You are currently browsing the category archive for the ‘central banking’ category.

It’s been a rather long time since I last took a look into developments in Euro area central bank balance sheets so it is a good opportunity to do an update.

General Trends

First, here’s a table (source eurocrisismonitor) with the Target2 balances for the major Euro area NCBs. What is quite evident is that during the second half of 2012 and the start of 2013, Target2 balances were reduced significantly, with the pace slowing down considerably in the last few months. Current levels seem to mostly be a function of current accounts (surpluses) rather than actual financial flows. Especially Germany has stabilized around €575bn with Spain and Italy at €280bn and €210-230bn (for a total of nearly €500bn).

Euro NCB Aug 2012-2013


The above stabilization is clear if one looks at the detailed data of the Spanish NCB (BdE):

BdE Balance Sheet Aug 2013

Lending to MFIs through MRO/LTRO has stabilized since May something reflected on the Target2 net balance. During the start of 2013, the drop in refinancing operations was accounted by the reduction in the deposit facility from €47.4bn in January to €3.1bn in June. This should be considered a positive development since it probably reflects the fact that Spanish banks do not need any large liquidity buffers any more and are able to more actively use private money markets. Nevertheless, current Target2 figures seem to be quite sticky with current(+capital) account balances driving any changes.


Bank of Greece actually provides quite detailed monthly financial statements which can be used to draw important conclusions:

Bank of Greece Balance Sheet Aug 2013

During 2013, the main BoG asset categories have been on a downward trend, although the pace is rather slow with a reduction of close to €10bn in the Jun-Aug period. The positive fact is that this decrease has been driven by lower demands of ELA, which has dropped from over €30bn in January to a bit less than €12bn currently. The main counterpart of the reduction has been the Target2 balance which went from €87bn in January to €54bn in August, a result of various factors such as the current account surplus of recent months, inflows from the Greek Loan Facility and some inflows of private deposits. Banknotes have not exhibited any significant changes.

On the other hand, data on collateral haircuts is quite disturbing. Collateral only dropped from €219bn to €188bn, with the overall haircut increasing from 51% to 61%. This is mainly reflected by ELA which currently carries a haircut of close to 90%! Given the current size of MRO (€61bn), its haircut (25%) and current securities holdings of Greek MFIs (close to €77bn), it seems that Greek banks have posted most of their securities holdings as collateral for ECB regular refinancing operations while keeping credit claims for ELA. Current ELA collateral is close to 50% of the loans registered on their balance sheets as assets.

The 90% haircut is obviously very worrying for two reasons. First, it means that the ability of Greek MFIs to cover any liquidity shocks (due to a capital outflows scenario for instance) is quite limited: At 50% haircut they should be able to provide collateral for additional €60bn in liquidity while the current 90% haircut will limit them to only around €10-15bn. Secondly, haircuts of this size, cast doubt on the asset quality of posted credit claims. If Bank of Greece only accepts collateral at 10% of its face value, that fact should provide a hint of what the correct recovery rate is for the relevant loans (in case of debtor defaults). In general, it seems that current asset quality of Greek MFIs is extremely low while Target2 liabilities are still quite high.

Paul De Grauwe wrote a very interesting article on Vox a few days ago on ‘Fiscal implications of the ECB’s bond-buying programme’. The article tries to clear a few misconceptions about the treatment of a central bank balance sheet and its monetary operations and analyzes the results of a possible activation of the OMT program.

Although I agree with a large part of the article, there are a few points where I think that the authors are not entirely correct: The treatment of a government bond default and the inflation tax.

Bond Default

The article consolidates the government and the central bank into the public sector thus making any bonds held outright by the central bank as irrelevant from an accounting point of view since one part of the government owes a ‘debt’ to the other, which can be easily just canceled out. Even if the asset side of the central bank (CB) balance sheet is reduced substantially, the CB can still create liabilities (in the form of banknotes and bank reserves) without any need for an injection of capital by the Treasury (since its liabilities can only be exchanged for other central bank liabilities in a floating exchange rate regime).

Although the above is true, the article implies that government bonds are the only assets generating income for a central bank (in which case it is not a problem to cancel out bond coupon/principal payments with central bank profit remittance to the Treasury). In reality and especially in the case of the ECB, a central bank earns income on other assets as well such as short/long-term loans to commercial banks (MRO/LTROs in the ECB case) and it might also hold outright other interest bearing securities not issued by the government (such as MBS and covered bonds).

A default on bonds held on its balance sheet would lead to a reduction of capital and if the latter was not enough to the creation of a new asset containing ‘future profits’ on its balance sheet. As long as the central bank held negative (or not enough) capital, no profits would be returned to the Treasury. As a result, although the Treasury would not have to pay any interest payments on the defaulted bonds to the central bank, it would lose any ‘seignorage’ profit from the other income generating assets held by the latter. On a cash flow basis, the overall Treasury position would become negative with an increased ‘deficit’ which should be covered by new sources of funds (higher taxes or new borrowing).

An easy fix is to just assume that the central bank would acquire (on a yearly basis) any government debt necessary to finance this deficit. Nevertheless, this negates the initial proposition that the government can default on its bonds held by the central bank without any real financial consequences. The reality is that any future negative cash flows are a true cost for the Treasury and should be included in any accounting exercise.

Inflation Tax

The article concludes by repeating the (long-dead) money multiplier concept and stipulating that, due to the current crisis/liquidity trap, the money multiplier is zero which provides the central bank with enough room to pursue any program of monetary expansion (through outright purchases of government bonds) without any threat to price stability. The reality is that this way of thinking is deeply flawed:

  1. As acknowledged by all central banks (Fed, ECB, BIS) and their chief economists (Bindseil) the money multiplier framework does not really apply to modern central bank operations.
  2. Since the ECB provides a deposit facility at a certain spread over its target interbank rate, its operations are always sterilized since that facility provides a floor under which interbank rates cannot fall (since no bank would lend in the interbank market for a lower rate than the risk-free deposit rate at the central bank apart from cases of ‘repo specials’).
  3. The ECB can always use other mechanisms to sterilize its purchases such as fixed-term deposits (as it does already) or even debt certificates, thus negating any increase in the monetary base.

As a result, price stability is not really a constraint on any bond buying program by a central bank and any OMT activation does not risk run-away inflation today or in the future.

Taking a look at recent ECB term deposit rates (the 7-day term deposits used to sterilize SMP liquidity) one will observe that rates have expanded somewhat from 1bp to 4-5bp (the latest auction settled around 5bp):

ECB SMP Term Deposit Rate

Total amount of bids has also fallen below €300bn (with SMP liquidity around €206bn). This indicates that overall excess liquidity has fallen considerably (based on the term deposits auctions to lower than €100bn) and is pushing short-term rates higher. This is reflected on overnight repo rates with the Eurex GC Pooling EUR Funding Rate climbing to 4-5bp. LTRO repayments and lower Target2 balances have started having an impact on money market rates. The latest auction only had €254bn in bids and is a good indicator of available funds in the Euro money market.

Falling excess liquidity will start pushing money market rates closer to the cost of funds (currently 75bp). Unless the latter is lowered by the ECB it is possible that money market lenders will start avoiding long repo maturities and focus on short-term deals (overnight/weekly). It also points to a higher probability of an ECB rate cut in the immediate future.

Since it’s been a few months since I took a look at periphery NCB balance sheets, it’s time to examine trends during 2013.


Spain seems to be the one driving overall Open Market Operations (OMOs) usage down:

BDE balance sheet Mar 2013

Since December, OMOs have declined from €357.29bn to €270.94bn (-€86.35bn) with 80% attributed to a fall in LTRO usage. It seems that Spanish banks are confident to repay a significant part of their LTRO borrowing from the ECB which, given the low interest rates of LTRO funding and the relaxed collateral rules, imply that market conditions have improved strongly. During the same period, the consolidated Eurosystem OMO funding to European banks has dropped €225bn which means that Spanish banks account for almost 40% of the relevant fall (with another large part accounted by French and German banks).

Looking into the balance sheet at more detail, one observes that use of the deposit facility was €44.2bn in December and only €10.94bn during March, reflecting much lower safety buffers for Spanish banks. This is linked to the drop in Target2 liabilities, from €352,4bn in December to €298.3bn in March (-€54.1bn). Still, liabilities continue to be high, almost 28% of Spanish GDP although they are much lower than their maximum of €428.6bn in August 2012 (a fall of €130.3bn).

The drop in Target2 liabilities is related to the increase in non-residents government debt securities holdings, with registered holdings increasing €30bn since October, from €209.6bn to €240.4bn in February (an almost 15% increase).

Overall, credit conditions have clearly eased during the last few months. Nevertheless, both the situation in the real economy and current NPL figures point to large risks ahead for the Spanish banking system. Spanish banks will probably keep low quality collateral (such as credit claims) parked at the ECB and only use high quality securities in order to borrow in the repo market at low interest rates   (since current repo rates are close to 0.02-0.03%).


The Italian case seems to be a bit more muted than Spain. Since December, bank borrowing from the central bank of Italy has dropped only marginally from €271.8bn to €268.2bn with a somewhat larger fall in LTRO usage from €268.3bn to €262bn. This is mainly explained from the fact that Target liabilities only dropped from €255.1bn to €242.9bn. It seems that Italy has decoupled from Spain, probably due to the results of the recent national elections and the inability to form a stable government as well as the fact that government debt figures are moving close to the 130% GDP figure.

This is also reflected on the general government debt statistics which illustrate the fact that non-residents holdings of securities have been extremely steady during the last few months and are much lower than 2011 figures.

It’s been a rather long time since I last took a look at central bank balance sheets so I think it would be interesting to see where we stand, especially in the case of Greece:

BoG - Dec2012 - Feb2013

Since 2012, the ECB started to accept Greek collateral in its regular refinancing operations which allowed Greek banks to move a large part of ELA loans (paying high interest rates around 3%) to the ECB MROs. Moreover, the fall in banknotes demand continued, lowering the corresponding Eurosystem liability by €2.3bn while the new loan package (which included cash to be used by the Greek government to pay arrears) and a return of deposits back home lowered Target2 liabilities by €20.2bn to €78.14bn (40% of 2012 GDP). As a result, total bank lending dropped by €24.81bn while also lowering their costs of funds significantly (although there was a €5bn increase in time deposits in the corresponding period).

The last fact is evident in the effective haircuts applied on posted collateral (calculated simply as 1 – loans/collateral). Regular refinancing operations have a mean haircut of 22% (the debt buyback probably played its role since it allowed Greek banks to swap Greek bonds with high quality EFSF notes), making the effective lending rate close to 1% while ELA financing has a haircut close to 80% (assuming that all of ‘Other Claims’ are ELA loans) with an effective rate of.. 15%. What is rather worrying is the fact that collateral posted on ELA was rather steady despite the 80% drop in its usage, increasing the effective haircut from 47% in December to 82% in February. Greek banks have probably posted very low quality credit claims as collateral (with their best assets held for the ECB MROs), leading to large haircuts and asset encumbrance.

Since the Greek banks asset quality is not destined to improve any time soon (rather the NPL share will probably increase further), it is very important that the positive developments in the Eurosystem liabilities (banknotes and Target2) continue. Total collateral posted (€212.75bn) is still very high, especially compared to total assets available (€240bn in loans after provisions and €76bn in securities – €23bn in securities of countries outside the Euro area). The Greek banking system remainsl on a fragile balance.

Since I had to write an essay on monetary policy I ‘ve posted the relevant short paper on SSRN. It examines Quantitative Easing in relation with payment of Interest On Reserves and how it can help the central bank to achieve its monetary targets. The main reference has been the work of Scott Fullwiler, especially his paper on the subject.

Lately, there’s (finally) some talk about a Large Scale Asset Purchase (LSAP) program from the ECB. I ‘d like to elaborate a bit on the implementation of such a program since the Eurosystem structure and the segmented government debt market pose certain issues.

Usually, LSAPs are used mainly to steer long-term interest rates and provide monetary stimulus to the economy by making financing of durable goods purchases and long-term investment (like housing) more affordable. In the Euro case though, the main problem is the disfunctioning monetary transmission mechanism with certain government securities not being perfect substitutes with bank reserves. This is evident if one compares yields with corresponding Overnight Index Swaps:

Euro government securities - OIS

Since there’s still excess liquidity in the Euro banking system, EONIA mostly arbitrages with the ECB deposit rate rather than the MRO rate (since banks with excess liquidity are not the ones borrowing excess funds). As a result, short-term government paper should be mostly considered as the equivelant of a fixed-term deposit at the ECB, making them close substitutes with bank reserves.

It is clear from the above table that core countries securities actually carry a premium (compared to unsecured interbank lending swaps) up to 1 year and arbitrage quite closely for longer maturities. On the other hand, Italian and Spanish paper carry considerable risk and are ideal targets for LSAP. Any GDP-weighted LSAP would waste a large part of resources without having any real ecnomic impact (since they would be a perfect asset swap, replacing assets with roughly the same ‘economic value’).

An important issue with LSAPs is which NCB will actually perform the purchases. As long as non-issuer country NCBs buy securities, any coupon payments will register as an increase in Target2 liabilities (for the issuer country) and contribute to the Eurosystem profits (minus the fixed-term deposit rate used to sterilize any purchases). Unless monetary profits are remmited back to the issuer Treasury, they will be removing domestic interest income and not contribute to any reduction in the current account income deficit of the issuer.

Taking an accounting view, LSAP will remove an interest bearing asset from bank balance sheets and replace it with bank reserves which currently pay close to zero (zero for the deposit facility and close to zero for the fixed-term deposits). The most probable impact will be a reduction in net lending from the ECB by the banks selling the assets, although the speed will depend on if their loans are short-term (MROs) or long-term (LTROs). The net income impact would probably still be negative while it would allow for collateral to be removed from ECB borrowing and made available in private money markets. Any capital gains would provide immediate income for sellers and lower collateral needs of banks borrowing from the ECB and using government paper as collateral (due to positive effects from daily mark-to-market). As a result, a LSAP announcement would probably be followed by excess liquidity banks trying to front-load the ECB.

Since the ECB uses weekly-term deposits to sterilize the impacts of any asset purchases, the net effect would be a reduction in liquidity available for private money markets and a probable push of interbank rates closer to the MRO rate. This means that any LSAP would be accompanied by a reduction of the MRO rate to lower levels such as 0.5% in order to avoid a defacto monetary tighting stance.

Overall, LSAP can be positive but can have several unintended consequences.

Bubbles and Busts had a very nice post on the Impossible Trinity (the fact that an economy cannot simultaneously maintain a fixed exchange rate, free capital movement, and an independent monetary policy) and how it is related with Interest On Reserves (IOR). Since IOR, especially in a floor system, allows a central bank (CB) to ‘sterilize’ the expansion of the monetary base (and keep control of the short-term interest rate) it ultimately can help avoid problems in some cases of exchange rate targeting.

In particular, while a currency peg with an external deficit implies that the central bank must be able to sell enough FX in order to maintain the peg (either by already having large FX reserves or by being able to borrow from official sources) thus making the peg unstable, a CB can avoid its FX rate from appreciating by buying FX and selling its local currency (a clear example is the Euro currency floor for the Swiss franc by the SNB). By paying IOR, the CB can maintain a non-zero interest rate target despite of the monetary base expanding (due to FX intervention). In any other case, the CB would have to use other measures to sterilize its operations such as increasing reserve requirements (the usual policy followed by PBoC) or selling securities (outright or through reverse repos) held in its portfolio.

What I would like to note is that, unless the capital inflows are a result of flight to quality and thus temporary (such as the inflows to the Swiss franc), they usually reflect strong underlying factors such as a current account surplus and/or large interest rate differences. In such a case, the currency appreciation will not really end unless the original cause is reversed. As a result, the CB might end up in a position where a large part of its asset base is in foreign currency and paying low interest rates while its liabilities are in the form of local currency and paying much higher rates. Ultimately, the question of avoiding the trilema is one of if the CB can accept long-term losses on its operations and even a negative capital base. This question has been asked sometimes (especially in the case of the SNB and ECB) but with no clear and definite answer (although the Asian central banks seem to be less concerned with the problem).

The BdE released its balance sheet data for December yesterday. There’s a significant drop in Target2 liabilities to €352.4bn (from €376.3bn in November). Overall, since a peak of €428.62bn in August, liabilities have dropped by €76.22bn, reversing private capital flows (mainly driven by the ‘Draghi-OMT effect’). As a result, Spanish banks managed to create a large deposit facility buffer of €44.2bn (from €24bn in November) and also to reduce their recourse to regular refinancing operations by €7.6bn to a total of €357.3bn. Banknotes and deposits of general government were somewhat lower during this month.

Spanish banks don’t seem to regard the current situation as perfectly stable, something evident from the fact that they increased their deposit facility holdings substantially even though excess reserves pay zero interest (and did not opt to pay down more of their borrowing from BdE). This indicates fear of a possible price shock on their main collateral (government bonds) which would increase their margins with BdE and risk current private capital inflows. Nevertheless, the fact that private financial flows have clearly reversed and the current account deficit has been closing and is now less than -€1bn/month makes the Spanish economy a rather ‘closed’ one. Lower money market and deposit rates as well as the large deposit facility buffer will be supportive of government debt auctions in the near future (although it is possible that the Spanish government might take the route taken during the first months of 2012 when the LTRO effect allowed it to auction off more funds than immediately required and increase its deposits at Bde – something that would lower the Spanish banks safety liquidity buffer).

The ECB released its weekly statement for the week ending at 21 December 2012. Since on 21/12/2012 Greek government titles began to be accepted as collateral in ECB refinancing operations this week’s statement should provide for a closer look at the magnitude of the change in Greek banks borrowings.

Asset Side

Looking into the asset side ‘Other claims’ were lower by €20.7bn which should correspond to the reduction of the use of ELA by the Greek banking system. Since MROs happen in the middle of the week, Greek banks would only have access to the marginal lending facility which shows an increase of €13.61bn. The almost €7bn difference should most probably be attributed to the (cash) part of the December Greek disbursement which would be used for budget needs (the rest of the funds were in the form of EFSF notes and bills). Regular operations (MRO and LTRO) were lower by €4.9bn which should be attributed to the generally positive climate since the OMT announcement.

Liability Side

On the liability side banknotes in circulation show a very large increase of €11.91bn, probably of seasonal (Christmas) nature. Bank reserves are 11.62bn lower and liabilities to other euro area residents by €18.27bn, both in the case of ‘General Government’ and ‘Other liabilities’. Liabilities to non-euro area residents are €3.57bn higher which is a negative development.


The latest MRO (which should include any borrowing by Greek banks from the marginal lending facility being moved to the regular lower rated weekly operation) shows an increase of €17bn from €72.8bn to €89.66bn. In general, it seems that Greek banks managed to move a large part of their ELA borrowing to the ECB regular operations while also lowering their Target2 liabilities (with the help of the disbersement). Things will be much clearer when BoG releases its December balance sheet.

On a related note, the ECB also included the Greek collateral haircut list in its decision of 19 December. The haircuts are quite heavy (15% for T-Bills of up to 1 year maturities) but will still allow for much lower yields in the forthcoming T-Bill auctions which could move close to 1% (from over 4%). Such a development would make current T-Bills a high return investment. Remaining Greek bonds will face haircuts of 56-57% which should place a rather high floor on their yields.

Greek collateral haircuts

About Me

Kostas Kalevras

LinkedIn profile

E-mail:kkalev AT gmail DOT com
My status
Follow on twitter
More about me...