Monday, May 03, 2010

Spain's Unemployment Problem

Well, according to a popular urban legend, Spain's unemployment rate - which is the second highest in the EU after Latvia - is currently running at something just a touch over 20%. Or is it? The unemployment problem I wish to address here is not the one of how to get to grips with actually putting all these people back to work, rather it is that of untangling what exactly Spain's real EU harmonised unemplyment might be, since, to say the least of it, some strange things have been happening in recent months.

But to start with some consensually grounded facts: The number of unemployed jumped by 286,200 during the first three months of the year - using the not seasonally adjusted labour force survey methodology - and hit 4.61 million or 20.05 percent of the workforce, as reported in the conservative Spanish newspaper ABC on Tuesday of last week. And how did ABC know last Tuesday that the first quarter unemployment rate was 20.3% given that National Statistics Office (the INE) was not due to publish the official figures till Friday. Well ABC knew the number since the figure was "accidentally" posted on INE's web site for several minutes on Monday. The incident - which is reminicent of the Mr Bean interpolation on the EU Presidency Website at the Moncloa in January - was subsequently confirmed by the INE itself, who issued a statement baldly stating that a technological "incident" had made "certain data" from its quarterly unemployment study visible on its web site.

Of course, in a country as given to conspiracy theorising as Spain is, this "technological incident" has lead to all sorts of speculation - that, for example, statistical staff at the INE (in a similar fashion to those employees in the statistical office in Argentina's Mendoza province who rebelled against Nestor Kirchner's use and misuse of inflation numbers) had simply gotten tired of seeing their data massaged for political objectives, and wanted to get the politically sensitive 20%+ number published before it was changed. Indeed none other than Economy Minstry Secretary of State José Manuel Campa had to come out publically to deny that anything untoward had happen.. As far as the Ministry was concerned there had been no leak. "The Employment Survey data was not leaked", he said "there was an error. There is absolutely no evidence that the information was leaked. Spanish insititutions are quality ones. To create doubt about this seems to me to be a major act of irresponsibility. I am convinced that mistakes can be made."

Unsurprisingly, there are many who remain unconvinced. "It is becoming clearer with every passing day that what happened on Monday was not an accident, but was a leak to try to ensure that the data was not manipulated even more," declared the conservative newspaper Hispanidad. "Fortunately", they went on to say, "in our statistics office we still have professionals who are not willing to accept just any old methodology" .

In fact, just this week the mystery (or war of leaks) deepened even further, since data apparently leaked on Saturday by government sources to the EFE news agency showed that the number of unemployed registered at government agencies (a different methodology) fell by a non seasonally adjusted 24,000 in April. The data was not scheduled for release until tomorrow (Tuesday).

Now a lot of this would be none-to-worrying, and might seem like a lot of the typical to-ing and fro-ing you tend to associate with normal political debate about unemployment, except, except.... well except that something rather strange does actually seem to have been happening, and except that, well, you know, there is rather a lot of concern around about the level of Non Performing Loans in Spain's banking system, and the econometric equation used by both the Bank of Spain and the IMF (more on this in another post) for their stress tests, well, the assesment is based on projections about Spain's actual unemployment rate. So there is more than just votes at stake here.

What, then, has been going on? Well the first thing that I find strange is the fact that the unemployment numbers do not really fit with other indicators we have, like the number of people affiliated to the national social security system - which is, if you want, a measure of the number of people actually employed. Basically since last September, when you could say that the funny things happening with Spain's unemployment data got even funnier, a seasonally adjusted 189,000 people have stopped contributing to the social security system (by March, see above chart). This represents something like 1.06% of total employment, so how the hell, we might like to ask ourselves, can estimated unemployment have only risen by 0.1%? Especially when, and according to the Labour Ministry's own data, the economically active population has risen by a seasonally adjusted 35,000 over the last six months. Something, somewhere just doesn't fit here.

Especially if you look at the chart I have made of the Eurostat data as it now stands, where it seems unemployment has been completely flat for several months.

Now, since I have been very irritated by this whole situation for some time now, with the data filed at Eurostat being constantly revised, I decided to do that really tedious and tiresome thing, and go back through all the relevant press releases from Eurostat. If you want to check for yourself you can go to this page, where you will find the entire file, complete with the relevant links.

Here, for the sake of convenience, I will just produce two extracts, the latest (March) data, and the data for November 2009. (Please click on images for better viewing).

Now, as can be seen in the November file (below), unemployment had been rising steadily at a more or less even pace since the spring, and had hit 19.4% by November, which made the sort of predictions that I personally was making for unemployment going up towards the 25% mark in 2010, if not completely scientifically valid, at least not simply wild speculation.

But if we now move on to the March 2010 file (below) we will see that the 19.4% level was never actually hit (in theory), and that unemployment is supposed to have peaked at around 19% of the population, and is now, of course, about to turn down.

Of course, they may have some problems with the seasonal adjustment methodology, but in which case they should say so. On the other hand, the social security affiliates data suggests a constant employment loss of around 35,000 a month for the last several months, and that the bloodletting continues relatively unabated, which means Spain is experiencing a "real" increase in its unemployment rate of around 4% a year. And if you then apply this input to a simple linear regression model based on earlier Spanish data (which gives a factor of 0.66 to apply to this percentage rate of increase), then we might reasonably expect the rate of distressed loans to go up by something like 3% this year.

The IMF Global Financial Stability Report on the other hand, using face-value unemployment data (see this file page 54) projects NPLs at commercial and savings banks will peak at 6.3 percent and 6 percent, respectively, in 2010:Q3, and then come down to 5.1 percent and 5 percent, respectively, by the end of 2011. That is really the importance of this whole debate, since (assuming other things to be equal, which in fact aren't but we'll see about that another day) if unemployment hasn't yet peaked, then NPLs won't peak in Q3 2010, or anything like it, as the authors of the IMF report themselves point out in their adverse case scenario.

Now, there is another possible explanation for the mismatch between unemployment and job loss, and it is one that I have explored in this post here in my blog on the Spanish newspaper Expansion: it could simply be the case that people - both young Spanish nationals and migrants - have left Spain.

No comments: