Graphs of the week: Tax evasion and votes for Syriza

Over the past year many commentators on the Greek political situation have claimed that Syriza, the radical left party that surprisingly won the Greek elections in 2015 (twice!), drew much of its support from tax-evasion areas. In other words the tax-evading upper-middle class voted for Syriza as they wanted to continue with their tax evasion (an endemic problem in Greece) and didn't want to accept the EU-imposed bailout deal which necessitates that Greek authorities clamp down on tax evasion. 

In a long overdue post I will briefly look into whether or not there is any evidence to back this claim. Below I show two sets of maps. The upper map shows results of the 2015 general elections held in January (right), and the bailout referendum (left) held in July, while the lower map shows the share of tax evasion per municipality (I've used this map before from a paper by Artavanis et al.). 

Before we make the visual comparison, a quick reminder on the tax evasion part from my earlier post: 

"...Their main finding is that contrary to popular belief that only the super-rich are avoiding to pay their taxes, it is actually the middle classes (the upper middle classes to be precise) which are dodging their taxes. This makes a lot of sense, as 30bn euros is too much to link it only to the super rich. The problem must be more widespread. And it is. Basically across the Greek society. In the map ... you can see that tax evasion is almost evenly distributed geographically...

The authors note that the primary tax evaders in Greece are doctors, engineers, private tutors, accountants, financial service agents and lawyers ... "Testing the industry distribution against a number of redistribution and incentive theories, our evidence suggests that industries with low paper trail and industries supported by parliamentarians have more tax evasion."

Going back to the 'tax-evaders voted for Syriza' argument. Let's compare the two maps visually (this is how such arguments are usually formed). If we look at, for example the island of Crete and the greater Athens region, the story clearly holds (even as a simple correlation). Even in the circled area (Larissa, an area that has the largest number of Porsche Cayennes in Europe), the voters supported Syriza and rejected the bailout deal. However, one can look at the map the other way around - in the Peloponnese peninsula, where tax evasion also tends to be high, Syriza didn't do so well, nor was the bailout deal rejected. Also I can easily spot a number of areas where tax evasion is low, and the votes for Syriza and against the bailout were high. This still doesn't prove anything.

Producing arguments based on a simple visual comparison of maps like these gets you nowhere. This is pure cherry-picking. You see what you want to see. The best way to actually test this is to take the tax-evasion dataset produced in the paper by Artavanis et al. and regress it against both election results, in addition to a number of country-specific controls (demographic, socio-economic, political). Using surveys is a bit harder to uncover this relationship as no one in their right mind will willingly confess to be a tax evader (even when people are arrested for tax evasion they still very often find ways to justify themselves by saying they didn't actually broke any law). 

To me the only thing obvious from the comparison of these two maps is that tax-evaders have NO causal effect on votes for Syriza or the votes for or against the bailout. As I've noticed in the earlier post, the tax evaders are evenly distributed geographically. Votes for Syriza not as much. The whole country has a problem of tax evasion (perhaps the paper even underestimated the total amount by imperfect measurement - it was an estimate after all (a good estimate but an estimate nonetheless), there's no actual data to measure this), so linking this to voting patterns in naive without a serious, in-depth analysis of actual voter choices in Greece.