RebelEconomist comes late to this debate, but since it appears to remain
active, his views may yet be useful. For what it is worth, RebelEconomist agrees with
Hans Werner Sinn's conclusions that the European Central Bank (ECB) is conducting a stealth bailout of the peripheral eurozone economies and that this bailout disadvantages German borrowers, although Sinn's explanation could have been better. The key point, which none of the articles on this subject read by RebelEconomist has made explicit however, is that, because the ECB has set apparently more lenient loan collateral standards than the market for the types of assets that peripheral countries' banks tend to hold, the eurosystem is effectively lending to these banks at a subsidised interest rate, meaning that the eurosystem has become the marginal lender that finances these countries' net euro payment outflows. At the very least, Sinn deserves kudos for raising to public attention such an arcane but significant issue, which had not occurred to RebelEconomist despite his interest in central banking.
In order to judge the arguments made by Sinn and his detractors, it is necessary to appreciate how a
cross-border euro payment directly and indirectly affects the balance sheets of the various financial institutions involved. This may be seen by tracing the impact of a typical cross-border payment through the eurozone banking system. Since Sinn was criticised for associating these effects with current account imbalances, I would offer instead a simplified version of his example of an Irish farmer buying a German tractor, in which the payment is exactly the same as it would be for a capital account transaction arranged by the farmer to move his deposit from an Irish bank of questionable creditworthiness to a safer German bank. Suppose that instead of needing to borrow, the Irish farmer, being flush with EU milk subsidies, has enough cash in his bank account to pay for the tractor outright, and just writes a cheque in favour of the German tractor dealer (in the case of deposit flight, the farmer would open a German bank account and write himself a cheque).
This transaction would result in the farmer's Irish bank current account being drawn down and the same sum credited to the tractor dealer's German bank current account, while the Irish bank would make a payment to the German bank via the interbank fund transfer system that settles in their respective current accounts at their central bank.
In most monetary areas, both banks would hold their current accounts at the same central bank, and settlement would involve a simple reallocation of current account credit from the payer bank to the payee bank, with no change in the nature of the central bank's liabilities. In a regionally distributed central bank system, however, like the Federal Reserve System or the Eurosystem, banks deal only with their regional central bank. In this case, the counterpart of the decrease in the payer bank's current account balance, which represents a liability of its regional central bank, is a negative contribution to that central bank's cumulative position with the rest of the central bank system, while the counterpart of the increase in the payee bank's current account balance is a credit to its own regional central bank's cumulative relative position. In the Eurosystem, these relative positions are recorded as balances with the ECB in the interbank fund transfer system TARGET2 (
Trans European
Automated
Real time
Gross settlement
Express
Transfer system
2). In the absence of an existing cumulative position, the farmer's tractor purchase would generate a TARGET2 deficit at the Central Bank of Ireland and a TARGET2 surplus at the Bundesbank.
Because these bank current, or reserve, accounts at their central bank play a vital role in monetary policy operations, how banks accommodate interbank payments depends on how monetary policy is conducted in their currency area. Since these indirect effects are central to the debate stirred up by Sinn, an understanding of monetary policy operations is necessary to evaluate the arguments, so some principles and details relevant to the present discussion are explained here. Readers who believe that they are already familiar with monetary policy operations may skip the following four paragraphs. For a more complete
non-technical explanation, refer to the first half or so of my "
Easing understanding" post.
Positive balances in banks' current accounts at the central bank represent a secure and liquid asset and are known as "reserves". However, any interest rate paid on reserves is typically relatively low, and is not always paid on the whole balance. Banks therefore usually try to minimise their holding of reserves, subject to a constraint provided either by a ban on overdrafts or by a requirement to hold an amount of reserves representing some minimum fraction (in the eurozone,
2%) of deposits for regulatory reasons, leading banks to aim to hold on average a slightly higher level of reserves than the constraint to allow for the unpredictability of payment flows.
Monetary policy ensures that the aggregate stock of reserves held by the banks is as large as needed, and the banks distribute this stock between themselves via a money market wherein banks with surplus reserves lend to those with a shortage, typically for one day at a time. As any transaction between the banks and the central bank is also settled with reserves, the central bank can use designated transactions to adjust the overall stock of reserves. And because there is, other things equal, a
one-to-one correspondence between the stock of reserves and the
market-clearing interest rate in the money market (assuming a downward sloping reserves demand curve), the implication is that the central bank can also control that interest rate, and by arbitrage,
short-term interest rates generally. In practice of course, central banks generally choose to explicitly set the interest rate rather than the quantity of reserves, as this approach best accommodates the impact of
non-policy variations in reserves supply and demand, but note that this equivalence means that it is immaterial to the present discussion whether the central bank uses its transactions to set the stock of reserves or a money market interest rate (contrary to what some of Sinn's
critics have argued). The type of transaction that central banks normally use for marginal adjustment of monetary policy is to trade debt; in particular to offer to lend to, or occasionally borrow from, the money market at a prescribed policy interest rate. These are known as open market operations (OMOs). In the eurozone, monetary policy is set by the ECB as the monetary area's central authority, but monetary policy operations are conducted by the national central banks. The policy interest rate is the ECB refinancing rate, charged on
week-long loans offered in the ECB's weekly main refinancing operation.
Naturally, if there are significant disparities in creditworthiness between banks, the money market may make allowance for this by requiring less creditworthy borrowers to pay a higher interest rate to compensate lenders for the greater risk of loss. Alternatively, loans may be secured in the form of repurchase agreements (repos) to practically eliminate credit risk. To ensure access to such loans as needed, banks tend to hold some debt securities which are readily acceptable as repo collateral, notably government and
government-guaranteed bonds. In a repo, the lender's objective is to take sufficient collateral to make the joint probability of the borrower defaulting and the collateral then being worth less than the value of the loan negligible. It follows that more collateral should be required if the value of the collateral offered is volatile and also that collateral with value that is positively correlated with the fortunes of the borrower is inferior. The conventional way of arranging this is to match the value of the loan with an amount of collateral valued using its market price reduced by some percentage known as a "haircut" specified for that type of collateral, with larger haircuts for collateral with a more volatile market price. Central bank lending is normally collateralised, allowing the central bank to set monetary policy by offering loans at a single rate to all counterparties without discrimination.
Although the importance of monetary policy means that OMOs tend to attract the most attention, the archetypal transaction between the central bank and banks is actually the supply of banknotes. While the aggregate stock of reserves is always greater than zero, in normal market conditions banknotes are by far the largest liability on the central bank's balance sheet, as banks draw down their reserves to pay for banknotes to meet customer demand. This demand for banknotes tends to be fairly stable, typically growing roughly in line with nominal economic activity. The result is that the central bank is structurally a net lender to the money market, regularly renewing its loans as they mature, which provides frequent opportunities to adjust interest rates. Simply renewing less than the full amount of loans maturing would usually be sufficient to adjust money market interest rates upwards; only rarely, between such opportunities, might the central banks need to borrow in the money market to adjust interest rates. As will be explained later, it is this structural central bank position as a lender to the money market that leads Sinn to argue that the size of the "stealth bailout" is limited, on the presumption that the ECB would be reluctant to become a large routine money market borrower.
The reader should now be in a position to appreciate the indirect effects of the payment generated by the Irish farmer's tractor purchase from Germany.
Assuming that the Irish bank had been managing its balance sheet to minimise its reserves holdings, in the absence of any other flows, the payment would leave the Irish bank with less reserves than it needs, because the drain on its reserves is equal to the contraction in its demand deposits and a bank's holding of reserves is typically a fraction of its deposits. The Irish bank therefore might be expected to borrow in the money market to replenish its reserves. The German bank, meanwhile, would be left with excess reserves. In the eurozone, although interest is paid on officially required reserves, no interest is paid on amounts in excess of this; the ECB does provide an overnight deposit facility, but this is intended for emergency use and, at present (
since 13 May 2009), yields ¾% less than the refinancing rate. The German bank therefore might be expected to lend reserves in the money market to earn a return on its excess holding. Note, however, that the sizes of the Irish bank's shortage of reserves and the German bank's surplus, though still equal, are a little less than the value of the tractor, because the Irish bank has lost deposits and therefore needs fractionally less reserves than before, while the German bank has gained deposits and therefore needs fractionally more reserves.
Ideally, it might be expected that a bid for funds by the Irish bank at a slightly higher interest rate than generally prevailing in the money market (assuming, for the sake of argument, that money market loans are secured) reflecting its need for reserves, and an offer of funds by the German bank at a slightly lower interest rate reflecting its excess reserves, would be sufficient to ensure that the money market rebalances via a loan from the German bank to the Irish bank. In that case, the payments imbalance arising from the initial transaction would be largely offset by the opposite payment imbalance arising from the subsequent
cross-border loan (note that interbank deposits are not subject to reserve requirements), without involving the central bank. From an Irish point of view, the contribution of the tractor import towards a current account deficit would be financed by a capital account surplus in the form of a private sector loan from Germany. Moreover, the contribution of the initial transaction to the Irish and German TARGET2 balances would also be largely reversed.
Since the onset of the financial crisis, however, eurozone money market conditions have not been ideal. The creditworthiness of Irish banks was in question right from the beginning, meaning that their money market counterparties became extra careful to ensure that loans to them were well collateralised. In an attempt to reassure lenders to Irish banks and forestall deposit flight,
the Irish government fully guaranteed Irish bank liabilities. Naturally, this undermined the creditworthiness of the Irish state itself, initiating a series of downgrades by the credit rating agencies which prompted the money market to require larger haircuts on Irish government and government-guaranteed bonds pledged as repo collateral. Unfortunately, despite the progress towards an integrated eurozone capital market, there is still a home bias in banks' bond holdings in favour of their own country's government and government-guaranteed bonds. For Irish banks, this has meant that the main repo collateral that they are able to provide is not only less valued in the money market generally, but also less acceptable from them in particular, because its credit risk is positively correlated with their own.
For whatever reason, the ECB treated the financial crisis as a liquidity problem, and chose to
continue accepting Irish government and government-guaranteed debt as collateral on much the same terms as before, as if private sector concern about Irish creditworthiness was exaggerated. The result has been that, for any given refinancing rate, the ECB is the least demanding lender in the eurozone money market in terms of collateral requirements to Irish banks. Irish banks have therefore made heavy use of ECB refinancing operations to replace deposits drawn down either by payments for net imports like tractors, or by capital flight. And in the face of this removal from the money market of a natural borrower of their excess reserves, German banks have instead tended to run down their excess reserves by not fully renewing maturing ECB loans. Consequently, TARGET2 payments from Ireland to Germany have on the whole not been offset by payments of loan capital in the opposite direction, implying a growing Irish TARGET2 liability and a growing German TARGET2 claim.
Of course,
Irish-German transactions are just one example; the same process has led to generally negative contributions to TARGET2 balances in other peripheral eurozone countries with relatively poor bank and state creditworthiness, notably Greece and Portugal, and generally positive contributions to TARGET2 balances in other countries with payments surpluses such as Luxembourg (which, for its size, has a large banking industry) and the Netherlands. In the absence of a full recovery from the financial crisis, this has led over time to the large TARGET2 imbalances highlighted by Hans Werner Sinn. By offering payment deficit countries' banks loans on terms that do not, if market expectations are rational, fully reflect their higher credit risk, the ECB is effectively subsidising payment deficit countries' borrowing and, since the ECB also provides an alternative outlet for the increases in reserves generated by payment surpluses in other countries, the ECB has become a kind of central counterparty for the eurozone money market, intermediating many of its
cross-border loans. The home bias of banks' bond holdings means that to some extent these
ECB-intermediated loans are also indirectly funding payment deficit country governments, which also find it difficult to borrow in the international capital market at acceptable interest rates because of their own perceived poor creditworthiness.
As Sinn and other commentators on this issue have noted, interest at the ECB refinancing rate is payable on TARGET2 liabilities and paid on TARGET2 claims, but this is should not simply presumed to be the consolidated national (ie public sector, including the central bank, plus private sector) return for the countries involved. The designers of European monetary union were scrupulous to ensure that the revenue derived from supplying the monetary base (ie reserves plus banknotes) is pooled and shared out between the national central banks following an agreed formula, rather than according to where banks choose to be formally established (and hence which national central bank they deal with), and the geographical vagaries of the demand for banknotes. According to Article 32 of the
Statute on the European System of Central Banks and the European Central Bank, the sum of the national central banks' "monetary income" (ie seigniorage) derived from "assets held against notes in circulation and deposit liabilities to credit institutions", less interest paid on reserves and operating costs, is "allocated to the national central banks in proportion to their
paid up shares in the capital of the European Central Bank" (ie the "capital key"). For the purposes of this redistribution, the national central banks are required to "earmark" their monetary assets and liabilities according to
ECB instructions which specify that positive and negative TARGET2 balances should be counted as monetary assets and liabilities respectively. The monetary income received by a national central bank is therefore independent of the size of its own TARGET2 balance and the interest rate on that balance. In fact, the consolidated national returns on payment imbalances derive from the way in which they are accommodated by private sector banks. Because the payment deficit bank and the payment surplus bank reset their reserves position by borrowing more and less respectively in the ECB refinancing operations, the return to their adjustments is indeed the refinancing rate. As explained above, however, the size of these adjustments is fractionally smaller than the size of the TARGET2 balances because of reserve requirements.
Although the payment deficit countries may pay the same interest rate on their
ECB-intermediated borrowing as the payment surplus countries earn on their
ECB-intermediated lending, the poorer creditworthiness of the payment deficit countries means that, for them, the ECB refinancing rate is lower in
risk-adjusted terms. In the absence of any major bank defaults in the payment deficit countries so far, the subsidy provided by the ECB has been in terms of the expected cost of the credit risk it is bearing without compensation. It is therefore important to know where that risk ultimately lies. As now seems to be well understood by all the participants in this discussion (if it ever was misunderstood), because any losses sustained on ECB OMOs are, like the revenue they raise, shared between eurozone countries according to the capital key, the subsidy is effectively provided by the whole eurozone rather than just the payment surplus countries like Germany. In practice, however, payment surplus countries' share of any losses is likely to be slightly higher than their capital key, because in the event of a major loss, the most vulnerable payment deficit countries might well be unable to bear their share of loss without triggering their own default.
Does the present configuration of TARGET2 balances represent the "
ECB's stealth bailout" of the payment deficit countries as Sinn contends? Though a bit polemic, the description is not unreasonable. Firstly, ECB lending is shielding those countries from the full market cost of their accumulated payments deficits. Without ECB lending, to cover its payment deficit in the eurozone money market a country like Ireland would have to sell assets to foreigners at whatever price can be obtained, such as
long-term debt or even gold reserves as suggested by Sinn, either to generate incoming euro payments or to raise money to buy safer collateral to support its money market borrowing. This effective increase in the interest rates faced by that payment deficit country would be expected to reduce its domestic demand, for instance for tractor imports, and thereby choke off its payment deficit. And the present ECB policy is not simply a case of monetary easing in response to overall eurozone economic conditions helping the weakest regions most. Because the ECB is more willing to accept the payment deficit countries' government and
government-guaranteed debt as loan collateral than the market, and because such debt tends to be disproportionately held by domestic banks, the ECB is effectively charging these banks a preferential
risk-adjusted interest rate, in violation of (at least the spirit of) the principle that "the Eurosystem’s monetary policy operations are executed under uniform terms and conditions in all Member States", as stated in Chapter 1 of the
Guideline of the European Central Bank. Secondly, while probably not deliberately so, this is also arguably a stealth bailout, because it derives from a technical decision to relax the collateral standards ostensibly for all OMO counterparties rather than accept the implications of lower credit ratings, and the details of how the bailout works are manifestly (ie judging by the protracted debate raised by Sinn) difficult to understand.
Sinn is also correct that the diversion of ECB refinancing towards payment deficit countries is constraining credit creation in Germany. German banks are not constrained in their access to as much base money as they want at the ECB refinancing rate. However, this interest rate is set to meet an inflation objective for the whole eurozone, so to the extent that the relaxation of the ECB collateral rules marginally eases monetary policy in the payment deficit countries, monetary policy must be marginally tighter elsewhere. In other words, the refinancing rate is set higher than it would be if no collateral concessions were made to payment deficit countries. This, given the
one-to-one correspondence between the money market interest rate and the stock of reserves, implies less credit creation in Germany. However, considering the present divergence of the strength of economic activity between the payment surplus countries like Germany and the payment deficit countries like Ireland, somewhat tighter monetary policy in the payment surplus countries may well be appropriate.
And as Sinn argues, the size of the stock of ECB refinancing does constitute a restraint on the total size of the TARGET2 deficits. Once the payment surplus country banks have paid down their entire initial share of the stock of ECB refinancing (the liability counterpart to their required reserves then being customer deposits only), it ceases to be possible for them to invest the reserves inflow from payment deficit countries any further in this way. Although the ECB's overnight deposit facility in theory provides an unlimited
interest-bearing outlet for excess reserves, the low interest rate offered on such deposits could be expected to prompt payment surplus country banks to either resume lending to payment deficit countries at interest rates they will accept or even to deter unprofitable potential depositors by measures such as by raising bank charges. Either response could be expected to slow the increase in TARGET2 imbalances. Naturally, if the ECB provided a more competitive deposit facility, such as by offering
fixed-term deposits at interest rates close to the refinancing rate, this restraint on the TARGET2 imbalances would be removed. Indeed, to meet any further expansion in the payment deficit countries' borrowing at the refinancing rate without easing monetary policy, the ECB would have to borrow money from somewhere.
Finally, assuming that the present chronic TARGET2 imbalances are problematic because the obscure subsidy involved is undesirable, how can they be constrained? Clearly, in a monetary union,
inter-regional payments cannot simply be suspended. The most suitable measure would be to withdraw, perhaps in stages, the borrowing subsidy that sustains the imbalances – that is, to bring the ECB's collateral standards in line with those in the market. Thereafter, to avoid a renewed
build-up of risk in future, some soft limit on TARGET2 deficits, say as a percentage of national GDP, could be set, beyond which a national government is expected to take some action of its choice to inhibit further increases in its country's deficit. If the burden that this would involve is considered unreasonable for any particular payment deficit country to bear, it would be more appropriate for the rest of the eurozone to explicitly either grant that country's government the money to solve its problem by, for example, recapitalising its banks, or lend that government (more) money at clearly subsidised interest rates, such as via the European Financial Stability Facility.
Addendum on July 12th 2011
To make the relationship between TARGET2 payments and the adjustment of banks' usage of ECB refinancing operations clearer, a numerical example of the changes in the balance sheets of the institutions involved may be helpful. Continuing with the example of an
Irish-German interbank payment motivated by either a tractor purchase or deposit flight, Figures 1 to 3 below show the balance sheets of the Irish payer bank, the Central Bank of Ireland (CBI), the German payee bank and the Bundesbank at various stages in the process. Figure 1 shows the situation before the payment is made, Figure 2 shows the effect of the payment alone and Figure 3 shows the balance sheets after the banks have adjusted their reserves position to accommodate the payment. In practice, since
day-to-day overdrafts at the central bank are not normally permitted, banks would adjust their reserve position at the same time or even in anticipation of the payment being processed, so Figure 2 is somewhat hypothetical, except perhaps briefly
intra-day.
The figures in this stylised example may be taken to be in some notional monetary unit like billions of euros. The German bank and central bank are, in balance sheet terms, about twice the size of the Irish bank and central bank. In the example, the interbank payment represents half of the existing stock of deposits held at the Irish bank. For simplicity, it is assumed that no TARGET2 balances exist before the payment is made. Both private sector banks and central banks have some equity which is invested entirely in securities which could be used as loan collateral as necessary. The only type of monetary asset held by central banks is refinancing loans to the banks. Although banknotes may have originally entered circulation by being purchased by the banks to meet customer withdrawals from their deposits, and were paid for at the time out of the banks' reserves borrowed from the central bank, it is assumed that all banknotes are now held by the public. The banks are subject to a reserve requirement of 20% of their deposits. In case it is not obvious, "secs" is securities, "depo" is deposits, "refi" is ECB refinancing loans, "res" is reserves and "o/d" is an overdraft in a bank's current account at the central bank.
Addendum on December 9th in reply to umarmung: Why debt default need not necessarily lead to euro exit
On 16 November, in response to a
comment I made on Olaf Storbeck's Economics Intelligence blog, commenter umarmung asked why I believe that Greece could remain in the eurozone after defaulting on its government debt, and whether banknotes can provide an unrestricted channel for capital flight, as suggested in an
FTalphaville post. Since my answer became quite long as I considered it, and since it may interest others who might not read the comments, I include it here as an addendum, rather than in the comments.
Thanks for your appreciation and good questions, umarmung. Sorry to be so slow to reply to your comment. I needed to
re-read the information on the
inter-regional settlement and banknote management arrangements in the eurozone before I could have confidence in my opinion.
The reason why I say that a eurozone member country which defaults on, or more realistically, restructures its government debt need not give up the euro is that such a credit event does not necessarily disrupt that country's participation in the euro payment system, including
inter-district settlement via TARGET2. As described in my post, any change in a country's TARGET2 balance is matched by an incipient change in the reserves liability position on its central bank's balance sheet (Figure 2 in the post). For private sector banks, reserves are an asset, which banks collectively obtain from their national central bank by selling debt (ie borrowing) secured by pledging a sufficient amount of eligible collateral to make that debt effectively
credit-risk-free. It follows that TARGET2 balances are
credit-risk-free. For example, if trade drains euros from Greek banks to German banks, the TARGET2 liability incurred by the Bank of Greece displaces some of its reserves liability, but the assets exchanged for the displaced reserves remain. As long as the ability of a country's banks to find sufficient eligible collateral to obtain reserves as needed is not impaired by their government's default, that country can remain a fully functioning member of the eurozone. In principle, a eurozone government default need not threaten its country's membership of the eurozone much more than the bankruptcy of one its largest private sector companies.
In practice, however, because of the
home bias of EU banks' government bond holdings, especially in the peripheral eurozone countries most at risk of default, a government default is likely to substantially depreciate, if not wipe out, the value of the collateral that its country's banks normally provide for central bank loans. Nevertheless, it may still be possible for the country's banks to carry on making euro payments. If a bank has sufficient capital to cover the loss on its government bond holdings, it may be able to supply supplementary collateral to top up the cover for its loans from the central bank. If the bank cannot provide sufficient new collateral, and cannot repay its loans from the central bank, the central bank is supposed to be able to liquidate the collateral already given by the bank to recover what it owes. Unfortunately though, the ECB's failure to tighten collateral standards and haircuts enough to reflect the true risk of eurozone government defaults means that the central bank is likely to suffer a credit loss in the event of bank failures triggered by a government default. This loss is shared by all the eurozone members in proportion to their capital key, but since the loss is then a bygone, there is no reason why the defaulting country's surviving banks should not continue to use the euro. Government defaults tend to be selective such that the government concerned keeps making some priority payments, and unlike companies, bankrupt governments are not liquidated, so that, if a defaulting government is determined that its country should remain in the eurozone, it may nevertheless be able to find the resources – eg by using its gold and foreign exchange reserves – to recapitalise its banks to allow them to replenish their stock of eligible collateral if necessary.
Of course if, regardless of such mechanisms, depositors do associate government default with withdrawal from the eurozone and the introduction of a weaker national currency, a growing likelihood of default can be expected to prompt capital flight from nominal assets like bank deposits. In that case, the banks have to find sufficient collateral to allow departing depositors to be paid off as long as the run lasts. How much money is required to ride out this run depends on the stock of bank deposits, the proportion of them which are demand or maturing deposits, and how long the run persists. Naturally, if the banks run out of sufficient eligible collateral and even the government cannot provide more, then in effect, the whole country, and not just its government, is bankrupt. Even then, the country could continue using the euro with exchange controls that restricted the value of outgoing euro payments to no more than the value of incoming payments.
I think the issue of banknotes is a red herring. As Clemens Jobst describes in the
VoxEU column mentioned by lostgen above on 22 July at 01.04, the balance sheet effects of
inter-district fund transfer by banknotes are very similar to those of deposit flight cleared through TARGET2. The small difference is that, instead of a TARGET2 liability (assuming no previous TARGET2 balance), the central bank of the country suffering capital flight, say Greece, incurs an extra liability to the eurosystem to add to that country's allocation of eurozone banknote circulation (equal to its capital key multiplied by 0.92 to allow for 8% of the banknote circulation to be allocated to the ECB itself) to reflect the fact that that central bank has issued more than its share of the eurozone's banknotes. Unless these banknotes are paid into commercial banks, and in turn into central banks, elsewhere in the eurozone, an offsetting adjustment – in their case an asset – appears on each of the other eurozone central banks' balance sheets in proportion to their banknote allocation key. If all the banknotes are paid into the banks and then the central bank of one eurozone country, say Germany, and redeemed, the resulting reduction in that country's net banknote issuance means that its central bank acquires the whole of the offsetting claim on the eurosystem on its balance sheet. I certainly think that the idea given in the alphaville post you mention that "Greece could theoretically continue to issue unrestricted amounts of euro banknotes" to effectively sustain eurosystem funding in the face of capital flight is wrong. Just as they lose reserves when making an
inter-district payment via TARGET2, banks have to pay in reserves for the banknotes that they draw, and would still need to pledge eligible collateral when borrowing from the Bank of Greece to replenish those reserves.
It is of course true that, if the ECB permits the government bonds of a dangerously indebted or even defaulted country to be pledged as collateral in eurosystem money market operations, that country can fund itself effectively without limit via the eurosystem by selling its government bonds to its banks which then borrow from the eurosystem using these bonds as collateral at their national central bank. But this would clearly be an abuse of the eurosystem which the other members would be foolish to allow to progress very far, and could stop by either capping or suspending the acceptability of the offending country's bonds as eligible collateral.
Hopefully that explains my remark on Economics Intelligence and answers your questions satisfactorily.