Skip to content

Invest1 publisher2 min readPublished

Goldman blames deficits and AI borrowing for a 30-year yield stuck at 5.2%

The bank's September 9 note says the forces holding the long end up are structural. It sends the hedge down to the five-year sector, where a hundred basis points of selloff costs under a third as much as it does at 30 years.

The Investor · Invest desk

What happened

  • The 30-year US Treasury yield is sitting at roughly 5.2 percent, its highest level in nearly twenty years.
  • Goldman strategists George Cole and William Marshall published a report on September 9 arguing for a persistently steep global yield curve driven by structural forces that Treasury buybacks are unlikely to fix.
  • Long-maturity yields in Japan and the UK stood at decade highs as of September 8, and Germany's were at their highest level since 2009.

Compiled by The InvestorSomething wrong?How this is made

Why it matters

  • decision An allocator who moves the hedge into the belly of the curve gives up the long bond's greater capacity to rally in a downturn in exchange for far less damage if yields keep climbing. Goldman's note forces that choice now.
  • constraint If official buying at the announced scale cannot move the long end, investors waiting for the Treasury to reprice duration have no lever to wait for.
  • exposure Government financing of the AI buildout puts bond holders on the funding side of a capex cycle they own no part of. The supply arrives at the maturities they hold.
  • contradiction The same note both raises the case for owning bonds after the selloff and warns off the long end, so a buyer has to pick which half applies to their book.

Duration is what the 5.2% has to pay for. A 30-year bond priced at par with a 5.2% semiannual coupon has a modified duration near 15, so another 100 basis points on the yield takes roughly 15% off the price [13]. At a 5.2% coupon, that is close to three years of income gone in a move the Treasury market has made inside a quarter before [15]. These are par-bond approximations, and the actual issue trading at that yield will not be exactly at par.

The five-year sector, at the same yield, carries a modified duration of about 4.4, which is under 30% of the long bond's price sensitivity [14]. Cole and Marshall suggest that part of the curve as the more effective hedge if growth or inflation dynamics shift [10].

The Treasury's response has been buybacks, up to $6 billion of securities in the 10-to-20-year range [8]. Goldman's strategists project that buybacks at that scale will not meaningfully dent yields being driven by much larger structural forces [9]. The maturity sitting at 5.2% is the 30-year, and the announced bucket stops at 20 [1][8].

Cole and Marshall expect energy-driven inflation to ease over the next six months, which they frame as a temporary reprieve rather than a structural fix [6]. If that easing lands alongside weaker growth, the same factor of 15 works for the holder instead of against, and 5.2% was the entry price [6][13]. That case against Goldman's caution is Goldman's own. The other one is fiscal: sustained deficits across developed economies top Goldman's list, a hangover from pandemic-era borrowing that governments have shown little appetite to unwind [4].

On this evidence I would rather own the five-year than the 30-year. The call is wrong if the long yield falls steadily while developed-market deficits keep running; that would mean the long end had been pricing energy and the cycle all along.

All of this reaches investors through Cryptobriefing's account of the September 9 report [2]. That account puts AI-related investment at roughly 1% of global GDP with governments borrowing heavily to finance their share; it gives no figure for that government share and does not identify the countries [5][12].

What to watch

  • Whether Treasury raises the buyback size above $6 billion or extends the bucket past 20-year maturities.
  • The energy-driven inflation path Cole and Marshall expect to ease inside six months, and whether the long end follows it down.
  • Japanese and UK long-maturity yields, which were at decade highs on September 8, and whether they keep setting new ones.
Loading claim ledger
Loading source directory links
Loading share composer
Loading topic controls
Loading related stories