EURO-AREA TRANSMISSION MAP
Who touches central bank money, at what price, and what they would do instead
Who touches central bank money, at what price, and what they would do instead
Since 18 September 2024 the Eurosystem steers with the deposit facility rate and offers unlimited one-week and three-month money 15 basis points above it, against a collateral list far broader than anything the repo market accepts. Whether that design works is not a question about the ECB's balance sheet. It is a question about a few dozen kinds of institution, the rate each one actually faces, what it would do instead, and how long it needs the money for. This page maps them.
Left to right is distance from the policy rate, measured in balance sheets crossed — that is the one spatial claim on this map. Vertical position is not a ranking: nodes settle where their connections pull them, so things that trade with each other end up near each other. The halo behind a node scales with its balance sheet. Switch to the column grid if you want the tidy version for export. Click an actor for the rate it targets, its outside options with prices, and its size. Hover an arrow for its tenor. live means the number was refetched from the ECB Data Portal when you loaded the page; snap is from the build-time snapshot; assumed is ours, and is never mixed into a measured column.
The whole graph at once is a picture of complexity, not an argument. Each channel below is a claim about how one thing reaches another, and selecting it dims everything that is not part of that claim. The two that matter most for the operational framework are the floor channel, which explains why €STR sits below the deposit facility rate, and the take-up channel, which is the subject of §05.
A US money market fund with cash to place overnight can place it at the Federal Reserve through the ON RRP. A euro-area money fund cannot place it anywhere but a bank, a repo or a bill. The Eurosystem's deposit facility is open to credit institutions and to nobody else.
So every euro of non-bank cash that wants to earn the deposit facility rate has to rent a bank balance sheet to get there, and the rent shows up as €STR printing below the DFR. That spread is not a stress indicator. It is the price of friction F1, and it is the single largest structural difference between the two frameworks.
Aggregate excess liquidity tells you nothing about whether the bank that needs reserves has them. What has actually happened as the aggregate fell is that term repo grew: reserve-rich dealer banks lend secured to everyone else, cross-border, against collateral that would not qualify as high-quality liquid assets.
This works, and the ECB has said so. It has one dependency, which is that dealer balance sheet stays cheap. Leverage-ratio and G-SIB reporting dates are precisely when it is not.
The same graph as a table, which is the form the take-up model actually consumes. Rates and sizes resolve live against the snapshot; click a row to select that actor on the map above.
| Column | Actor | Rate it targets | Level | Tenor | Size | Alternatives | Frictions |
|---|
A transmission map with no frictions is just an accounting identity. These are the places where the identity holds but the behaviour does not. live means it is binding now; structural means it is a permanent feature of the design; latent means it is not binding yet and would be the thing that broke first.
This map exists to make one calculation possible. It does not yet perform it. What follows is the specification, so that what is missing is explicit rather than implied.
Each point is a calendar month. Horizontal axis: the spread between Eurex GC Pooling and the MRO rate — how much cheaper the market is than the central bank. Vertical axis: MRO outstanding. The 3-month LTRO is deliberately excluded, because until December 2024 that series carried TLTRO-III, and a €1.3trn subsidised programme is not a point on a demand curve. Two things are worth noticing: it slopes the way a demand curve should, and the September 2024 corridor narrowing moved the euro area a long way along it in a single step.
A bank chooses the cheapest way to fund a euro of assets for a given horizon, subject to what collateral it holds and what its binding regulatory constraint is. Bidding the MRO is one option among six. Its all-in cost is
Only the first term is observed. The map supplies the second from the collateral-specific repo series, the third from the published haircut schedule, and leaves the fourth and fifth as the two free parameters that a take-up model has to identify from data.
The fifth term is the one most often left out. Secured funding from the domestic central bank maturing inside thirty days carries a 0% LCR outflow rate whatever the collateral; market repo of the same maturity against collateral that is not HQLA carries up to 100%. For a bank funding illiquid assets at short tenor, that is worth more than fifteen basis points. Note it is an LCR effect and not an NSFR one — the NSFR's encumbrance relief runs off the remaining term of encumbrance rather than off the counterparty, and so does not discriminate between the MRO and a short market repo at all.
Take-up is small because the outside option is cheaper. GC repo has printed at or a little above the deposit facility rate, which is still comfortably below the MRO — so a bank that owns general-collateral-quality bonds has no reason to bid, and a bank that does not own them has no collateral advantage large enough to overcome the spread.
What would change the answer is not the level of excess liquidity but the GC-to-MRO spread, which is why that spread is the first number in the header of this page. The interesting threshold is not zero: it is the point where the spread, net of collateral opportunity cost and LCR relief, crosses the 15 basis points the Governing Council chose.
The graph is a versioned artefact, not a drawing. It lives in
scripts/transmission_map.py, validates on build (no orphan nodes, no dangling edges,
every series key present in the snapshot), and emits data/transmission_map.json.
Nothing on this page is typed into the HTML.
Every rate and size is a key into the same data/snapshot.json that drives the
excess-liquidity page, and therefore carries the API URL it was fetched from. Four daily rate
series — DFR, MRO, MLF and €STR — are refetched from the ECB Data Portal when the page loads,
because the headline arithmetic in the header should be today's and not the build date's.
Where a number is not an SDMX series it carries a published value and a date. Where a number is ours it is flagged assumed and appears in its own column, never averaged into a measured one. There are five such numbers on this page, all of them credit spreads on bank term issuance, and all of them are inputs the take-up model in §05 will have to identify rather than assume.
It is not a quantity map. Edge widths where a size exists are stocks or turnover on very different bases, comparable within a flow type and not across them. Turning the width scale off is the honest default for anything you intend to publish.
It is not twenty banking systems. Every node in the banks column is an aggregate over institutions whose collateral, funding mix and reserve position differ enough that the aggregate can mislead — which is exactly the thing that makes the take-up question hard.
And the structural portfolio and structural longer-term operations are drawn because they have been announced, not because they exist. As of the build date neither has been specified.
The decision this page describes: demand-driven floor, MRO−DFR cut to 15bp from 18 September 2024, structural operations and a structural portfolio to follow "at a later stage", key parameters to be reviewed in 2026. ecb.europa.eu
Sets the sequencing that makes the take-up question load-bearing: structural operations begin only after a persistent, broad-based rise in take-up of the standard operations. Also the source for the €600bn–€2.2trn range on steady-state reserve demand, and for the argument that the structural portfolio should be tilted short. ecb.europa.eu
Why the eligible list is as broad as it is, and the precautionary logic that makes the MRO an asset-transformation instrument rather than only a price. ecb.europa.eu
The empirical anchor for the take-up model: participation scales with funding need and with the ability to transform illiquid assets into reserves by pledging them — the collateral motive, not the price motive. ecb.europa.eu
The comparative case that euro-area stigma is low, and the reason offered: the ECB has never described its operations as a backstop. Friction F4 is the argument that this is contingent on reserves being ample rather than intrinsic to the design. newyorkfed.org
The mechanism behind friction F3 and the collateral channel: central bank bond holdings, not central bank lending, are what move specific-collateral repo away from the policy rate. esrb.europa.eu
