• REMEMBER THAT NUMBERS ARE PEOPLE LOOK FOR COCK-UP BEFORE CONSPIRACY • ALWAYS CITE PRIMARY SOURCES


Friday, 16 September 2016

COUNTING THE FINGERS OF THE INVISIBLE HAND

'Greece needs EUR100bn to get it out of today's recessionary spiral. We know how to release it.'

I'm wary of Big Numbers. They tend to stick regardless of their source, the methodology involved in calculating them, or any caveats well-meaning originators attach to them. Citing Big Numbers is thinking big, maybe. But it's also asking for trouble. So who came up with this one? Step forward Kyriakos Mitsotakis, the leader of Greece's opposition.

The figure, to be fair, predates Mitsotakis' ascension to the ND leadership- it was first mulled by the Hellenic Federation of Enterprises in Dec 2015, only to be resurrected by him in June 2016. And it's not hard to see where the numbers come from; here are the figures on Gross Capital Formation in Greece going all the way to 2015, and a breakdown by industry and asset is available going from 1995 to 2014. You can also look at the figures for business investment by detailed sector in tradeservices and construction and industry, although again the data only go up to 2014 (2013 if you want all of those yummy 4-digit industries!).

As far as I can tell, the HFE researchers have used the data above to compare pre- and post-crisis figures for fixed capital formation other than construction of dwellings, then calculated the shortfall between actual investment and some counterfactual extrapolated from pre-crisis numbers. This isn't a bad idea, as long as we know what we're working with. For the avoidance of doubt, this isn't 'economics' and neither is my commentary. It's playing with numbers to assist our thought process.

Extrapolating from a pre-crisis trend assumes that the level of investment by Greek businesses in the run-up to the crisis was sustainable and growth could have continued much as it had before - avoiding the excesses of 2007-8 but still on the same basic trajectory. That's unlikely. If we assume that the level of investment would remain constant at 2008 level turns out to be harsher than just mean reversion. Assuming a linear trend is more generous. Both estimates converge on a shortfall of some EUR89-EUR109bn between 2009 and 2014 (the last data available to researchers when the estimate was first produced) or EUR137bn to EUR108bn between 2009 and 2015.* That explains where the EUR100bn number came from.

An alternative might be to assume that 2008 was the peak of a credit bubble and use 'bubble free' figures that allow for a correction. I'm not going to bother with something detailed for the purposes of a thought experiment, so let's just take the 2001-08 (post-Euro) average as the benchmark for what 2008 should have looked like and assume that investment would then grow in a linear fashion based on the pre-crisis trend. This would produce a 2015 gross capital formation figure (ex dwellings) not far from the 2008 average. It would also mean a shortfall that is only about three quarters what the HFE estimated - about EUR73bn between 2009 and 2014 and EUR95bn between 2009 and 2015.

Here's what the various estimates look like when plotted against historical data:


Now if you find these counterfactuals a useful place to start, the logical next step is to apply the same kind of thinking on a sector-by-sector basis. Here a few things become glaringly obvious:
  • The big difference between my 'bubble-free' estimate and the simpler ones is precisely investment by real estate firms. By my estimate, there is no real net shortfall; the bubble has burst and that's all. By the estimates I think the HFE carried out, there is a huge shortfall. I think I'm right and they are wrong.
  • A lot of the investment shortfall is in the public sector. By my reckoning, EUR31bn of the lost EUR76bn of investment were in the broad public sector (including public administration, education and health) and about EUR21.5bn in public administration alone.
  • The other big contributors to the shortfall are shipping, sectors directly linked to domestic consumption - wholesale and retail and agriculture, and sectors linked to the real estate market. I would argue that this investment, totalling a net EUR40bn of the missing EUR76bn, cannot return while trade and the labour market remain subdued, but will likely be attracted to the sectors naturally as the former recover. 
  • This is some interesting arithmetic - the missing EUR100bn is actually EUR76bn and could in fact be mostly due to public investment cuts. I cannot be certain that all of this investment was productive or needs to be restored. It may take a much smaller, and cheaper, boost to trigger investment into much of the private sector- as long as it is allowed to reach the businesses that can put it to good use.
  • One sector's investment was, nevertheless, boosted tremendously by the crisis - "membership organisations" have been investing in commercial property and other non-housing construction in an explosion of spending that started in 2009. Since "membership organisations" or s.94 in the NACE classification include a) the church b) political parties c) professional and business lobbyists d) unions, I can come up with any combination of conspiracy theories. Unfortunately I can't get much more detail than this as Eurostat doesn't publish detailed enterprise statistics for NACE s.94.
To confirm some of this thinking I looked at the Conference Board's Total Economy Database  and its estimates of how disinvestment has contributed to the recession. Again, I have not tried to retro-engineer the Conference Board's Growth Accounting but used simple extrapolations instead. What I got what this: essentially disinvestment in capital itself accounted for just under 10 percentage points of lost growth. Most of the damage was done by a loss of multifactor productivity, which means that the Greek economy has, over the years of the crisis, become fundamentally worse at allocating resources. Surely, disinvestment has contributed to this, but it's hard to know by how much.




* I can create extrapolations up to 2015 for capital formation excluding dwellings because, even though I don't have an asset x sector matrix for 2015 I know that almost all the investment in dwellings during the crisis era has come from the real estate sector. I can therefore make a pretty accurate guess of the 2015 matrix.