The 5% Treasury and the $1 Trillion Question

Five percent can describe two very different things. For an investor, it is the yield available on newly purchased government debt. For the government issuing that debt, it is a price that will enter public accounts over years, sometimes decades.

The distance between those two clocks is where the more interesting economic question begins. The sovereign bond supplies the benchmark used to price other assets, while the government must finance the balance sheet behind that benchmark.

The misleading simplicity of yield

At the end of September 2026, the ten-year Treasury yielded 5.29%. The S&P 500 offered a dividend yield of approximately 1.1%, a trailing earnings yield of 3.94%, and a forward earnings yield of 5.18%.

The last figure makes equities appear remarkably close to Treasuries. It comes from inverting a forward price-to-earnings multiple of 19.3, compared with a trailing multiple of 25.4. Both calculations are arithmetically correct. Their economic interpretations require considerably more care.

Dividend yield measures cash distributed to shareholders; earnings yield measures accounting profits relative to market value, including profits retained for reinvestment; forward earnings yield reflects estimates. A ten-year Treasury yield is derived from a defined schedule of nominal payments, subject to the issuer meeting its obligations and the investor’s holding period.

There is also a methodological trap: trailing reported earnings and forward analyst estimates use different accounting and time conventions. The 32% arithmetic difference between the implied earnings denominators cannot be interpreted as a clean forecast of earnings growth.

Inflation adds another distinction. On 30 October, the ten-year nominal Treasury yielded 4.94%, while the corresponding inflation-protected Treasury yielded 2.60% in real terms. The title’s 5% is thus a recent regime marker rather than the latest spot reading. Corporate profits can rise with inflation, offering potential protection against the erosion of purchasing power. But an earnings yield is a snapshot of profits relative to price, not a promised return: the investor’s real outcome depends on future earnings, distributions, valuation changes and inflation.

The more revealing comparison starts with actual payments.

The Coupon Crossover

In fiscal 2023, federal net interest expenditure reached $658 billion. Over the corresponding twelve months, S&P 500 companies distributed $580 billion in cash dividends. The federal interest bill was 1.13 times those dividends.

This coupon crossover is the return of a relationship seen in earlier decades, including the 1980s and 1990s. By 2025, federal net interest expenditure amounted to approximately $1.46 for every dollar of S&P 500 cash dividends, compared with 71 cents in 2021.

Dividends, however, capture only one route through which corporations return capital. Adding share repurchases changes the comparison: in 2025, government interest amounted to 58 cents for every dollar of dividends and buybacks combined. The table shows both measures.

Federal figures are for fiscal years ending September 30; S&P 500 figures cover the same trailing twelve-month windows. Payouts include gross buybacks, some of which offset equity issuance or employee compensation. The S&P 500 is a defined corporate universe, rather than the entire private economy. Ratios compare the scale of payments; they do not measure a transfer between these two groups.

The crossover concerns dividends alone. Once buybacks are included, federal net interest remains below total shareholder distributions throughout the period. Yet the broader ratio has more than doubled, from 28% to 58%. The significance lies in the speed at which the public interest bill has grown relative to corporate payouts.

Three growth rates explain the movement. Between 2021 and 2025, federal net interest expenditure increased by 175%, S&P 500 dividends and buybacks by 36%, and federal receipts by 29%.

Aggregate dividends and buybacks increased by approximately 36% between September 2021 and September 2025, while the S&P 500’s aggregate market value rose by approximately 56%. Its combined payout yield consequently declined from 3.40% to 2.95%: investors received less cash distribution per dollar of market value.

Meanwhile, the government’s interest obligations grew much faster than the revenues supporting them.

Sovereign Interest Persistence: two clocks, one liability

The mechanism begins with a distinction between the market yield on government debt and the average interest cost of the debt already outstanding. A ten-year yield of 5% never turns the entire Treasury stock into a 5% liability overnight.

Existing fixed-rate securities retain their contractual coupons. Higher market rates enter the public accounts progressively as debt matures, deficits are financed, and securities are replaced at prevailing borrowing costs. Each refinancing decision leaves a record in the future payment schedule.

Sovereign Interest Persistence describes the delayed transmission of market yields into debt-service expenditure, and the persistence of that expenditure as financing costs accumulate across a large stock of debt. It follows a simple sequence: market repricing → refinancing and new issuance → accumulated interest cost.

If market rates subsequently decline, higher-cost securities already issued continue paying their coupons until maturity. Relief arrives over time as refinancing occurs. The speed of adjustment depends on the composition of the debt stock, the share financed at short maturities, new issuance, and the path of interest rates. A persistent burden need not be irreversible.

Three measures trace the pressure on government finances: how far market yields stand above the average rate on existing debt; how that average rate responds when yields rise or fall; and whether the interest bill is growing mainly because debt is larger or more expensive.

The first two use Treasury’s published average interest rate. The third calculates effective cost by dividing annual marketable-debt interest expense by the corresponding average debt stock, including inflation and accounting adjustments. Federal net interest serves a different purpose: after deducting interest receipts and applying other budget adjustments, it measures the burden on government revenue rather than the financing cost of marketable debt. The results follow in the US application, with formulas and limitations in the technical panel.

The fiscal consequences are observable. Net interest absorbed approximately 8.7% of federal receipts in 2021 and 18.5% in 2025. By fiscal 2026, net interest spending had reached $1.052 trillion, around 19.5% of federal receipts. The additional $82 billion relative to 2025 represents an 8.5% increase in a single fiscal year, even though the market yield observed at year-end applies only to debt currently being priced or refinanced.

This makes the interest-rate environment an increasingly important determinant of fiscal flexibility. As interest absorbs more government revenue, less remains for other spending unless taxes or borrowing increase. The effect on Treasury yields depends on how investors expect that pressure to be managed. A rising interest bill does not automatically push yields higher, but it makes the credibility of future tax, spending and borrowing decisions more consequential for the price of government debt.

The destination of interest payments also matters. A rising interest bill generates income for bondholders that can support domestic spending and investment—but only to the extent that it circulates within the issuing economy. Foreign investors held ca. 30% of US federal debt held by the public, ca. 30% of UK gilts in September 2025, ca. 55% of France’s marketable central-government debt in June 2025, and ca. 12% of Japanese government bonds and Treasury bills in December 2024. Although dates and debt definitions differ, these figures illustrate how widely ownership varies. Foreign holders may reinvest locally, while domestic holders may invest abroad. The government incurs the interest expense regardless; the domestic economy does not necessarily receive a corresponding boost. Any benefit also depends on the taxes, borrowing or spending adjustments used to finance those payments.

Even interest paid to a domestic central bank does not automatically return to the government. The Federal Reserve normally transfers its net earnings to the US Treasury (the government department responsible for public finances), allowing some interest paid on Fed-held government bonds to flow back into the budget. But the Fed also pays interest on bank reserves. After 2022, those costs rose faster than income from its securities, generating losses and interrupting these transfers. By July 2026, losses that would need to be covered by future earnings before transfers could resume (the Fed’s “deferred assets”) totalled approximately $236 billion, although some Reserve Banks had already recovered their losses and resumed payments. What matters is therefore not simply who holds the debt, but how much interest income remains after the holder’s own financing costs.

Empirical application: the United States and Japan

The two countries provide complementary evidence. US data show how interest costs have already increased; Japan’s budget shows how further increases are anticipated.

United States: repricing first, debt growth increasingly thereafter

Treasury records show the combined effect of a larger debt stock and a higher effective interest cost:

Effective cost equals annual marketable-debt interest expense divided by average marketable debt, including inflation and accounting adjustments. These figures differ from federal net interest, which includes interest receipts and other budget offsets.

The midpoint decomposition separates the increase in annual interest expense into two contributions:

Calculations use unrounded inputs. Each row compares its own endpoints; the full-period decomposition is not a sum of annual decompositions.

Higher effective cost explains 67% of the increase between FY2021 and FY2025 because most of the repricing had already occurred by FY2024. Effective cost rose from 1.80% to 3.29%, then increased much more slowly, to 3.37% in FY2025 and 3.44% in FY2026. As those cost increases moderated, continued debt growth became the main source of additional expense.

The findings are therefore consistent: higher effective cost explains most of the accumulated increase over FY2021–FY2025; debt growth explains roughly three-quarters of each of the latest annual additions. A slowdown in repricing has changed the source of pressure without stopping the interest bill from growing.

The monthly regression provides no stable evidence that borrowing costs respond more strongly to rising yields than falling ones. Specifications using 12 and 24 monthly lags find no statistically significant difference; the 36-month version finds a stronger response to falling yields. The results depend on the specification and do not establish causation. Full estimates and limitations appear in the technical panel.

Japan: refinancing pressure enters the budget before it is fully realised

Japan’s FY2026 budget projects ¥13.0 trillion in interest payments, compared with ¥10.5 trillion in the FY2025 initial budget. The Ministry of Finance attributes approximately ¥1.5 trillion of the increase to changes in the outstanding bond portfolio, including refinancing and debt growth. A further ¥1.0 trillion reflects an increase in the assumed rate applied to scheduled issuance, from 2% to 3%.

The 3% figure is a prudent budgeting assumption, rather than the rate paid on every new bond. Alongside an average rate of approximately 0.8% on outstanding general government bonds, it illustrates how legacy financing costs can remain low while the budget prepares for more expensive borrowing.

Japan’s figures are budget projections; the US figures record realised expenditure. Their fiscal years also differ: April–March in Japan and October–September in the United States. Direct numerical comparisons would therefore be misleading.

Together, the cases reveal two stages of the same process. Japan’s budget anticipates the cost of replacing cheaper debt. US accounts show how, once much of the repricing has occurred, an expanding debt stock can sustain further increases in interest expense. The rate cycle can lose momentum before the fiscal consequences do.

The balance sheet behind the benchmark

The central finding is a change in the composition of fiscal pressure. Higher effective cost explains two-thirds of the increase in annual marketable-debt interest expense over FY2021–FY2025; debt accumulation explains roughly three-quarters of each of the latest annual additions. This does not mean repricing is complete. Further increases in market yields would progressively affect a debt stock that is itself expanding. The larger liability increases the dollar cost of each additional percentage point of financing expense.

This brings the opening bond–equity comparison into sharper focus. In September 2026, the ten-year Treasury yielded 5.29%, slightly above the S&P 500’s 5.18% forward earnings yield. The figures are not equivalent returns: equities offer uncertain future growth, while conventional Treasuries promise defined nominal payments. But they run on different clocks. A change in Treasury yields can immediately affect the discount rates investors apply to equities and other private assets; its effect on the government’s interest bill emerges gradually, as existing debt is refinanced and new debt is issued.

The consequence for valuation is that the benchmark and the assets priced against it share a fiscal exposure. If higher financing costs are met through taxation or spending restraint, the adjustment can affect corporate earnings. If they are met through additional borrowing, the effect may instead reach the discount rate through changes in Treasury supply and the return investors demand.

Comparing an equity earnings yield with a Treasury yield therefore captures only part of the relationship. A narrow gap between the forward earnings yield and the Treasury yield should be assessed alongside the assumptions supporting forecast profits, including how those profits might respond if fiscal policy or financing conditions differ from expectations.

That brings the trillion-dollar question into focus.

How does the expanding fiscal cost of supplying the risk-free benchmark influence the future price investors demand for holding it?

The answer the evidence points to is a gradual one, working less through the interest bill itself than through doubt about who will absorb it. A rising cost of debt can be met through higher taxes, lower spending, inflation, or further borrowing. So far, the United States has largely chosen the last: debt growth now drives most of each year’s increase in interest expense, which means part of today’s interest is effectively financed by tomorrow’s debt. If that continues, the price of holding the benchmark will increasingly include compensation for uncertainty about the issuer’s fiscal path, not only about the path of rates. The risk-free rate would remain risk-free in its payments, while becoming less neutral in its price.

Relative valuation requires consistency between the fiscal assumptions embedded in bond prices and those supporting earnings forecasts. Treasury pricing that depends on fiscal restraint warrants closer scrutiny if equity forecasts depend on sustained government support. These possibilities may already be reflected in prices and forecasts; the evidence does not establish that markets hold conflicting expectations. It shows why the comparison should extend beyond headline yields.

The analysis does not show that fiscal pressure pushes yields higher mechanically, nor does it support a blanket preference for bonds over equities. It supports a more precise conclusion: the return offered by government debt and the growth expected from equities cannot be assessed independently of the fiscal path linking them.

Similar yields do not necessarily describe the same economic future. The benchmark is more than the return equities must compete with: its financing consequences help shape the conditions in which their expected earnings must be delivered.

Measuring persistence: technical panel

1. How much repricing may remain?

The refinancing gap compares the ten-year Treasury yield with the average interest rate on outstanding marketable debt:

$$G_t = y_t^{10\mathrm{y}} - r_t^{T}$$

Here, y is the ten-year yield and rᵀ is Treasury’s published stock-average rate, both measured at month-end.

September 2026: 5.290% − 3.518% = 1.772 percentage points, or 177 basis points.

If market yields remain elevated, replacing older, cheaper debt can continue to raise the average borrowing cost. This is a benchmark signal rather than a forecast: the actual effect depends on which securities mature, the maturities of new issuance and rates across the yield curve.

2. Do borrowing costs rise more readily than they fall?

A monthly regression separates the effects associated with rising and falling ten-year yields:

$$\Delta r_t^{T}=\alpha+\sum_{k=0}^{K}\beta_k^{+}\,\Delta y_{t-k}^{+}+\sum_{k=0}^{K}\beta_k^{-}\,\Delta y_{t-k}^{-}+\varepsilon_t$$

Δ denotes a monthly change. The + and − terms separate yield increases from decreases, retaining their signs. The β coefficients measure the estimated response at each monthly lag; K sets the maximum lag, α is a constant and ε captures unexplained movements.

Responses are sums of the estimated coefficients. For example, the 12-month specification associates a one-percentage-point yield increase with a cumulative 0.531-percentage-point increase in the stock-average rate, holding other regressors fixed. The p-value tests whether rising and falling responses are equal.

The 12- and 24-month specifications find no statistically significant difference. The 36-month specification finds a stronger response to falling yields. The evidence does not establish a stable upward asymmetry.

The 1.395 falling-yield response should not be read as a direct refinancing effect. It may also capture policy-rate movements transmitted through bills and floating-rate debt, a possibility not tested here with a short-rate control. With 75 coefficients estimated from 164 observations, the longest specification also warrants particular caution.

Method: January 2010–September 2026 raw data; monthly changes in Treasury’s stock-average rate and GS10 yields; a common February 2013–September 2026 estimation sample. Ordinary least squares includes a constant, contemporaneous yield changes and K lags, without additional controls. Equality tests use Newey–West covariance estimates with 12 lags, Bartlett weights, the n/(n−p) correction and asymptotic chi-square p-values. These are exploratory associations, not causal estimates; even 36 lags cannot capture the full refinancing cycle.

3. Is the increase coming from more debt or higher cost?

Effective interest cost is annual marketable-debt interest expense divided by average debt outstanding:

$$c_j=\frac{I_j}{D_j},\qquad j\in\{0,1\}$$

The change in expense can then be separated exactly:

$$I_1-I_0=\underbrace{(D_1-D_0)\,\frac{c_1+c_0}{2}}_{\text{debt-stock contribution}}+\underbrace{(c_1-c_0)\,\frac{D_1+D_0}{2}}_{\text{effective-cost contribution}}$$

I is annual marketable-debt interest expense, D is average marketable debt and c is effective cost. Subscripts 0 and 1 identify the earlier and later periods. The midpoint formula shares the interaction equally between the two contributions and leaves no residual. The results appear in the decomposition table in the US application above.

Calculations use unrounded inputs and twelve-month averages of month-end marketable debt. Each row compares its own endpoints; the full-period decomposition is not a sum of annual decompositions. Interest expense covers marketable bills, notes, bonds, TIPS and floating-rate notes, including inflation compensation and discount/premium adjustments. Debt includes the small intragovernmental and Federal Financing Bank components within Treasury’s “Total Marketable” total. Effective cost absorbs remaining timing or perimeter differences; it is neither an average coupon nor Treasury’s published stock-average rate. Federal net interest is a separate budget measure after offsets.

Sources

1. U.S. Treasury, Monthly Treasury Statement
2. FRED, Federal Receipts (FYFR)
3. CBO, The Federal Budget in Fiscal Year 2025
4. S&P Dow Jones Indices, S&P 500 Buybacks, 2021
5. S&P Dow Jones Indices, S&P 500 Buybacks, Q3 2022
6. S&P Dow Jones Indices, S&P 500 Buybacks, Q3 2023
7. S&P Dow Jones Indices, S&P 500 Buybacks, Q3 2024
8. S&P Dow Jones Indices, S&P 500 Buybacks, Q3 2025
9. S&P Dow Jones Indices, Q3 2026 Scorecard (conventional S&P 500 row)
10. U.S. Treasury, Daily Nominal Treasury Rates
11. U.S. Treasury, Daily Real Treasury Rates
12. Federal Reserve, Monetary Policy Report, July 2026
13. U.S. Treasury, Monthly Statement of the Public Debt
14. U.S. Treasury, Interest Expense on the Public Debt
15. U.S. Treasury, Average Interest Rates on Treasury Securities
16. FRED, 10-Year Treasury Constant Maturity (GS10)
17. Japan Ministry of Finance, FY2026 Breakdown of National Debt Service
18. Japan Ministry of Finance, Debt Management Report 2026
19. Japan Ministry of Finance, Debt Management Report 2025
20. Congressional Research Service, Foreign Holdings of Federal Debt
21. UK Debt Management Office, Quarterly Review, Q4 2025
22. Agence France Trésor, Bulletin mensuel, July 2026