Rates justify their rally, and AI isn't a credit problem yet
Volatility has spiked in financial markets over the past couple of weeks, driven by the tech and AI-related sell-off, US midterm elections and falling oil prices. In fixed income, spreads have been volatile but better supported than equities. As in equities, the split of “winners and losers” has varied by industry. Rates have also rallied significantly (more on this below), so both credit and government bonds have delivered positive returns through this volatility.
The moves in rates have been particularly interesting. Exhibit 1 below shows the 10-year Treasury Inflation Protected Securities (TIPS) yield, or real 10-year rate, and the 10-year inflation breakeven, which is calculated by subtracting the 10-year TIPS yield from the 10-year Treasury yield. The peak in 10-year real rates coincided almost perfectly with the peak in the Nasdaq index, while the 10-year inflation breakeven peaked just a couple of days before Brent oil prices fell from their recent highs of $120 per barrel.
Worries about the profitability of the AI trade and the speed with which data centres will come online have led to a rally in real yields (falling yields have pushed up bond prices) due to three interrelated reasons. First, markets have realised that funding for data centres and other projects might not be as easily available as previously expected. Second, this means there might be less supply of bonds and other forms of credit. Real yields are conceptually the rate that brings savings and investments in an economy into equilibrium, so all else equal, less supply (which means less investment) than previously expected should take real yields down. Third, if there is less supply and less investment, then growth might slow down, and lower growth brings about lower real yields. These are the ingredients that typically cause a "risk off" move, with lower growth expectations prompting central banks to relax monetary policy in the future and inflation expectations falling in line with a less rosy outlook.
Oil prices reached $118 per barrel on October 30 and then fell sharply, this provided another boost to 10-year Treasuries inflation expectations as shown in Exhibit 1. We strongly believe this is the main reason for the move in inflation expectations rather than investors pricing in a more recessionary outlook. However, at 2.25%, the 10-year inflation breakeven is at the low end of the post-pandemic range and it is therefore difficult to see this falling much further in the absence of a recession. We would also point out that inflation breakevens are not the only mechanism through which lower oil prices can lower nominal yields. To the extent that higher oil prices pushed monetary policy rate expectations up earlier in the year, those expectations have also come lower as oil prices have eased.
Growth and the labour market have remained robust, and we have not seen economists revising their growth projections down at this stage. We think this makes sense, and it is in great part behind the fact that spreads have been relatively resilient. However, we highlight that index-level spreads hide a lot of dispersion. Lower rated credit has moved materially wider in some places, and investors have zero appetite for holding onto more challenged positions. Fast money accounts sense this too, which makes it an easier choice to lean on these lower rated credits because no one steps in their way.
As the situation currently stands, our interpretation is that equity markets are reassessing how profitable the AI trade is and who will be entitled to those magnificent returns on equity that are being projected for coming quarters and years. As a result of this reassessment, those business models that look weaker, less sustainable once the current compute glut sorts itself out, or too highly leveraged (given that raising debt has become more difficult), are being challenged by investors. This is consistent with markets being more discerning about the AI trade, which is why the sell-off has not been uniform across all sub-sectors.
Going forward, fixed income managers are in the somewhat uncomfortable and unusual position of watching the Nasdaq for clues on what rates and spreads might do in the short term. Our big-picture view is that this is, for the most part an equity market issue. Expectations that every single AI investment will make trillions of dollars in short order are probably too optimistic and need to be tamed, not least because data centres may well come online more slowly than previously expected. While this would necessarily mean slower growth projections for 2027, we do not think at this stage that the potential downward revisions will end up being severe enough to cause a recession. Therefore, our views that spreads of highly rated corporates and banks represent reasonable value, and that defaults will not spike higher, remain unchanged. Our relatively positive views on government bonds have not changed dramatically either. Real rates have rallied significantly from recent peaks, but this was for good reason, and they remain close to 25-year highs anyway.
However, we are aware that markets can overshoot, and when there is a rally of the magnitude that tech equities have had in recent years, at some stage it becomes harder to justify holding onto these. With our credit analyst hat on, the best time to buy credit is when spreads widen on contagion that does not necessarily translate into higher defaults, or deterioration in broader credit metrics. If we did have a major correction in spreads due to tech stocks continuing to move lower, we would see this as a great opportunity to increase exposure to credit (sector and rating band depending on how price action develops). This is a scenario that we definitely do not rule out, and it would be almost guaranteed to be accompanied by a rally in government bonds. In fact, last week felt slightly panicky at times, with sharp moves lower in rates, which are usually indicative of a "sell now, ask questions later” environment. This is a reminder to investors that government bonds are likely to work as a risk-off asset, at least in the short term, even if budget deficits and debt-to-GDP ratios look poor.
For the time being, our preference is to rotate positions by taking advantage of new issue premiums that have made a comeback as investors have become more cautious, while maintaining sufficient liquidity to act quickly if needed. Although lower rated credit has moved materially wider, we are not tempted to increase allocations in this space. At this stage, in spite of the decompression in the last few weeks (months for certain sectors), there is little evidence of indiscriminate selling. When we peek under the hood, we find that the weakness in credit has been concentrated in the names we are already cautious on, while the names we favour have not sold off dramatically. Even so, the additional spread on offer in those unloved credits is not yet enough to bring the love back.
