Federal Debt: Exploring Solutions
What would it actually take to bring debt per person down?
Pick policies, change the assumptions, and watch gross federal debt per person (in 2026 dollars) respond in real time. Every option uses a published score from CBO, JCT, CRFB, TPC, Penn Wharton or GAO. Anything that isn't published is labeled as an estimate.
Today, gross federal debt is about $118,000 per person ($40.6 trillion on 13 Nov 2026). On current law, CBO projects it will keep rising faster than the economy. This page projects that baseline forward and lets you layer on deficit-reduction options. The default growth assumption is that real GDP keeps growing at its 12-year moving average (2.47% a year to start). That is more optimistic than CBO's 1.74%, and you can switch to CBO's path below.
Explorer
Real gross federal debt per person (2026 dollars)
Baseline (current law) With selected options Target
Debt as a share of GDP
Gross, baseline Gross, with options Held by the public, with options
Net interest as a share of GDP
Baseline With options
Options
Savings are shown as a share of GDP once fully phased in (2036 → 2046), with the dollar equivalent at CBO's projected 2036 GDP ($46.7T). Each option phases in over three years starting in FY2027 and is held at its 2046 level after 2046. Tap “Why / source” for a plain-language explanation and the citation. Est. marks figures that are extrapolated or assumed rather than published.
Assumptions
Inflation follows CBO's path (CPI-U about 2.3% a year long run) and is used only to express results in October-2026 dollars.
How hard is $40K per person by 2046?
- Under default assumptions, the model needs about 5.9% of GDP a year of deficit reduction beyond the baseline. That works out to an average primary surplus of about +3.4% of GDP over FY2027–2046, against a primary deficit of about −2.5% today.
- Softer goals: $80K needs about 4.3% of GDP, $60K about 5.1%, and just holding debt per person at today's level about 2.9%.
- History: the best U.S. primary surpluses were 5.9% (FY1948) and 4.5% (FY2000). Canada swung from −4.0% (1992) to +5.8% (2000) and Sweden from −9.9% (1993) to +5.2% (2000). Eichengreen and Panizza found only three episodes anywhere of primary surpluses of 5% or more held for ten years.
- Interest rates matter more than growth. One point higher on rates raises the requirement to about 6.6% of GDP. Half a point more or less growth moves it by only about ±0.2.
Method
- Starting point. Treasury Debt to the Penny, 30 Sep 2026 (end of FY2026): $32.43T held by the public plus $7.75T intragovernmental, for $40.17T gross. Population is 343.1M (Census, via FRED POPTHM). Dollars are October-2026 dollars (CPI-U 337.511).
- Baseline. Revenue and non-interest spending (% of GDP) follow CBO's 2026 Long-Term Budget Outlook (Feb 2026), with revenue lowered by 0.17% of GDP a year for CBO's 20 Aug 2026 tariff update. That update found trade policy through 31 July 2026 adds $0.7T to primary deficits over FY2027–36 (publication 62704). After FY2056, the end of CBO's window, revenue and non-interest spending continue their FY2046–56 linear trends; interest rates, CBO growth and inflation are held at 2056 values. That extension is the model's assumption, not CBO's.
- Debt each year. Public debt = last year's debt − primary balance + net interest. Net interest = k × r × last year's public debt. Here r is CBO's average rate on federal debt (3.45% in 2027, rising to 4.15% by 2046), and k is calibrated so the model reproduces CBO's net interest on CBO's own path.
- Interest feedback. r rises 2 basis points for every percentage point that debt held by the public exceeds CBO's projected ratio, and falls when it is below. CBO uses this “debt sensitivity of interest rates” of 2 bp in its long-term projections (CBO Working Paper 2024-05). In 2022 CBO put the range at 2–3 bp. The slider lets you choose 0–5 bp.
- Growth. By default each year's real growth is the mean of the previous 12 years of annual real GDP growth, seeded with BEA data for 2015–2026 (2026 = first three quarters vs the 2025 average). Projected years then feed back into the average.
- Gross vs public debt. Gross debt includes Treasury securities held by trust funds. Intragovernmental debt follows CBO's gross-minus-public path. Social Security savings first build up the OASI trust fund (which leaves gross debt unchanged) and lower gross debt only once that fund would otherwise be exhausted. CBO projects exhaustion in FY2032. After that the model assumes an OASI cash shortfall of 1.34–1.60% of GDP, an estimate derived from CBO's figures.
- Scores. Each 10-year score is converted to a share of GDP: the score divided by cumulative GDP over its window, or the final-year score divided by that year's GDP where one is published. Options phase in over 3 years, and some grow toward 2046 (marked Est.). Package checks, with interaction discounts off and default assumptions, match the offline Python model.
How the two new options are scored
92% estate tax on estates above $5 million (estimate). The estate and gift tax raised $29.5B in FY2025, or 0.097% of GDP (CBO). The Tax Policy Center finds that returning to the pre-OBBBA exemption (about half of today's $15M) would raise about $230B over 10 years, or 0.060% of GDP, for roughly 0.157% of GDP in total. The model assumes a $5M exemption adds 15% more (an assumption). TPC counts 12,300 taxable estates at the 2009-level exemption ($5.17M today), against 3,900 now. Raising the rate from 40% to 92% multiplies the static yield by 2.3, and the result is scaled by the net-of-tax ratio ((1−0.92)/(1−0.40))ε.
With Kopczuk–Slemrod's ε ≈ 0.16, the net gain is about 0.20% of GDP. At the default ε = 0.33 it is about 0.12%. At ε = 0.5 it is about 0.05%, and above ε ≈ 0.72 it loses money. Avoidance at a 92% rate is far outside the range ever measured. The top rate was 77% from 1941 to 1976, and that is the closest precedent. The gift tax, trust and charitable-bequest planning, and income-tax interactions are not modeled.
Replace the individual income tax with an Automated Payment Transaction Tax (APTT, Feige) (estimate). The model removes the individual income tax entirely (CBO: about 9.1–10% of GDP) and adds APTT revenue = rate × transaction base × (1 − haircut). The base is 2025 large-value payments: Fedwire Funds $1,148T plus CHIPS $503T, or $1,652T. That is 53.5 times calendar-2025 GDP, and the base grows with GDP. The haircut covers double counting, financial activity that leaves or shrinks, and avoidance. Its 50% default is an assumption. Published rate estimates run from about 0.3% to 4.3%. Feige (1996 base of about $445T) found that 0.3% could replace all federal taxes because it keeps the huge financial-transaction base. CRS (RL32266) assumed large-value and financial transfers would flee and taxed only about $57T of retail payments, so it needed about 4.3%. The APTT is mutually exclusive with options that change individual income-tax rates, deductions or capital-gains rules. Those options are disabled while it is selected.
Caveats
- Static scores. Options are scored without macroeconomic feedback, except where the source is dynamic (the Penn Wharton wealth tax) or the option is itself a growth proxy (immigration). Large tax increases or cuts would change growth.
- Pre-OBBBA scores. Most CBO options were scored in Dec 2024 against a baseline in which the TCJA expired. OBBBA made those cuts permanent, so some revenue options are likely understated, and the Medicaid cap overlaps cuts that are already law.
- Interactions. Real packages interact. Discounts (when checked) are rough assumptions: TCJA expiration plus higher top rates (top-rate gain × 0.8); wealth tax plus billionaire minimum tax (× 0.5); wealth tax plus estate taxes (estate × 0.7); two or more Social Security benefit cuts (each × 0.9); 92% estate tax plus taxing gains at death (gains × 0.9); APTT plus the employer health cap or IRS enforcement (× 0.5, since much of their value comes through the income tax).
- Revenue and spending stay at CBO's share of GDP regardless of the growth assumption, so faster growth helps mainly through a bigger denominator and lower interest relative to GDP.
- Past 2056 the baseline is an extrapolation, so treat 30–50-year results as illustrative.
- Tariffs are in the baseline and carry no extra package credit. One-time asset sales, permitting reform and fiscal rules have no score and are not included.
Sources
- CBO, The Budget and Economic Outlook: 2026 to 2036 (Feb 2026) and the 2026 Long-Term Budget Outlook; data from CBO's GitHub data repository.
- CBO, Updated Budgetary Projections of Tariffs as of July 31, 2026 (20 Aug 2026); CRFB summary.
- CBO, Options for Reducing the Deficit: 2025 to 2034 (Dec 2024).
- CBO, Revisiting the Relationship Between Debt and Long-Term Interest Rates (WP 2024-05); Budgetary Costs of Higher Interest Rates and Inflation (2022).
- Treasury, Debt to the Penny; BEA real GDP via FRED GDPC1; Census population via FRED POPTHM; BLS CPI-U.
- Tax Policy Center, Who Would Be Affected By A Decrease In The Estate Tax Exemption? (Aug 2026); Fixing a $40 Trillion Problem. Kopczuk, Estate Taxation (NBER Reporter, citing Kopczuk–Slemrod). PWBM, Sanders estate tax (77% top rate 1941–1976).
- Feige, The Automated Payment Transaction Tax; CRS, RL32266: A Transaction Tax; Federal Reserve, Fedwire Funds annual statistics; The Clearing House, CHIPS annual statistics.
- Eichengreen & Panizza, A Surplus of Ambition; IMF primary balance data; GAO, improper payments FY2025.
- Individual option sources are linked under each option above.