SO WHAT DID GREECE SPEND THAT DEFICIT MONEY ON? A NAIVE ANALYSIS

Some time ago, I wrote a post examining the puzzle of Greek public procurement, in which I argued that there's simply not enough dodgy public procurement in recent Greek history to justify calling all of our debt 'odious' even by a long stretch of the definition of the term.

I was called out for this by @talws, who -correctly- argued the following:

1. Α counterpoint: 70% of my personal income goes to eating, renting a house and basic expenses. 20% goes to gasoline, recreation etc and another 20% goes to gambling and 5% to binge-driniking.
In order to sustain my personal 10% deficit I have reverted to loans and credit cars, which I obviously can't pay up anymore as they are by now increasing my deficit.
According to the reasoning above, my main problem is that I eat too much and move to a cheaper flat. There is a fallacy here :-)

That hurt my fuzzy neoliberal feelings, so I thought I'd run some figures following his logic. He's right; when trying to get to the causes of supposedly odious debt, it doesn't matter what we spend on. What matters is what we spend deficit money on. Primary deficit money, to be precise. 

What I did was, admittedly, a naive analysis. I downloaded a breakdown of our spending broken down by COFOG categories, each representing a distinct purpose of government in the dramatically expanded statist universe. I figured that any type of spending that rose as a share of total primary expenditure when the primary deficit rose may have had something to do with causing the deficit. That should help us separate the wheat from the chaff - or the housing and eating from the binge drinking and gambling. I cannot disaggregate primary deficits into constituent parts because there aren't enough observations (hence this is a naive analysis), but I can run a simple linear regression and see what returns a large enough R square value - i.e. the ability to explain a substantial share of the variation in government deficits. 

COFOG figures are available from Eurostat as shares GDP, but to me that's inadequate. Realistically, ministers don't divide shares of GDP among themselves - they can't control that precisely enough. Instead they divide shares of total spending, although in some sectors, such as education, spending aspirations are typically benchmarked as shares of GDP. 

So I ran this analysis with COFOG categories expressed as % of total spending as opposed to % of GDP, which is simple enough to do as Eurostat also provides expenditure estimates as % of GDP. I also restated the deficit figures as % of total primary spending, to ensure consistency. 

Family-oriented spending and medical supplies are the types of spending best correlated with primary deficits, but agriculture and housing are also high on the list. So is the suspiciously named 'Recreation, culture and religion not elsewhere classified' - which by the way does not include church wages since they are elsewhere classified. Note that things such as survivor benefits also rank high in their correlation with the primary deficit, but the relationship with primary deficits is negative - essentially in times of fiscal relaxation governments would cut survivor benefits to fund waste. 

Now suppose we group together medical equipment, family and children, agriculture, recreation/culture/religion and housing into a 'Big Five' of suspicious spending. It would look a bit like this (watch out for the two axes):

Here' a more detailed breakdown of the Big Five:

Impressive, no? That's our culprits then. But wait - at their worst, the Big Five only accounted for some 9% of spending. The 2009 primary deficit was more than twice that. And besides, no government can cut all spending on medical supplies unless it privatises all healthcare. Similarly with other parts of the Big Five. But let's suppose we take the average of 2002 and 2003 as a 'normal' level of spending - that was the only two years in recent history when we ran primary surpluses after all. Back then, the Big Five made up 4.7% of spending. The 'excessive' spending on the Big Five is smaller than the total spending, and peaked at just 4.5% in 2009 - when, admittedly, it made up 49% of all spending on the Big Five.

In fact, here's what would have happened if spending on the Big 5 had frozen, as a share of total expenditure, at the 2001-02 levels. Remember, all figures are % of total primary spending, not GDP:

All in all, this tells me once again that there's just not enough dodgy spending to justify the 'odious debt' accusation. It may be that if we could trace figures all the way back to 1980, we'd find more evidence of odiosity, but I just don't have the figures.   

You can download my results from here. or just check them out below.