Sovereign Debt

The ultimate faith trade

The recent sell-off in government debt in the U.S., U.K., Europe and Japan has caused many commentators to conclude that investors are now suddenly worried about state balance sheets. With debt/GDP ratios in excess of 100%, ballooning deficits and facing a change in the economic world order, the fact is that none of these issuers are capable of paying off their debts.

But investors have known this for a long time. This has not stopped them from devoting large percentages of their asset allocations to government debt, with the main destination being the U.S. Treasury market.

The level of interest rates is a reflection of more than just fear. Now governments are not the only game in town for long-duration fixed income securities, and the term premium (the spread between the long-dated bonds minus the short-dated bonds) is no longer suppressed by central banks. Moreover, there is the growing suspicion in the market that the stock of debt will continue to rise because of the promises made by administrations to the electorate in terms of spending and benefits. What we are seeing is the repricing of the faith trade that has been running for decades.

No, investors haven’t just “woken up” to the risks

I am not buying the accepted narrative that investors have suddenly got religion and woken up to fiscal risk. If that were indeed the case, the past fifteen years would have looked very different. Japan has lived with 200% for a long time by borrowing from itself in its own currency. Italy’s debt/GDP ratio has rarely been less than 130% in the post-crisis era. The United States crossed 100% after 2008 and has not looked back. And France and the U.K. have now joined the HIN Club ( “ Highly Indebted Nations”). Moreover, these number don’t take into account the billions of off-balance sheet liabilities run up by sovereigns which include the unpaid receivables due for goods and services to the private sector.

Germany is the cleanest dirty shirt among the G7 nations, but even Berlin is borrowing more as its industries are facing an energy crisis brought on principally by the Ukraine war[1] . Reinhart and Rogoff (2010) Growth in a Time of Debt found that across two centuries and 44 counties the debt-growth relationship is weak below a 90% of GDP threshold. In other words, instead of boosting growth, high debt/GDP rations actually erodes growth. In a 2012 paper following-up their previous analysis, the economists founds that in prolonged periods of high debt rations, growth was over 1% point lower than other periods[2] .

This is the negative debt spiral: as indebtedness rises, so do interest payments as a percentage of the government budget rise, inhibiting spending on pro-growth policies. This is the picture of the G7, and investors have known this for a while, and yet they still bought the bonds anyway because sovereign debt is part of the financial plumbing — it is what serves as collateral and the regulators tell us it is a “safe asset” (“risk free” in the jargon of financial modeling). Again: it has been about faith not ignorance.

The risk hasn’t changed. It’s the competition

The level of yields on sovereign debt are also indicative of where investors are putting their money. There is renewed competition in the fixed-income debt space. Let’s take the new new thing: financing the AI build-out with bonds issued by the hyperscalers. In 2025, Alphabet, Meta, Microsoft and Oracle issued ~$120 billion of bonds, and forecasts for 2026 are running at about $200 billion. Bonds linked to the AI ecosystem are estimated at ~30% of net new dollar investment-grade supply[3] . This new issuance is finding its way into investors’ fixed income portfolios and replacing U.S. Treasuries (which have to be sold to make room).

Many of the bonds issued to support the AI build-out are not just senior, unsecured obligations of the hyperscalers themselves. They are often issued by special purpose vehicles, such as Beignet Investor in the Meta and Blue Owl $27 billion “A”-rated deal. The bonds are serviced by lease payments on a 5-gigawatt datacenter, guaranteed by Meta, which also provides a “residual-value”guarantee in the event the lease payments terminate early. The structure incorporates a sinking fund structure which repays the interest. The investor is taking the risk on a contracted cash flows from the tenant paid into a bankruptcy-remote SPV[4].

This is not a new piece of financial engineering. Structures such as these are commonplace in the commercial real estate and project finance world allowing the sponsors such as Meta to keep the debt off their balance sheet, yet provide an interesting return to bondholders.

According to Morgan Stanley estimates, securitized data-centre credit could supply around $150 billion between 2025 and 2028. Since 2018 the stock of that paper has already grown past $50 billion[5] . These bonds come at an attractive spread over Treasuries and offer exposure to a high-growth area of the economy, and while sovereigns have the monopoly on the printing press, they cannot compete with a sector which will dominate everybody’s lives for years to come.

Inflation. What inflation?

Saying yields are rising in the sovereign debt market because of inflation is a convenient ruse. The market for inflation swaps (where you can pay or receive the rate of inflation) is telling us otherwise. It is true that energy prices mainly due to the war with Iran are keeping inflation prints sticky, particularly in the eurozone, with the rate for August near 3.3%. But this is one print. It is not the same as looking at where the market prices forward inflation. The U.S. 5-year-5-year forward inflation rate has barely changed from around 2–2.3% over 12 months. The 10-year breakeven has hovered around 200 bps. Whereas the front end of the inflation swap curve does move with energy prices, the long end hasn’t so far[6] which tells me that investors are seemingly indifferent. Otherwise, the long-end breakevens would have moved higher.

The term premium has risen: the curve has steepened reflecting fiscal risk and a concern for increase supply in the long-end. Because that’s the thing with sovereign debt: you are not lending to an entity with assets as collateral for your loan. You are literally banking on its ability to refinance the debt in future.

Europe and the UK

The real concern is in Europe and the United Kingdom, where left-leaning governments are making promises that can only be kept with an increase in financing. Defense spending, once for years benefiting from U.S. profligacy in its NATO contributions is of a real concern.

In 2025, (most) NATO allies committed to 5% of GDP on defense and security by 2035–3.5% on core defense and 1.5% on related infrastructure, cyber and resilience. The UK intends to lift core NATO spending toward 2.7% of GDP this decade although this may face some internal Labour party opposition , with a political commitment to 3% in the next parliament and 3.5% by 2035. The Defence Investment Plan puts almost £300 billion through the Ministry of Defence over four years, with the annual budget heading toward £80 billion by 2029[7] .

These increased requirements are bound to put pressure on the gilt and Eurozone bond markets, together with social programs, and healthcare spending.

Pensions: the ticking fiscal time bomb

Pubic pension systems in Europe are pay-as-you-go promises dressed up as institutions, which have largely incestuously invested in government debt. The circular nature of the payments (as in Japan), may have guaranteed that pensioners will not suffer large drawdowns (unless there are defaults), but they also won’t get the benefits of the higher returns afforded by investments in equities. In addition, European public pension systems suffer the bond math conundrum: as rates fall the mark-to-market of the assets rise, but the returns on bond the portfolios fall . The UK had a solution for this: pension funds used their sovereign bonds as collateral for total return swaps in order to invest in riskier assets promising higher returns. Except when Liz Truss’ mini budget came out in September 2022 and gilt yields rose rapidly, the pension funds were caught with margin calls and declining collateral values.

Aside from the lower returns, the math is demographic. Europe is getting older. The working-age/retiree ratio in Europe is declining from ~3 workers/retiree to a predicted value of less than 2 by the mid 2040s. Italy spends ~15% of GDP on pensions, with France and Greece not far behind at 14%. In Germany, it is estimated that around a third of federal tax revenues will be needed to plug the pensions deficit in a single year. The UK’s fiscal watchdog, the Office for Budget Responsibility says that the state pension will rise from the current 5% of GDP to just under 8% by the early 2070s — and that number doesn’t include increases in health and social care which are also bound to rise.

If things continue as they are in the UK, the debt/GDP ratio could rise to more than 200%[8] . In France, audit bodies have sketched a pension deficit that could widen from a rounding error today toward €15 billion by 2035 and €30 billion by 2045 without further reform[9] .

Get Return Seeker’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

This doesn’t look good for bond investors: the pension systems across the developed world are basically open-ended political claims driven by lower contributions as workers get older and the demographics don’t collaborate. The idea to import new, young, foreign contributors may have sounded like a solution, but it creates other social problems as we are seeing in the UK and Spain. But that’s another story. In a nutshell, as pension deficits widen, it only points to thinner future benefits and more government debt.

Let’s talk tariffs

While we debate who has won or lost the trade war, we forget the effects of higher tariffs on European and UK exporters which will undoubtedly reduce the volumes of exports to the world’s largest consumer market as well as squeeze margins, impacting corporate profits and thus reducing tax revenues for the state[10] .

UK goods exports to the U.S. fell about 11% in 2025; the effective tariff on UK goods into America has been running around 8%, with the EU higher.

This doesn’t mean that investors necessarily are thinking that a default is around the corner, but it does mean that there will be more supply coming on stream pretty soon.

European governments have be loath to pool their borrowing as this would be really seen as taking away a large amount of their individual powers. Yet, behind the scenes it has been happening with bond issues from the European Community, funds from the EU to cope with the economic devastation caused by the Covid-19 pandemic and other bits and pieces around the edges which effectively are joint issuance projects. The real issue is the fact that the NextGenerationEU plan of rolling over the ~of €650 billion of recovery stock really does kick the can down the road: the current generation gets all the benefits but future generations have to pay the price in the hope that fiscal probity and growth will soften the debt load.

This is called “refinancing” by the bond market and has become an accepted part of sovereign debt. Nobody in government really wants to be the little boy and tell teh world that the Emperor has no clothes. The answer is essentially a gigantic shrugging of the shoulders and a higher yield, a larger term premium, perhaps a longer maturity.

The faith trade

You cannot analyze sovereign debt in the same way as you perform a credit analysis on a corporation. When you lend to a company on a senior, unsecured basis, you are first in line to receive anything after taxes and wages have been paid out from the carcass of the firm. If a sovereign defaults, it is much harder to seize an asset (although that has been tried with Argentina reportedly, with the jet fuel inside the planes of Aerolineas Argentinas). For a reserve currency borrower like the U.S., the state can always pay in dollars (that’s why CDS on the U.S. are quoted in euros, for example). If a country like Japan defaults, it can always pay in yen. The UK can always pay in sterling. Russia, interestingly enough, in 1998 chose to default on its domestic currency debt — the debt that everybody thought it would not default on, because a lot of the short-term bills had been repackaged and swapped into dollars with the FX swap provided by local Russian banks.

With G7 debt, allocators are seemingly oblivious to rising debt/GDP levels, even when the ratio climbs above 90% and enter the Rheinhart-Rogoff danger zone. Everybody knows that the U.S. is never going to redeem the principal of its bonds with cash (what a funny thought that would be!). The faith trade is that the debt will always be rolled over, and despite the contrast with corporate issuance (where it is expected that the company repays its debts) it is perceived as the less risky place to store your money and gain duration to cover your long-dated liabilities. And if bad comes to worse, the central bank will step in to buy back the long-end, repressing yields to a level which makes it acceptable for the government to borrow at. They are the bid of last resort in a sense. But the question is have they spent all their ammunition? The ultimate faith trade is still being made every time a pension fund, a reserve manager or a bank treasury rolls a 10-year government bond. The congregation is simply demanding a higher vig.

Footnotes

1. See IMF, Fiscal Monitor, April 2026, general government gross debt. For context, the same source puts Japan at ~204% of GDP, Italy ~138%, the United States ~126%, France ~118%, Canada ~111%, and the United Kingdom ~104%.
2. Carmen M. Reinhart, Vincent R. Reinhart and Kenneth S. Rogoff, “Public Debt Overhangs: Advanced-Economy Episodes Since 1800,” Journal of Economic Perspectives 26, no. 3 (2012): 69–86. In episodes where public debt stayed above 90% of GDP for at least five years, growth averaged about 2.3% versus 3.5% in other periods (roughly 1.2 percentage points lower). The 90% threshold itself is from Carmen M. Reinhart and Kenneth S. Rogoff, “Growth in a Time of Debt,” American Economic Review 100, no. 2 (May 2010): 573–78.
3. Bank of America and other dealer compilations put 2025 issuance by the large hyperscalers at about $121 billion. 2026 forecasts range from roughly $175 billion to $300 billion depending on definition; Reuters (September 2026) reported about $220 billion year-to-date from the five largest names. AI-related paper has been estimated at around 30% of net new US dollar investment-grade supply.
4. Beignet Investor / Hyperion: special-purpose vehicle financing of about $27.3 billion for Meta’s planned ~5 GW campus in Richland Parish, Louisiana, co-owned with Blue Owl. Bonds are serviced by lease payments, with Meta providing lease support and a residual-value guarantee; a sinking fund is designed to repay principal (not only interest).
5. Morgan Stanley research on securitised data-centre credit: more than $50 billion raised since 2018, with an estimated $150 billion of potential ABS/CMBS supply over 2025–28. Comparable structures include QTS-Microsoft campus bonds and Hut 8 project bonds.
6. Federal Reserve Bank of St. Louis (FRED) and related market data: US 5-year, 5-year forward inflation expectation about 2.33% on 4 September 2026; 5-year breakeven about 2.37%; 10-year breakevens in the low-2% area through mid-2026.
7. NATO Hague summit, June 2025: Allies committed to 5% of GDP on defence and security by 2035, of which 3.5% core defence and 1.5% related infrastructure, cyber and resilience. UK Defence Investment Plan, June 2026: almost £300 billion through the Ministry of Defence over four years, with the annual budget heading toward about £80 billion by 2029 and core NATO spending toward 2.7% of GDP this decade.
8. Office for Budget Responsibility, Fiscal Risks and Sustainability, July 2025. On current policy the OBR’s long-term scenario still has UK debt rising onto a path above 200% of GDP (and, in the published central long-term exercise, toward ~270% by the early 2070s) as ageing, the state pension and health and social care compound. State pension spending is projected to rise from about 5% of GDP to about 7.7% by the early 2070s.
9. French audit-office / Cour des comptes estimates reported in 2025–26 coverage: the pension-system deficit could widen toward about €15 billion by 2035 and about €30 billion by 2045 without further reform. Broader European pension-cost figures: Italy a little over 15% of GDP; France and Greece above 14%; German estimates that a large share of federal tax revenue is already absorbed plugging the statutory pension system.
10. UK parliamentary reporting and ONS trade data: UK goods exports to the United States fell about 11% in 2025, from roughly £66 billion to £59 billion. The effective tariff on UK goods into the US has been running around 8%, with the EU higher (around 10% in comparable tallies). Tax Foundation and related work on 2025–26 tariff incidence find a material share of the burden falling on foreign exporters, which squeezes overseas margins and, in principle, corporate tax receipts in the exporting country.

References

• Morgan Stanley Research. Bridging the Data Center Gap (securitised credit estimates).
• Bank of America Securities / LSEG market compilations on hyperscaler bond issuance, 2025–26.
• Federal Reserve Bank of St. Louis (FRED). 5-Year Breakeven Inflation Rate; 5-Year, 5-Year Forward Inflation Expectation Rate.

Originally published at https://returnsseeker.substack.com on September 7, 2026.