Long-term yields, a proxy for the cost of capital, tend to track GDP growth, a proxy for the rate of return on capital: the US has gone through sustained periods of higher yields with higher growth than currently.
This article is only available to Macro Hive subscribers. Sign-up to receive world-class macro analysis with a daily curated newsletter, podcast, original content from award-winning researchers, cross market strategy, equity insights, trade ideas, crypto flow frameworks, academic paper summaries, explanation and analysis of market-moving events, community investor chat room, and more.
- Long-term yields, a proxy for the cost of capital, tend to track GDP growth, a proxy for the rate of return on capital: the US has gone through sustained periods of higher yields with higher growth than currently.
- The 400bp increase in long-term yields since the lows of 2020 reflect a combination of economic normalization, higher growth, market fears of a long-term move away from zero interest-rate policy (ZIRP) and quantitative easing (QE).
- While the US is more leveraged now than during the era of high yields, strong private sector balance sheets and the dollar’s global status suggest the US economy can withstand higher long-term yields.
- The relationship between long-term yields and GDP growth suggests yields have more upside.
- Federal Reserve (Fed) speakers have signalled that staying on hold in November is likely, but I think an extra 2023 hike remains the Fed’s base case.
- A trigger for a renewed bond selloff could be the Fed signalling that the extra 2023 hike is still likely at its December meeting.
Higher Yields Reflect Higher Growth
Long-term yields are back to the levels of the early 2000s, and investors are wondering whether this is sustainable. In this note, I argue that over the longer run, yields are likely to rise further.
During 1970-2000, 10yr yields were higher than they are currently, yet real GDP growth was also higher at a 3.3% annual average vs 2.1% during 2000-19 (Chart 1).
This reflects that nominal/real yields and nominal/real GDP tend to track. A regression of nominal GDP over 10yr yields during 1970-2023 yields an adjusted R2 of 75, and a regression of real GDP over 10yr TIPS yields during 1997-2023 yields an adjusted R2 of 0.62. The intuition is that the cost of capital, proxied by long-term yields, should equal the return on capital, proxied by GDP growth (this is the essence of the neoclassical growth model – more details here).
Simultaneously, the positive spread of GDP growth over yields is a recent phenomenon: up to 2000, yields tended to be higher than GDP growth. The spread increased after the GFC.
In previous research, I have argued that changes in the GDP/ yield spread reflect largely financial repression. Up to the 1970s, several regulatory measures (e.g., Regulation Q or the Interest Equalization Act) repressed yields. These were eliminated in the 1970s, when yields rose above GDP growth.
With the low inflation era of the 2000s, yields fell again below GDP growth, possibly because low inflation brought about a negative bond-stock correlation that supported demand for bonds. The GDP growth/bond yields spread widened after the GFC, likely due to QE, in my view a form of financial repression.
Adieu QE, Hello Higher Yields
In 2020, the GDP/yield spread reached its highest level since the 1970s (Chart 2). Since then, the spread has narrowed due to a combination of slower growth and rising yields. The 400bp increase in 10yr yields since the low of 2020 could reflect three factors:
Normalization: Up to about 2021, there was some normalization of yields associated with the end of the COVID-19 pandemic panic and associated extraordinary policy easing.
Recession expectations falsified by data: Since 2022, there has been a tug of war between ‘recessionistas’ and ‘inflationistas’, with the recessionistas consistently underestimating growth dynamism and overestimating the risks of large Federal Funds Rate (FFR) cuts. By mid-2023, the data had turned decisively for the inflationistas with, for instance, a marked acceleration in GDP and Fed hikes rather than cuts.
Regime change: Over 2023, there have been growing fears that the long-term macro backdrop will remain too inflationary for the Fed to resume ZIRP and QE.
Moving away from the ZIRP would first manifest as an increase in the long-term FFR. Several market participants expected Fed Chair Jerome Powell to discuss an increase in R* in his Jackson Hole speech. This did not happen, but the September FOMC decision to reduce the number of 2024 cuts to two from four, together with the new FOMC mantra of ‘higher for longer’, has likely contributed to the bond selloff. San Francisco Fed President Mary Daly is likely to have fuelled market concerns further by stating that the long-term FFR could be as high as 3% (but not 5% she said), compared with 2.5% in the SEP.
I think the current level of the long-term dot is out of step with growth and inflation trends (Chart 3). I therefore expect Fed speakers to build towards an increase in the long-term dot in the runup to the December meeting (when the Fed will produce a new SEP).
Long-term, market expectations that the era of ZIRP and QE is over could see long-term yields rise above GDP growth, the situation prevailing up to the 2000s. However, this would raise debt service burdens.
Higher Leverage Is Sustainable
Compared with the 1970-2000 period, the US economy is more levered but overall seems likely to withstand higher yields (Chart 4).
The US Treasury has seen the sharpest rise in indebtedness during the past 20 years, more than doubling to 105% of GDP from a low of 38% in 2001. Until now, the increase in debt has been offset by a decrease in yields (Chart 5). This has started to change even though current weighted average maturity (WAM) of marketable Treasury debt, at 73 months, is higher than the historical average of 63 months.
Interest as a share of revenue, a better measure of public debt sustainability than debt service since governments do not have to repay their debts, is roughly at the level of the late 1990s, still well below the peak of the early 1990s (Chart 5). Also, the dollar’s global status suggests a low risk of a US debt crisis. That said, high Treasury indebtedness will eventually impact the value of the dollar and US yields as well as reduce US fiscal flexibility (I will discuss Treasury debt sustainability in greater detail in future research).
By contrast with the US Treasury, household debt relative to GDP is roughly the same as in the 2000s, about 75% of GDP, though this is much higher than the 45% of GDP in 1970.
Simultaneously, the household debt service ratio, which measures the actual impact of leverage on spending, is likely to remain insulated from higher yields since 70% of household debt comprises residential mortgages, and households have locked in low long-term mortgage rates.
By contrast with households, non-financial corporations (NFCs) have steadily increased their leverage. In Q1 2023, it was about 15ppt of GDP higher than in 2000 and 30ppt higher than in the 1970s. However, NFC debt service ratios have remained in a band of about 35-45% of corporate income (i.e., profits) during 2000-23, despite long-term yields falling to a low of 0.5% in 2020 from above 6% in 2000. This suggests a low sensitivity to yields, possibly due to NFC locking in low rates. Supporting this benign view, while business bankruptcies are rising, they are doing so from below normal levels that reflect the extraordinary Fed support deployed during the pandemic (Chart 6).
Finally, while many banks are struggling with profitability, the combination of implicit guarantee of all deposits and strong Fed support has reduced their balance sheet exposure to higher yields. This is shown by the stability of borrowings under the Fed Bank Term Funding Facility.
In Q2 2023, YoY nominal GDP growth was 5.9%, and the three year average was 5.7%. By contrast, 10yr yields are currently about 4.6%. Continued sustained growth and inflation, together with growing market belief that QE is gone for good, create more upside for yields over the medium-term.
In the short term, FOMC speakers have signaled that the Fed would remain on hold in November, triggering a bonds rally. I think such a hold is meant to assess how the economy is adapting to higher yields rather than to forego an extra 2023 hike altogether.
Indeed, in the latest FOMC minutes, ‘Participants stressed that current inflation remained unacceptably high’ and that ‘further evidence would be required for them to be confident that inflation was clearly on a path to the Committee’s 2 percent objective.’
This suggests that core PCE would have to be significantly below the Fed’s expectation of 3.7% Q4/Q4 for the Fed to give up on an additional hike. I therefore expect that while pausing in November, the Fed could signal a December hike. If so, this would likely see long-term yields resume their climb.