Sunday, November 20, 2011

The ECB and Last-Resort Lending

There is a lot of talk recently about the possibility that the European Central Bank (ECB) could act as a "lender of last resort" to mitigate the European sovereign debt crisis. The Germans are against it, and the southern Europeans (who might stand to gain from such as policy) are for it. France may be sympathetic.

What is a lender of last resort? A key role for a central bank of course, is in acting as a lender-of-last-resort to the private banking system. The conventional view of banking is that the key function of banks - transforming illiquid assets into liquid liabilities - leaves an individual bank open to runs. According to the standard logic an otherwise sound bank could fail due to an illiquidity problem. Depositors run to the bank to withdraw their deposits under the assumption that everyone else will do so. The bank is unable to sell its assets at their "full value" so as to satisfy withdrawal demand, and it fails. However, a central bank willing to accept the bank's assets as collateral can lend to the bank, allowing deposits to be converted into currency, and this can quell the panic. In a full-blown systemic financial panic, the central bank can extend this credit to the entire banking system.

The key problems for a central bank are in determining what will qualify as eligible collateral for a central bank loan, what the haircut might be on such collateral, and at what rate the central bank should lend. Moral hazard comes into play, and central banks are leery of extending the lives of banks which are actually insolvent and not simply illiquid.

The Diamond-Dybvig model is thought by some to justify a lender-of-last-resort role for a central bank, but that is incorrect. The original model, and its extensions, does not incorporate anything that resembles central banking - indeed the basic model is not monetary (Diamond Dybvig also has no role for deposit insurance, but that's another story). Some things though, such as the the liquidity transformation role of banks and moral hazard problems in banking, I think are well understood.

In general, there are not many complaints about the Fed's lender-of-last-resort role in the financial crisis. Though the Fed may have been overzealous in lending in unconventional ways to unconventional borrowers, mainstream opinion seems to be that the Fed's lending during the crisis was necessary.

But are demands that the ECB play a "lender of last resort" role in Europe simply requests for the central bank to perform its conventional role? The argument seems to be that Italy and Spain, for example, because they are not running primary deficits, are like the bank that is suffering from an illiquidity problem, and not like the bank that is actually insolvent. According to this argument, bondholders are "running" on Italy in the sense that they are demanding very high interest rates. In this sense, an Italian default could be self-fulfilling, just as failure could be self-fulfilling if there is a run on a bank. Then, according to the argument, it makes sense for the ECB to step in and buy Italian government debt - or the debt of any other European government subject to this "liquidity" problem.

But hold on here. In the case of conventional central bank lending, for example as one might envision in response to a pre-Federal Reserve or Great-Depression-era banking panic, the central bank is replacing the liquidity that the public has lost confidence in - bank deposits - with liquidity that it views as roughly equivalent - currency. Is that the case if the ECB buys debt issued by European governments? Well, maybe so. Deposits at the ECB are not quite the same as the safe government debt which the bondholders want, as central bank deposits are not as widely traded as government debt (not all financial market participants can hold an account with the ECB). But, anyone can hold a deposit with a bank in the EU, and banks in the EU hold reserve accounts at the ECB, so this does not seem to be a problem. Thus, it seems the ECB can indeed convert "illiquid" government debt into liquid central bank liabilities.

But to actually quell the panic, the ECB must be able to lower the bond yields on the debt it is purchasing. Is this actually possible? Let's take a look at the current ECB balance sheet. The size of the balance sheet is about 2.3 trillion Euros, as compared to about 1.5 trillion Euros in January 2008. Thus, the expansion in the ECB balance sheet is nothing like the tripling that occurred in the United States, but there are features that are qualitatively similar. For example, there has been an expansion in the debt obligations of ECB member countries held by the ECB(much like QE2, though a smaller intervention) and the ECB currently holds reserves in excess of requirements. On its most recent statement, the quantities apparently in excess of reserve requirements are 144 billion Euros in the deposit facility and 183 billion Euros in term deposits.

These quantities of excess reserves are not as large as in the United States, but I think you can see the effects of these reserves on interest rates in the European overnight market. Recent data shows the overnight rate near the bottom of the interest rate "channel," with the lower bound determined by the interest rate on the ECB's deposit facility, currently at 0.5%. Note, for example in early 2008, that the overnight rate would typically be close to the middle of the band.

Thus, my working hypothesis is that, given a sufficiently large stock of excess reserves in the EU, the interest rate on ECB deposits is determining the overnight rate in Eurpope, much as the interest rate on reserves in the United States is determining short-term interest rates here. Just as in the US then, purchases of government securities by the ECB do not matter, at the margin. The ECB can buy government debt, but in spite of the fact that the debt they are buying may be risky and the liabilities they are issuing may be much less so, the ECB has no advantage over the private sector in intermediating Italian debt, for example. Buying Italian debt will not change the path for prices, and cannot change the prospects for a default on Italian debt.

If the ECB were to lower the interest rate on its deposits, this would indeed raise the price level and allow all EU members to implicitly default on a piece of their debt outstanding. Some EU members obviously want this. Others, like Germany, understand that the reason they can borrow at low rates is because the ECB made a commitment in its charter to price stability.

In any event, just as quantitative easing is currently not a solution to anything in the United States, "lender-of-last-resort" lending by the ECB will do nothing for Europeans.


  1. The trouble with a lender of last resort is that solvent banks don't need one. Insolvent banks, on the other hand, will be just as insolvent after a loan as they were before. What they need is a GIVER of last resort, and all a gift does is transfer the bank's losses from its own shareholders to the poor saps (i.e., Germany) who were dumb enough to act as lenders of last resort.

  2. Stephe,
    I dont understand your argument. Why is it that buying italian government debt does not change the relative price of debt of italian bonds to German bonds? I can kind of see that the ECB can not change prices if government is viewed as one European aerea as you point out for the US, but here we have say two governments and the ECB takes a risky position using German credibility to help italy. Why isnt that a redistribution which might affect relative prices and therefore iralian default probabilities? thanks in advance for a short elaboration

  3. I think the asset swap does not matter, in the current context. It seems to me that the ECB can only move the prices of government debt in the EU by taking actions that will change the price level, which in turn changes the real debt obligations of the Italians, Spanish, French, etc. And the ECB can only affect the price level if it changes the interest rate it pays on its deposits.

  4. "It seems to me that the ECB can only move the prices of government debt in the EU by taking actions that will change the price level"

    if i understand you correctly, the ECB will only have an effect if the purchases are permanent, which is effectively monetizing the debt, thus it will increase the price level?

  5. Could it be that there is disagreement regarding the expected repayment of sovereign debt in those countries? In that case, purchases by the ECB would drive some pessimistic investors out of the market.

    Recent bans on short-selling could be further impairing the ability of those agents to affect bond prices.

    By no means, I am implying that this is a desirable policy intervation.

  6. If the ECB has the power and the resources to buy 100% of Italy's debt, how is it possible that they can't impact the price of that debt?

    If the ECB has the power, the resources and the willingness to buy 100% of Italy's debt, how is it that market participants will not drive up the price of Italy's debt in expectation of those purchases?

  7. Two questions. Let's stipulate that right now there is zero liquidity premium in the marginal demand for reserves in the EU.

    First, why do you believe the ECB can't have an intermediation advantage over the private sector in some circumstances? Why can't the ECB occupy a unique focal point the market's signaling web such that its actions (and mere statements of intent to act) have a much higher probability versus the private sector of jolting the market from a run equilibrium to a stable equilibrium?

    Second, do you dismiss the possibility that central bank actions which currently provide no liquidity transformation might nonetheless have liquidity implications through the expectations channel? Or, in English, let's take the typical New Keynesian type of future commitment from the Central Bank (something akin to "we will be too low for too long.") And let's say we are suspicious that such a commitment can be credible. Isn't it possible that unconventional monetary policy like buy Italian bonds in distress -- even if it provides no current liquidity transformation -- might add credibility to such a commitment? Shouldn't the asset risk they are taking on have implications for their ability to remove liquidity should the demand for reserves recede to the point where they again include a liquidity premium? You might not like the implications of creating inflation expectations through selective credit risk, but why don't you think it should be expected to affect the price level and thus the real debt burdens of Italy?

  8. "If the ECB has the power and the resources to buy 100% of Italy's debt, how is it possible that they can't impact the price of that debt?"

    Yes, if they do that, it certainly matters. Buying some of it at market prices should not make any difference.

  9. DLR,

    Expectations about the future matter of course. However, I think what matters are expectations about future policy actions that actually matter. Unless you think you are in a multiple-equilibrium world where actions with no extrinsic impact can make a difference, I don't see why taking meaningless actions does anything.

  10. "Yes, if they do that [buy 100% of Italy's debt], it certainly matters. Buying some of it at market prices should not make any difference."

    True, but a commitment to buy sufficient amounts to end a self-fullfilling debt spiral can break that cycle, no?

  11. "True, but a commitment to buy sufficient amounts to end a self-fullfilling debt spiral can break that cycle, no?"

    Sure, if they buy it all, then we're in a different ballgame, involving redistribution from Germany, for example, to the southern Europeans. The Italians could, on their own, also provide the fiscal commitment that would end your "self-fulfilling debt spiral."

  12. "Yes, if they do that, it certainly matters. Buying some of it at market prices should not make any difference."

    What if they only buy 99.9% of it, Steve?

    At what point do their purchases stop affecting the price?

  13. Interesting post.

    I second Chris the regression runner's point though. You seem to be trying to have it both ways... that purchases have no difference and purchases create an entirely different ballgame. What makes some purchases different from others?

  14. The marginal buyer determines the price, right?

  15. "Sure, if they buy it all"

    Do they really have to?

    What if the ECB announced it was going tender for an unlimited amount of Italy's debt next week at yields averaging 5%, or even a fixed spread to German bonds? If what Italy is really experiencing is a liquidity crisis brought about only by market concerns regarding it's ability to pay currently high interest rates, this solves that problem. The credit outlook of Italy improves because of ECB's willingness to backstop Italy's rates at manageable levels, credit investors should return, and the ECB may never actually have to buy a single bond.

  16. "What if the ECB announced it was going tender for an unlimited amount of Italy's debt next week at yields averaging 5%..."

    I think that means buying it all.

  17. "I think that means buying it all."

    Then what you are implying, I believe, is that Italy has a solvency problem, not a liquidity problem.

  18. I'm saying:

    1. This is really not the same as a liquidity problem faced by an "illiquid" banking system.

    2. If the problem is that there could be a self-fulfilling default by Italy, the ECB may not have the power to stop it.

  19. Speaking as a credit investor, I would lend to Italy at 4%, because the country's finances are stable at that cost of borrowing. I would not lend to them at 7%, because the country's finances are not stable at that cost of borrowing. At 7% and above and with no prospect for ECB intervention, yields need to reach 20% or higher before the investment makes sense.

  20. steve,

    what do you make of the stated intention of the existing SMP, i.e. to restore a broken monetary policy transmission mechanism?