#1. EUR front-end repricing is being driven by QT’s steady reserve drain, not by an outright funding shock. Mild impact on front-end spreads so far.
Eurosystem excess liquidity has fallen from ~€4.7trn at the peak to ~€2.5trn, with QT now the dominant structural drain as TLTROs have fully rolled off. On current roll-down assumptions, excess liquidity trends towards ~€1.9trn by end-2026 and ~€1.5trn by end-2027. This gradual erosion matters because marginal liquidity tension shows up first in spreads and term premia, well before aggregates look “scarce.”
Our core monetary policy message is a transition: from abundant liquidity to a system that increasingly has to price the marginal unit of reserves. Higher repo/term rates closer to refi are framed as a feature of the transition, not a trigger for stopping QT.
While front-end euro spreads have widened since the liquidity peak in 2022, including repo-€str and Euribor-€str, euro market rates remain below comparable ECB’s lending rates, which has kept usage of the central bank’s open market operations low so far. This reflects the redistribution of reserves within the euro area, including from 1) cash-rich banking systems to cash-poor banking systems, and 2) a shift of deposits by euro area governments from the Eurosystem to the banking system after remuneration changes by the ECB.
#2. The ECB’s demand-driven framework is designed to let market spreads widen until OMOs become economically compelling.
With a 15bp corridor (MRO above DFR) and FRFA in standard operations, the ECB can cap volatility while letting price signals do more of the allocation work. OMO usage is still very low (~€20–25bn), but our base case is a gradual climb towards ~€120bn by end-2026 and ~€400bn by end-2027 as reserves fall.
Structural LTROs / a structural securities portfolio are explicitly sequenced after a durable balance-sheet re-expansion, so 2026 “early backstops” look unlikely. Practically, the ECB can tolerate higher secured and term funding rates as part of the transition, provided control over the corridor is preserved.
#3. Reserve scarcity risk is about distribution and regulation as much as the headline liquidity number.
Excess liquidity remains highly concentrated (Germany and France still hold ~55–60%), and the intra-country dispersion across entities is equally important. TARGET2 redistribution has cushioned the system so far, but it becomes harder to rely on as QT proceeds and buffers shrink.
Liquidity regulation (LCR/NSFR) creates a structural bid for Level-1 HQLA, with reserves uniquely valuable for payments, optionality and operational simplicity. That is why “scarcity-like” behaviour can emerge even when reserves still look large in absolute terms—spreads reprice before aggregates do.
#4. We prefer the HQLA/LCR lens—not the €STR-DFR model—as the cleanest way to time when reserve scarcity starts to bite.
€STR-DFR is a noisier scarcity proxy because it is dominated by balance-sheet costs, segmentation, and banks’ pricing power in unsecured intermediation. By contrast, HQLA/LCR is the binding constraint that ultimately forces banks to retain a minimum stock of reserves, regardless of microstructure.
Despite ~€2.2trn of reserve decline, aggregate HQLA buffers have stayed broadly stable as banks increased holdings of EGBs, SSAs and high-quality covered bonds. The share of cash/reserves in liquidity buffers fell from ~77% (Jun-2022) to ~44% (Sep-2025), while central government bonds rose from ~16% to ~30% over the same period. Since early-2024 to Jun-2025, banks added ~€451bn of euro-area bonds, with large cross-country dispersion that highlights both capacity and limits to further rotation.
The ECB scenario analysis puts “excess reserve demand” in a wide €600bn–€2.2trn range, and a static LCR/HQLA approach implies ~€1.8–2.0trn as a plausible stress-zone.
This approach also makes country- and bank-level heterogeneity explicit, which is precisely how the Eurosystem transitions from “ample” to “tight at the margin.”
#5. Repo should grind “cheaper” as collateral abundance persists and the cash margin tightens, pulling secured rates toward MRO over time. We see German GC cheapening up to a DFR+5/+6bp target range by 2H26 under a non-disruptive scenario assumption.
The GC curve has structurally shifted away from the cash-rich/collateral-scarce configuration, with repo pricing normalising as sovereign supply rises and the Eurosystem footprint shrinks. German 1-day GC is currently around DFR+2bp, with the central path for 2026 pushing it up to DFR+5/+6bp (≈ refi-9bp).
Mechanically, collateral abundance (record net EGB supply, plus QT returning paper to free float) shifts pricing power toward cash lenders and lifts repo rates. At the same time, HQLA substitution increases the need to finance government-bond inventories, raising the structural demand to borrow cash against collateral.
We see German GC “less special” and more persistently positive, with risks skewed to the upside as cash becomes scarcer vs collateral. This is the logical equilibrium of a system moving toward reserve sufficiency.
Risk to our view: a prolonged market risk-off phase, which could trigger some demand for German collateral.
#6. Specialness compression is becoming structural, shifting repo RV from scarcity to flow- and calendar-driven dislocations.
As free float rises and the central bank footprint shrinks, fewer bonds can sustain persistent “special” status across the curve. Cross-country GC dispersion is currently very tight (order of ~2–3bp), consistent with abundant collateral and still-ample liquidity—yet it is vulnerable to episodic funding pressure.
DMO behaviour matters more: in Italy, a ~€10bn monthly liquidity drain/injection is presented as worth roughly ~0.5bp in GC, making public cash management a first-order driver.
CTD rolls and futures-related hedging flows can create sharp, bond-specific micro-tensions even when the macro liquidity backdrop looks benign.
#7. Repo–€STR dislocations persist because secured vs unsecured arbitrage is structurally incomplete.
Many €STR lenders (MMFs/NBFIs) lack direct access to the DFR and face operational, legal and balance-sheet frictions in scaling cleared repo. Repo is a morning market tied to collateral settlement cycles, while unsecured cash often arrives later when secured liquidity is thinner and dealer balance sheets are set.
Central clearing disproportionately benefits top-tier sovereign collateral, reinforcing segmentation between “best” collateral and the rest of the stack. The practical implication is that repo can reprice faster than €STR, and €STR remains an imperfect read-through of marginal reserve conditions.
#8. We see Euribor-€STR and Repo-€STR spread wideners driven by rising reserve competition as €STR stays pinned slightly below DFR.
We see €STR-DFR stabilising around -3/-4bp (from ~-6.7bp recently), because unsecured €STR funding delivers limited regulatory value and banks preserve intermediation margin. A sustained move of €STR-DFR above 0bp is unlikely outside stress: banks have a DFR outside option for liquidity, and paying above DFR for unsecured O/N is uneconomic given leverage costs.
As reserves drain, banks compete harder for term liabilities and seek to term-out funding for NSFR/precautionary reasons, pushing Euribor higher vs €STR (beyond the 15bp corridor implication). In parallel, collateral abundance + increased financing demand lifts GC repo vs €STR, so both Euribor-€STR and Repo-€STR widen as the system migrates toward a marginal-reserves regime.
Risks are that banks demand less reserves than we expect and/or that the ECB introduces a structural securities portfolio or structural credit operations with very favourable terms sooner than we expect.
The key implication is that continued QT increasingly requires either (i) higher market funding usage, (ii) eventual ECB OMO take-up, or (iii) a meaningful change in banks’ liquidity preferences.
#9. Credit collateral is re-entering the repo complex, with covered bonds and SSAs gaining relevance as workhorse HQLA alongside sovereigns.
For overnight, core SSA repo typically trades about +3–4bp vs Bund GC, reflecting liquidity, balance-sheet and haircuts versus sovereign collateral. Covered bonds add another +7–10bp vs SSAs (i.e., roughly +10–14bp vs Bund GC) as they consume more balance sheet and benefit less from clearing-driven demand.
Covered Bonds/SSAs repo spreads vs GC can widen with higher issuance/free float, reduced ECB programme-related scarcity (post-CBPP3), risk-off preference for core sovereigns, and tighter dealer balance sheets. On the other hand, they can tighten when volatility is benign and carry demand rises, when haircuts/clearing conditions improve, or when relative scarcity returns in specific high-quality lines.
#10. Backstops matter most for tails: The revamped EUREP facility reduces tail risk, but not the regular GC pricing.
EUREP is framed as a repo firewall for non-euro central banks—designed to prevent forced selling and disorderly funding spirals in stress. Because it is a backstop facility, its equilibrium impact is on distribution of outcomes: fewer extreme specials, fewer air-pockets, and less contagion to broader money markets.
In the base case, the pricing signal still comes from market funding and the ECB corridor, with OMOs absorbing pressure only once it becomes economically rational. Thus, EUREP would help in a lower-tail-risk regime, with fewer existential repo breaks.
#11. Bund-swap spreads should tighten in 2026 as collateral abundance and QT cheapen Bund financing and tilt EGBs to underperform swaps.
We see further tightening across 2Y/5Y/10Y German swap spreads (around ~10bp), with repo cheapening itself worth ~3–5bp of that in the belly. In 10Y, the target is Bund-swap revisiting ~-5/-10bp in 2H26, but with limited scope for materially tighter levels beyond that.
Drivers are fundamentally supply-led (record net issuance, higher free float, full-pace QT), which are also mechanically reinforced by less special repo and higher GC levels. Offsetting forces include banks’ HQLA needs, foreign central bank demand, risk-off episodes, and ALM/mortgage hedging flows that can intermittently re-widen spreads.
#12. EURUSD XCCY basis has a “zero ceiling”: we see limited scope for sustained positivisation and a range-bound bias as 2026 catalysts fade.
The EURUSD XCCY basis is structurally negative because the dominant pressure is to receive USD/pay EUR (synthetic USD borrowing), with arbitrage constrained by dealer balance sheets and regulation. Both ST and LT basis tightened through 2025 in benign conditions, but longer tenors struggle to turn sustainably positive—hence the “zero ceiling.”
2026 tightening factors include relative ECB vs Fed balance-sheet momentum and potential easing of reporting-date frictions, but these supports look less powerful as the year progresses. Cross-border issuance (Yankee vs reverse Yankee), risk-off balance-sheet constraints and quarter/year-end window dressing remain the main swing factors for renewed widening risk.