The ECB’s new interest rate framework/the Italian problem
• With sizeable TLTRO maturities this year and TLTRO-created reserves
all but disappearing less than a year from now, reserves in the Italian
banking system will fall to uncomfortably low levels before an acute
shortage emerges in early 2024. Already next June, we estimate that
the Liquidity Coverage Ratio of significant banks in Italy will fall by 50
percentage points By March 2024, it will be below 100. The ECB will
judge this situation to be unwanted and unwarranted, forcing it to act.
• The ECB will find itself at a crossroads by the coming summer. We see
three options ranked in order of likelihood:
• In the second half of 2023, the ECB will take a leaf out of Bank of
England’s playbook. The ECB will align its deposit rate with its main
borrowing rate, the MRO rate. This will allow the ECB to seamlessly
cover both anticipated and unanticipated reserve shortfalls in parts of
the Euro Area banking system while shrinking its balance sheet with
Quantitative Tightening and TLTROs as they come due. More
importantly, the ECB will prevent a bifurcation of money market rates
between member states where bank reserves are relatively high and
where they are relatively low. At the same time, the ECB will keep its
operational framework simple and prevent US-like liquidity crunches;
• Stick with the current framework, which will further fragment
monetary policy transmission, but which will still prevent a liquidity
crunch while preventing potential communication problems
surrounding aligning the deposit rate with the MRO-rate;
• A blast from the past: 1-week deposit tenders or their newer version:
1-week ECB certificates. The ECB would pay a heavy price for mopping
up reserves where they are most plentiful: in the core Euro Area
countries. The national central banks will experience increased
operating losses, while risks of a US-like money market breakdown rise.
• In each scenario, we see odds of the ECB instituting a standing facility
for nonbanks as very, very remote.
• Regardless, because of TLTRO repayments and Quantitative
Tightening, the ESTR-deposit rate spread should narrow markedly in
the second half of the year.
At first sight, liquidity conditions in the Eurozone banking system appear to remain plentiful even when taking into
account maturing TLTROs. At the end of the first quarter of 2023, excess liquidity (excess reserves would be the
better wording) stood at 3,868 billion euros. While that’s quite a bit below the late 2022 peak of 4.7 trillion euros,
the decline in reserves hasn’t moved the dial with regards to ESTR settlements. The ESTR-deposit rate spread
averaged 9.67bps in the fourth quarter of 2022. In the first quarter, the spread averaged 10.0bps despite the
aforementioned decline in reserves in the Eurozone banking system.
Looking ahead, reserves are set to fall further because of ECB Quantitative Tightening and TLTRO maturities. The
table below shows remaining borrowings of the seven pandemic TLTRO tranches:
By June, banks will repay 477 billion euros, leading to a corresponding one-on-one decline in excess reserves.
If we add to TLTRO repayments Quantitative Tightening and assume so-called autonomous factors1
to remain broadly stable; then by the end of the year excess reserves will fall to 3 trillion euros:
We assume 15 billion euros a month in QT in the second quarter, and the pace to double to 30 billion euros a
month in the second half of the year. Based on the past relationship between excess reserves and the ESTR – deposit rate spread, the spread should
narrow to 6-7bps at when excess reserves are around 3 trillion euros:
On the other hand, the linear relationship between reserves and the spread suggests the spread should narrow
by 2bps from its current level of 10bps. Note that the relationship between reserves and the ESTR – deposit
rate spread is linear as long as reserves are ample. When reserves start to become scarce(r), marginal decreases
in reserve result in proportionally stronger increases in ESTR settlements (see here and here).
In any case, recent settlements of ESTR have been unusually low relative to the ECB deposit rate despite the
aforementioned less easy liquidity conditions in the Euro Area banking system. The recent relatively low ESTR
settlements are a topic for another research note. At this point, we can only speculate about the causes, ranging
from bank balance sheet leverage constraints to the still very elevated level of bank deposits over bank loans in
the banking system (i.e. deposits remain too plentiful because of ECB Quantitative Easing), both of which
incentivize banks to ‘underbid’ for deposits.
If we breakdown the level of reserves in the banking systems of the big five member states, a picture of plentiful
reserves seems to be confirmed:
But scratch below the surface, and a very different picture emerges. In the chart below we have netted current
account/deposit facility holdings with borrowings from the ECB. We dub this the net debtor/creditor position of
the banking system in that member state vis-à-vis the Eurosystem:
Banks in the core countries (and even Spain these days) are net creditors vis-à-vis the Eurosystem.
The net creditor status is purely the result of the mechanics of ECB QE. Nonbank holders of core government
bonds that sold their holdings to the ECB deposited the proceeds in the banking systems of the core member
states. Thus, the corresponding reserves that the ECB created with QE stayed within the national banking system.
For example, a German pension fund that sold German government bonds to the ECB deposited the proceeds
with a German bank. The deposits and corresponding reserves that the ECB created with QE stayed in Germany.
However, sellers of Spanish and Italian government bonds tended to deposit the proceeds in the banking
system of Germany in particular. The resulting migration of bank deposits and corresponding reserves resulted
in the literal blowout of Target2 imbalances:
Since the ECB ended QE almost a year ago, Spain has seen a net inflow of capital into its bond market, resulting in
a decline in Target2 liabilities of the Bank of Spain and a corresponding increase in reserves in the Spanish banking
system. Italy saw no such net inflow of capital. Thus, Target2 liabilities haven’t registered a notable decrease. As
a result, the level of reserves in the Italian banking system has stayed relatively low. In fact, if the ECB called in all
TLTROs overnight, the Italian banking system would have a reserve shortfall of 66 billion euros. Put differently,
Italian banks would have to borrow from the ECB to pay back the ECB; or borrow reserves from outside of Italy:
either directly by borrowing from other banks, or indirectly by issuing bonds to foreigners or attract foreign
Fortunately, the ECB won’t call in the TLTROs overnight. But when the last TLTROs mature over the course of
2024, Italian banks will be short of reserves. Which begs the question, when will the level of reserves in the Italian
banking system become too low for comfort, which will force the banks to act?
Recall that in June, the biggest TLTRO tranche worth a remaining 477 billion euros will mature. How will the
repayment affect the Italian banking system? The chart below shows the evolution of TLTRO borrowings in the
Italian banking system, with that of Spain provided for comparison purposes:
For example, if an Italian bank attracted corporate deposits from, say, the Netherlands, then the Dutch banking system
would lose wholesale deposits (a liability) and corresponding reserves (an asset). The Italian bank would record an increase in
wholesale deposits (a liability) and reserves (an asset) on its balance sheet. At the same time, the Dutch central bank would
see its Target2 claims (an asset decline), and see bank reserves decline (a liability). The Italian central bank would see an
increase in bank reserves (a liability), and a decrease in Target liabilities.
By June 2020, when all Eurozone banks splurged on TLTROs to the tune of roughly 1.5 trillion euros, Italian banks
had taken up 351 billion euros in long term loans from the ECB. By February 2023, the latest data point available,
that had fallen to 330 billion euros. So, by definition it must be true that Italian banks rolled older TLTROs (those
from June 2020 or older) into newer tranches. Since there is now publicly available information of TLTRO
borrowings per member state for individual tranches, we have to make several assumptions. We are going to
make the most conservative assumption, namely that Italian banks have been quite aggressive extending the
maturity of the TLTROs by rolling older tranches into newer ones at the fixed quarterly repayment dates.
Between June 2020 and December 2021, all Eurozone banks repaid EUR155 billion in TLTROs. We simply assume
that half that amount represents Italian banks aggressively rolling older TLTROs into new one. That’s 77.5 billion
euros, or 75 billion euros for simplicity’s sake. Furthermore, we know as a fact that Italian banks increased their
TLTRO borrowing by EUR75 billion in 2021. So, out of the remaining 330 billion euros in Italian banks’ TLTRO
borrowings, 150 billion euros will mature in 2024. The remaining 180 billion euros will mature this year, with the
overwhelming bulk – say 150 billion euros – maturing in June. How will the 150 billion bumper repayment affect
the liquidity position of the Italian banking system?
According to the ECB’s MFI data set (apologies for the jargon, MFI stands for monetary financial institution, or
bank to you and me), reserves held by the Italian banking system stood at 264 billion euros by the end of
February. If we assume a 150 billion repayment for Italian banks, and simply assume that banks in the other big
member states will repay all their borrowings, reserves to total assets/liabilities will remain above pre-pandemic
levels even in Spain:
But in Italy, the level of reserves to total assets/liabilities will barely be above pre-pandemic levels.
Furthermore, the level of reserves will be far below levels in Spain and especially the core countries. For what
it’s worth, at the time of the SVB failure in the US last March, reserves of small US banks had fallen to 6.8 percent
of total assets/liabilities (big banks were at 11.6 percent).
For our regulatory-minded readers, we can also calculate the effect of the fall in Level 1 High Quality Liquid Assets
(i.e. central bank reserves) as a result of TLTRO maturities. According to ECB SDW data, the estimated decline in
central bank reserves of 150 billion euros by June will cut the stock of Level 1 HQLA by about 25 percent,
resulting in a 50 point decline in the LCR:
Note that the other series aren’t adjusted for TLTRO repayments. Also note that GSIBs report the lowest LCRs
while small(er) banks report the highest LCRs, a situation that is the mirror image of the US situation. In any
case, we judge the estimated LCR of 133 percent not to cause any problems for the Italian banking system.
However, when all TLTROs have matured by the end of 2024, the LCR of the Italian banking system will fall below
the benchmark level of 100.
Some historical perspective on the expected fall in the reserves-to-total assets/liabilities ratio. In 2018 and 2019,
when bank reserves started to fall because of the ECB tapering QE and TLTRO repayments, Italian banks were not
reserve-constrained in their lending to the real economy. Analyzing the data shows that there is no correlation
between the level of reserves and bank lending in the biggest five member states in the era of excess reserves
that started with the TLTROs in 2014 and QE in 2015. Furthermore, if Italian banks were reserve constrained, that
would have shown up in Italian banks bidding up money market rates: wider Euribor-OIS spreads and FRA-OIS
widening. No such thing happened for a very good reason. Since overall bank lending on the eve of the pandemic
remained relatively subdued, TLTROs were widely seen as a permanent feature of ECB policy. Simply put,
everybody expected the ECB to keep offering TLTROs indefinitely under the pretext of too tepid bank loan growth
across the Eurozone.
In the present, however, the writing is on the wall for the TLTROs so to speak. The ECB will almost certainly not
renew the TLTROs. Furthermore, by March 2024 the last big TLTRO tranche, on which Italian banks are heavily
reliant for reserves, will mature. But before that key date, another estimated 180 billion in TLTROs to Italian banks
will have matured. Plus, ECB Quantitative Tightening and the ‘natural’ growth in autonomous factors will drain
additional reserves from the Italian banking system.
Real bank lending rates in Italy moved in lockstep with real lending rates in the other five economies. What’s more, the level
of real interest rates in Italy was in line with the other member states.
To conclude: the outlook for reserves in the Italian banking system in the present is very different compared to
2018-2019. Back then, bank credit creation wasn’t constrained by reserves; and reserves were expected to
remain cheaply available because of the TLTRO backstop. But in the next twelve months, reserves will become
scarce, and that will very likely negatively affect bank lending through higher lending rates compared to member
states where reserves will remain relatively plentiful.
There is another factor at play, namely the other holder of reserves in Italy besides the Italian banking system: the
state Treasury. It turns out it is certainly not the case that the Italian Treasury is cash-rich – to use the market
parlance. In February, the most recent available data point, the Treasury’s cash balance at the Bank of Italy had
fallen to just 5 billion euros, the lowest level since the pandemic outbreak:
We are not predicting a cash/reserves crunch in Italy. In Italy, such a thing can always happen as politics has a
habit of intruding. The point we want to make, as we will show below, is that the ECB will have to act. In part two
we will show how.
Part II: the new framework and solving the Italian problem
The expected reserve shortfall in bank reserves in the Italian banking system will very likely cause what the ECB
dubs fragmentation of its monetary policy. Fragmentation, or the postal code problem, means that the Italian real
economy will face tighter credit conditions not because of fundamentals of its corporates and households, but
simply because its banks are short of reserves. The ECB has a habit of responding to fragmentation: from former
President Draghi’s “whatever-it-takes” bond buying program to President Lagarde’s pandemic bond buying
program and the aptly-called ‘Transmission Protection Instrument’, which was launched last summer.
Enter ECB Board Member Isabel Schnabel. In a both important and interesting speech dated March 27, Schnabel
floated several trial balloons regarding money market interest rate control in the era of maturing TLTROs and
Quantitative Tightening. Foreseeing the reserve shortage in Italy (we hope), Schnabel compared the money
market and bank reserves frameworks of Sweden, the United States, and the United Kingdom. She specifically
signaled out and praised the Bank of England framework, whereby the central bank offers reserves at the same
interest rate that it pays for excess reserves, as having particular relevance for the Euro Area. The reason is
simple: the UK situation most closely resembles that of the Euro Area because of the importance of banks in
providing credit to the real economy in both jurisdictions. Plus the notion that a shortfall of reserves in parts of
the banking system could cause unwarranted tightening of financial conditions (i.e., shock-like reductions in bank
lending instead of gradualism that central bankers prefer).
The Bank of England framework is designed from the outset to seamlessly cover both anticipated and
unanticipated reserve shortfalls that result from the central bank shrinking its balance (through Quantitative
Tightening and ending TLTRO-like schemes). By aligning the deposit rate with the lending rate, the BOE wants
to achieve two things: avoid stigma being attached to borrowing from the central bank; and securely anchor the
overnight rate to one central bank rate. Simply put: the BOE wants to convey the message that depositing
reserves at the central bank and borrowing reserves from the central bank are both business as usual. Now
compare the Bank of England framework with the US situation. The Federal Reserve only found out it had a
reserve shortfall in the banking system after the fact: namely when SVB Bank failed.
Lending reserves to the banking system is business as usual for the ECB. The ECB lends reserves to banks regularly
since time immemorial: the Main Refinancing Operations with a one-week maturity; and the Longer Term
Refinancing Operations with a three-month maturity. Recent uptake has been marginal, amounting to several
billions of euros.
As part of the well-telegraphed redesign of its operational framework, the ECB could take a leaf out of the BOE’s
playbook by aligning the MRO with the deposit rate. Without such an alignment of the central bank overnight
rates, we would end up with a situation whereby the MRO rate becomes the de facto anchor for money market
rates in low reserve member states (Italy, and possibly others); while the deposit rate remains leading in
member states with still high levels of excess reserves.
To manage the supply of reserves in the BOE-like framework, the ECB could tweak its collateral rules or reserve
remuneration rules. Put differently, the ECB will want to avoid a situation whereby banks ‘overborrow’ reserves
and invest reserves in higher yielding assets – carry trades.6 After all, the ECB’s monetary policy aim still very much
is to gradually shrink its balance sheet.
Note that uptake at the Bank of England lending operations has been modest, running at several billion pounds. See:
2&C=IS5&C=2DF&Filter=N6 To put it bluntly, the ECB will want to steer the amount of carry trades undertaken by the Italian banking system. Not too
much (which could be seen as ‘giving the Italians a free pass), and not too little, which would put undue pressure on Italy.
The BOE-like framework has the added benefit that the ECB can continue to focus on (aggressively) shrinking its
balance sheet via Quantitative Tightening while keeping its money market framework simple with one single
overnight rate. Of course, money markets rates will still increase relatively more in member states where reserves
are relatively low compared to member states where reserves are relatively high. As it should be, it must be
emphasized. But the ECB would limit such divergences by aligning the MRO rate with the deposit rate. More
importantly, the ECB would prevent a shock-like increase in money market rates and broader interest rates in
member states where the MRO becomes the leading interest rate when the MRO still stands 50bps above the
Time is short for the ECB to overhaul its operational framework. By June, the relative level of reserves in the
Italian banking system will have returned to its pre-pandemic level, while it will remain elevated in the other big
member states. By March 2024, when the next big TLTRO tranche matures, Italian banks will be short of reserves.
If one expects no sizeable net capital inflow into Italy, and banks not being able to cover the expected reserve
shortfall with private borrowing, then the Italian banks will have to turn to more costly ECB funding at the MRO
rate (which will be 50bps over the deposit rate). The ECB will consider that situation to be unwarranted
fragmentation of its monetary policy. Hence, expect the ECB to align the MRO rate with the deposit rate in the
second half of the year. That’s our base case scenario.
Of course, there is a communication problem with aligning the MRO rate with the deposit rate if the ECB isn’t able
to do so as part of rate hikes over the course of two monetary policy meetings. Put differently, if the ECB can’t
increase the deposit rate relative to the MRO rate so that the two are equal, it will have to simply cut the MRO
rate. The ECB could find it difficult to convince the market and the public of the merit of cutting the MRO rate
while at the same time promising to keep interest rates high for long(er) to fight stubborn inflation. Aligning the
two interest rates could easily be seen as the start of the easing cycle, or the ECB ‘going soft’ on Italy (again).
A workaround would be an apt communication strategy, rename the MRO rate or simply put in place a new
standing lending facility with the interest rate equal to the deposit rate. Of course, the ECB would needlessly
complicate its operational framework, a framework that has served it well since its inception.
If the obstacles to aligning the MRO rate with the deposit rate prove to be insurmountable, the deposit rate will
remain the leading interest rate in the bulk of the member states, while in Italy – and perhaps others – the MRO
rate will be the leading interest rate. Or: a floor system for most, and a corridor system for some. If that would be
the case, it would be best if the ECB managed to sneak in a corridor-narrowing interest rate increase in the
monetary policy meetings in the third quarter (or perhaps even fourth quarter). That would soften the blow for
Italy and other member states with relatively low bank reserves.
As an alternative to the BOE-like system of aligning the central bank deposit rate with the central bank lending
rate, we have seen trial balloons that all come down to the following: let the ECB pay more for central bank
reserves as to tighten money market conditions in the high-reserve member states in particular. At the March
2023 Money Market Contact group meeting, trial balloons pertaining the usual suspects were floated: ECB
certificates and/or a standing facility for nonbanks. Regarding the former, think of the weekly deposit tenders
that the ECB held from 2010 to 2014 that paid 50bps over the deposit rate. So, equal to the MRO rate. If that’s
the case, ESTR would settle below the 1-week interest rate/MRO rate, with the spread dependent on the size of
the new facility and the amount of reserves in the banking system. The bigger the size of the 1-week facility (or
the greater the amount of ECB certificates outstanding, and the lower the amount of reserves in the banking
system, the tighter the spread between ESTR and the MRO rate/1-week rate.
ECB 1-week certificates or a 1-week deposit facility are less likely than the MRO rate/deposit rate alignment.
For starters, the ECB will be forking over even greater amounts of interest payments on excess reserves to the
banking system. That’s a problem as many – if not most – national central banks are suffering heavy losses on
their bond portfolios that yield far less than what they pay on bank reserves. Furthermore, the notion of ECB certificates will set off alarm bells ringing in core countries as they will be seen as euro bills in disguise. Better call it weekly deposit tenders then.
Thirdly, a 1-week facility will likely exacerbate the problem of low reserves in Italy, with the damage increasing
the greater the size of the facility. Italy would become unnecessary and unwarranted collateral damage in
mopping up reserves in the core.
Fourthly, the ECB will need to estimate ex ante what the level of reserve scarcity is. Then, the central bank will
need to fine-tune the size of the facility to prevent a blow-out of money market rates or a money market seize up
like the 2019 US repo market carnage, or the most recent turmoil surrounding SVB and other smaller banks.
Fifthly, why would the ECB come up with a 1-week facility when TLTRO-created reserves will all but disappear
in the next twelve months, and when Quantitative Tightening has just begun? Better wait and see. Finally, the
ECB will needlessly complicate its operational framework while a simple and elegant solution is available: the
aforementioned alignment of the MRO rate with the deposit rate.
On a final note, we see odds of a standing facility for nonbanks as very remote. For starters, the ECB is already
working on the panopticon that is the ‘digital’ euro, which will give nonbank financials quasi access to the ECB,
but on a marginal scale though. Furthermore, the problem of excess wholesale deposits acting as a drag on banks’
balance sheet and profitability has started to ease. The chart below shows the ratio between all wholesale
liabilities to total/assets liabilities. The ratio peaked in December 2021 and has been on a marked decline ever
since. QT will accelerate the decline as the nonbank private sector will hold an increasing amount of (government)
debt securities, thus easing the pressure on bank balance sheets.
AFS GROUP AMSTERDAM
All opinions and estimates expressed in this document are subject to change without notice. AFS does not accept any liability whatsoever
for any direct or consequential loss arising from the use of this document. This document is for information purposes only and is not, and
should not be construed as, an offer to buy or sell any securities or derivatives. The information contained in this document is published for
the assistance of the recipient, but is not to be relied upon as authoritative or taken in substitution for the exercise of judgement by any