the inflation-adjusted spending is $107.94 trillion ($317,472 per person) and the real multiple is 10.8×.

Outlays for a fiscal year are invested at that fiscal year's end: June 30 for FY1946–FY1976, September 30 for the July–September 1976 transition quarter, and September 30 for FY1977 onward. The contribution earns subsequent monthly total returns, starting the next month. The transition quarter (July 1–September 30, 1976) is a real outlay period between FY1976 and FY1977. It is included in cumulative spending and in the invested value. It is not a separate November 1 row.

Full calendar years 1946–2025 are scaled so the twelve monthly gross returns compound to Damodaran's S&P 500 total return for that year, including dividends. Monthly shape through January 1988 is Robert Shiller's S&P composite: (price + dividend/12) / prior price. From February 1988 the shape is the S&P 500 total-return index month-end level. January–October 2026 is the unscaled index change, because a full-year Damodaran return for 2026 is not published yet. Calendar-year checkpoints include 26.06% in 2023, 24.88% in 2024, and 17.78% in 2025. The histretSP HTML table publishes 17.78% for 2025. The companion spreadsheet on the same site shows 17.72%. This file uses 17.78%. From the December 2025 month-end level (15,220.45) to the October 2026 month-end level (17,011.141), the total-return index changed 11.77%. The long-run geometric average used in the projection is 10.02% over 1928–2025.

FY1946–FY1961 are OASI plus DI payments for individuals from OMB Table 11.3 (millions of dollars). FY1962–FY2026 are Table 8.1 Social Security (billions, one decimal). At FY1962 the two series are $14.049 billion and $14.0 billion. Table 8.1 first reports Medicare outlays in FY1967 ($2.5 billion). Benefits began July 1, 1966. No FY1966 Medicare outlay is published in this table, so none is entered. Table 8.1's Medicaid line begins in FY1962 and includes predecessor medical assistance before the 1965 statute. Those early amounts are used as published. The category is not extended before FY1962. Other means-tested entitlements are published in Table 8.1 from FY1962. Earlier years are not estimated.

FY2026 in the FY2027 Historical Tables (April 2026) is an estimate, not a final actual. It is included in the November 1, 2026 snapshot and labeled as an estimate. Each November 1 value uses the October month-end total-return index, the last month-end on or before November 1. October 31, 2026 was a Saturday, so the October 2026 level is the last trading day of that month.

Table 11.1 payments for individuals are broader than the four categories (they include veterans, unemployment, federal retirement, and other payments). They are context, not part of the sum. Table 11.1 shows $5,828,000,000 in FY1946 and $4.84 trillion in FY2025.

In FY2025, Table 8.1 lists programmatic mandatory outlays of $4,316.2 billion, equal to Social Security $1,575.1 billion, Medicare $987.7 billion, Medicaid $668.1 billion, other means-tested entitlements $517.8 billion, and other programmatic outlays $567.5 billion. The columns add without counting Medicaid twice.

Nonprofit figures are obligations, not outlays. Obligations (aggregated_amount). total_outlays was null in the response, so these are not outlays. 02 block grant, 03 formula grant, 04 project grant, 05 cooperative agreement. Retrieved 2026-11-06. On November 6, 2026, USAspending showed about $7.17 billion of FY2026 obligations for this filter, well below a finished year. The year is omitted rather than treated as complete. Nonprofit obligations are not added to the four-program total. Some of those dollars are already inside program outlays, and USAspending also misses subawards. Government grants (contributions), Form 990 Part VIII line 1e. Includes grants from state and local governments, not federal grants only. Government grants were 10.07% of total revenue. Applying that share to assets and net assets is a proxy, labeled as an estimate on the card. It is not a count of federally funded assets.

Debt held by the public was $32,765,175,912,225.50 and gross debt was $40,550,230,047,609.15 on 2026-11-04, from Treasury Debt to the Penny. November 1, 2026 was a Sunday, so the debt print is the November 4 release named in the data file.

Check our math

Anyone can run these numbers

If the invested value looks too large, or the spending total looks too small, run it yourself. The arithmetic is public. The yearly rows are in a spreadsheet. The links at the end of this section are the same sources already used on this page.

How the spending total is built

Add four lines of federal spending. Do not count a dollar twice. The lines are Social Security, Medicare, Medicaid, and other means-tested entitlements. Medicaid is its own line. It is not also inside other means-tested entitlements. Grants to nonprofits are on their own card and are not in this total.

Social Security for fiscal 1946 through fiscal 1961 is old-age and survivors insurance plus disability insurance, from OMB Historical Table 11.3, which is printed in millions of dollars. From fiscal 1962 through fiscal 2026, Social Security is the Social Security line in OMB Historical Table 8.1, printed in billions of dollars with one decimal. In fiscal 1962 the two readings are $14.049 billion and $14.0 billion.

Medicare in Table 8.1 starts in fiscal 1967, at $2.5 billion. Benefits began July 1, 1966, but the table does not publish a fiscal 1966 Medicare outlay, so none is entered. Medicaid in that table starts in fiscal 1962 and includes medical assistance from before the 1965 statute. Those early amounts are used as published. Nothing is filled in before fiscal 1962. Other means-tested entitlements are also published from fiscal 1962. Earlier years are not estimated. The Historical Tables introduction lists the accounts in that line. Medicaid is not one of them.

FY2026 in the FY2027 Historical Tables (April 2026) is an estimate, not a final actual. It is included in the November 1, 2026 snapshot and labeled as an estimate.

Between fiscal 1976 and fiscal 1977 there was a transition quarter, July 1 through September 30, 1976. That spending is real. It sits inside the cumulative total and inside the invested value. It does not get its own November 1 row.

That year’s spending is the four lines added together. Cumulative spending adds the years from fiscal 1946 through the fiscal year that ended the September before that November 1. Divide by 340,000,000 people, the population used on the rest of this site, to get the per-person figure. That count is today’s population on every row, including 1945.

How the invested value is built

Take those same dollars and grow them as if they had been invested in the S&P 500, dividends reinvested. This is an illustration of what the spending could have been worth. It is not a portfolio the government held, and it does not subtract the benefits people received when the money was spent.

Each fiscal year’s outlays go in at the end of that fiscal year: June 30 for fiscal 1946 through fiscal 1976, September 30 for the 1976 transition quarter, and September 30 from fiscal 1977 on. The balance starts earning the next month’s return.

The return is the S&P 500 total return, which means the price change plus dividends. For each full calendar year from 1946 through 2025, the twelve monthly moves are lined up so that, multiplied together, they equal Aswath Damodaran’s return for that year, including dividends. Through January 1988 the month-to-month shape is Robert Shiller’s stock series. Take the price, add one month of dividends, and divide by the prior price. From February 1988 the shape is the month-end level of the S&P 500 total-return index. January through October 2026 is the change in that index, with no full-year adjustment, because a full-year 2026 return is not published yet.

Each November 1 value uses the October month-end total-return index, the last month-end on or before November 1. October 31, 2026 was a Saturday, so the October 2026 level is the last trading day of that month. The invested value is every included year’s outlays, each grown from its own investment date to that October.

Nominal figures add the dollars as they were spent and compound them at those nominal returns. Inflation-adjusted spending is in September 2026 dollars, the latest CPI-U month in the file: all items, U.S. city average, not seasonally adjusted. Each year’s outlays are multiplied by the September 2026 index divided by the index in the month the money is invested. October 2025 was not published, because a lapse in appropriations stopped the data collection, so the November 1, 2025 row uses September 2025. October 2026 was not yet published, so this November 1 uses September 2026. The portfolio’s nominal value is already in those dollars. Growing the inflation-adjusted outlays at real total returns, the nominal return divided by the change in the CPI, lands on the same invested value.

If the return is lower

The projection on this page compounds at 10.02%. That is the geometric average of nominal S&P 500 total returns from 1928 through 2025: the single yearly rate that ends at the same wealth as the actual up-and-down years. It is not reduced by a fee.

Freedom Accounts, as the rest of this site describes them, charge a 0.25% management fee on the account value each year. Take that fee at the end of each full year. Multiply the year’s growth by 0.9975, which is 1 minus the fee. Do that for every year in the 1928–2025 average and the geometric average falls from 10.02% to 9.74%.

The table uses 1% of fiscal 2025 outlays for the four programs, $37,487,000,000 a year ($110 per person). That is the low end of the 1% to 5% range in the projection on this page. Deposits go in at each year-end, so the last deposit has not yet earned a return. One column compounds at the unadjusted average. The other compounds at the lower, fee-adjusted average.

One percent of fiscal 2025 program outlays compounded at the long-run geometric average and at the fee-adjusted average
Years10.02% nominal9.74% after the fee
10$597.97B$1,759 per person$590.14B$1,736 per person
20$2.15 trillion$6,328 per person$2.09 trillion$6,134 per person
30$6.19 trillion$18,199 per person$5.87 trillion$17,277 per person

The large invested number on this page is not that average. It is the actual year-by-year path. Apply the same fee to each full calendar year on that path. Leave January through October 2026 as published, because that year is not finished and has no full-year return to scale. On November 1, 2026 the invested value is $1,058.54T instead of $1,169.67T. Spending does not change: $66.52T nominal, and $107.94T in September 2026 dollars. The multiple is 15.9× the nominal spending instead of 17.6×, and 9.8× the inflation-adjusted spending instead of 10.8×. Per person, the fee-adjusted stock is $3,113,351 instead of $3,440,197. On this November 1 the invested figure is already a current market price, so the nominal amount and the September 2026 amount match in both cases.

Yearly figures

The spreadsheet is generated from the same November 1 calculation as this page. The production build writes the file. It is not a separately typed table. There is one row for every November 1 from 1945 through 2026. Dollars are plain numbers, rounded to the cent. The population on every row is 340,000,000.

Download the yearly figures (CSV)

To check a row, add the four programs’ cumulative columns and compare the sum with the cumulative total. Do the same for inflation-adjusted spending, and for the invested value. Divide by the population column and compare with the per-person columns. Divide the invested value by cumulative spending and compare with the multiple.

From 1946 on, the invested value equals the previous November’s invested value grown forward (the column prior_invested_grown_to_this_november) plus that year’s spending grown to this November (spending_that_year_invested), plus the transition-quarter column. That last column is zero except in 1976. The grown-forward column is the previous invested value times this October’s index divided by the previous October’s index. There is no October index on the 1945 row. That row is the baseline, before fiscal 1946 outlays. Rounded to the cent, the pieces add back to within a dollar.

Adding the “that year” spending column down the whole file leaves out the transition quarter. Add the 1976 transition-quarter dollars once.

Sources

Every public source used on this page is listed here. These are the same links as “Where every figure comes from.”

People and the Republic would be better positioned with Freedom Accounts than with the ineffectual spending we do now.

Limits

What this doesn’t capture

  • The money was spent. It paid Social Security benefits, medical care, food assistance, tax credits, and the other payments in these accounts. Those payments had value to the people who received them, and they had economic effects. This page does not net those out.
  • A buyer the size of the federal government, purchasing equities year after year, would have moved prices. The S&P 500 return is the historical return on the index, not the return available on a sovereign-scale book. Treat the invested value as an opportunity-cost comparison, not as a tradable portfolio.
  • Markets fall. 2022’s total return in this series was -18.04%. Past returns are not future returns. Nothing here is investment advice or a promise that Freedom Accounts would earn the long-run average.
  • Social Security and Medicare are insurance, including disability, survivors, and health coverage, not only a cash balance. Sequence of returns matters for people who need the money on a fixed date. A trust-fund depletion date is a financing projection under stated assumptions.
  • The Census Supplemental Poverty Measure counts noncash transfers and refundable credits. Those programs reduce measured SPM poverty. The official poverty rate does not count them, so the two rates answer different questions.
  • The nonprofit card is federal award obligations to recipients USAspending calls nonprofit. It leaves out private donations. It undercounts federal dollars that pass through states and other prime recipients before they reach a nonprofit. IRS line 1e, used in the asset estimate, includes state and local grants. The asset figure applies a revenue share to the whole balance sheet. It is an estimate.
  • Dollars are nominal. Population is today’s 340 million on every row, including 1945. FY2026 program outlays are an OMB estimate from the FY2027 Historical Tables. USAspending does not publish this nonprofit series before FY2008 in the response used here, so that card starts where the data starts.