NUMBERS ARE PEOPLE COCK-UP BEFORE CONSPIRACY • CITE PRIMARY SOURCES OR GO HOME


Sunday 20 December 2015

THE GREAT UNPAID

Acknowledgement 5/1/2016: This post, and the mixed income post that triggered it, owe much to the thinking of A. Doxiadis and his book, Το Αόρατο Ρήγμα, on the structural characteristics of the Greek economy. That's not to say he has endorsed the post of course. All errors etc are my own.

Are You Into Detailed Sector Statporn?

Veteran readers know that I am a big fan of the Eurostat website. I have no interest to declare here; I am not an employee or contractor or associate of Eurostat or any of the national statistics agencies it oversees. I do, however, perform regular fact-checks and so I appreciate two things that Eurostat does to make my life easier.

One is of course their database's bookmark function - the ability to capture a set of tables as I have customised them and share them in the precise same layout to anyone else. I'm amazed at how rare this is (eg the Bank of International Settlements publishes Bankstats with almost the same functionality - the OECD is nowhere near), and how little it is used by commentators.

The second thing I like about the Eurostat database is the sheer coverage of national-level statistics. Greece's ELSTAT produces a shit-ton of data that the average Greek can never access except through Eurostat. Of these, my firm favourite is the SILC dataset on poverty and living conditions (see eg here) - as grim as it is to read the findings. However, I have recently found a close second to SILC in Eurostat's annual detailed enterprise statistics.

I urge you to consider this dataset because it really exposes the guts of the economy - both by providing rich data (number of businesses, employment, value added, investment, etc) but also by providing data by 456 detailed sectors (4-digit NACE, not including agriculture, government and social and community services). Want to find out how many call centre workers there are in Greece, and how much they get paid? Done. Want to know what the finances of Greek newspapers look like? Done. Maybe how much was invested into Greek winemakers (because allegedly much of this activity was money laundering)? Done.

I started looking into these data again the other day a propos of my post on mixed income and its relationship with tax revenues. The figures on mixed income are very interesting, but where do they come from? Who are all these self-employed people? What do they do? More importantly, does the decrease in mixed income as a share of GDP represent structural reform as I had originally suggested? Or was there something else at play?

A Mixed Income Scoreboard

So I ran the detailed tables for industry, construction, trade and services - and prepared a comparison between 2009 (the last pre-Memorandum year) and 2013 (the latest available data). I would ideally have picked 2008 but there are substantial gaps in that year's figures. Even with 2009 I have had to extrapolate once or twice (eg taking averages between 2008 and 2010).

The sectors included here accounted for 2.54m of 4.6m working persons in 2009 (of which 37% were unpaid) and 2.11m of 3.5m working persons in 2013 (of which 36% were unpaid). Two notes: first- these figures do not include the entire private sector, as explained above. Second- the 'unpaid' counted by Eurostat were not employees who hadn't been paid what was owed to them. They were people not drawing a salary from the business - working proprietors and their families, perhaps a few volunteers too. That's not to say they weren't earning an income from the business, however: their income might eg be a share of the profits of the business, as opposed to a salary.

So I ranked all of the 456 detailed sectors according to the absolute number of unpaid workers in each of them and kept the top 20 in each year. I avoided a ranking based on relative ranking because a) in turbulent times a small, niche sector's workforce can change more dramatically than that of a major sector and b) my intention is to track Greece's reliance on household mixed income - so the larger sectors are a priority.

As you can imagine, there's a big overlap betweem the 2009 and 2013 rankings; there have only been two movers in each direction; so I've opted for a top 20 + 4 leaderboard.  These 24 3-digit sectors (see list below) accounted for 21% of all employment in 2009 (971k). Of this 971k, 499k were unpaid (51%). Not one of the 24 top sectors had more than 64% of its workforce on the payroll in 2009. In 2013, the top 24 sectors accounted for 23% of all employment (798k); of this total, 427k were unpaid (54%). Only one of the top 24 had cleared the previous 64% ceiling for payrolled workers (the sector in question being the Greek equivalent of off-licences or 7-11s, which now had 69% of its workers on payroll).



This is Not the Reform Proxy You're Looking For

The key outcome here is that the industries that are core users of unpaid work (ie producers of mixed income) have not reformed quite so drastically. The improvement in the balance of mixed income to GDP seems, instead, to have come from the reduced income of these sectors. That's not really what I had in mind when wondering whether the fall in mixed income as a share of GDP could be a possible proxy for structural reform.

This is clear if you look at the supposedly 'reforming' sectors, in which reliance on the unpaid fell: bakeries, homeware stores, architects' practices, off-licences, and legal practices. The relative share of the unpaid fell by more than 38%. Yet only construction companies and law firms improved their labour productivity at the same time. On the other hand, a number of the top 20+4 sectors saw the role of the unpaid expand: look at hotels, specialist retailers, car repairs. All of these saw their reliance on the unpaid grow by 43%  or more, as business grew too tight for anyone with a payroll. All of them also saw labour productivity fall.

Now for the big picture. In 2009, the top 20+4 produced EUR16bn of value-added* (7.5% of all gross value added). clearly underperforming the rest of the economy. In 2013, they underperformed even further, producing EUR9.5bn. (5.9% of all gross value added). Their apparent labour productivity fell by 27%.



What does this mean? Far from reforming, Greece's mixed-income generating industries are withering on the vine. It's unlikely that one explanation alone can account for this, but clearly turning failing, near-informal businesses into companies takes substantial amounts of credit, which Greek banks still struggle to provide, and equity, which even well-off households are running out of. Moreover, an antiquated insolvency framework and treacle-slow courts make it harder to dissolve failed businesses. Perhaps, of course, the fall in apparent labour productivity is simply a reflection of subsdued demand, and a recovering economy will reveal structural reform of the kind I was hoping for. We will have to wait and see.

*I've had to tweak one figure in all of this - the value added number for bakeries, which may or may not have been inflated by money laundering. I'm not sure but the numbers don't stack up. See for yourselves. The difference between my 'extrapolated' figure (average of 2008 and 2010) and the one Eurostat records is about EUR200m.

TO BE CONTINUED

Tuesday 1 December 2015

UNDER-TAXING GREECE

Acknowledgement 5/1/2016: This post owes much to the thinking of A. Doxiadis and his book, Το Αόρατο Ρήγμα, on the structural characteristics of the Greek economy; as well as to conversations on tax with Gregory Farmakis. That's not to say either has endorsed the post of course. All errors etc are my own.

One of the stylised facts of the Greek crisis has been that Greece never truly overspent (barring perhaps 2009); if anything, it under-taxed. It's not hard to understand where this argument comes from - just compare revenue and spending as a % of GDP between Greece and the EU or the Eurozone average - the data are available here.

Clearly, in 1995 government spending in Greece was well below the EU average. It caught up quickly, but even so was still very close to the EU average around the 2005-7 period, especially after allowing for higher interest costs. Revenues, on the other hand, lagged the EU average by a persistent 4 percentage points each year since accession. On the face of it, it's more credible to look at Greece as a story of under-funding of the state than one of overspending.

Advocates of the under-funding hypothesis also point to the (unfortunately, now discontinued) dataset of tax by economic functions. This generally suggests that Greece has been taxing, eg., consumption more than the rest of Europe, while taxes on capital, once above the EU average, have fallen steadily and payroll taxes on employers have been persistently lower than the EU average (data here and here). Greek governments, the narrative goes, gave tax breaks to some industries, turned a blind eye to avoidance and evasion of social security contributions by others, then let taxpayers foot the bill.

This narrative eventually made its way, via the Trade Union movement, to the Greek Parliament's debt audit report - the definitive account of that kangaroo court that called itself the 'Debt Truth Committee.' It was one of the better-documented claims made in the report.
And it is wrong.

1. Greece is not what you think it is

Upset about Greece's low government revenues? Spare a thought for Tunisia's government, whose revenues are only ca. 24% of GDP- just over half the EU average. Is Tunisia's government under-funded? To be sure, it is less good at extracting tax from its population than the average EU country; but no one would dream of using an EU average as the yardstick for Tunisia.

Yet, by virtue of being an EU country, Greece is often benchmarked against the continent.The assumption is that Greece is a similar sort of country as its EU peers and the Greek state should rightfully expect similar revenues. If there is a difference in revenue, it is due to lower tax effort: a combination of lower tax rates or a lower level of tax compliance, which the state tolerates. In reality, this comparison is invalid. Greece may not be like Tunisia, but it is also not a mini-Germany, a mini-Spain or a large Portugal for that matter. It is structurally less well placed to yield large tax revenues.

The reasons are simple. All other things being equal, it's harder to tax the self-employed; of whom we have proportionately many more than other European countries. It's harder to tax the poor; of whom we also have more; and harder to tax retirees, of whom we also have more; finally, it's hard to tax chronically unprofitable micro-enterprises; of whom we also have more.

To illustrate how big the effect of these structural characteristics is, I wanted to focus on one example. Let's look at the components of market output that a government can reasonably expect to tax - the operating surpluses of financial corporations; the operating surpluses of non-financial corporations; the operating surpluses of organisations serving households; and the mixed income of households. You can find all of these figures here; unfortunately the figures are not expressed as % of GDP, which has to be done manually by comparing with these figures.

Once you run the figures, one set of numbers stands out - household mixed income. As pg 200 of Eurostat's ESA2010 manual explains, this is the income of self-employed people working in unincorporated businesses (ie not companies); it includes the income of business owners and their families. Hence the term 'mixed': these are part wages and part profits, and the two can only be separated arbitrarily.

Greece, as you might expect from my introduction, has the second-highest share of mixed income as a share of GDP in Europe, and had the highest bar none pre-crisis. But, whatever lazy journalists tell you, this structural issue is not common to all of the PIIGS countries. Greece stands alone amongst peripheral Euro countries in having its operating surpluses distributed in this way. At 23.7% of GDP in 2008, mixed household income as a share of total output was over three times higher than in Cyprus; more than twice as high as in Spain; nearly twice as high as in Portugal; and over 40% higher than in Italy. Spain was more or less on a par with Germany on this metric. In fact, to find economies similar to Greece's in their dependence on household mixed income, you need to go as far as Poland and Slovakia.





But what does this mean for tax? Well, there is a good correlation in Europe between mixed income as a share of GDP and the amount a government can raise from taxes on income and profits (data on this here). In fact, all of the high-tax/high spend Nordic economies have tiny mixed income contributions to GDP; while those with high mixed income contributions tend to be recent accession countries. The correlation isn't perfect of course, and I am guilty of cherry-picking; the correlation becomes less neat after 2008. I believe the reason for this is the increased tax effort of some economies over others during and after the crisis; as tax effort rose most amongst those countries with the lowest tax revenue, it makes sense for the correlation to weaken. In fact, in later years the curve tends to flatten as all countries apparently aim to raise at least 5% of GDP in income and corporation tax.



Now look very closely at the revenue/mixed income graph. First- it suggests that tax effort in Greece may, if anything, have been relatively high pre-crisis. In a slightly different study looking at structural influences on tax revenues, the World Bank's researchers have found much the same thing, although for completeness I should note their colleagues at the IMF and IGC have found the opposite.

Yet another way of approaching this would be to consider how far along the Greek laffer curve we were in 2010. As veteran readers know, we have an estimate of this from Trabandt & Uhlig (2012), who found that 2010 Greece could only increase its tax take by 2.4% of GDP by raising capital taxes before it started falling again. It could increase it by a maximum of 4.8% of 2010 GDP by also maximising labour taxes. True enough, Greece has never since managed to get to that peak, but even without a major recession it seems the best we could ever do would be to catch-up to the EU average.

Now the tax to mixed income graph we discussed earlier suggests that, if the mixed income contribution in Greece were to fall to where Spain's is (ie by 12 percentage points - which is to say, very substantially), we might expect income and corporation tax revenues to rise by a good 4% of GDP - incidentally the same margin by which our government revenues have chronically lagged those of the rest of Europe.

Does all of this not sound familiar? This over-reliance on self-employment and very small businesses is precisely the same distortion that is responsible for the bulk of Greece's lead over the rest of Europe in terms of hours worked per person. Isn't it time we reviewed this properly? I know you've heard it said a million times that small businesses are the backbone of the economy, and that entrepreneurs will save Greece/Europe/the world; I have worked in small business advocacy for years and I have some sympathy for this view. But not all self-employment is enterprising and not all small family businesses are small and family-run for the right reasons. A Greek self-employed pharmacist may be an entrepreneur; but some of their colleagues are also effectively civil servants with a profit margin. A Greek freelancer may be a flexible go-getter; or they may be an employee whom the employer doesn't want on their books so they can avoid national insurance contributions.

Here's a heretical proposal; why not adopt the change in mixed income as % of GDP as a simple indicator of structural change? Whether it is a good thing, I cannot say. But it is worth noting that, by this token, Spain, Croatia and Bulgaria are the star post-crisis reformers; much more so than Greece.

UPDATE 22/12: I've looked into why the mixed income of Greek households has fallen as a share of GDP here. In summary, and with the exception of construction and law firms, there is no evidence that this is a sign of 'reform' - Greece's mixed income-generating sectors are withering on the vine rather than formalising.

An ideological aside

I don't wish for readers to interpret the above as a suggestion that the Greek economy must be reformed into a shape that maximises tax revenue. Paying taxes is nobody's idea of the meaning of life or the purpose of economic activity. But we need to accept that the current setup leads to low revenues, almost inescapably. We can choose to accept a low-revenue, low-spend equilibrium; tax ourselves to the gills to achieve a high-spend, high-ttax equilibrium with very low output, or aim for a low-revenue, high-spend and high output equilibrium and accept the hardship that will surely come whenever the credit line next dries up. There is no other choice.

2. Tax revenue is not what you think it is

In addition to misunderstanding what kind of country Greece is, the discussion of Greek tax effort also ignores what tax revenues are, and why ours are different than many other EU countries'.

Tax revenue isn't just money the state takes. Some tax revenue is also money the state makes. It is a return on past public investment. By this I don't mean tax revenues resulting from fiscal multiplier effects, which should be small on a cyclical basis if the public finances are consistently well managed. I am referring to tax revenues resulting from the deepening of physical, social or human capital as a result of public spending.

Essentially, when a government invests productively, it builds public capital which in turn boosts the productivity of the private sector. Better roads and ports enable trade. A nation-wide electricity grid makes appliances more reliable and homes more valuable. A fibre-optic network brings more people within reach of their peers and of online retailers. More educated, informed, empowered or healthier people are both more productive and more demanding. Simply- and well-regulated industries are more trustworthy and productive. At the extreme, the state can sometimes build entirely new kinds of intellectual capital; it can become a venture capitalist of sorts, creating whole new fledgling industries for the private sector to explore - the premise, after all, of Mariana Mazzucato's Entrepreneurial State.

The better an investor the state is, and the more it allocates funds to investment rather than consumption, the more its tax revenue will tend to grow as a share of GDP, holding tax effort and state capacity constant. The share of its tax revenue that reflects returns on investment will grow, while the share of revenue that reflects coercion and rents will shrink. If, on the other hand, government spending fails to build valuable capital and improve productivity across the economy, then over time returns on public spending will fall as a share of GDP - forcing the state to increase its tax effort or its capacity just to stay still.

The right amounts, the wrong places

Is there reason to believe that this has happened to Greece in the run-up to the crisis? Yes. We invested broadly the right amounts, but got very little by way of returns. Government gross fixed capital formation was not low in the pre-crisis era, and with the exception of the year 2005 it was almost unwavering at ca. 3.5% of GDP (Eurostat claims that none of this was defence spending, by the way). That put us ahead of places like Norway or Sweden, but just short of Spain or Ireland and way behind Portugal.


The problem then was not the allocation of public spending between consumption and investment, but rather the quality of public investment and its complementarity with private investment.
There are some very good long-term calculations for Greece, in two studies in particular. Kamps (2004) finds very strong returns on public investment between 1961-2001, more so than in other EU countries; but these are what are known as partial returns - ie they look at the returns on public spending without considering the losses from crowding out of, or by, private investment. Using data from 1960 to 2005, Alfonso and St Aubin (2008) confirm the finding of strong partial returns on public investment, but find negative total returns due the negative response of public investment to private investment. The actual, total effect of public investment in Greece was negative .

 A second ideological aside 

In discussing the money that government makes I do not pretend that all government spending builds capital; that all of it is efficient, or has high or even positive returns; or that the private sector would not, left to its own devices, have used the same funds better. All I am saying is that, undeniably, some tax revenue is a return on public investment.



Friday 27 November 2015

Ο ΛΑΟΣ ΔΕΝ ΞΕΧΝΑ, ΟΚ. ΑΛΛΑ ΤΙ ΣΗΜΑΙΝΕΙ (ΚΕΝΤΡΟ)ΔΕΞΙΑ;

Ο λαός, λέει το σύνθημα, δεν ξεχνά τι σημαίνει δεξιά. Όπως όλα τα συνθήματα, έτσι κι αυτό είναι αστείο. Ο λαός ξεχνά και ξαναθυμάται κατά το συμφέρον του, και τη μανούβρα αυτή μπορεί να την κάνει και σε μετα-γνωσιακό επίπεδο: ξεχνά πως κάποτε ήξερε, ή θυμάται πως κάποτε ξέχασε, τι σημαίνει Δεξιά. Έτσι κάπως η ΝΔ ανακαλύπτει το αντιμνημόνιο που ξέχασε μετά το 2012 και η πρώτη φορά Αριστερά συγκυβερνά με Ανελ και κάνει (προσωρινά) υφυπουργό τον Πόρτα-Πόρτα.

Εντούτοις, ενόψει των εσωκομματικών εκλογών στη ΝΔ, με τον Καμμένο άλλη μια φορά κυβερνητικό εταίρο και την ΧΑ σταθερά τρίτο κόμμα (αν και ευτυχώς σε απόλυτους όρους η ψήφος της πέφτει) το σύνθημα είναι τουλάχιστον επίκαιρο αν εκφραστεί λίγο διαφορετικά. Κυρίως δε γίνεται επίκαιρο το τί σημαίνει κεντροδεξιά: ένα ερώτημα-δείγμα παρακμής από μόνο του, όπως και η συζήτηση περί κεντροαριστεράς που προηγήθηκε της τελικής διάλυσης του ΠαΣοΚ.

Προσωπικά δεν με ενδιαφέρει πώς ορίζουν την (κεντρο)δεξιά οι επίδοξοι ηγέτες της. Τα λόγια είναι τσάμπα εξάλλου, και όλοι έχουν κίνητρο να ανοιχτούν σε ένα πλατύ κοινό. Ούτε με ενδιαφέρει πώς την ορίζουν οι αντίπαλοί της. Με ενδιαφέρει μόνο το πώς καταλήγει κανείς να θεωρεί τον εαυτό του (κεντρο)δεξιό - ποιές αρχές, αξίες, μνήμες, συνθήκες διαβίωσης και παραστάσεις ωθούν κάποιον στο να ταυτιστεί με αυτή τη λέξη. Ευτυχώς υπάρχει τρόπος να το ελέγξουμε.

Τα στοιχεία μου εν προκειμένω προέρχονται από την Ευρωπαϊκή Έρευνα Αξιών του 2008 - την πιο πρόσφατη δυστυχώς. Περιμένω πώς και πώς την επόμενη έρευνα του 2017, στην οποία έμαθα με μεγάλη ανακούφιση ότι θα συμμετάσχει η Ελλάδα. Η σύγκριση των δύο ερευνών θα μας πει πιο πολλά για την κρίση, τα αίτια και τις επιπτώσεις της από δέκα διδακτορικά οικονομολόγων πάνω στο θέμα.

Το δείγμα της Έρευνας Αξιών είναι σταθμισμένο και αντιπροσωπευτικό: 1.208 Έλληνες άνω των 16 (η έρευνα περιλάμβανε γύρω στους 1.500 αλλά τόσοι απήντησαν στις ερωτήσεις που με απασχολούν - δείτε παρακάτω). Το ερωτηματολόγιο μπορείτε να το δείτε εδώ. Η πολιτική τοποθέτηση των ερωτηθέντων μετριέται σε κλίμακα από 1-10, όπου 1 είναι ο πιο αριστερός Έλληνας και 10 ο πιο δεξιός. Σε καμμία φάση της συνέντευξης δεν δίνεται ορισμός σε αυτές τις έννοιες. Πέρα από την ερώτηση περί πολιτικής τοποθέτησης, υπάρχουν και πολλές άλλες που εστιάζουν στις νοοτροπίες, αξίες και ιδανικά του συνεντευξιαζόμενου, αλλά και στην καταγωγή, την οικογενειακή και οικονομική κατάστασή του, και τη θέση του στην κοινωνία γενικότερα.

Η λογική είναι απλή: αν βρούμε τις νοοτροπίες, τις ιδέες και τα ιδανικά που τείνουν να ενστερνίζονται περισσότερο οι αυτοαποκαλούμενοι 'δεξιοί' από τους υπόλοιπους Έλληνες (και τα οποία τείνουν να ενστερνίζονται σε μεγαλύτερο βαθμό όσο πιο 'δεξιοί' θεωρούν ότι είναι), τότε έχουμε στα χέρια μας έναν εμπειρικό ορισμό και της δεξιάς, και της 'κεντροδεξιάς.' Και επειδή αυτό που στην πραγματικότητα αναζητούμε είναι μια σειρά από στερεότυπα, σκέφτηκα να χρησιμοποιήσω μια απλή μέθοδο - decision trees. Ουσιαστικά, η μέθοδος αυτή επιτρέπει να δοκιμάσουμε πολλές τέτοιες νοοτροπίες, ιδέες και ιδανικά, μία-μία, για να βρούμε ποιές χωρίζουν πιο ξεκάθαρα τον Ελληνικό λαό σε 'δεξιούς' και 'αριστερούς'. Μόλις βρεθεί η καλύτερη διαχωριστική γραμμή, η διαδικασία επαναλαμβάνεται για κάθε ένα ξεχωριστά από τα δύο μέρη στα οποία χωρίστηκε ο πληθυσμός με όλες τις μεταβλητές που δοκιμάσαμε και νωρίτερα - και ούτω καθεξής, μέχρι να καταλήξουμε σε υπο-ομάδες που παραείναι μικρές για να διασπαστούν παραπάνω, επειδή τα επιμέρους μέρη δεν θα μπορούσαν να συγκριθούν μεταξύ τους με στατιστικά σημαντικό τρόπο.

Εφαρμόζοντας αυτή τη λογική πάνω στο πλήρες δείγμα της έρευνας, βλέπει κανείς ότι μερικές ερωτήσεις χωρίζουν το δείγμα με ξεκάθαρο τρόπο σε περισσότερο και λιγότερο δεξιούς. Σημειώνω με (+) τους παράγοντες που σχετίζονται με πιο 'δεξιά' άτομα, και με (-) αυτούς που σχετίζονται με πιο 'αριστερά' άτομα. Τα 'επίπεδα' που σημειώνω υποδηλώνουν πόσες φορές έχει υποδιαιρεθεί ο πληθυσμός όταν εμφανίζεται για πρώτη φορά ως σημαντική διαχωριστική γραμμή μια μεταβλητή.
  • Επίπεδο πρώτο: πόσο εμπιστεύεται κανείς την Εκκλησία ως θεσμό (+); 
  • Επίπεδο δεύτερο: Πόση εμπιστοσύνη έχει στους δημοσίους υπαλλήλους (-) Έχει ποτέ συμματάσχει σε διαδήλωση (-); Πόση εμπιστοσύνη έχει στα εργατικά σωματεία και τους συνδικαλιστές (-); 
  • Επίπεδο τρίτο: Θεωρεί αποδεκτή την έκτρωση για μια ανύπαντρη γυναίκα (-); Θεωρεί αποδεκτή συμπεριφορά το περιστασιακό σεξ με αγνώστους (-);  Θεωρεί ανεπιθύμητους ως γείτονες τους Ρομά (+); Θεωρεί επικίνδυνο το να επεμβαίνει ο άνθρωπος στο φυσικό του περιβάλλον (+); θεωρεί σημαντικό η δουλειά του να τον φέρνει σε επαφή με (ενδιαφέροντες) ανθρώπους (-); θεωρεί ότι είναι δουλειά των ιδιωτών ή του κράτους να φροντίζουν τους πιο αδύναμους; (-)


Δοκίμασα, βέβαια πολλές ακόμη ερωτήσεις που δεν αποδείχθηκαν καλές στο να ξεχωρίζουν τον πληθυσμό σε 'δεξιούς' και 'αριστερούς.' Δεν τις παραθέτω εδώ για λόγους συντομίας, αλλά περιλάμβαναν και πολλές δημογραφικού χαρακτήρα μεταβλητές όπως ηλικία, φύλο, οικογενειακό εισόδημα, καταγωγή και λοιπά. Θα προσέξετε ότι η ανάλυσή μου δεν εξετάζει την επίδραση των διαφόρων μεταβλητών ταυτόχρονα - δεν υπάρχουν controls. Οπότε δεν απαντά στο ερώτημα 'ποιές επιρροές κάνουν έναν Έλληνα δεξιό/αριστερό;' - μόνο στο ερώτημα 'τι σημαίνει στο μυαλό του Έλληνα δεξιά/αριστερά;'

Μιλήσαμε λοιπόν για το πλήρες δείγμα. Όμως και η αριστερά έχει τη 'δεξιότερη' μεριά της, και η δεξιά επίσης έχει την, αχέμ, 'δεξιότερη' μεριά της. Τι σημαίνει 'δεξιά' σε αυτό το πλαίσιο; Για να το εξηγήσουμε αυτό επαναλαμβάνουμε την ίδια ακριβώς ανάλυση, ξεχωριστά για 'δεξιούς' (6-10 στην κλίμακα αριστερά-δεξιά) και για 'αριστερούς' (1-5).

Για τους αριστερούς τα πράγματα έχουν ως εξής (τα θετικά πρόσημα υποδεικνύουν παράγοντες που σχετίζονται με την κεντροαριστερά):
  • Επίπεδο πρώτο: πόσο εμπιστεύεται κανείς την Εκκλησία ως θεσμό (+); 
  • Επίπεδο δεύτερο: πόσο σημαντικό μέρος της ζωής του είναι η πολιτική; (+) πόσο θεμιτό θεωρεί το να παίρνει κανείς επιδόματα πρόνοιας τα οποία τυπικά δεν δικαιούται; (-) πόσο σημαντικό θεωρεί ότι είναι για την υγεία ενός γάμου το να έχει το ζευγάρι ένα καλό σπίτι; (+); πόσο σημαντικό θεωρεί για τη ζωή του το να έχει άφθονο ελεύθερο χρόνο; (+); 
  • Επίπεδο τρίτο: Θεωρεί αποδεκτή την έκτρωση για μια ανύπαντρη γυναίκα (-); πόση εμπιστοσύνη έχει στις ένοπλες δυνάμεις (+); θεωρεί σημαντικό για την υγεία ενός γάμου να έχουν οι δύο σύντροφοι ίδιες θρησκευτικές πεποιθήσεις; (+)


Για τους δεξιούς τα πράγματα έχουν ως εξής. Όπως πάντα τα θετικά πρόσημα υποδεικύουν παράγοντες που σχετίζονται με τη δεξιά, εν προκειμένω με τους αυτοχαρακτηριζόμενους ως πολύ δεξιούς ή και ακροδεξιούς. Άρα τα αρνητικά πρόσημα υποδηλώνουν συσχετισμό με την κεντροδεξιά:
  • Επίπεδο πρώτο: πόσο σημαντικό θεωρεί ότι είναι για την υγεία ενός γάμου το να ζει το ζευγάρι ξεχωριστά από τους γονείς των δύο συντρόφων; (-)
  • Επίπεδο δεύτερο: όταν τον ρωτάς αν η συντήρηση των φτωχότερων είναι καθήκον των ιδιωτών ή του κράτους, και όταν τον ρωτάς αν ο άνθρωπος έχει δικαίωμα να διαχειρίζεται τη φύση όπως νομίζει, απαντά με απόλυτο τρόπο (σίγουρα ναι ή σίγουρα όχι;) (+)
  • Επίπεδο τρίτο: θεωρεί με απόλυτη βεβαιότητα ότι οι άντρες πρέπει να κάνουν παιδιά για να ολοκληρωθούν σαν άτομα; (+)
[Το γράφημα μου βγήκε λίγο λάθος - το διορθώνω σύντομα]

Συμπέρασμα - το 2008, οι Έλληνες θεωρούσαν 'δεξιά' μια θρησκευόμενη, κοινωνικά συντηρητική μερίδα του πληθυσμού που έβλεπε ως απειλές προς τον τρόπο ζωής της μια σειρά από επιστημονικές, κοινωνικές και γεωπολιτικές εξελίξεις και προέβαλλε ως αξία την άμυνα ενάντια σε αυτές. Η 'δεξιά' στο μυαλό του μέσου Έλληνα έβλεπε κάποιους συνανθρώπους ως ανεπιθύμητους, έβλεπε με δυσπιστία το συνδικαλισμό, τον ακτιβισμό , και είχε επιφυλάξεις απέναντι στο κοινωνικό κράτος και την πρόοδο της επιστήμης. Έβλεπε δε τη δουλειά πολύ περισσότερο ως μέσο βιοπορισμού και λιγότερο ως χώρο έκφρασης ή αυτοπραγμάτωσης.

Με βάση την ίδια λογική, οι δεξιότεροι Έλληνες θεωρούσαν 'Κεντροδεξιά' το κομμάτι της δεξιάς που φοβόταν τους δογματισμούς σε περίπλοκα ζητήματα, και έδειχνε ανοχή σε όσους αμφισβητούσαν τις δομές και αξίες της παραδοσιακής (εκτεταμένης;) οικογένειας.

Οι αριστερότεροι Έλληνες από την άλλη θεωρούσαν 'κεντροαριστερά' το κομμάτι της αριστεράς που ήταν θρησκευόμενο ή θεωρούσε την εκκλησία σημαντικό θεσμό, δεν εξέφραζε αντιμιλιταριστικές θέσεις, είχε μεσοαστικές αξιώσεις (οικονομική άνεση, πολιτική έκφραση και ελεύθερο χρόνο), και επιπλέον το ενοχλούσε η σπατάλη στις κοινωνικές δαπάνες.

Πού να είναι άραγε τώρα όλοι αυτοί οι κεντροαριστεροί και κεντροδεξιοί; Η πρώτη μου αντίδραση είναι ότι οι κεντροαριστεροί του 2008, με τις μεσοαστικές και (γιατί όχι;) υλιστικές αξίες τους, πρέπει να τσαλακώθηκαν περισσότερο από όλη την υπόλοιπη χώρα την εποχή της κρίσης και να εκφράζονται πλέον εντελώς διαφορετικά - ο χώρος τους δεν υπάρχει. Οι κεντροδεξιοί του 2008, από την άλλη, έχουν ελαφρύ ιδεολογικό στίγμα και θα μπορούσαν πλέον να ψηφίζουν ό,τιδήποτε.

Μια τελευταία σημείωση: Το 2008, από τους ερωτηθέντες που δέχτηκαν να πουν πού τοποθετούνται στην κλίμακα δεξιά-αριστερά, το 61% ήταν από κεντροαριστεροί (5) ως @conclavios (1). Το 20% προτίμησε να μην απαντήσει. Το 'κέντρο' (5-6) συγκέντρωνε το 40% της ψήφου, ανώ τα άκρα (1-2 και 9-10) το 18%.

Tuesday 10 November 2015

WE NEED TO TALK ABOUT GREEK PRIVATE EDUCATION

In the early years of the Greek crisis, I used to feel a stirring of guilty pleasure when foreign media were forced to take deep dives into Greek politics. I would imagine young journos doing their meticulous research with a smug look on their faces and then suddenly being hit with a depressing realisation: "wow, this story goes so deep and so far back; I don't know who to trust; and every policy decision past or present hurts someone! - is that how it feels to be Greek?"

I rarely feel this way anymore, but the Economist's recent piece on Greek private education brought it all back for a moment.

VAT on Private Education: the story so far

As readers may know, in the aftermath of its capitulation in July and as details of the Third Memorandum were being ironed out with the Troika Institutions Quadriga, the Greek government found itself looking for alternatives to a VAT hike on beef. This idea, nominally popular with creditors, had run into stiff opposition within the ruling party, and speculation persists that pressure was being applied on the government by the French, eager to protect their beef exports (context here and here).

The Greek counter-proposal was to raise VAT on private education to an eye-watering 23%, and the story offered to the electorate at the time - that this was a specific demand from Greece's creditors - turned out to be a lie. On 'discovering' this in September, Syriza (now campaigning for re-election) pledged to reverse the measure if re-elected. However, on returning to power, they found little to offer the creditors in return and mooted a counterproposal for a three-tier VAT regime (0% for primary and pre-primary, 6% for vocational and cramming schools and 13% for private secondary schools). Unfortunately, the three-tier proposal was illegal. The VAT Directive lists services to which a two-or three-tier VAT regime may apply but this does not include education (for the entire context, read Articles 98, 132-133 and Annex III here). Effectively, the Greek government has only ever had a choice between applying a full VAT rate or a zero VAT rate to all private education.

With opposition to the VAT hike growing, and a number of private schools already in significant difficulty as a result, the Greek government will now, perhaps more appropriately, raise equivalent from gambling instead. Turns out the owners of newly-privatised OPAP aren't as good at lobbying as private schools, or maybe they did their lobbying too early.

Unfortunately, too little actual evidence was used in debating this issue; which is a shame because the facts on the ground tell us a fascinating (and often tragic) story about Greek society and how it's coping with the crisis.

An objection in principle

Before I go into the statistics, I need to clarify one thing: I believe that true education should not be taxed, and definitely shouldn't be subject to VAT of all taxes. Education, whether private or public, is not consumption; it is an investment. The time to tax is is when the human capital it creates starts generating income. There is a significant debate about how good the returns on investment in education are and whether any of the mechanisms that we assume produce its returns actually work (see this gem from Pseudoerasmus for example), but there is no doubting the purpose of most such spending; it is an investment in human capital.

I say most because not all spending with education providers purchases education as such. Parents may eg pay a premium for kids to be looked after a little while longer while they're at work. This extra schooling might build no human capital, but instead simply buy employers and employees additional flexibility. Rich parents might pay for access to a social elite - an investment in social capital but also (in less meritocratic societies) in future economic rents - which many libertarians would happily agree should be taxed. Parents afraid of the stiff competition their kids will face in getting into university or finding a job market may be paying for teaching-to-the-test even though they know it does not build human capital; as a kind of insurance for their children.

How education spending should be taxed or subsidised, whether it is investment or consumption, and whether it ought to be promoted or suppressed through better co-ordination really depends on these questions. It is perhaps natural for the ideological Left (whatever's left of it in Syriza) to despise private provision of a public good. However it's also worth bearing in mind a historical irony: in Greece's modern history, the expulson and, later, exclusion of teachers with open communist sympathies from public schools contributed strongly to building the supply of private tuition; this may also help explain its ability to specialise in serving lower-income groups.

How many Greek households use private education anyway?

The obvious starting point is the actual share of pupils enrolled in private education. 'Private education' is a very wide term, so it helps to speak more precisely by level of education, and to distinguish between provision in private schools and tuition in private institutions such as frontistiria, or by private tutors. Each of these sectors is a whole different kettle of fish, and taxability varies. Private one-on-one tutoring in particular can go underground in the blink of an eye - good luck collecting VAT on that.

About 7% of Greek pre-primary and primary school pupils go to private schools (see p 416 here or raw data here). The percentage falls as children grow, from 5% in lower secondary (gymnasio) to 4% in upper secondary (lykeio). By most countries' standards this is actually a small share of the population - only three OECD countries have (marginally) less privatised education systems. On last count (2012) there were ca 75,500 pupils in Greek private schools excluding nurseries and pre-schools (on which more details will follow), and an additional 11,500 in the latter.

Not only is the Greek private school population small in relative terms, it's also not growing; in fact (again, excluding nurseries) it peaked in 2003 and has fallen relatively consistently over the years, both in absolute numbers and as a share of the pupil population. Demand for upper secondary schools is positively falling in the long term. Overall, the private school population was down 12% from its peak even in 2007, and down 18% in 2012. This suggests to me that, all other arguments aside, private school fees make for a rather poor tax base, unless they are somehow a fantastic proxy for undeclared income (which I expect they are not).


But schools are not the only kind of private education out there. Private foreign language tuition is common. Over half a million Greek children were enrolled in 6,500 language schools as of 2013 - roughly one in three Greek children in education of any kind.

Then there is private tuition across a range of academic subjects, whether remedial or as a top-up for students cramming ahead of their final exams. PISA findings reveal that the majority of Greek secondary school pupils attended private lessons outside school in 2012 - 56% in the case of mathematics, though fewer when it came to other subjects. That might sound like a lot but the equivalent figure was 74% in 2009 and 2006; clearly parents cut down during the crisis. Possibly because of this, the overlap between different kinds of private lessons isn't as big as one might think. If you can get hold of the raw PISA 2012 data (I did) you can combine these figures to reveal that 72% of Greek high school kids had private lessons of some kind or other in 2012.

Shockingly, this included 69% of Greek kids in single-parent households. This is shocking because half of all single-parent households in Greece were, by that time, struggling to afford food. Let me repeat this in case it did not penetrate; in 2012 a good percentage of Greek single parents (38%, if you assume single parents have on average as many kids as two-parent households) were to some extent willing to prioritise paying for private lessons over food. One does not do this sort of thing on a whim; these people were no doubt convinced that private lessons were crucial if their children were to have any hope of getting out of poverty.

What about primary education?

There is one type of private education that is in particular demand among Greece's lower middle class: private nurseries, primary schools and pre-schools. As already demonstrated, the percentage of pre-primary school children going to private institutions doubled during the peak of the crisis (2011-12). As early as 2010, spending on primary and pre-primary education was already shooting up among the top 20% highest-earning households.

This is due to a combination of push and pull factors. The crisis forced more Greek women to become economically active, most of them working part time or only occasionally (on which much more detail here). This added, over exactly the years of the pre-primary boom, between 3 and 4 hours of childcare per week for the average household. This increase did not come from existing users taking on more hours, but from more families leaving their kids (especially those under 3) with nurseries. An even more important factor was the pilot operation of 801 all-day schools (see p 40 here) supported by EU structural funds. The pilot was meant to have a demonstration effect, with the Greek government taking over the cost of the scheme once it was convinced of its practicability and benefits. It did not, and the private primary education boom proved to be short-lived - in fact since 2012 the numbers have been slowly reverting back to normal as fewer and fewer people can afford nurseries, or alternatively fewer women can find part-time work.

Is it true that private schooling makes up for failings in the educational system?

There are two ways to approach this question. One is a matter of efficiency, as assessed by Koutsampelas (2015). I cite this study with apologies to the authors, who clearly are still working on the paper and don't want it used as is by other researchers until it has been finalised. They find [...] household willingness to pay €2,182 (annually, in 2009 prices) in 2009 and 2,517 (annually, in 2013 prices) in 2013 per school-age child for substituting state for private education. The corresponding figures for government cost per school-age child is €4,33915 and€ 3,70716 or 2009 and 2013 respectively, suggesting that from the consumers’ point of view the public provision of education in Greece might be inefficient.

You can check Koutsampelas' sums here. If you combine Eurostat's data with the OECD's figures on pupil numbers, the result is ca EUR3600 per pupil for primary and pre-primary and EUR4400 per pupil for secondary school as of 2012. Of course these figures are down from a peak of EUR4100 and EUR5600 respectively in 2009 and, assuming no change in pupil numbers in 2013, they would be 3300 and 3900 respectively.

The second way is to ask whether private school pupils do better than their state-educated peers, and why. The OECD's PISA assessment finds a persistent, statistically significant difference across all areas, with private schools performing better. However, the PISA 2009 assessment also found that, once the effects of pupils' social backgrounds, school independence and competition for pupils are taken into account, private school pupils actually do marginally worse. Adding 'independence' to the mix is not trivial though. Greek schools have probably the least discretion in deciding on their own curricula in the developed world (see IV.4.2. here).

Then there is the question of whether private out-of-school tuition makes up for failings of public schools specifically as opposed to those of the educational system. The answer is likely to be no. Going back to the PISA 2012 data, you can see that pupils in private schools still use private tutoring as much as public school students. When it comes to non-traditional subjects (ie not language, science or mathematics) they arguably use more private tuition.

UPDATE 15/11: PISA 2012 included three trick questions, in which pupils were invited to rate their familiarity with the made-up concepts of "proper numbers"; "declarative functions"; and "subjunctive scaling." Only ca. 3% of Greek pupils claimed to understand all three concepts well, but about 37% claimed at least some familiarity with all of them. This isn't bad by international standards. However, what is really interesting is the correlation between such 'overclaiming' and out-of-school tuition. Greek pupils that received out-of-school tuition in mathematics in 2012 were significantly more likely to over-claim (ca a quarter of a s.d.), and pupils that received 6 hours of out-of-school tuition per week or more were even more so - a full s.d above average. It could be reflex: when taught to the test, a pupil knows it's best to try some answer and show familiarity than none at all. It could be psychological pressure; a child taught on the insistence of parents may be eager to please. Or it could be confusion: a child desperate to keep track of new concepts may genuinely feel one is vaguely familiar, even if it is not.




Do poorer Greeks use private education?

The latest data we have suggest that, in 2010, and among Greek two-parent families with children, education made up ca. 5% of consumption spending. Among single parents, this went up to 10%. This generally includes all spending on education; books tend to be free but a wide range of accessories are not. Even so, private schools and tutoring are likely the major driver.

Eurostat doesn't break these figures up by income bracket and household composition at once, but there is a breakdown by income quintiles across the whole population here. This suggests that the top 20% of households by income spent 2.3 times as much on education as the bottom 20% in 2010 - but the differences are greatest for pre-primary and primary education, where the top 20% spent 6x as much. So we know that private education spending is skewed towards higher incomes - but is it the top end of the distribution driving this or the bottom end?

PISA confirms that private schooling in Greece is skewed towards higher incomes - more so than other countries, but suggests that this is mostly due to the exclusion of the very poor, not exclusivity to the privileged (see fig. 2.1 and 4.2 here). Private lessons and cram schools are, as we saw, widely used by lower income families and even charitable shadow education is becoming increasingly common. But again this is a story of the very poor missing out on a near-necessity, not the well-off enjoying a luxury: on PISA's standardised socio-economic status index non-users score a very significant 0.4 of a standard deviation lower than users.*

It's worth noting that Greek society is changing rapidly during the crisis. The findings of Koutsampelas (2015) suggests that Greek state schools are now receiving an influx of children from newly-poor, once middle-class families. This group dominates the flow out of private education to such an extent that the progressiveness of public education spending has actually increased during the crisis even though the class composition of state schools has widened to include better-off people.
The downside of this is that the government has rarely been able to budget for the increased demand for public schooling, leading to widespread teacher shortages and putting the Greek education budget under further pressure.

Epilogue

In Greece, attendance of private schools is rare, and falling. In fact, the crisis has pushed previously middle-class families into state education, leading to mass teacher shortages. Use of private daycare is more common and an important, if dysfunctional, contributor to labour market flexibility. But private lessons, cram schools and tutoring are extremely common, even though they too have taken a hit during the crisis. There is no suggestion that private tutoring in Greece is economically efficient; it is a bad, path-dependent solution to a poor education system. Because of this function, it's also a desperate necessity. Private schooling, on the other hand, is efficient up to a point; it does not make up for failings in public schools as such, simply for the lack of school autonomy and choice in the educational system as a whole.

The cases of private nurseries and of single parents using private tuition make me think of all the times critics have told me to put 'people over numbers' and check my figures against their (sometimes atypical, and almost always second-hand) slice-of-life anecdotes. The numbers are people, guys. The numbers are always people. Telling their story well is the same art as that of putting a tearful first-hand account into context. And if you can't be bothered to do the one, you probably can't do the other correctly either.


* [Bear in mind, PISA's index is a composite, and uses posessions, immigrant status, parents' jobs and parents' education to create a status proxy; this means it is not as variable as family income - if neither parent has become unemployed it is likely that a child's PISA status will not have changed throughout the crisis.]


PS: What if private schools do not confer an advantage after accounting for socio-economic status?

I can't know without running the OECD's regression whether school independence makes more of a difference ot the private/public performance gap than socio-economic status. But what if its impact turns out to be negligible? What would that mean? I have three pet theories.

1. Some of the Greek private school system may not be in the business of producing 'education'. It may be efficient in terms of meeting parents' actual requirements, which may not match a reasonable person's idea of 'good education.' Parents may be willing to spend money in order to ensure their kids are supervised while they work long hours; or are allowed to coast or get away with poor conduct; or are spared from mixing with 'the wrong sort' or have a chance to make their way into the elite.

2. Demand for private schooling could be compensating specifically, but inefficiently, for the lack of independence and competition among state schools. Parents may be willing to pay for a more tailored curriculum, or for non-mainstream teaching methods. This tailoring, however, may be inefficient because of Greece's fragmented geography and relative scarcity of children - which means that parents can rarely find exactly the 'alternative' education they need, and the schools may themselves struggle to find the specialist labour that they need. In other cases, parents with a demand for tailored schooling may genuinely want a good education but have preferences as to what this entails that don't prioritise academic achievement (eg they might want an education that provides religious indoctrination; or an environment that nurtures creativity).

3. Greek private schools may, by virtue of being private, have access to inferior inputs. If people who train as teachers in Greece value secure jobs more than marginal differences in pay and are willing to wait and/or relocate to get such jobs, they might prefer to pass up offers from private schools, especially if they are highly qualified on paper. Similarly, people moving into teaching (or a specific kind of teaching) after switching careers out of necessity may also be more likely to go into private schools for similar reasons. Parental effort is also an input. Parents who opt for private schooling because it helps them trade off money for working time may have less time to devote to supervising or encouraging learning.

Sunday 11 October 2015

WE NEED TO TALK ABOUT GREEK SURVEYS

UPDATE 21/11/15: Join me in protesting the Greek government's attempt to discredit and punish staff at the polling unit of the University of Macedonia; the Independent Greeks' conspiracy theories and personal vendettas have hurt to many people already. Sign the petition here.

The days leading up to the second Greek election of 2015 saw a rash of new polls suggesting Syriza and ND were neck-and-neck ahead of the second election of 2015. No less esteemed an organ than the FT was citing Greek polls to inform its readers that ND's Meimarakis (whose mustache alone should have given him away as a throwback to the party's deep past) had refreshed the party's image and was proving such 'a formidable opponent' for A. Tsipras that ND was in with a chance again.

It was, we now know, bollocks. So were surveys conducted ahead of this summer's referendum, whose errors were made that much more humiliating for pollsters by the fact the actual result fell outside even the envelope of projections published in previous days. No less successful were predictions of the January election result, which again showed a dead heat to the very end, despite a comfortable lead for Syriza in real life. The graph below shows the evolution of polls, with large circles representing the actual election results.

"ElectionMonthlyAverageGraphGreece2015" by Impru20 - Own work. Licensed under CC BY-SA 3.0 via Commons.

Now readers know how this blog works. I explore cock-up before conspiracy. And I believe in surveys. I deeply respect those who produce them. They are tools with limitations, but they are good tools nonetheless. When surveys start to malfunction our window into the real world narrows, and we become more dependent on the confirmation bias factories in our heads and in our social circles.

So what can account for the errors we've seen? The following explanations have been put forward.
  • non-response (non-contact and refusal): an increasing share of the population mistrust pollsters and this attitude is correlated with particular, non-mainstream political views. More importantly, more and more people reject landlines altogether in favour of mobiles and the internet. Finally, social and economic upheaval makes people's contacts less reliable as households fall apart, people emigrate, and domestic life is disrupted. The people affected by these phenomena are systematically different in their politics from those unaffected, and are under-represented in what should be 'representative' samples. There is certainly enough evidence that these influences are important. See for instance UK results, and US findings on the matter, and some of my own work on emigration. To put it simply, rising non-response is a symptom of social deterioration; reduced fertility rates, reduced neighborhood cohesion, precarious employment and social isolation.
  • fluid public opinion/unit non-response: It is possible that individuals are becoming increasingly unwilling to discuss their politics specifically with pollsters, either out of mistrust or out of disillusionment, and that this attitude is correlated with a predisposition to vote for non-mainstream parties. For an empirically founded overview of this argument, in Greek, see here. To this explanation should be added the fact that the context of recent Greek elections and even more so the Greek referendum was extreme; public opinion may have been genuinely fluid to such an extent that late surges can make a difference. For an argument and data in support, see HuffPo's surprisingly good coverage here. Alternatively, pollsters' failure to allocate 'don't know responses' to a likely preference (eg based on their previous voting record, which would usually have favoured Syriza) may have been at fault (an argument to that effect here)
  • abstention: with abstention on the rise, Greek survey results are becoming increasingly sensitive to people's propensity to vote and small variations in nationwide voting intention can be magnified in real elections.
  • herding: pollsters are worried about being the odd-one-out and weed out or suppress survey results that put them at odds with their peers. Nate Silver makes a good case for this here. Ironically, New Democracy staffers caught on to this theory and spent much of the eve of the last elections briefing opinion formers (of whom I am not one) that ND was indeed ahead in surveys but the pollsters were self-censoring out of fear. They got caught.
  • tampering: pollsters are in cahoots with political parties whom they insist on painting in a favourable light. There were multiple accusations of such tampering in the aftermath of the 2015 elections, reviewed in this excellent post on Dateline: Atlantis. My feeling is that this may be true of some, but not nearly enough pollsters to explain the outcomes we've seen (see below on the performance of 'non-establishment' polls)
How to deal with these problems?
  • Stick to cross-tabs. The Press Project ran its own crowdfunded survey ahead of the September elections. Despite a stated intention to compare its findings with those of other polls, in the end TPP simply chose to report on cross-tabulations only (full video here and more commentary here). Although this was likely motivated by the relatively small sample size achieved, I would hazard a guess that TPP found themselves faced with the same methodological issues as everyone else. 
  • Invite scrutiny. Pollsters ProRata (who came closest of all to the actual September election results) have opened up their survey datasets to researchers. [Or did they? See Comments section!]
  • Extrapolate from biased samples. Ironically, the innovators here were populist blogs such as Tromaktiko (see a sample here), who quickly realised how unrepresentative their readership is and developed simple rules of thumb in order to work backwards from this to a more 'normal' population. At least one company (the unknown and ungoogle-able Clear Foundations Institute) has tried to combine a telephone survey with a web survey (see here). These approaches so far seem to have failed as badly as the old-fashioned methods, but that's not to say that there aren't valid ways of extrapolating from biased samples (example here).
  • Build the sampling frame or weights matrix around referendum results. Pollsters have repeatedly claimed that their raw data were of the highest quality, but errors crept in as the data were weighted. In particular, weights based on past electoral preferences became unreliable as the Greek electorate became more fluid and deranged. A solution would be to use the recent referendum as the reference for weighting. The sampling frame approach was, for instance, suggested here. ProRata also appear to have used respondents' referendum voting record in weighting responses (as opposed to or in addition to past electoral voting record).
An even bigger problem 

Ultimately, political polls may not matter much; I've yet to see convincing evidence that undecided or tactical voters are swayed by them, although private party funding may be influenced.

More to the point, political polls will always be subject to accusations of tampering. At the back of the Greek people's heads, they are part of politics and thus to be taken with a grain of salt. But what if similar problems were impacting the surveys used to produce national statistics?

National statistics surveys have an advantage over polls in that they are allowed to be much more expensive than political surveys, they have to follow a common methodology, and are, in some cases, compulsory for the poor bastards chosen to provide a response. And while there are potentially incentives to tamper with, say, unemployment rates or poverty rates, tampering with more obscure findings like IT usage statistics is pretty inconceivable.  

It's not difficult to find the non-response rates associated with national statistics surveys. ELSTAT publishes a good deal of this information under each theme's 'methodology' section (see an example for the Labour Force Survey here). Additionally Eurostat summarises the national quality reports for each survey on an annual basis (see an example for EU-SILC here). In the graph below, I've put figures together for some of my favourite national statistics (on employment, household spending, living conditions and IT use). Plotting them all together makes for grim viewing.


First of all, it's plain to see that there is a long-term trend towards rising non-response rates, mirroring the experience of pollsters in Greece and abroad. However, this trend clearly accelerated after 2010, the first memorandum year, which gives credence to a 'social breakdown non-response' theory. Second, the years 2011 and 2012 seem to have been a particularly sharp deviation from this trend - which I believe is due to the fact that this was the period of mass disruption - e.g. Greek's sense of physical safety and food poverty were at their worst. Third, not all surveys have seen non-response develop in the same way. EU-SILC, a survey designed to tease out signs of poverty and its implications, has managed non-response a lot better, and appears to be returning to normal. How is SILC different? I am not sure - bear with me as I try to find out.

The Labour Force Survey provides even more detail on non-response. It shows, for instance, that it's the 'uncontactables' that account for the rise in non-response. They also show that major urban areas have substantially higher non-response rates. Attica, in particular, has a non-response of around 40%, twice its pre-crisis level. This should get ELSTAT worried.


One might recoil in horror from such numbers - we get our unemployment figures from a survey that has 25% non-response? WTF?! It's important to remember, however, that statisticians don't just let non-response be; they contact different households fitting the same general description (e.g. local area, age of reference person) until they fill their quota for each box on their grid. When non-response is random, or displays no systematic pattern that is related to the subject of the survey, it does not bias the resulting figures, even if it's really quite large. Greek election polls had ca. 80% non-response even back when they were accurate (data here).

Ultimately, I'm not even sure that the Syriza surge was correlated to the rise of the uncontactables. Electoral data are available here, and a regional breakdown of non-response is available, at least for LFS. The interesting thing is that the correlation between transience and Syriza vote growth is quite weak




TO BE CONTINUED