An ECB QEsplainer

This is Euroland and nothing is simple in Euroland. But here goes for an explanation of QE, ECB style.

This week the ECB finally introduced its long-awaited Quantitative Easing program (pro-tip for journos and talking heads: Quantit-AY-tive easing is much easier to say than Quantit-It-ive easing.) The transcript of Mario Draghi’s press conference is here and the press statement on the details is here.

The Fed and Bank of England have had QE programs for years so you’d imagine everyone knows exactly how the ECB program is going to work. Alas, this is Euroland and nothing in Euroland is simple. Here’s my attempt to explain what the ECB’s QE program is and what its implications will be.

How Much Will the ECB’s QE Program Buy?

The program will see the ECB and the national central banks (NCBs) of the Eurosystem purchase €60 billion in securities each month. This includes the already-announced purchase programs for asset-backed securities and covered bonds. In terms of how the future purchases will be split between new and existing programs, Draghi has said “What you could look at is basically the past behaviour as an inference for the future behaviour of our purchases.” The existing programs have purchased about €10 billion per month in asset-backed securities and covered bonds, so we should probably expect about €50 billion per month in purchases of new classes of assets.

The purchases will begin in March and Draghi described the intended duration of the program as follows:

They are intended to be carried out until end-September 2016 and will in any case be conducted until we see a sustained adjustment in the path of inflation which is consistent with our aim of achieving inflation rates below, but close to, 2% over the medium term.

If we use the €50 billion figure for additional asset purchases, this translates to a minimum of 19 months of purchases, meaning €1.114 trillion in purchases, of which €950 billion will be “additional asset purchases”.

What Will the QE Program Buy? And Who Will Buy What?

Of the €950 billion in additional asset purchases, 12% or €114 billion be purchases of securities of European Institutions (such as EFSF, ESM, EIB) so the remaining €836 billion will be purchases of national sovereign bonds. The bonds purchased will have a maturity between two and thirty years.

The ECB itself will make 8% of the additional asset purchases (equalling €76 billion, split between €9 billion on securities of European Institutions and €67 billion on national sovereign bonds.) The remaining 92% of the purchases will be done by the NCBs in proportion to their ECB capital key. (This 8% share mirrors the ECB’s role in sharing profits from banknote issuance via the “banknote allocation key”.) This would entail NCBs buying €769 billion in national sovereign bonds and €105 billion in securities issued by European Institutions

Perhaps surprisingly, as far as I can see, there is no clear statement that each NCB will allocate all of its sovereign bond purchases to its own country’s sovereign debt. However, this is what everyone is assuming is the case and, based on this assumption, this table from Frederik Ducrozet summarises the allocation of purchases and compares them with likely sovereign debt issuance this year.

Is This a Big Program?

Well, €1 trillion is about ten percent of euro area GDP but that’s still a good bit smaller than the other major international QE programs. Here’s a nice picture from Bruegel that puts that ten percent in context: It shows the size of various central bank balance sheets relative to the respective area’s GDP. The Fed and BoE QE programs both purchased 25% of GDP and the BoJ’s latest rounds of bond buying will be larger than those programmes. So, for all the talk of shock and awe and beating expectations, this is still, for now, a modestly-sized program.

Are There Limits on the QE Sovereign Debt Purchases?

Yes, though they are fairly mysterious. Firstly, there’s the general area of “eligibility”. Draghi’s statement said the program will purchase investment-grade securities, which would seem to leave out various euro area member states such as Greece. But the press statement says

Securities that do not achieve the CQS3 rating will be eligible, as long as the Eurosystem’s minimum credit quality threshold is not applied for the purpose of their collateral eligibility. Moreover, during reviews in the context of financial assistance programmes for a euro area Member State, eligibility would be suspended and would resume only in the event of a positive outcome of the review.

Don’t worry about what CQS3 means. Basically, this means that if the ECB is willing to accept the bonds as collateral then they can still be purchased as part of QE. And if you have a financial assistance program review, you need to behave yourself. It’s pretty standard ECB heavy-handed moral suasion and mission creep but the overall policy is less strict than it might have been.

Then we have the “quantitative limits to quantitative easing.” The press statement tells us

Securities purchased under the expanded asset purchase programme that are not covered by the ABSPP or CBPP3 will be subject to an issue limit, an aggregate holding limit and other operational modalities specified, in particular, with the aim of preserving market functioning and allowing the formation of a market price on a given security. Moreover, the limits ensure that the application of collective action clauses for a bondholder decision is not obstructed.

But the authors of the press statement were too busy\tired\confused\[Insert your own explanation] to actually tell us what these issue limits were.

This left it to Draghi to explain:

we have two limits. The first one is an issuer limit, which is 33%, and another one is an issue limit, which is 25%. In other words, we won’t buy more than 25% of each issue, and not more than 33% of each issuer’s debt. The 25% limit, by the way, is the one foreseen in order not to be a blocking minority in the collective action clause assemblies, basically, bond holders’ assemblies, and it’s the basis for us to be able to say, there is going to be pari passu.

The mathematically-minded among you may have noticed something here. If the Eurosystem can’t buy more than 25% of each specific bond issue, then how could they ever get over this level in terms of total holdings? Why would the 33% matter?

The answer appears to be the following. The 25% relates to new purchases, while the 33% relates to total Eurosystem holdings including previous purchases via the Securities Market Program (SMP) and other holdings (such as bonds acquired in the course of bank liquidations.)

Draghi pretty much confirmed this when he said

There are obviously some conditions before we can buy Greek bonds. As you know, there is a waiver that has to remain in place, has to be a program. And then there is this 33% issuer limit, which means that, if all the other conditions are in place, we could buy bonds in, I believe, July, because by then there will be some large redemptions of SMP bonds and therefore we would be within the limit.

This shows existing holdings via SMP are being counted towards the 33% limit. Draghi also suggested that there was nothing new in these rules:

all these are not exceptional rules. They were rules that were already in place before.

Most likely he is referring to the mysterious Agreement on Net Financial Assets (ANFA) which has never been published but features regularly in the ECB’s dark mutterings on monetary financing.

What Does this Mean For Ireland? They’re Already in “Monetary Financing” Troubles, Right?

On the face of it, the 33% rule would seem to limit the amount of Irish government bonds that the QE program would purchase. The Central Bank of Ireland acquired €25 billion in Irish bonds in mind-numbingly complicated circumstances in 2013.They also ended up owning a €3.5 billion Irish government bond issued in 2012 in literally-nobody-understands-this circumstances. Then there’s the SMP, which as of February 2013 meant the Eurosystem had another €14.2 billion in Irish bonds though presumably SMP holdings are now a bit smaller than this.

That’s a lot of Irish government bonds for a country that only has €116 billion in total government bonds. Of course, it could be that the 33% rule applies to the total of all government debt, not just government bonds, in which case the relevant figure would include debts to the official sector, bringing the total Irish public debt up to €176 billion.

A “source close to the Central Bank in Dublin” (usually that means they’re so close they’re inside it) has assured the Irish Times on this issue

There had been concerns in Government circles that less Irish debt would be acquired as a result of large bondholdings in the Central Bank in the wake of the deal to scrap the Anglo Irish Bank promissory note scheme.

However, a source close to the Central Bank in Dublin said the deal would not undermine Ireland’s eligibility to benefit in full from the ECB plan.

It’s a pity that a program that is supposed to be transparent has started with under-written press releases and anonymous sources providing unsubstantiated assurances. The ECB should explain these rules clearly as soon as possible.

Can the ECB Really Force National Central Banks to Buy Sovereigns?

This is a bit less clear than you might think. On the one hand, Draghi was very keen to explain that this was the Governing Council’s plan and it was being directed from Frankfurt, e.g.

The modalities, the amounts, the rules, the limits that you just asked me about have been decided here in Frankfurt.

The Governing Council is the sole decision-maker, and the decisions are meant to affect monetary and financial conditions across the whole euro area.

This still seems to me to fall short of saying that the ECB has decided to force an NCB to buy exact amounts of specific securities. Most likely I’m wrong and the Bundesbank is about to start buying very large amounts of German government bonds but it seems like there might just be the slightest amount of wiggle room here, e.g. the Bundesbank could decide to use its QE allocation just to buy covered bonds.

In terms of legalities, I think the ECB can probably direct any NCB to do as it asks. Article 14.3 of the ECB Statute states

The national central banks are an integral part of the ESCB and shall act in accordance with the guidelines and instructions of the ECB. The Governing Council shall take the necessary steps to ensure compliance with the guidelines and instructions of the ECB, and shall require that any necessary information be given to it.

That seems very broad to me. So I believe the ECB can direct NCBs to make specific purchases. That’s not quite the same thing as saying that it will.

How is Risk Being Shared? What About Profits?

This was the big issue that generated lots of debate prior to the announcement, most of it best ignored. As flagged, any risk from the “additional asset purchases” will not be shared apart from the 12% of purchases used to buy debt issued by European institutions. The ECB will also be making 8% of the purchases under the program and it is jointly owned by the NCBs, so effectively this means 20% of the purchases will feature risk sharing.

However, the vast majority of the purchases — national sovereign bonds bought by NCBs—will not have risk shared. Thus, if a national government defaults, the NCB that purchased its debt will not be compensated by the other central banks.

On the flip side, no risk sharing also means no profit sharing. The ECB already has some history with programs that don’t feature risk-sharing, with Emergency Liquidity Assistance (ELA) being the most obvious one. And we know that the profits from ELA operations are not being shared around the Eurosystem. For example, the Central Bank of Ireland had a very large ELA program until 2013 and it passed large profits from this operation back to the Irish government.

Draghi was asked somewhat tentatively about this at the press conference:

Can you give us an idea of what will happen with this money actually that is going to come from those securities paid on a monthly or a yearly basis? I mean is that going to be held as reserves by the national central banks, and can it be ruled out completely that it might be used in the future for possible fiscal accommodation?

Draghi gave a non-response, ruling out the idea that profits passed back from NCBs could allow for fiscal easing. But Draghi knows well that the way NCBs operate is that profits are repatriated to central governments. A QE program will substantially increase the profits earned by NCBs and there is no reason to believe that this won’t increase remittances by NCBs back to governments. See Eric Lonergan’s post on this for some numbers relating to Italy.

So the Eurosystem Isn’t Going to Be a Senior Creditor?

No it’s not. But be careful here. The Eurosystem wasn’t a senior creditor in Greece either. They bought the same kinds of bonds that all other investors did and the legal terms of these bonds didn’t give the Eurosystem any special status.

The Eurosystem avoided losing money on the SMP purchases of Greek bonds because Trichet’s ECB was able to use their significant leverage (via their role in threatening to cut off loans to Greek banks) to get themselves a better deal.

There is also the issue of whether the ECB can take part in a voluntary restructuring or whether this would be considered monetary financing. The consensus seems to be that voluntary participation in a restructuring would be monetary financing. But there is nothing preventing the Eurosystem losing money on its sovereign bonds in a coercive default or in a restructuring organised via collective action clauses.

This is where the 25% comes in—the ECB doesn’t want to have the power to block a restructuring because they feel they will have to behave as a holdout of those circumstances arise. By staying below 25% for each issue, the Eurosystem will just have to go along with the outcome of CAC negotiations.

Overall, Draghi’s comments on seniority are best seen as a promise not to behave as Trichet did when future debt restructurings are being considered. We are being lead to believe that a Draghi-led ECB will not throw its rattle and threaten to cut off funding to a country that is negotiating a sovereign debt restructuring with its creditors. Amen to That.

Doesn’t Central Bank Ownership of Sovereign Bonds Raise Risk for Private Sector Investors? Won’t This Raise Yields?

This has generated quite a lot of debate. Even with Draghi’s assurances, many people are convinced that, when it comes down to it, national governments won’t default on their own Central Banks. For example, they may issue their NCB with new bonds and then make private investors bear the full brunt of the default. In this way, the QE purchases could raise market yields on sovereign bonds—the exact opposite effect of what is intended.

These concerns have largely focused on the idea that the NCB losing money is a very bad thing for a national government and that the government will have to re-capitalise the central bank afterwards. A couple of points on this:

1. There isn’t anything in European law that says that NCBs have to have a particular level of capital, i.e. have to have assets exceeding “liabilities” in their accounts by some specific amount.
2. A large part of central bank liabilities take the form of banknotes, which are a notional liability with no actual cost. So central bank “capital” is a fairly meaningless number and certainly very different to the corresponding capital for private banks.
3. Indeed, many central banks have carried on for years with assets less than liabilities on their balance sheets without anything bad happening.
4. The main potential item that leads to costs for central banks is making payments on reserves deposited with them by banks. As I discuss below, QE will increase the stock of central bank reserves. However, at present, the Eurosystem charges banks for having money in its deposit accounts, so this isn’t a cost that will be relevant for some time.
5. Finally, as Draghi himself noted, “most central banks have adequate buffers”. The Eurosystem has over €400 billion in capital and reserves and re-valuation account.

But leave all that aside for a moment. Suppose a government did decide that it had to re-capitalise its central bank. How costly would this be? The government would need to provide the central bank with sovereign bonds and this would increase their gross government debt after the debt default. However, the payments from these bonds would increase the profits of the NCB and this increase in profits would translate, on average, one-for-one to higher dividends paid back to the government. So, on a consolidated basis, re-capitalising a central bank has no net cost. It’s a left-hand, right-hand operation.

To summarise on this topic, there is no particular need for governments to recapitalise central banks and, if they had to do so, it would on net be costless. So the idea that it is impossible for a government to default on a central bank (and thus QE increases risks for the private sector) does not hold.

Still, prior to the announcement, there were lots of people saying that a program without risk-sharing wasn’t “full QE” or how the absence of shared risk “undermined the effectiveness” of the program. Most of these arguments weren’t backed up with much in the way of economics. Perhaps what people had in mind was that it would be easier for a country to default on its debt if they knew the money their NCB had created via QE was mainly safely invested in other people’s bonds.

This misses a key point made by Lorcan Roche Kelly: If the bonds were shared around the Eurosystem, that would greatly increase the chance that the Governing Council would use its powers to act as a de facto senior creditor (as happened in Greece). All told, this is probably the safest realistic option for private creditors concerned about default risk.

Is this Effectively Monetisation of Sovereign Debt?

It depends. As long as the QE bonds are held, the interest payments will end up shuffling back to government, making the bonds effectively cost-free. When the bonds mature, if the Eurosystem decides that the money printed to buy them should be “retired” then the government will need to raise private sector funding to replace them and the debt will ultimately not be fully monetised. Alternatively, if the Eurosystem chooses to be buy more government debt when its purchases mature, then the debt will have been monetised.

Any decisions on rollovers is a while off. With a minimum maturity of two years, it will be at least March 2018 before a QE bond matures. At that point, the ECB Governing Council may still be doing new QE purchases, in which case NCBs will buy replacements for the maturing bond. Alternatively, the ECB could decide to stay in a “holding position” and keep its stocks of sovereign bonds fixed for some time, which would still involve purchasing a new bond to replace those maturing.

If the US or UK experiences are anything to go by, it seems unlikely that the ECB will be retiring any of the money created in this program for quite some time.

How Could QE Boost GDP and Inflation?

Perhaps surprisingly this many years into the QE experience, I have heard a number of commentators discuss how the money created via QE will produce additional bank lending. It won’t. At least not in any direct fashion.

The basic money multiplier simply doesn’t describe how actual banks behave. European banks are being very careful with bank lending for a whole host of reasons (Basel 3, concerns about credit risk, caution based on past mistakes) and they don’t behave like the banks in Principles of Macro who seek to constantly satisfy minimum reserve requirements. Instead, they set the size of their balance sheet with capital adequacy requirements in mind and QE has little impact on this. Note the US money multiplier plummeted when QE was introduced and is now well below one.

So what does QE do? Lots of careful research (such as this) has been done on the effects of the Fed’s QE program. Ben Bernanke has joked “The problem with QE is it works in practice, but it doesn’t work in theory.” And, indeed, QE really isn’t something that had to work but the evidence suggested it did reduce yields on Treasury bonds and mortgage-backed securities. At the same time, it didn’t reduce the targeted yields by a huge amount so it’s best not to over-hype its potential in Europe based on the US experience.

Still, QE may do a bit more to reduce sovereign yields in Europe because there is a margin that it can work on that didn’t exist in the US: Default risk. Unlike OMT, it’s not a specific program designed to stop euro area member states from defaulting but it’s another sign of Draghi’s “whatever it takes” commitment and we may well see sovereign yields of high-debt countries decline by more than the effects estimated by the Fed studies.

So there will be some effect in reducing sovereign yields. This will reduce costs for euro area governments and may get passed on a bit to private sector lending rates. The net effect of these lower yields on growth and inflation will be positive but probably fairly small.

Probably more important than the kind of “liquidity premium” effect generally stressed in the Fed research is that QE provides a new kind of “open mouth operation” to influence expectations. It provides a way of structuring policy so that markets can see that you really are not going to raise rates soon—the ECB has always had such a ridiculously high level of “credibility” that many market participants have consistently assumed that it is only a small upward tick in inflation away from getting “vigilant” again (a la the rate hikes of 2008 and 2011).

QE thus signals the following: “We’re not going to increase rates until our QE program finishes. And watch this space, we might even increase the pace from €60 billion per month if we’re not happy with inflation developments.” These signals help increase the credibility that Draghi is now concerned about: The credibility of the ECB’s commitment to get inflation back close to two percent. This should contribute to raising inflation expectations, which can pave the way for an increase in actual inflation.

The immediate impact of the QE announcement in triggering a long-overdue swoon in the value of the euro suggests that the program is already having some of the desired impact.

So what will be impact on the economy? Modest enough, I would guess. But it crosses a Rubicon of sorts and leads to the possibility of even more “radical” policies later. To go into cliche overdrive, it may be too little and it’s definitely too late but better late than never.

How Will This Affect the Balance Sheets of European Banks?

Finally, there has been very little commentary on how QE will afffect the balance sheet of the euro area’s banks. The ECB will buy very large amounts of securities. (“From whom?” you might ask but there’s always turnover in financial markets so you can always find a seller.) It will pay for the bonds by crediting the reserve accounts of the banks that sellers keep a deposit account with.

Thus, the euro area’s banks will soon be flush with deposits with the Eurosystem. And these days, the Eurosystem demand that banks pay for this privilege. At 20 basis points per year on €1.1 trillion, that’s an annual “tax” on euro area banks of over €2 billion per year. They won’t like it. Which could be another sign it’s a good thing.