The Return of the Bond Bear Market? Globalisation, Real Rates and the Structural Repricing of Long-Term Government Bonds

Hypothesis. The structural argument runs as a chain: globalisation reverses; global disinflation weakens; supply chains become redundant; strategic investment and defence spending rise; investment demand and the natural rate rise; inflation uncertainty and the inflation risk premium rise; the term premium rises; long yields settle in a higher range. The cyclical alternative is simpler: inflation rose, central banks tightened, yields followed, and once inflation normalises yields drift back.

Why the distinction matters. Under a cyclical regime, yields rise with the policy rate and fall when it is cut. Under a structural regime, long yields can stay high after policy easing because the equilibrium real rate, inflation uncertainty, the term premium and fiscal risk have moved up together. This paper is not about whether central banks keep rates high for longer. It is about whether the long-run price of duration has changed.

2. The Four-Decade Bull Market and Why It Was Unusual

Evidence. The BIS notes that the 10-year US nominal yield fell by roughly 13 percentage points between 1981 and 2020, with real yields falling substantially in parallel across developed economies. The 10-year constant-maturity yield was about 15.8% in September 1981 and reached a record closing low near 0.5% in August 2020 (FRED; approximate), and the 10-year TIPS yield was about −1.2% in 2020–21. Today the 10-year Treasury stands near 5.1–5.3% and the TIPS yield at about 2.8% (Trading Economics, 14 October).

Interpretation: was the pre-2020 regime exceptional? The 2010s are best read as a combination of forces, not a permanent equilibrium. Disinflation after Volcker, falling inflation volatility, globalisation and the integration of China, high global saving and reserve accumulation, ageing, regulation that favoured safe assets, and from 2008 zero rates and quantitative easing all compressed yields. Recent BIS work suggests monetary policy played a larger role in the long-term decline than the conventional saving-glut story implies. The regime was therefore partly structural and partly policy-made. Treating 2019 or 2020 yields as the natural resting point is a choice, not a fact.

Adaptation. The key point for financial stability is that institutions adapted to low rates: mortgage borrowers, leveraged buyers, private equity models, infrastructure valuations, commercial property cap rates, pension and insurance duration, government debt stocks and equity multiples all assumed cheap capital. The end of that regime is a balance-sheet event, not only a bond-market event.

3. Globalisation, Saving and Investment, and the Natural Rate

Mechanism. In an integrated world, savings cross borders: surplus countries finance investment elsewhere and the system behaves like one capital market. If fragmentation makes capital allocation more selective, with flows shaped by alignment, sanctions, supply-chain security and national-security rules, capital recycling becomes less efficient. Domestic investment demand may rise (defence, energy, grids, semiconductors, critical minerals) while the supply of internationally recycled saving shrinks. That would raise the equilibrium real rate. The IMF has examined the link between globalisation and the natural rate, and the logic can be reversed. The counter-forces are strong: ageing and slow labour-force growth raise desired saving, weak productivity depresses investment demand, high debt can restrain private spending, and population decline lowers growth.

r* is not observable. It is estimated from models, and estimates are wide, revised and method-dependent. Table 2 compares them. On the evidence the equilibrium real rate has probably risen somewhat since the pandemic on several models, but the increase is modest in the main institutional estimates and the confidence intervals are wide. A cyclical increase and a structural one are not separable with present data.

Interpretation. The gap between model r* (about 1%) and the market’s real yield (about 2.8%) is the most important fact in the debate. Either r* is higher than the models say, or the term premium and other components explain the difference, or both. The New York Fed’s finding that a falling convenience yield explains a large part of the post-pandemic rise implies that the increase could reflect weaker demand for safe assets, which is a risk-premium story, not a productivity story.

4. COVID-19: Structural Break or Accelerator?

COVID-19 exposed the fragility of an efficiency-optimised model: supply-chain disruption, fiscal and monetary expansion, inflation, then energy and geopolitical shocks, strategic autonomy and higher investment requirements. But the sequence did not start in 2020. The 2008 crisis weakened the case for unfettered financial integration; the US–China trade conflict from 2018 was an earlier turning point for supply-chain strategy; the 2022 energy shock and the 2026 Hormuz disruption (the second major energy shock in four years) were accelerants. The evidence is consistent with COVID as an accelerator of a pre-existing trend, not as a single structural break, and the data do not allow us to date the break with confidence. The Supreme Court’s 20 February ruling against emergency-powers tariffs, the replacement of the temporary Section 122 duty by Section 301 tariffs from 24 July, and an effective tariff rate of 6.7% (a partial July reading) against 7.7% in 2025 show that trade policy remains in flux, not a one-way retreat.

5. From Efficiency to Resilience: the Resilience Premium

Concept. The resilience premium is the extra economic cost accepted for redundant supply chains, domestic manufacturing, strategic inventories, energy security, semiconductor capacity, critical minerals, defence and technological autonomy. Concentration is real: China refines more than 90% of the world’s gallium, graphite, manganese and rare earths (IEA, 2026). Resilience requires capital, so it can raise investment demand and, if saving does not rise in step, the equilibrium real rate.

Counter-argument. The effect has two horizons. In the transition, building capacity raises demand for capital, materials and labour and is inflationary; in the long run, redundant capacity and diversified supply can be disinflationary and reduce shock volatility. The long-term sign is not obvious, and a durable disinflation from new capacity would work against higher rates. The thesis is strongest for the transition period, which could last a decade.

6. The Inflation Risk Premium and the Return of the Term Premium

Decomposition. A nominal yield is expected short rates plus inflation compensation plus a real term premium plus other risk, fiscal and liquidity components. Inflation compensation is not expected inflation. A breakeven (the nominal yield less the TIPS yield) contains expected inflation, an inflation risk premium, a liquidity premium and technical effects, and should not be read as “the” inflation expectation. The 10-year breakeven was 2.36% on 2 October; euro-area professional forecasters see 2.0% longer term and consumers’ five-year expectations are 2.5% (ECB). Longer-term expectations are broadly anchored while the real yield has risen, so the rise in the nominal yield is mostly real-yield and term-premium, not inflation expectations. A wider distribution of inflation outcomes could raise the inflation risk premium even with a stable central expectation, but model-based inflation risk-premium estimates are not re-pulled in this paper, and we do not claim a measured change.

Term premium. The premium on duration was compressed in the 2010s by quantitative easing, central-bank credibility, safe-asset demand, regulation and low inflation volatility. Table 3 shows that it has returned from negative levels: the ACM premium rose from 0.66% in May to 0.89% in September 2026, Kim–Wright was 1.07% on 9 October and the San Francisco Fed model 1.49% on 5 October, with about 70% of the rise since the 16 September FOMC attributed to the term premium. The BIS links the latest rise in term premia to fiscal pressure more than inflation concerns. This is a return to a more normal level, not an unprecedented one: the Federal Reserve Board’s decomposition of the 2023 peak had premia between 0.4% and 2.0%. Why can long yields rise when expected policy rates do not? Because the premium prices duration risk, supply, inflation uncertainty and fiscal credibility, none of which the policy rate controls.

7. Fiscal Supply and Sovereign Debt

Evidence. CBO projects a US deficit of 5.8% of GDP in fiscal 2026, debt held by the public rising from 101% to 120% of GDP by 2036 and net interest from 3.3% to 4.6% of GDP, and more than $2 trillion by 2036. The IMF puts the US general-government deficit at about 7.5% of GDP in 2026 and gross debt rising from about 126% to about 142% of GDP by 2031, the largest rise among large advanced economies; it also sees US interest payments approaching 5% of GDP by 2031. For comparison, the IMF has euro-area debt near 88% of GDP, Germany rising from about 65% to 74% with its defence and infrastructure programme, France near 118–121%, Italy about 138% falling slightly, Japan falling from about 204% to 193% and the United Kingdom about 104%. Global debt is 93.9% of GDP in 2025 and projected at 100% by 2029 (Fiscal Monitor, April 2026).

The feedback loop and when it does not occur. Higher yields raise interest expenditure, larger deficits mean more issuance, more issuance can lift the term premium, and the term premium lifts yields. But the loop is conditional. The effective rate on US marketable debt is 3.475% and 33% of it matures within a year, so repricing is gradual. If nominal growth stays above the effective rate (r below g), the debt ratio can still stabilise. Fiscal consolidation, productive public investment and strong growth break the loop. The decisive variable is not the level of debt but the gap between the marginal cost of new debt (about 5% on 10-year paper) and nominal growth, and the primary balance.

8. Foreign Demand and Japanese Capital

Evidence. Foreign holdings of Treasuries were $9.25 trillion in July, with official holdings of $3.77 trillion down from $4.01 trillion in February (Japan down $135 billion and China $76 billion; part valuation). Japanese investors made net sales of foreign bonds of ¥969 billion in September and about ¥5.1 trillion so far in 2026, the most since 2022, and sold ¥4.7 trillion of US bonds in the first eight months (Ministry of Finance, Bank of Japan, as reported). The 10-year JGB yield has passed 3% for the first time in three decades, the 30-year is near 4%, and the 10-year JGB now exceeds the FX-hedged 10-year Treasury yield. Life insurers, who manage about $2.5 trillion with a quarter overseas, account for only ¥289 billion of the sales.

Interpretation. The relevant mechanism is the marginal bid, not liquidation. If Japan’s domestic yields rise relative to hedged foreign returns, new savings are allocated at home, and maturing hedged foreign bonds are not replaced. Industry analysts expect a gradual shift, not a broad sale. The Treasury market is deep and domestic institutions are crucial, so Japan alone is unlikely to cause a sustained structural repricing, but a weaker marginal bid from Japan, official reserve managers and the euro area together raises the yield needed to clear supply. This is a risk-premium channel. De-dollarisation of flows, reserve diversification, reduced marginal demand, liquidation and a dollar-system crisis are different things; the data show the first three at the margin only. The dollar is 56.7% of allocated reserves.

9. Strategic Investment, Defence and the AI Question

Evidence. EU defence spending is €454 billion (2.4% of GDP) in 2026 and NATO allies are committed to 5% of GDP by 2035; ECB staff put additional green investment needs at 2.7–3.7% of EU GDP a year to 2030; and the Lagarde hearing of 28 September noted that borrowing linked to technology investment is about a quarter of euro-area credit growth to firms. Geopolitics itself therefore creates demand for capital, which is the link to r*. But whether this lifts the equilibrium rate depends on whether saving rises as well and on the productivity of the spending: grid and semiconductor capacity are not equivalent to untargeted transfers.

AI and productivity: theoretically ambiguous. Technology investment could raise r* by raising expected returns on capital, investment and potential output. It could lower r* if productivity gains need little extra capital, if corporate and household saving rise (profits, inequality, precautionary saving), if labour demand falls or if deflationary pressure appears. Evidence is limited and early. Equity markets already price a large payoff: the S&P 500 forward earnings yield (about 5.3%) roughly equals the 10-year yield, and the ECB has warned about valuations concentrated in a handful of firms financed increasingly by debt. We treat this as one of the largest unresolved variables, and we do not assume it points one way.

10. The Causal Chain, Tested Link by Link

11. Real Estate, Equities, Credit and Banks

Real estate. The 30-year mortgage rate was 7.40% on 8 October, the highest since November 2023 and up from 6.30% a year earlier (Freddie Mac). Valuations adjust slowly because fixed-rate mortgages, long leases, infrequent transactions and gradual loss recognition delay recognition: a property financed on a 2–3% assumption looks different when refinanced at 5–7%. Commercial offices and leveraged property are more exposed than industrial assets and regulated infrastructure with inflation-linked income. A higher cap rate is a repricing, not necessarily a collapse. Detailed cap-rate and transaction data are not re-pulled here, so this section is a mechanism, not a measurement.

Equities. Higher real rates and strong equity performance can coexist if productivity and earnings growth justify them. The S&P 500 forward price-earnings ratio was 19.0x on 30 September, slightly below its five- and ten-year averages (FactSet), so the compression of the equity premium reflects yields, not multiples. But the tailwind of ever-lower discount rates has gone; a 100bp rise in r−g cuts the value of an equity with a 4% r−g by about 20%.

Credit, private markets and banks. The low-rate regime allowed leverage outside traditional banking, and the end of cheap capital affects refinancing, hurdle rates, valuations, exit multiples and defaults in investment grade, high yield, private credit, leveraged loans, private equity, infrastructure and venture capital. The ECB’s May stability review judged energy-intensive sectors most vulnerable and noted that non-banks hold over half of long-dated euro-area government bonds. US banks held $326.7 billion of unrealised securities losses at end-June (5.5% of amortised cost), with return on assets of 1.37% (FDIC). A gradual rise in yields is manageable for banks, because net interest income and credit can adapt, while an abrupt rise is not.

12. Three Structural Scenarios and the Floor

The floor matters more than the peak. A transparent way to frame it: floor ≈ r* + inflation target + sustained term premium. With r* of 0.5%, a 2% target and a zero premium the floor is about 2.5%; with r* of 1%, a premium of 0.5–1% the floor is 3.5–4%; with r* of 1.5% and a premium of 1.5% it is about 5%. In the 2010s yields ranged roughly 1.4–3.2% (excluding the 2020 low). A floor of 3.5–4.5% would mean permanently higher debt service, mortgage rates, hurdle rates and required equity returns, even if yields never reach 7–8%.

Where the evidence points. Current readings are closest to Scenario 2, with a leaning toward 3 on term premium and supply and toward 1 on expectations and growth. A high yield is not itself evidence of scenario 3. The 10-year was 4.45% on 29 May, 5.28% on 7 October and about 5.1% on 15 October (Treasury; Trading Economics), against floors of about 1–2% in the 2010s.

13. A Secular Bear Market: Defined Carefully

A secular bear market in bonds does not mean yields rise every year. It means equilibrium yields are higher, duration offers less predictable capital gains, bond prices face a structural headwind and rallies occur within a higher range. The 10-year could fall from 5% to 3.5% in a recession without returning to the regime in which it traded persistently below 2%. A cyclical bull market within a higher range, 4.5% to 3.5% to 4.5%, is a different regime from 4.5% to 1.5% to 2.5%.

Productive versus risk-premium rates. The same yield has two meanings. Where rates are higher because the economy is more productive, with investment and productivity rising, higher returns on capital, higher potential growth and healthy real yields, the regime is compatible with growth. Where they are higher because risk premia and fiscal uncertainty are rising, with deficits, volatile inflation, weaker foreign demand and crowded-out investment, the regime brings higher borrowing costs, weaker growth and financial stress. Current evidence has features of both: strong US growth and technology investment on one side, large deficits and a rising term premium on the other.

14. What Would Falsify the Thesis?

• Real long yields falling persistently toward pre-pandemic lows.
• Term premia compressing again while deficits stay high.
• Fiscal consolidation and lower issuance, or a durable fall in net interest as a share of GDP.
• A renewed global saving surplus, strong demographic saving or safe-asset demand.
• Re-acceleration of globalisation, or weak investment and productivity stagnation, or technology that proves disinflationary.
• Inflation expectations staying firmly anchored, and stronger foreign demand for duration.

The most informative test is the next easing cycle: if long yields fall with policy rates to the 2–3% area, the thesis weakens materially; if they stay above 4%, it strengthens.

15. Cross-Country Comparison and the Link to the Sovereign Stress Series

Our stress tests ask what happens under a severe 450bp ECB and 525bp Fed shock: in the Treasury paper, the 10-year reaches about 8.25% from 4.45% in May, in the OAT paper about 7.75% from 3.65%, and in the BTP paper about 7.5% from 3.67%. This paper asks a different question: what if the structural equilibrium underlying long-term yields has changed? Treasuries are the global benchmark and reserve asset; the OAT carries the European fiscal and sovereign-risk repricing; the BTP is more sensitive because of its debt stock, spread dynamics and the sovereign–bank nexus. A higher structural floor would shift the starting point of all three stress tests upward.

16. Institutional Monitoring Dashboard

Conclusion

Has the world moved from an era of structurally falling discount rates toward one in which the price of long-duration capital is structurally higher? The evidence supports a more limited conclusion than the headline question. The compressed term premium of the 2010s has partly gone; fiscal supply is heavy; the marginal foreign bid, notably from Japan, is weakening; and strategic spending is adding to investment demand. These make a return to the lowest yields of 2020 unlikely in the near term. But r* estimates are low and contested, expectations remain anchored, and demographic, debt and technology forces could still pull yields down. The most probable reading is a higher and more volatile range with a floor above the 2010s, not a straight-line bear market.

The post-pandemic rise in long-term yields is therefore best described as a cyclical normalisation with a structural component whose size remains unproven. The test is the next easing cycle. If yields fall back toward 2–3%, the cycle is dominant; if the floor holds above 4%, the world has entered a regime in which capital is no longer structurally cheap, and the most important repricing in housing, equities, credit, public finance and sovereign duration may still be ahead.