AI-related investment and US fiscal incontinence add to the existing tendency for the global rates structure to average higher than pre-pandemic. Expect more US government action to contain debt-servicing costs.
- Bond yields have been rising in recent weeks, especially at the long end of the maturity spectrum. We have long believed that the global structure of interest rates would have a higher centre of gravity in future than it did in the period between the GFC and the pandemic. The factors often cited for the recent rise in yields – larger government deficits globally, the investment boom from AI – were already behind this judgement.
- The question then arises: have we allowed enough for these forces? Getting the direction right is one thing; quantifying the impact is another. In doing so, though, we must take care not to confuse trend and cycle. Some of the recent rise stems from the exceptional demand for funding data centre construction. This effect of the technology wave could last a while, but it is more a cyclical development than a permanent structural one.
- For the Australian economy, developments at the long end of the yield curve tend to be less relevant. Most financing happens at the short end of the maturity spectrum, especially for households. But there are borrowers and investors who are affected by long-dated yields, especially governments.
- Higher yields mean higher debt-servicing costs, particularly for the US government. Expect more manoeuvres like the US Treasury supporting the yen by selling euros not dollars, as well as the buy-back operation this week. These are merely stop-gaps rather than a lasting solution like genuine fiscal consolidation.
Westpac Economics has long held the house view that the global structure of interest rates would be higher in the future than it was in the period between the GFC and the pandemic. Central bank policy might influence the short end, but over a longer period, what matters for interest rates globally is the global balance of saving and investment. Pre-pandemic, saving rates were high, thanks in part to official sectors accumulating reserves. Meanwhile, investment was weak. Confidence had been frayed by the crisis, and some governments were in austerity mode. That decade’s new technologies did not spark the same kind of productivity-enhancing investment boom as computers and the internet did in the late 1990s and early 2000s. And in any case, banks in North America and Europe were recapitalising and less inclined to lend.
There were good reasons to see that period as an aberration, though, including the very long-term evidence we noted two years ago. As we highlighted at the time, a range of forces have been pushing “neutral” rates higher, along with the structure of interest rates more broadly. These include higher government spending as populations age, as well as the private sector needing to invest in both energy transition and AI. Higher defence spending was also in the mix, especially after Germany changed its constitution last year to allow more deficit spending for this purpose. The general fiscal incontinence of the US government was already evident two years ago and worsened under the current administration. All these factors imply greater bond supply, and so higher yields.
To unpack this further, a bond yield is normally thought of as being the combination of several components that are easier to name than to measure. Nominal government bond yields are composed of (expected) inflation plus real yields, where real yields are normally assumed to reflect expected future (real) policy rates plus a “term premium”. The term premium is in principle the compensation investors require for the risk that things turn out differently from their expectations of the other two components, and in practice is a residual grab bag of everything else that might affect demand and supply in the bond market. Corporate bonds add a risk spread to this, reflecting that a company might default but a sovereign issuing in its own currency will never need to.
The higher global structure of interest rates can be seen in all these components. Inflation generally undershot central bank targets in the 2010s; since the pandemic, it has at best been around target, with global supply shocks periodically lifting it above target. Real yields are higher, and with shorter-maturity bonds now paying positive real yields, we are seeing less of the “search for yield” behaviour that compressed both term premia and risk spreads in the 2010s.
But how much?
It is one thing to say rates will be higher on average. We also need to ask: by how much? This is where the AI boom comes in. The scale of the planned global investment in data centres is nothing short of mind-boggling, and it is creating substantial demand for debt finance. The “hyperscaler” tech firms (Alphabet, Amazon, Meta and Microsoft) used to be enormous cash generators. Now they are responsible for some of the largest debt issuances in the market, including Alphabet’s super-sized $A deal this week. Issuance in Australia and other non-US jurisdictions makes sense for these firms. They want to access a diversified pool of investors, and many of the assets they are financing are outside the US, especially now that data centres have become unpopular in many US communities.
The scale of this greater investment call on global savings is one reason to expect longer-dated yields to average even higher in the near term than their range since the pandemic so far.
We must take care, though, to distinguish trend from cycle. This is particularly relevant for the data centre boom, which is likely to be a big thing for the next few years, but settle down as the technology matures and computing capacity expands (and becomes cheaper).
The US government is also making an exceptional call on global saving, and this looks more like a longer-term structural trend than the AI boom does. The US federal government has been running deficits in excess of 4% of GDP for a decade, aside from a brief period in 2022, and more than 6% since the beginning of 2023.
In recent weeks, rumblings of investor discontent about the US fiscal position have become evident in yields. And those hyperscaler bonds, including the $A-denominated ones, are offering a higher yield for companies with similar ratings to the US sovereign. The bonds of governments in better fiscal positions, including Australia’s, also start looking attractive. To this we must add that Japan now looks to be becoming a “normal” economy with positive inflation and interest rates, and an undervalued currency that might tempt some investors to bet on an appreciation. While it remains the case that There Is No Alternative to the US Treasury market for depth and liquidity, some diversification out of USD assets is starting to look both more feasible and more attractive.
The other consequences
For the Australian economy, developments at the long end of the yield curve tend to be less relevant than those at the short end. Most financing happens at the short end of the maturity spectrum, especially for households. But some borrowers and investors are affected by long-dated yields, especially governments, and the local firms involved in data centre construction.
Offshore, we can see higher yields constraining policy decisions by the US government and can expect more on this front. The US government’s debt-servicing costs are becoming more burdensome as large deficits persist and yields rise. Just in the past few weeks, we have seen some unusual currency interventions by the US Treasury, followed by buy-backs designed to shorten the maturity profile of US government debt and lower the term premium paid. And while we expect the Fed will not fold in the face of pressure to keep rates low, we can well imagine the content of the phone calls between FOMC Chair Warsh and President Trump. Expect more manoeuvres by the US authorities along the lines seen recently. These are merely stop-gaps, though, not lasting solutions like genuine fiscal consolidation.
