What's not to like about rising bond yields?
AI and duration risk
The continued run-up in global bond yields is widely seen as problematic alongside stretched equity market valuations. Those looking to bonds to hedge equity exposures have to wear the drag on performance, although bonds are also increasingly attractive on a relative value basis. 2026 is already landing on the negative side of the historical distribution of annual returns.
The rise in yields is variously said to be a function of the supply shock coming out of the Gulf, expectations for tighter monetary policy and concerns about long-run fiscal sustainability. You can weave a consistent narrative around all of those themes, but that leaves the rise in yields narratively over-determined. Any one of those factors would have been seen as a serious challenge to equity prices on other occasions.
We can also tell a more positive story around rising bond yields. Higher bond yields reflect expectations for higher future economic growth and inflation. The AI-related investment boom is placing increased demand on global capital markets relative to the supply of savings. The expected productivity gains from AI adoption should be reflected in higher yields. In this context, monetary policy tightening is not necessarily a tightening, but an endogenous adjustment to a rising equilibrium real interest rate and might not be an effective tightening at all.
Expectations of higher productivity and economic growth are positive for equity prices, helping reconcile those valuations with higher yields. This more positive narrative could be derailed by various shocks, but so far has proven resilient to a very big, albeit slow moving, shock coming out of the Gulf.
As a recent paper from the Dallas Fed noted, AI-related investment demand is introducing a significant amount of duration risk into rates markets and the Treasury market is where duration risk is supplied and traded. This no doubt accounts for some of the yield curve steepening we have seen recently. The rising term premium is perhaps telling a similar story. The authors caution against attributing non-Treasury duration supply effects to fiscal, monetary or regulatory policy.
Basil Halperin would perhaps argue that the run-up in yields hasn’t been large enough if AI is going to be transformative for economic growth, but the increase to date could reflect reasonable doubts about the speed and scale of AI diffusion. Perhaps the bond market is landing somewhere between various scenarios. US Treasury yields are still not all that cheap relative to nominal GDP growth, although the post-COVID valuation gap has closed substantially.
Outside the US
There are various country-specific stories told around rising bond yields, but for these to be convincing, we need to see significant changes in relative bond market performance not otherwise straightforwardly explained by fundamentals. For example, Australia’s 10-year bond yield spread to the US maps fairly neatly onto the spread between Australian and US nominal GDP growth.
Japan is an interesting case, where the rise in yields arguably reflects the successful reflation of the Japanese economy and not the long-standing concerns about fiscal sustainability that pre-date the post-Abenomics rise in bond yields. The success of Japan’s monetary policy-led reflation is denied or dismissed by many market participants who still view QE as an exercise in financial repression rather than successful monetary accommodation. Japanese equities are outperforming the US and the world ex-US even as Japanese bonds underperform.
Fiscal positions are a poor predictor of bond yields
The least convincing part of the bearish narrative around rising yields is the claim that they are the canary in the coal mine for fiscal sustainability and an indication of looming fiscal crises. Government spending and budget deficits add to the call on global capital markets, but sovereign risk premia are small relative to other determinants of bond yields. As the chart below suggests, government debt-to-GDP ratios are a very poor predictor of bond yields and volatility on a cross-country basis, especially in the DM sovereign credit space.
Fiscal positions are also not a great predictor of bond yields over time. As noted here previously, US bond yields were higher in the late 1990s when the US was running large fiscal surpluses and when productivity growth was also strong. Australian long bond yields had a wider positive spread to the US in the mid-2000s when Australia had a negative government net debt position. It is perfectly reasonable to ring the bell on long-term fiscal challenges, and rising yields do make those challenges more problematic, but it is not a useful way to think about the determination of bond yields or the role of bonds in portfolios.
AI and fiscal sustainability
It has been suggested that AI-driven increases in interest rates could be more problematic for fiscal sustainability outside the US given the role of the US as an effective price-maker in global capital markets. If higher US yields spillover into other jurisdictions, but AI-driven productivity gains do not, then yields outside the US could rise relative to economic growth rates. That potentially tips the balance between the r and g terms that determine long-run fiscal sustainability. This is a much-discussed issue in the EU, where officials are concerned that higher yields and investment demand in the US are channeling EU savings into US assets and the US tech sector at the expense of the EU, while leaving the EU exposed to US financial cycles.
The Australian experience in the late 1990s is potentially instructive. Australia was a leading adopter but negligible producer and net importer of ICT goods, but still saw a significant productivity uplift as a result. Australia also benefited on the income side, as ICT goods saw a secular decline in price while Australia’s commodity export prices rose on the back of China’s industrialisation and urbanisation, delivering what now looks like a permanent uplift in the terms of trade and national income.
Whether we can pull off a similar trick in response to AI is an open question. Regulatory settings are arguably less favourable than they were in the late 1990s. It is an often noted irony that EU efforts to regulate big US tech firms only served to entrench US tech dominance by creating a hostile regulatory environment for innovation on the part of European firms. Big US tech may have been the target, but the EU paid the price. Like everything else, the concern about the impact of AI on fiscal sustainability outside the US hinges on views about AI diffusion, for which we do not yet have good answers.
From an asset allocation perspective, bonds still have a role in cushioning against the negative demand shocks that could be expected to weigh on equity market performance. Given a richly priced equity market, that role seems more, not less important.