Research Digest — 2026-09-18

Verdict

The September FOMC delivered a 25bp hike and signalled more tightening than DB expected, with the long-run dot edging to 3.2% and 16 officials anticipating additional tightening in 2026 (P008). CACIB forecasts the 2Y Treasury at 4.9% and the 10Y around 5.00% by year-end, with curve flattening through Q1 2027 during the tightening cycle (P004). WF independently expects one more hike to 4.00-4.25% by year-end 2026 with no changes in 2027, arguing it is hard to justify hiking beyond that when excess inflation is largely supply-driven and the labour market is not overheating (P028).

Analysis

The causal chain runs from a hawkish reaction function to a higher discount rate. DB's read is that the dot plot skewed more hawkish than anticipated, with a strong consensus around another hike this year and next year's projections split between 50bps and 75bps of total tightening (P008). Chair Warsh framed the hike as removing a 'dose' of accommodation against limited evidence that policy is restrictive, and DB continues to expect 50bps more of tightening with 25bp increases in December and next March (P008). CACIB's forecast is more aggressive on the front end, expecting the 2Y to trade towards 4.9% in the coming months given additional Fed rate hikes, and the 10Y around 5.00% by year-end, with yields stabilising by mid-2027 assuming the Fed is done then (P004). The counterargument comes from WF, which caps its forecast at one more hike to 4.00-4.25% by year-end 2026 and no changes in 2027, on the grounds that excess inflation is largely supply-driven and the labour market is not overheating (P028). The market consequence is a higher floor under long-term yields: WF argues the floor under longer-term interest rates has moved higher, reflecting solid growth expectations, robust corporate and Treasury bond issuance and monetary policy uncertainty, with sensitive sectors like housing bearing the brunt (P028). For equities, GS frames the risk as a pace problem rather than a level problem: stocks have usually generated positive returns alongside rising interest rates unless the pace exceeded 2+ standard deviations, and today a 2-standard-deviation move in 10-year Treasury yields would equate to about 50bps over a month (P014).

Tension

DB and CACIB see scope for more tightening than WF, which caps the cycle at one more hike and cites supply-driven inflation and a balanced labour market as reasons not to go further (P008, P004, P028).

What changed

The dot plot skewed more hawkish than DB anticipated, with the long-run dot edging higher to 3.2% (P008).

Desk views

DB reads the September FOMC as a hawkish delivery: the dot plot skewed more hawkish than anticipated, 16 officials anticipate additional tightening in 2026, and the long-run dot edged to 3.2%. DB continues to expect 50bps more of tightening with hikes in December and next March.

Evidence IDs · P008

CACIB expects rates to stay high over the coming months, forecasting the 2Y Treasury towards 4.9% and the 10Y around 5.00% by year-end, with curve flattening through Q1 2027 during the Fed tightening cycle.

Evidence IDs · P004

WF expects one more hike to 4.00-4.25% by year-end 2026 with no changes in 2027, arguing it is hard to justify hiking beyond that when excess inflation is largely supply-driven and the labour market is not overheating.

Evidence IDs · P028

Investor implication

A higher-for-longer US rate floor raises the discount rate applied to long-duration equities and keeps the front end of the Treasury curve elevated, while the pace of yield moves rather than the level is the proximate equity risk (P014).

Trade expression

Not established in cited evidence.

Invalidation

A sharper tightening of financial conditions and/or an unexpected softening in inflation or the labour market are DB's dovish scenarios (P008).

Verdict

The BoJ raised its policy rate by 25bps to 1.25%, the highest since 1995, but the 7-2 vote and dovish communication undershot market pricing of nearly 100bps of further tightening over twelve months, triggering yen weakness (P007, P022, P024). CACIB argues the acceleration in the pace of hikes from roughly once every six months to once a quarter is temporary and externally driven by crude oil concerns, expecting a return to semi-annual hikes from 2027 as growth slows (P002).

Analysis

The mechanism is a reaction-function gap: the BoJ delivered the hike but not the hawkish path the rates market had priced. MUFG notes the bar was low for the BoJ failing to meet hawkish pricing and that the initial reaction to the decision to hike by 25bps to 1.25% underlined the risks that the BoJ could well fail to meet expectations of nearly 100bps of tightening over the coming twelve months (P022). Two factors drove the initial sell-off: a 7-2 vote with Toichiro Asada and Ayano Sato dissenting, both appointed by the Takaichi administration, and the removal of the description that real interest rates were 'negative' to 'remained at low levels' (P022). DB reads the same dissent as a signal of less political support for the hike than indicated by US Treasury Secretary Bessent, and reports the yen -0.72% lower at 157.10 having been around 153.40 at the start of the week, with the JGB curve steepening (P007). Natixis reports USD/JPY up 0.8% to 157.28 and notes the dovish-leaning development extinguished immediate market pricing for subsequent hikes (P024). The counterargument is CACIB's: the acceleration in the pace of rate hikes is temporary, driven by external crude oil inflation concerns, and with real GDP growth expected to slow to 0.6% in 2027 from 1.2% in 2025 and 0.8% in 2026, slower growth, falling crude prices and stabilising inflation are likely to return the pace of hikes to around once every six months from 2027 onwards (P002). CACIB also notes the policy rate is expected to reach a terminal rate of 2.5% by 2029 (P002). The market consequence is a yen that remains vulnerable to further hawkish repricing if the BoJ later meets market expectations, but near-term FX direction is dominated by the communication gap rather than the hike itself (P019, P022).

Tension

CACIB models a reversion to semi-annual hikes from 2027 on weaker domestic demand and falling crude prices (P002), while MUFG and ING treat the dovish communication as the dominant near-term FX signal (P022, P019).

What changed

The BoJ delivered its second hike of 2026, taking the policy rate to 1.25%, the highest since 1995, but the 7-2 vote and dovish statement language undershot hawkish market pricing (P007, P022, P024).

Desk views

MUFG argues the bar was low for the BoJ to fail hawkish pricing, and the 7-2 vote plus the change in real-rate language drove the initial yen sell-off; the BoJ could well fail to meet expectations of nearly 100bps of tightening over twelve months.

Evidence IDs · P022

CACIB argues the acceleration in the pace of hikes is temporary and externally driven by crude oil concerns, expecting a return to semi-annual hikes from 2027 as growth slows to 0.6% in 2027.

Evidence IDs · P002

DB reports the BoJ delivered another 25bp hike to 1.25%, the highest since 1995, but describes it as a more dovish hike than expected with two high-profile dissenters, and notes the yen -0.72% lower at 157.10.

Evidence IDs · P007

ING notes the yen has weakened back to summer levels following the BoJ's dovish rate hike and continues to see upside risks for the dollar, with oil prices likely to become the predominant driver in the short term.

Evidence IDs · P019

Investor implication

A yen that remains vulnerable to hawkish repricing keeps USD/JPY upside risks alive in the near term, while the JGB curve steepening and the dissent signal raise questions about the durability of the hiking path (P007, P022).

Trade expression

Not established in cited evidence.

Invalidation

If the BoJ later meets hawkish pricing, the yen weakness thesis would be challenged (P022).

Verdict

The MPC voted 6-3 to hold Bank Rate at 3.75%, with Pill, Greene and Mann dissenting in favour of a 25bp hike, and Bank staff now expecting inflation to rise to slightly above 4% in 2027Q1 (P009). The annual QT decision agreed a multi-year plan to reduce the gilt stock to zero by 2034 through £20bn per year of sales, but the Bank will pause APF auctions while it considers a new model (P009, P013). GS reads the pause as the near-term surprise outweighing the hawkish QT-for-longer effect, and sees the setup skewing risks towards a flatter curve (P013).

Analysis

The mechanism is a reaction function that is bracing for a hike without delivering one yet. Citi reads the minutes as more hawkish than the July MPR because the Committee now sees material risks to headline CPI continuing to surprise on the upside and seep into core components via indirect effects, and maintains its view that the MPC will likely hike Bank Rate by 25bps in Q4, most likely in the November MPR round (P005). Citi also argues the pivot to hike will not be gradual, expecting the vote to suddenly switch from a 6-3 hold to a 3-6 or 2-7 hike, with the movers being the Governor joined by at least two of his deputies (P005). GS's interpretation is that the MPC will probably hike in November if energy price pressures persist, maintaining a forecast for a 25bp hike at that meeting, although a hold remains possible with lower energy prices or weaker data, and it maintains its baseline that the MPC will refrain from tightening further as the labour market remains soft, underlying inflation rises only modestly, and wholesale energy prices start to ease in early 2027 (P010). The QT plan is the second lever: GS notes the net reduction in gilt holdings of £46bn until 2034 was higher than the market expected, at least via the latest BoE dealer survey, which is likely to lead to greater Short-Term Repo facility use over time given that reserves look to be stabilising even as gilt holdings continue to drop (P013). But GS argues the hawkish effect of QT-for-longer will be discounted by the market and should be outweighed by the near-term surprise of pausing active sales, because the total effect of QT is unlikely to be priced in full given potential programme revisions over such a long implementation window (P013). The counterargument is that the QT pause is a technicality rather than a policy easing: the plan still signals a long horizon for QT, and GS notes it is unlikely that the new QT plan will have a material impact on the macro outlook for gilts, but directionally it pushes towards the 5% end-year 10y gilt forecast (P013). The market consequence is a flatter curve skew and a gilt-positive near-term reaction, with the 10Y gilt yield around 5.20% after the QT pause (P024).

Tension

Citi expects a Q4 hike with a sudden vote switch (P005), while GS maintains a baseline that the MPC will refrain from tightening further after November as the labour market remains soft and energy prices ease in early 2027 (P010).

What changed

The MPC minutes read more hawkish than the July MPR, with increased upside risks to inflation and Bank staff expecting inflation to rise to slightly above 4% in 2027Q1 (P009, P005).

Desk views

Citi reads the minutes as more hawkish than July, expects a 25bp hike in Q4 most likely in November, and argues the pivot to hike will not be gradual with the vote switching from 6-3 hold to 3-6 or 2-7 hike.

Evidence IDs · P005

GS notes the net reduction in gilt holdings of £46bn until 2034 was higher than the market expected, but argues the hawkish QT-for-longer effect will be discounted and outweighed by the near-term surprise of pausing active sales, skewing risks towards a flatter curve.

Evidence IDs · P013

GS reports the MPC voted 6-3 to hold with Pill, Greene and Mann dissenting, and Bank staff now expecting inflation to rise to slightly above 4% in 2027Q1, with Bailey, Breeden, Lombardelli and Ramsden signalling openness to raising Bank Rate.

Evidence IDs · P009

GS interprets the communications as the MPC probably hiking in November if energy price pressures persist, but maintains a baseline that the MPC will refrain from tightening further as the labour market remains soft and wholesale energy prices start to ease in early 2027.

Evidence IDs · P010

Investor implication

A hawkish hold with a gilt-friendly QT pause compresses the near-term gilt risk premium while keeping the front end anchored, skewing risks towards a flatter curve (P013).

Trade expression

Not established in cited evidence.

Invalidation

A hold remains possible with lower energy prices or weaker data (P010).

Verdict

CACIB argues the tightening cycle anchors long-term inflation expectations but the possibility of more real yield pain persists, and expects repricing to shift from the 3-5Y segment towards longer maturities, especially if fiscal impulses go unchecked (P003). It has turned decisively bearish across the EUR rates curve, seeing no constructive case for duration or EGB spreads as markets adapt to a structurally higher neutral rate, and targets 10Y Bund at 3.95%, OAT at 5.05% and BTP at 5.00% by December 2027 (P003).

Analysis

The mechanism is a repricing of the term premium and the neutral rate rather than a simple front-end story. CACIB sees room for bear steepening of the EUR IRS curve, with the 2Y IRS drifting higher but undershooting forwards and the belly bearing the most pain (P003). It also argues EGB spreads look tight versus the rate environment, and that rising refinancing costs mean credit curves have further room to steepen (P003). The fiscal overlay is explicit: CACIB embeds 2027 election risk as an explicit overlay, with OAT-Bund peaking at 120bp in Q2 2027 around the French elections, with BTPs relatively more resilient (P003). Natixis provides the fiscal context: the French 2027 budget centres on an overall fiscal effort of roughly €54bn designed to bring the deficit to 5% of GDP in 2027 without raising headline tax rates, but the government now expects the 2026 deficit to exceed 5% of GDP, and returning to a 3% public deficit by 2029 appears out of reach on current trajectories (P023). Natixis also notes the scale of the effort is unprecedented outside of crisis recovery phases and raises real credibility questions, especially as the National Assembly remains deeply fragmented with no majority to pass the measures as designed, and the government's 2027 growth assumption of 1% looks optimistic against Natixis's own 0.8% call and the Banque de France's 0.9% (P023). The counterargument is that the front end remains anchored: Natixis reports the 10Y Bund yield fell below 3.5% but the front-end remained anchored despite lower energy prices, leading to a flattening of the curves (P024). The market consequence is a curve that may steepen at the long end even as the front end stays pinned, with EGB spreads vulnerable to fiscal credibility questions (P003, P023).

Tension

CACIB sees bear steepening with the belly bearing the most pain (P003), while Natixis reports the front-end remained anchored despite lower energy prices, leading to a flattening of the curves (P024).

What changed

CACIB has turned decisively bearish across the EUR rates curve, seeing no constructive case for duration or EGB spreads as markets adapt to a structurally higher neutral rate (P003).

Desk views

CACIB argues the tightening cycle anchors long-term inflation expectations but real yield pain persists, expects repricing to shift to longer maturities, and has turned decisively bearish across the EUR rates curve with targets of 10Y Bund at 3.95%, OAT at 5.05% and BTP at 5.00% by December 2027.

Evidence IDs · P003

Natixis notes the French 2027 budget centres on a €54bn fiscal effort to bring the deficit to 5% of GDP, but the 2026 deficit is now expected to exceed 5% of GDP and returning to a 3% deficit by 2029 appears out of reach, raising real credibility questions given a fragmented National Assembly.

Evidence IDs · P023

Natixis reports some relief in EUR rates driven by a slight decline in Brent crude prices and the term premium, with the 10Y Bund yield falling below 3.5%, but the front-end remained anchored despite lower energy prices, leading to a flattening of the curves.

Evidence IDs · P024

Investor implication

A repricing that shifts to longer maturities raises long-end yields and steepens curves, with EGB spreads vulnerable to fiscal credibility questions in France (P003, P023).

Trade expression

Not established in cited evidence.

Invalidation

A resolution of energy price pressures or a credible French fiscal consolidation would challenge the bearish EUR rates view (P003, P023).

Verdict

GS argues the recent strength of S&P 500 earnings growth has raised concerns of an earnings bubble, but its base case is for earnings growth to decelerate, not collapse, forecasting EPS growth of 11% in both 2027 (to $415) and 2028 ($460) (P018). The S&P 500 forward P/E multiple sits at 19x, in line with its 10-year average, but the market multiple on trend earnings has only been exceeded in the past several decades during the peak of the Dot-Com Bubble (P018).

Analysis

The mechanism is a fading AI investment tailwind transitioning into a productivity boost. GS notes S&P 500 EPS grew by 51% in Q2 and by 26% during the past four quarters, lifting earnings well above both the long-term trend and the historical relationship with economic growth (P018). The AI investment boom has accounted for nearly half of S&P 500 earnings growth this year, and this tailwind should begin to fade next year even as capex spending continues to grow; GS estimates AI investment will transition from an S&P 500 earnings tailwind of 11pp this year to a marginal drag in 2028 as capex growth slows and depreciation expenses mount (P018). The counterargument is that the earnings strength is genuine and balance sheets are strong: GS notes technology profits remain very robust, interest coverage ratios for the aggregate S&P 500 and its median stock rank in the 99th and 68th percentiles relative to the past 20 years, and demand for compute is accelerating and outstripping supply, with Microsoft recently declaring intent to triple data centre capacity in 6 years (P014). GS also notes Nvidia reiterated its 2030 AI TAM outlook to $3tn-$4tn and supply constraints continue to be broad based (P014). The market consequence is that any slowdown in profit growth in an environment of a much higher cost of capital could put downward pressure on equity prices, reducing confidence in future cash flows across the ecosystem from the hyperscalers to the 'pick and shovels' that have been benefiting from the capex boom (P014). GS remains neutral on equities alongside other asset classes on a 3-month horizon but continues to be overweight equities over 12 months, expecting ongoing growth in nominal GDP to underpin equity markets and the earnings cycle even without valuation expansion (P014).

Tension

GS's base case is decelerating but not collapsing earnings growth (P018), while it also flags that any slowdown in profit growth in a higher cost-of-capital environment could put downward pressure on equity prices (P014).

What changed

S&P 500 EPS grew by 51% in Q2 and by 26% during the past four quarters, lifting earnings well above both the long-term trend and the historical relationship with economic growth (P018).

Desk views

GS argues the market is over-earning but not in an earnings bubble, forecasting EPS growth of 11% in both 2027 and 2028, with the AI investment boom accounting for nearly half of S&P 500 earnings growth this year and transitioning to a marginal drag in 2028.

Evidence IDs · P018

GS notes technology profits remain very robust, interest coverage ratios rank in the 99th and 68th percentiles, and demand for compute is accelerating and outstripping supply, but warns any slowdown in profit growth in a higher cost-of-capital environment could put downward pressure on equity prices.

Evidence IDs · P014

Investor implication

A decelerating but positive earnings path supports equities over 12 months, but the transition from an AI investment tailwind to a drag in 2028 and the higher cost of capital create a narrower margin for error (P018, P014).

Trade expression

Not established in cited evidence.

Invalidation

A sharper slowdown in profit growth or a more hawkish Fed path would challenge the constructive 12-month equity view (P014, P018).

Verdict

GS maintains its $5,400/toz end-2027 forecast despite the Fed hike and its economists now expecting an additional hike in October, expecting the impact of tighter monetary policy to be felt primarily through a slower near-term appreciation path rather than a lower terminal gold price (P016). It now estimates year-end fair value at $4,650/toz, below its previous $4,900/toz forecast but above the current spot price of ~$4,350/toz (P016).

Analysis

The mechanism is a structural central bank bid that offsets the rate-sensitive ETF drag. GS notes continued central bank diversification remains the main structural driver of its constructive gold view, contributing nearly all of its expected 23% appreciation through end-2027, with its central bank nowcast pointing to purchases running at ~91 tonnes/month on a three-month seasonally adjusted basis, well above the pre-2022 average of 17 tonnes/month (P016). Reflecting this acceleration, GS raises its central bank demand assumption to 60 tonnes/month on average through 2026-27, versus 50 tonnes/month in 2026 and 40 tonnes/month in 2027 previously (P016). The counterargument is a more hawkish Fed path: GS notes a significantly more hawkish Fed path could generate a sharper-than-usual correction, and if the Fed were to deliver three additional hikes by year-end and signal a higher terminal rate, demand for gold as a macro-policy hedge could partially unwind as market questions about DM central bank independence fade further, with gold prices falling toward a near-term floor of $4,070/toz before gradually recovering toward ~$4,200/toz by end-2026 (P016). The market consequence is a gold price that grinds higher in the near term but with a wider distribution of outcomes depending on the Fed path, with much of the expected tightening already appearing priced into ETF demand (P016).

Tension

GS's base case is a slower near-term appreciation path with a higher structural floor (P016), while its hawkish scenario sees gold falling toward $4,070/toz before recovering (P016).

What changed

GS lowered its year-end fair value estimate to $4,650/toz from $4,900/toz but maintained its $5,400/toz end-2027 forecast and raised its central bank demand assumption to 60 tonnes/month on average through 2026-27 (P016).

Desk views

GS maintains its $5,400/toz end-2027 forecast, expects tighter policy to slow rather than derail the rally, and raises its central bank demand assumption to 60 tonnes/month on average through 2026-27, with central bank purchases running at ~91 tonnes/month versus a pre-2022 average of 17 tonnes/month.

Evidence IDs · P016

Investor implication

A structural central bank bid raises the floor under gold prices even as higher rates weigh on ETF demand, making the near-term path slower rather than lower (P016).

Trade expression

Not established in cited evidence.

Invalidation

A significantly more hawkish Fed path with three additional hikes by year-end and a higher terminal rate signal would challenge the constructive gold view (P016).

Verdict

UniCredit argues most of the reduction in UST exposure by Japan over the last few months has come mainly from selling T-bills and appears to be related to FX intervention, while underlying demand for longer-dated USTs remains resilient for now (P026). Japan's stock of US government bonds has declined by just over USD 80bn since the start of 2026, but the decline has been driven by T-bill sales rather than long-term USTs (P026).

Analysis

The mechanism is a compositional story that matters for the term premium. UniCredit notes Japanese investors are central to the debate on US Treasury demand, with just over USD 1.1tn in holdings, accounting for around 3.5% of the whole market, and that investors are concerned Japanese holdings may fall further due to the need to stem currency weakness and to deploy savings to absorb a growing supply of domestic government debt as the government engages in expansionary fiscal policy (P026). The key evidence is that the decline in the stock has been driven by T-bill sales related to FX intervention, while underlying demand for longer-dated USTs remains resilient for now (P026). The counterargument is that any sustained reduction in Japanese demand for Treasuries could represent an additional headwind for USTs (P026). The market consequence is that the term premium impact of Japanese selling may be smaller than the headline stock decline suggests, but the risk of a shift from T-bill sales to long-end selling remains the key uncertainty, particularly if the BoJ's hiking path and yen weakness persist (P026, P022).

Tension

UniCredit's compositional read is reassuring for long-end demand (P026), but it explicitly flags that any sustained reduction in Japanese demand could represent an additional headwind for USTs (P026).

What changed

Japan's stock of US government bonds has declined by just over USD 80bn since the start of 2026, but the decline has been driven by T-bill sales rather than long-term USTs (P026).

Desk views

UniCredit argues the reduction in Japanese UST exposure has come mainly from selling T-bills related to FX intervention, while underlying demand for longer-dated USTs remains resilient for now, though any sustained reduction could be an additional headwind for USTs.

Evidence IDs · P026

Investor implication

The term premium impact of Japanese selling may be smaller than the headline stock decline suggests, but the risk of a shift from T-bill sales to long-end selling remains the key uncertainty (P026).

Trade expression

Not established in cited evidence.

Invalidation

A shift from T-bill sales to long-end selling would challenge the resilience thesis (P026).