A leaky bucket ….

2010 and all that — Relitigating the Greek bailout (Part 2)


Part 2 of a series — read Part 1 here

In Part 1, we looked at the financial stability and bank bailout aspects of the 2010 Greek package. It was important to deal with these first because, as we’ll see below, the bank bailout element was very important to the overall package. I’m also going to, in some cases, amalgamate the 2010 and 2013 packages when it makes more analytic sense to do so. The current part of this multi-part post (it is now going to stretch to at least three sections …) aims to deal with probably the most important question of all if you’re relitigating the 2010 bailout, which is:

Part 2: Could the package have been structured to provide more primary deficit financing for Greece?

I think the best way to start thinking about this is to look at a really clever and intriguing argument due to Simon Wren-Lewis. I don’t totally agree with the conclusion (spoiler alert!) but it certainly got me thinking when I first read it, and going through it in detail is going to help set up the analytical framework. It’s actually a simple enough point to make in a couple of paragraphs:

A very interesting argument that set me thinking about the whole subject

On the numbers, this looks pretty plausible to me (the IMF might even have contributed a bit more, as its total promised contribution to the 2013 package was EUR19.8). To add a little bit of detail, here is a breakdown of the Greek primary deficits for the years 2010–2013:

It was surprisingly difficult to get this breakdown. I found the lead for it here, so thanks due to Simon Davis

It fits. (Note that for 2013, excluding the bank bailout cost, Greece actually ran a small primary surplus. This is something I’ll come back to in part 3). So on this basis, it looks like Simon Wren-Lewis’ implicit point might be right — if Greece had defaulted entirely in 2010, meaning that it only needed to finance the primary deficit, it could have taken the IMF money, ignored the rest of the troika and had a no worse outcome than it actually had. Or if the European “partners” could have been persuaded to cough something up for the cradle of democracy, it could have had a much smoother adjustment path and less loss of output. Seems too good to be true, doesn’t it?

Unfortunately, I think it is. Ignore questions like “what sort of conditionality would the IMF have demanded on its own?”, and “would the IMF really be prepared to make such a massive over-quota loan if it didn’t have the Eurogroup alongside it?”. The issue is the actual cost of the bank bailout.

As you can see from the table, the bank bailout costs are already the big driver of the financing need in this counterfactual. But the costs shown here understate that financing need substantially, mainly due to an issue of accruals versus cash.

By which I mean, the ESA95 deficit, correctly, counts the cost of a bank recapitalisation as (oversimplifiying) the amount of money disbursed from the Hellenic Financial Stability Fund, minus the estimated value of the securities received — think of these as the guesstimated future privatisation proceeds. That’s the only sensible way to account for it.

But, pretty obviously, this isn’t the amount of money that needs to be borrowed. The actual loans into the HFSF were EUR48bn. And it gets worse. As you can see, the big bank recap cost came in 2013, after Private Sector Involvement (PSI) wrote down the value of their GGB holdings by 75% (I don’t know why it hits in the year after the PSI took place). But if Greece had taken the hardcore default option in 2010, the losses would have been 100%, and they would have happened on the EUR39bn of holdings in that year, not the EUR26.6bn that they had been reduced to (per the Merler & Pisani-Ferry dataset) by the beginning of 2012. On a crude envelope calculation, taking (39.1 — (26*.75)), this adds nearly EUR20bn to the cost of the bank bailout.

So the actual breakdown of the financing needs of this strategy looks more like this:

All these tables are in EURbn, if I forget to do the labels, btw

So even if you can get the whole EUR50bn from the IMF, you have a financing gap — on what I’d consider a more realistic EUR30bn, you need EUR53bn from the troika. And nearly all of this needs to be disbursed right up front as it is mainly to deal with the immediate consequences of default.

But … when you compare these numbers to the actual EUR272bn actually committed in the two packages, you can see the basis for this critique of the program. It very much looks like the same effects could have been achieved with rather less than half the money — or alternatively, that the same money could have been spent better, in a way that could have provided nearly 50% of Greek GDP in additional primary deficit funding.

These are from the Commission website — I don’t know why the IMF column adds up to EUR50bn

Soooooo … I think this shows why I’ve been referring to this as “the intelligent critique”. It might be right. I think there are a few more complications though, before we can declare the relitigation over and the case proved beyond reasonable doubt. Let’s now set up the analytical framework I promised earlier.

The analytical framework

Basically, after going through the work on SWL’s argument above, I think it makes sense to split the bailout into three components, in increasing order of size:

1) Primary deficit financing for Greece (roughly EUR16bn).
2) Financing for a bailout of the Greek banking system (roughly EUR 50bn).
3) A commitment to “nationalise” Greece’s outstanding debt and restructure it at some future date. (roughly EUR200bn).

This was — and this is the kernel of truth in what I regard as a fairly misleading rhetorical point — not really a deficit funding program for Greece. It was a small deficit funding program, combined with a large bank bailout program, combined with a huge assumption of debt. That debt was later restructured on NPV-reducing terms, and in my view is likely to be further reduced, but I think that falls outside our current topic.

So the question we’re really addressing is whether some of the commitment made in 3) (the debt nationalisation program) could be transferred into 1) (the primary deficit financing program). Doing this would reduce the up-front austerity, smooth the adjustment path and in all probability allow a somewhat better outcome for Greece.

Let’s pretend that “normal” Greek GDP is of the order EUR250bn — we can then say that every EUR10bn we can shift out of one category into the other allows roughly a percentage point of GDP in terms of easing of austerity over the four years under consideration. That would suggest that if you could shift even a quarter of it, then you would be making a really material difference to the deficit adjustment path.

But there are complications …

First, there is the issue I alluded to above, of the timing of the disbursement. Two hundred billion euros is a lot of money — it’s about 6–7% of German GDP and about 1.5% of GDP for the Eurozone as a whole. Coming up with that money in a single year is a lot more difficult than promising to disburse it over a period of time.

Second, as we discussed in Part 1, the creditor economies might themselves have been affected economically by a Greek default — depending on your view on systemic risk — and would very likely have been affected politically. Going to the Bundestag for a program aimed at keeping Greece from defaulting was difficult enough in 2010. Going there in 2013 once they had defaulted was more difficult. Writing a big cheque in 2010 for a Greece that had just unilterally defaulted would have been a lot harder.

And finally, the financing needs in my table above are very heavily lowballed, because default is costly and they don’t make any allowance for the deadweight costs of default. Or more accurately, the “deadweight financing needs”, because what we’re concerned with here is not the actual eventual losses, but the additional amount of up front money that would need to be found in order to deal with the fact that the Greek economic and financial system would be affected by a decision to repudiate the national debt. I’m thinking of things here like:

1) costs of recapitalising pension funds, insurance companies and other entities outside the banking system
2) effects on tax collection rates from the associated political chaos
3) negative effects on investment from the uncertainty and perceived risk of Euro exit

but most importantly

4) liquidity provision for the banking system in a world where GGBs were no longer viable collateral, particularly as this is 2010, with Jean-Claude Trichet in charge of the ECB and he doesn’t even recognise that the ECB has a role as lender of last resort

[A spoiler alert — 4) above is going to figure heavily in Part 3, and Trichet will be taking the place usually occupied by Schaeuble in people’s Pantheon of Eurocrisis Villains. Schaeuble himself will be there too, don’t worry, mainly because 3) above is very much his fault]

There’s a huge amount of uncertainty here, and I don’t want to magic it away by boiling down all of these issues to a single set of guesses. Instead, here’s a ready-reckoner sort of calculator. What I’ve done here is

a) taken the two political points about the political consequences of default, and the difficulties of up front disbursement, and represented them by a single number — “How much up front funding would the Eurogroup contribute to a hypothetical financing package for Greece if it had defaulted in 2010?”. This is on the vertical axis.

b) taken a range of possibilities for the “Financing needs occasioned by bankruptcy” on the horizontal axis. This goes from zero (the “everyone was prepared” view) to EUR200bn (a scenario in which the international community would have had to take nearly the entire role currently provided by ECB ELA).

I’ve assumed as a base case that the IMF would provide EUR45bn, and then subtracted the EUR15.8bn of primary deficit financing actually provided. I’ve then, in every cell, calculated the amount of additional financing available, and expressed it as a percentage of Greek GDP over four years, based on a “true underlying” GDP of EUR250bn/year. Because I’m an analyst, I’ve added some Excel conditional formatting, to show which cases allow material extra funding for the Greek primary deficit, and which cases result in meltdown because the available financing isn’t enough to cover what they actually provided.

If this does not contain at least one serious spreadsheet error, it will be more or less the first time in my life. Expect numerous subtle revisions, and if you want to use the idea, for heaven’s sake do the calculations yourself!

I think you can see a number of things from this table:

1) If you think that the Eurogroup would have been able to put up the whole amount from the two packages as an up front sum in the event of default, then 2010 was definitely a mistake; they could have got a lot more bang for their buck in terms of helping Greece.

2) If you think that the EUR77bn in the 2010 package could have been made available, then as long as you think the additional financing needs caused by default were less than EUR50bn, then you would still be a supporter of the 2010 default option.

3) If you don’t believe any story about additional financing needs as a result of default, then the Wren-Lewis argument more or less goes through — even with no contribution from the EU at all, the package for Greece is within 5% of what they actually got.

4) But if you’re at all scared about systemic risk, then you have to make really strong assumptions about the political availability of European funding to think that default in 2010 would have been a viable solution.

5) And if you’re really scared of it (note that the scale could very easily have been extended to the right, particularly if Greece triggered a liquidity crunch elsewhere in Europe), then the decision isn’t even close. There’s a big risk asymmetry here.

6) As you’d expect, from the fact that I normalised it to show the most interesting ranges, if you take a middle-of-the-road view that the Eurogroup might have contributed as much as EUR100bn for a clean break, and that the deadweight cost of default was in the region EUR75bn, then it probably didn’t make much of a difference.

So my conclusion is that both points of view are defensible, but that the one actually taken by the Eurogroup authorities was very much the more risk-averse one, and involved less up-front contribution. I think this kind of demonstrates that re-litigation exercises like this are always heavily driven by hindsight, but you can also see that the “2010 was a mistake” view is entirely defensible on reasonable assumptions. That’s probably enough for now.

Coming soon … ish … Part 3, in which I tie up some loose ends and look at the political economy of why a bolder strategy wasn’t followed.