"Bad times, hard times, this is what people keep saying; but let us live well, and times shall be good. We are the times: Such as we are, such are the times."

Thursday, 20 August 2015


This is the third part of my series on the 2008-9 revision of the Greek deficit and the events that led to it; it considers the specific allegations related to the 2008-9 deficit revision. Part 1 of the series is available here, and Part 2 is available here. Part 4 is underway.

The truth is out there

It's hard to overstate just how much of what goes on in the world (even the unsavoury bits) is a matter of record, if not entirely in the public domain. Conspiracy theorists' claims are, as a result, always much more testable than they realise. This is true of Greece's 2008-9 deficit revision.

Here's what you need to make your way through the revisions:
  • Eurostat's rules on Government deficit and debt calculation are in the public domain. The ones in place at the time of the 2010 methodological visits are recorded here, while the revised rules of 2012 are available here. The European System of Accounts (ESA95) which was in force at the time is explained in detail here. We've since moved on to ESA2010.
  • Eurostat keeps a record of its decisions on how methodological matters should be dealt with, and a record of guidance visits and advice to member states. 
  • All notifications under the Excessive Deficit Procedure (EDP) are available hereThe ones most relevant to the 2008-9 deficit review are April 2008, October 2008April 2009October 2009 (the first one, of 2/10), April 2010 and October 2010
  • It's important to remember that the Oct 2009 Notification was fatally flawed; no one accepts it; not Eurostat, not ELSTAT, not the old ESYE, not the conspiracy theorists. No one.
  • Eurostat's report on the 2010 revision of Greek deficit statistics is available here with a very helpful annex here. Their report on the 2004 revision is available here. And here  (pp 31 and 32) is the commission's breakdown of the revisions made back in 2004.
  • Eurostat has also published reports on their methodological visits to Greece in 2006 (Spring), 2006 (Autumn)  and 2010.
  • The 2010 Greek expert report into the reliability of fiscal statistics is available here (in Greek)
Using the EDP notification tables, it is relatively easy to present a history of our 2008 and 2009 deficit forecasts and estimates, by vintage. All timings refer to the EDP notifications in which each estimate was made, but, since EDP notifications don't go back more than four years , I've also added the figures currently cited by Eurostat. For comparability, these are all ESA95 stats.



As you can see, the flawed Oct 2009 notifications conducted by ESYE, ELSTAT's non-independent predecessor, revised our 2008 and 2009 deficits enormously. Yet, while the Oct 2009 notifications were both severely wrong on many levels, their estimates of the 2008 and 2009 deficits weren't overly high - the figures Eurostat and ELSTAT agree on today, and which I think are methodologically sound, are in both cases significantly worse than what came out of the Oct 2009 notification. This means that, unless one rejects all subsequent revisions, the Oct 2009 notification did not inflate the deficit. Conspiracy theorists can only, perhaps, claim that it drew attention to the deficit at the wrong time. Which is true enough, but not treason.
To help you assess the timing of notifications, I've put together a simple timeline, with help from Wikipedia:

  • 2 October 2009: The first October 2009 EDP notification is submitted, revising Greece's 2009 deficit forecast from EUR9.3bn (3.7% of GDP) to EUR30.1bn (12.5% of GDP).
  • 4 October 2009: PASOK wins the 2009 legislative elections
  • 21 October 2009: A second October 2009 EDP notification is submitted.
  • 22 October 2009: Fitch downgrades Greece from A to A-
  • 8 December 2009: Fitch downgrades Greece from A- to BBB+
  • 16 December 2009: S&P downgrades Greece from A- to BBB+
  • 22 December 2009: Moody's downgrades Greece from A1 to A2
  • 2 January 2010: An expert report into the reliability of Greek fiscal statistics is published
  • 8 January 2010: Eurostat publishes its report on Greek fiscal statistics
  • 9 March 2010: Greece's new statistical law comes into force, establishing ELSTAT as an independent entity.
  • 29-31 March 2010: Eurostat methodological visit.
  • 1 April 2010: The April 2010 EDP Notification is submitted, revising Greece's 2009 deficit from EUR30.1bn (12.5% of GDP) to EUR32.3bn (13.6% of GDP).
  • 9 April 2010: Fitch downgrades Greece from BBB+ to BBB-
  • 21 April 2010: Bailout talks begin
  • 22 April 2010: Moody's downgrades Greece from A2 to A3.
  • 23 April 2010: Greece formally requests a bailout from the IMF, EU and ECB.
  • 27 April 2010: S&P downgrades Greece from A to BB
  • 2 May 2010: The first bailout is signed
  • 21-22 June: Eurostat methodological visit
  • 26 July 2010: Regulation no 697/2010 comes into force, giving Eurostat auditing powers
  • 2 August 2010: A. Georgiou takes over as president of ELSTAT.
  • 27-29 September 2010: Eurostat methodological visit
  • 11 October-9 November 2010: Eurostat methodological visit
  • 15 November 2010: Eurostat publishes Greek fiscal data from 2004-2009 without reservations for the first time.

The origins of a revision target - accounting for issuance

As readers know, I do not deal in 'insider' accounts. I am not privy to what went on within ESYE or ELSTAT except through what accounts there are in the public domain. However, I'm willing to bet that, come October 2009, the old ESYE's leadership knew that producing a deficit figure below ca. EUR30bn for 2009 and EUR20bn for 2008 would lose them the last of the international community's remaining goodwill, and I'm willing to bet they were under orders to reach those levels while still allowing their bosses (first ND, then PASOK ministers) to look reasonably good. The reason was simple, and it was not treachery; foreign analysts could see for themselves how much debt Greece was issuing, and it was way out of line with our deficit projections.

It is not impossible to manipulate deficit statistics in order to exaggerate government deficits; however, it is impossible to manipulate the numbers on bond issuance. Bond issues are, after all, probably the most public financial transactions in the world. Eurostat estimates net issuance (all securities issued minus all securities repaid) of EUR23bn in 2008 and EUR38bn in 2009. It would be hard to believe that our deficits (in cash terms, at least) were orders of magnitude smaller than this.

If you're a conspiracy theorist, you may not trust Eurostat on this matter; but bond-watching analysts' estimates at the time might convince you. By Nov 2009, gross issuance for the year was estimated at EUR59bn by Barclays Capital and later at EUR61.3bn by Danske Bank; and since EUR25.9bn of Greek bonds matured or were repaid over the year, that leads us back to a net issuance of EUR35.4bn for 2009. Similarly, as of Dec 2008, gross issuance for the year was estimated at EUR44bn; suggesting net issuance was way ahead of the projected deficit.

It is, of course, possible for a country to net issue more bonds than it needs to, if it is tactically overborrowing to take advantage of low interest rates. But Greece was raising funds in the teeth of literally the biggest bond sale in human history. It was common knowledge that the debt we were raising at the time would cost us dearly. It is also possible, in a conspiracy theorist's mind, for a government to intentionally overborrow in order to bring about a fiscal crisis, but then the bulk of the 2009 issuance was not carried out by the new PASOK government at all. It was carried out by the outgoing ND government which protested the new estimates and started the 2009 deficit myth in the first place!

In any case, it's worth comparing the EUR35.4bn of net bond issuance for 2009 to the conspiracy theorists' own upper-bound estimate for the 2009 deficit: 7.9% of GDP (based on the Oct 2010 notification) before accounting for what they allege was a major GDP underestimate. This produces an utterly impossible net borrowing requirement of EUR18.5bn - these people truly believe the Greek government had issued some EUR17bn worth of bonds just for fun in 2009.

The timing and impact of revisions

Timing is key to the inflated deficit myth: it is, after all, alleged that Greece's 'falsified' deficit figures triggered our debt crisis, hastened the involvement of the IMF in our bailout, or were used to build support for austerity policies domestically. The goalposts keep shifting, so one needs to be careful.

The first EDP notification to significantly review the 2008-9 deficits was the Oct 2009 notification, which came in two parts - the original notification of 2 October, and a follow-up notification on 21 October. Not only was this not prepared by ELSTAT (which did not exist at the time) - the first shocking set of figures was submitted ahead of PASOK winning the 2009 legislative elections. While the timing doesn't tie in well with conspiracy theories, it does tie in with the narrative of the crsis. As you can see here, our October 2009 notification, combined with Dubai's unfolding sovereign debt crisis, was the signal for foreign banks to up sticks and get out of Greek bonds. Greece had put up its hand to ask for a loo break just as investors were wondering who would be the next sovereign to go bust. This is tactically awful, of course, and had I been a PASOK minister I might have tried to stall as long as possible. But the massive deficit and international scrutiny were both already there; a budget would be drawn up in early 2010 no matter what. Greece could not put this off forever.

While foreign banks were individually heading for the exits post October 2009, in other ways the market was slow to catch on. Greece CDS premia were only beginning to diverge from the rest of the periphery in Q4 2009, and it took credit rating agencies, on average, until Q1 2010 to take Greece below investment grade (graph taken from here). Similarly, Google search trends for the Greek deficit only start to really pick up in March 2010. Throughout all of this, the revision of the 2008 deficit hardly made it into the international financial press at all; it was the dramatic revision of the 2009 deficit that was cited most.

Thursday, 13 August 2015


In this second part to my post, I try to explain the factors behind the decline and corruption of Greek statistics leading up to the 2009 deficit revision, and what it can teach us about Greece, Europe, and the State.

Goodhart's Law

Let's start with the basics: Goodhart's Law. It's reason number #45608 why centrally planned economies do not tend to work:
Goodhart’s “law” [...] stipulates that “any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes” [...] (Koen and Van den Noord 2005)
"when a measure becomes a target, it ceases to be a good measure."
Dame Ann Marilyn Strathern 
Normally, Government deficits are among the most closely monitored government statistics out there, and come with monthly targets; if Goodhart's Law holds, then it makes sense that they should be some of the most prone to manipulation. We don't know how they compare to other statistics but we know they are tampered with, at least in the Eurozone, as a result of political incentives subject to both economic and political cycles, and within the scope allowed by incomplete fiscal transparency. A number of studies demonstrate this conclusively, but I would single out Koen and Van de Noord (2005), Beetsma et al (2011), de Castro et al (2011) or Alt et al (2014) as the most useful. Readers may have others to contribute; all are welcome.

It takes constant vigilance to keep deficit figures relevant and free from interference - in fact, it is a job for institutions of fiscal governance and transparency, not for the occasional do-gooder. You might think Greece's historical record in this has always been poor, but you'd be wrong. As the IMF acknowledges, fiscal transparency was written into Greece's first constitution and elements of it were in place even during the Revolution. This is because the Greek state, born as it was out of war and a concerted Western nation-building effort, was dependent on foreign loans from day minus one. Subsequently, we spent decades under fiscal monitoring and monetary straitjackets of some sort or other (on which, read Tuncer (2009) and Lazaretou (2004)).

The Stability and Growth Pact

But what happens when institutions are as wrong as the people they're supposed to control? In Greece's case, the key institution was the Stability and Growth Pact (SGP) and its infamous 3% deficit target. Pina and Venes (2007) demonstrate that the SGP increased the tendency of governments to flatter their deficit forecasts. Alt at al (2014), moreover, demonstrate that the influence of the SGP reduced headline deficits but increased stock-flow adjustments, and particularly the disguising of deficits as equity injections into state-controlled enterprises. This stock-flow adjustment effect only occurred in countries with low levels of fiscal transparency. Unfortunately for us, Greece's recent record in this regard was poor, whatever our history might have prepared us for.

Differences of degree and of quality

These were in fact Eurozone-wide problems. All studies mentioned so far find the same problems when Greece is removed from the data. So what was special about Greece? For one, there are differences of degree. Tables 4 and 5 here  and Figure 1 here demonstrate that Greece has been an outlier in terms of downward revisions to the deficit figures, ever since 1997, with revisions typically doubling our deficits. The effect of accounting distortions on Greek deficit figures was typically three times the size of the distortions of the next worst-performing country.

As you can see in the graph to the right, the SGP (in effect from 98 onwards) was an effective constraint mostly on Greek governments' planned deficits - these were indeed never above the 3% ceiling. First releases were also subject to SGP; and although these would always revise the planned balances downward (2000 and 2006 were the sole exceptions), the size of the revision was more or less random, or subject to unforeseen circumstances such as the 2004 Olympics running much further over-budget than expected. Such revisions occasionally breached the 3% target. But it was the ex-post reviews, based on methodological visits and Eurostat interventions, that restored Greek deficits to a persistent, downward trajectory. The reason for this is that ex-post deficit figures weren't just about revisions due to random events or
within the bounds of good practice. They were all about gimmicks. Pg 28 here details the full list of accounting gimmicks used in the years leading up to 2005, and it really takes the whole page to go over them.

So what was the hard constraint on our deficits?

If the SGP was not a hard constraint on Greece's true deficits, did politicians see anything as a hard constraint? An OECD review of budgeting in Greece, prepared on the very eve of the crisis and two years ahead of the 2009 deficit revision, is clear on how things worked. Budgeting was a bottom-up, line-by-line as opposed to programme-by-programme process, planning for only one year at a time, leaving almost no role for Parliamentary control and with no provision for ex post oversight. Accounting became increasingly poor as one moved away from central government. Even the OECD had to concede that accrual accounting was not an immediate priority since the state was bad enough at cash accounting. Audit needed to be strengthened. But perhaps most telling is the way the OECD describes the relationship between the SGP targets and the actual budget (see p 14):
"for the medium term (t+2 and t+3), forecasts are done annually for the Stability and Growth Programme that the Greek government must deliver to the EU in the autumn each year. The medium-term forecast is not updated as part of the budget preparation process in the spring. The overall position of the central government finances is updated centrally using the new forecast. One feature of the forecasting process is the overall fiscal targets that the Greek government decides to reach in the medium-term Stability and Growth Programme forecasts. If the fiscal targets (deficit, expenditures, revenues) are not reached according to an updated medium-term forecast, unspecified or partly specified “reforms” are added (such as a reduction in tax evasion or government expenditures), without these reforms being specified in concrete detail. The macroeconomic forecasts are not used in the line ministries’ budget preparations; rather, as discussed below, they develop their own forecasts. This practice naturally hampers the use of the estimates and indeed undermines the integrity of the budget." 
As K. Featherstone, a long-time Greece-watcher argues, SGP compliance did make a difference but of a very different kind: it strengthened the hand of Greek finance ministers against their colleagues, at least in their short term:
The Maastricht convergence criteria and the Stability and Growth Pact set clear policy parameters and created an external discipline for monetary policy in Greece. At home, the government was empowered: the legitimacy of the EU and the precision of the convergence criteria carried a difficult process of adjustment forward. [...] ultimately the strength of the domestic reform initiative would very probably have run aground without the Maastricht constraint. It was telling that, [...] in 2002 [...][t]he Simitis government did not call for a lessening of this external discipline: presumably, it saw advantages in having the corset. It was a means of strengthening its domestic position when pressing for difficult reform.  
I would argue that, in the post-Maastricht era, the ultimate constraint was not the SGP targets, or the government's deficit forecasts. The true targets were the Government's cash targets; these were constrained by a combination of tax revenue and 'safe' borrowing. Notice, in the OECD's 2008 review of Greek budgeting, how much more regular, robust, integrated and closely monitored the cash targets were than the actual budget and the SGP targets.
The process of cash management includes the preparation of the “budget expenditure implementation plan” and of the “cash plan”. Both plans are backed up by the monthly cash limit decision. The “budget expenditure implementation plan” shows monthly forecasts of expenditures. It is prepared for the entire fiscal year, and is updated and rolled over on a monthly basis. The plan is based on the budget appropriations. The monthly forecasts are prepared by using the assumptions underlying the budget preparation and monthly historical data. The “cash plan” puts the “budget expenditure implementation plan” in the context of the revenue forecasts. It is on a pure cash basis and shows daily cash inflows and outflows from the “Single Treasury Account”. [...] The “cash plan” is reviewed and updated every day for the whole month and every month for the whole year. [emphasis mine] The monitoring system includes a continuous flow of data from the Treasury’s departments, the Central Bank, the Fiscal Audit Offices, and the local Tax and Payment Offices. The “cash plan” is a tool for ensuring that there will be adequate cash balances to meet the budget obligations. The forecasts of the cash plan are used for decisions on borrowing and for investing the cash surpluses. The forecasts are elaborated and a ministerial decision is issued, defining a monthly cash limit for every unit involved. Fiscal Audit Offices and Tax and Payment Offices are required to ask for special approval before payments above a certain amount are made (EUR 3 million). The limits are checked against the monthly outcome data and crosschecked against information received on a daily basis by the Central Bank.
The cash constraint was much harder than the SGP's 3% target. But it was soft in another, more insidious way. Tax revenues may have looked steady but they were vulnerable to erosion and the political cycle; market financing was based on a colossal, global mispricing of risk.

We can test some of this insight empirically. A reasonable number of studies have looked into the causal link between tax revenue and government spending in Greece. The question common to all is whether we followed a 'tax and spend' model, whereby government sets its spending target based on what it can raise through taxes, or a 'spend and tax' model, whereby government sets its spending target based on politics and then scrambles to raise the taxes to pay for it. You can see my selection of studies for yourselves below:
This is by no means the last word on the matter, but it seems to me that Greece operated a 'spend and tax' model (i.e. government set a spending target first and then adjusted taxes to fund this) for most of our modern history - but  switched to a strange type of cash-and spend policy post-Maastricht, which counted any borrowing we thought we could draw on without inviting undue attention as equivalent to tax revenue. It was the sum of this plus actual tax revenues which led government spending post-Maastricht.

Un-gaming Europe's deficits

People often wonder why, if Greece was only an extreme case of a much wider problem, Eurostat called for changes to European countries' deficit calculations in such a piecemeal manner, review by review, rather than demand that everything be restored to the appropriate level of accuracy at once. Part of the story has to do with the fact that Eurostat's powers and capabilities changed as a result of the Greek crisis - it did not always know what changes needed to be made, nor could it impose changes.

You see, back when our original 2009 deficit figures and the first revised 2009 deficit figures were released, Eurostat did not have auditing powers over national statistics agencies. It only got those in June 2010, because, and I quote, 'in 2005 [when this was originally proposed] several key member states were opposed to a strengthening of Eurostat's powers.'

This new demand for auditing powers for Eurostat came, appropriately, from the European Parliament, and this time, strengthened by the evidence of statistics gone wild in Greece, it managed to get past the Council. Eurostat's September 2010 visit to Greece, which resulted in the final deficit figures which are currently being questioned, was the first time ever that the Directorate made use of its new auditing powers. This resulted in an unprecedented ability to zero in on unreported or misclassified spending and liabilities.

Even in the days leading up to June 2010, the struggle to deny Eurostat its new auditing powers and maintain Governments' 'right' to lie to their citizens was fiercely defended by the Council:
[...] ministers have watered down aspects of the Commission's original proposal. The Commission wanted to require member states to punish their civil servants with “effective, proportionate and dissuasive” sanctions if they deliberately misreported data to Eurostat. Ministers have removed this requirement, because they felt it was an unacceptable infringement of national sovereignty.

The Commission also wanted to place a mandatory obligation on member states to provide Eurostat with “experts in national accounting”. These experts would work with Eurostat on a temporary basis, to help it prepare visits. This was also removed by finance ministers.

The Commission had protested against the changes, but backed down because it did not want to threaten the chance of the legislation being adopted. 
The Commission also wanted to place a mandatory obligation on member states to provide Eurostat with “experts in national accounting”. These experts would work with Eurostat on a temporary basis, to help it prepare visits. This was also removed by finance ministers. The Commission had protested against the changes, but backed down because it did not want to threaten the chance of the legislation being adopted.
Who were the 'key' member states so opposed to further scrutiny of their accounts? Why only the UK, France and Germany. There is no record of how Greece voted but, by the looks of it, that first bunch of proposals was dead on arrival.