Sunday, 19 April 2015


When will the Greek government run out of money? It's hard to know how to begin answering the question, but this has not stopped anyone from trying to guess over the last two months.

First, Bloomberg ran with February 25th; then Reuters ran with April the 9th; then Reuters ran with the 20th of April and Bloomberg ran with April 24th. The Beeb have their sights on the 12th of May, when the IMF needs to be repaid. Bruegel did the math on our financial assets here and found that Greece could stay afloat till the summer. The WSJ narrowed this down to July 20th, when a large amount of ECB debt becomes due, on the back of a major round of T-bill repayments. The Guardian seems to agree.

This speculation is sometimes driven by statements by Greek officials (typically unsourced and subsequently denied), but mostly by analysts poring over our schedule of repayments to the EU, the ECB and the IMF and trying to work out whether we'll have scraped enough money together from our tiny surpluses, predicted to come in at EUR200m-300m per month in the second half of the year, and T-bill auctions (which our own banks can no longer be forced to absorb due to the ECB) to pay these off. The speculation suits both sides, and is part of their negotiation tactics; since neither side wants Greece to default, underlining just how close the country is to default is important for both. There are reports of cautiousness and fatigue around such estimates, but they just keep coming.

You can compare and contrast the main repayment timelines here, here and here. You can also see the Economist's very recent and complete estimates of pending payments below, and contrast them with Greece's recent record of monthly surpluses. The reason why March, April and May were such popular targets for default theorists is partly that a very substantial amount of debt came due in March and partly that the three months are persistently cash negative for the Greek government, and there's three of  them in a row. Even Greece's original bailout request came in April 2010. But the barrage of large repayments in the summer is even more compelling.

If you ask me, this method of comparing payments due with monthly surpluses is a little misguided. It works well for people used to commenting on highly leveraged organisations such as major banks. Instead, Greece is far more likely to run out of  money the way a medium-sized business would - in denial, tapping suppliers for credit, leaving staff unpaid and dodging debt collectors. This kind of default is messy - suppliers will down tools and refuse to provide further goods and services; staff will do likewise, e.g. through industrial action; and creditors will use their influence to cause difficulty and embarrassment. As a rule, messy defaults tend to amplify the original cash shortfall, as planned revenues remain unrealised or uncollected and expensive patches are needed to deal with interruptions in vital services.

The first sign of a messy default was the Greek Government's preparations to seize the surpluses and/or reserves of General Government entities. You can see these cash positions (currently up to Q3 2014, which isn't helpful) here, or check out the Bruegel article cited earlier. The Greek Government is already working in this direction, e.g. passing a law allowing it to force pension funds to buy T-bills, and mulling raids on the cash reserves of state enterprises. and other entities. But it's not just suppliers that can be tapped for credit - it's workers, too.

What might surprise outsiders is that Greece had a tradition of leaving general government contractors unpaid for a long time even in the good days - especially when their programmes had complex funding structures. But pockets of unpaid labour have started to spring up in recent months, including hospital doctors, auxiliary doctors, and general practitioners tied to the health service, local authority tourism staff, social workers, local authority staff on disability inclusion programmes, psychologists working for immigrant detention facilities, archaeologists, or exam invigilators.  Most, though not all, of these have been unpaid since before Syriza came to power.

This is horrible news for the people involved, but good news for the Greek Finmin: it is much harder for him (us!) to run out of money than people expect. Once a sovereign enters this cynical looking-glass world, and learns to view their own staff and suppliers as a source of finance, all expenses become opportunities to raise cheap funding - by paying late.

Here are the payroll and procurement expenditures of the Greek government from early 2006 to Q3 2014. I use intermediate consumption as a close proxy for procurement, for which figures are not released with this frequency. The sum comes up to a cosy EUR7.3bn per quarter.

How much of this are we already tapping? In our quarterly financial accounts, you can find both the stocks and flows on a quarterly basis under 'other accounts payable' (see here for the figures*, and here for definitions). The figures show that, for the first time since at least 2006, the Greek state was net paying off creditors throughout 2014, to the tune of around EUR1bn per quarter - a shift in policy that was equivalent, in cash terms, to a small fiscal stimulus.

Unfortunately, we don't have recent figures beyond Q3 2014, but at least we can establish the orders of magnitude involved. The Greek state has never been able, post-crisis, to get away with borrowing more than a tenth of its procurement and payroll expenditure from staff and trade creditors. If it were to go back to this level now, it could count on about EUR1.8bn per quarter, albeit only for a very short period of time. It was only able to squeeze that much out of creditors back when its was creditworthy after all. Additionally, government agencies are now bound by the Late Payment Directive, which limits the terms of credit they can demand to 30 days (60 for hospitals).  A more realistic target would be to simply get back to being cash neutral, which is still a stretch and would yield closer to EUR1.1bn. That would be enough to cover our summertime debt payments, and see us through to the traditionally cash-positive second half of the year.

But here's the deal. If Greece is to use this tactic to pay off the debt coming due in June and July, it needs to start delaying payments right now - and Syriza's populist government cannot afford a buildup of many months' worth of selective defaults on its own people. Generally speaking, this kind of internal borrowing is hair-raisingly risky, and cannot be sustained for long without risking total collapse.

So the bottom line is, behind the scenes the Greek government should be drawing up a list of the people it can afford to piss off for a couple of months, and who can afford to wait. If it manages to draw on suppliers and state employees for the credit to see it through July, and if it manages to control spending thereafter despite its pre-election promises, it just might scrape by the rest of the year, after which our repayments schedule becomes quite uneventful (see below, from here). It could work, but it's fraught with threats to the economy and to social cohesion.

*I notice that, bizarrely, the stock and flow numbers don't reconcile after Q2 2014 - i.e., there are changes to the stock of trade creditors that cannot be accounted by transactions, i.e. by the state paying off its debts to suppliers. I have no idea what this means but it looks odd.

Wednesday, 1 April 2015

On Greek working hours and structural reforms

Did you know who the hardest-working Europeans are? Go on, guess. It's the Greeks. Crazy right? Don't you feel bad for thinking they were lazy until now?

Three times already I've looked into this subject (here and here and here), and three times the story has turned out to be fatally flawed; yet with each year's 'Weekly Hours Worked' release from Eurostat a new batch of clickbait articles pop up tantalising the few people left who haven't yet been exposed to this supposed mythbuster with some variation of 'you'll never guess who the hardest-working Europeans are!'

If you feel vindicated by such articles against unfair stereotypes of supposedly lazy Greeks, I can't blame you. A proper look at (appropriate) data confirms that we are not lazy. But if you actually take these articles at face value, you have been woefully misled. You've also missed out on an opportunity to understand the Greek labour market in more detail. The only lazy people here are content mills like Statista.

The latest iteration - 2013 working hours and the Independent

This time, the story made it into the Independent, after trickling down from other outlets (see here and the original, here), until the Greek edition of Vice ended up covering the story under the headline 'should we be proud that Greeks work the longest hours in Europe?' The Vice_Gr story featured an interview with one Mr. Poupkos, Youth Secretary of the Greek tertiary Union, who is actually not an entirely unreasonable guy - however, on this occasion his commentary focused on curbing illegal overtime and unpaid work; reading the stats and the interview together, one might be excused for thinking that the reason Greeks work so much more than other Europeans is because of unscrupulous bosses (of which, I do not dispute, we have no shortage).

In fact, the truth is very, very different. Comparing like for like, Greeks don't work insane hours at all. But the Greek labour market itself is deeply, structurally unhealthy - and the whopping working hours total is merely a statistical artifact of its structural defects. To understand why this is, you have to understand what figures these publications are quoting, how they compare across countries, and what drives them.

About the hours worked data

The comparative figures on hours worked in Europe come from three sets of publications - actual hours worked in main job; usual hours worked in main job, and actual hours worked in second job. In theory, usual hours for second jobs should be reported somewhere but I've struggled to locate them.

All three of these datasets are compiled by Eurostat on the basis of the Labour Force Survey, which is carried out across the EU, and you can see in minute detail how these were defined up to 2013 here. LFS is the same survey we rely on for employment and unemployment figures, and is carried out pretty uniformly across the EU member states. If you see anyone referring to OECD hours worked figures, like Statista does, note that these too are sourced from the LFS when it comes to EU countries. For non-EU countries, the data sources can vary, and the implications of such variation can be very significant, so country figures are not comparable (as the OECD duly points out). For some reason, Statista's figures seem to be aligned to 'usual' as opposed to 'actual' hours worked,  even though they cite 'actual' hours worked data as the source. The effect is to slightly overestimate the actual time worked by Greek employees - citing 42 hours per week in people's first jobs, as opposed to the 'actual' figure of 40.3. Adding people with second jobs yields an 'actual' figure of 41.2 hours a week (see here).

These hours are self-reported by individual respondents -not employers, as the LFS is a household survey. They include overtime but not lunch breaks or travel to and from work. They are not unduly seasonal, as they incorporate four quarterly measurements, with multiple survey waves per quarter.

Comparing like with like

Whole-labour-force figures like the ones published by Statista are very crude: they lump every working person together, add up their hours and divide by the number of people. So unless the economies being compared are very similar, these sorts of comparisons end up not comparing like for like. Full-time workers work longer hours than part-time workers. Men work longer hours than women. The self-employed work longer hours than employees, particularly if they own businesses of their own.

When you do compare like for like the differences don't look so huge anymore. The Greeks who genuinely work longer hours than their Eurozone counterparts are everyone with a second job (by a long stretch), and people who work part-time for a family business. Additionally, male part-time employees and part-time own account workers work longer hours than their German counterparts, though not their Eurozone counterparts.

So the contribution of unscrupulous bosses to hours worked in Greece is very limited, and the bosses in question are most likely to be business owners employing their own families, employers of second-jobbers (who are, presumably neutral with regards to whether their employees have another job but prefer more qualified candidates), and employers unable to offer their staff full hours, who end up employing people part-time when everyone would prefer a full-time arrangement.

Using usual v. actual hours worked where possible doesn't change the big picture, although it does tend to amplify the differences between Greece and other countries.

Note that the discrepancy that is so obvious when one looks at hours worked carries over to other areas of work that are far less documented - holidays, sick leave, retirement age, the lot. The self-employed and family workers, for instance, take fewer days off and retire later, if at all. 

A structural issue

So what accounts for the huge discrepancy in total hours worked, if not unscrupulous employers or workaholic Greeks? The structure of the labour market, that's what.

Simply put, Greece has fewer part-time employees* (less than half the EU average despite an upward trend); fewer economically active women (a participation rate ca. 12% lower than the EU average); more self-employed people (including the 25.4% of our workforce who are own-account workers, v. 10.8% EU-wide and the 6.6% who are employers, v. 4.3% EU wide); more contributing family workers (three times the EU average); but not more people with second jobs - in fact, the share of employed people with second jobs (deduced from here and here) peaked at 3.4% in 2009, 10% lower than the EU average, then fell to 1.8%, or less than half the EU average, by 2013; as might be expected when jobs are in short supply.

So the question of 'why do Greeks work so many hours', put in its proper context, becomes one of why we have so many unproductive family businesses, why there are so many own-account workers in our workforce, why our workplaces find it so hard to accommodate part-time workers, and why we can't get more women into the labour market. Not all of these are necessarily problems, but they are structural characteristics of the Greek economy. Changing them requires that magical catch-all phrase, 'structural reforms.'

* Note that LFS definitions of part-time employment differ among member states. 

What about productivity?

The usual counterarguments offered to the kinds of crude working hours  statistics provided by the Indepedent are a) yes, but productivity is lower b) yes but it is possible to spend hours at work without doing much c) yes but it's possible to work very hard producing products/services that no-one wants. All three are roughly, though not exactly, equivalent, and all three generally lead back to the structural questions raised above.

Since these objections are prompted by a discussion of hours worked, the obvious point of departure in comparisons of productivity is output per hour worked. As you can see here, the Greek economy produced about a quarter less per hour worked than the EU as a whole, even after accounting for our lower purchasing power.

However, this tells us very little about Greek workers themselves: individuals are not 'productive' as such, because labour is only one factor of production. The productivity of labour is a function of many things, including private and public capital. So while it is meaningful to ask how many hours the average Greek works, it is pointless to ask how productive he/she is and attribute this to him/her as a personal characteristic. To do that, we need to ask a different set of questions altogether, and there are two alternatives.

We can measure the quality of labour used by the Greek economy and its stock of human capital; or how much of its output cannot be explained by inputs such as hours worked and capital employed alone. Each question is methodologically fraught but attempts have been made to answer them.

With regards to human capital, sources such as the WEF's Human Capital Report, the WEF's World Competitiveness Report or the UN's Human Development Index are not terribly helpful because they include too many labour market outcomes in their calculations - what we're looking to capture are inputs, such as skills, motivation, resilience and self-discipline, and they only include education inputs up to a point. The best use one can put them to is in measuring the quantity and quality of education - where you'll find that Greeks do well for quantity but less so for quality - and even then a lot of the lag in quality comes in after one enters the job market - which is an employer's fault at least as much as it is the employee's.

One can also draw on social psychology research, and particularly comparative studies focusing on the Big Five personality traits, to measure some intrinsic qualities that make up the profile of a hard-working personality - conscientiousness in particular is a trait known to correlate strongly with work performance, and which fits the kind of attitudes involved in the 'hard working Greeks' theme. One comparison finds that Greeks score reasonably well and there is little difference in this regard between Greeks and Germans; in another, we score slightly higher. Similarly, the OECD PISA comparisons of perseverance and motivation among pupils show that Greek pupils do really quite well (check the data here, there is too much to summarise). 

TFP calculations are problematic because they require detailed data on the various factors of production (including different kinds of capital), which the Greek authorities do not produce regularly. Both the OECD and EU-KLEMS projects include no TFP calculations for Greece, although EU KLEMS provided enough data for others to fill in the blanks. We now know, for instance, that Greek TFP did not rise from 1999 to 2007, nor indeed did TFP in any PIIGS country, or in Belgium for that matter (but see here too for the opposite finding). Greek TFP was also losing ground against other developed countries from the early 80s onwards, possibly even the 60s, and continued to lose ground throughout the early part of the crisis

But TFP figures us nothing about the attributes of Greek employees - only that they appear to be saddled with crap technologies (in the broad sense) of production. Indeed, TFP is a residual - the measure of our ignorance.