The Project

This is a joint project with Lídia Farré from Universitat de Barcelona and Francesco Fasani from Queen Mary University London.

In May 2011 massive protests and ongoing demonstrations in public squares shook Spain. Even though demonstrators formed a heterogeneous and ambiguous group, they shared a rejection of unemployment, welfare cuts and Spanish politics. Motivated by these protests we asked ourselves: What are the social effects the economic crisis? What has happened to health, crime, migration, violence, education, etc. in the most affected groups? This webpage serves to circulate some of the findings of the extensive data-collection effort which we financed with a RecerCaixa project.

Unemployment in Spain

The economic crisis of 2007 had severe consequences on the Spanish labor market, as the unemployment rate escalated from around 8% in 2007 to 25% in 2011. The shock to employment was highly concentrated on the construction sector. Over 60% of all jobs in construction were lost between 2008 and 2013. In some provinces this was a substantial part of all available jobs.  The Figure below shows the evolution of the unemployment rate before and during the crisis. The skyrocketing of unemployment rates after the crisis hit was visibly more severe for Spanish regions with large levels of construction. Unemployment has remained high until today with only a subtle fall most recently.

Unemployment in the Spanish Crisis

What made and makes the Spanish particularly dramatic is that the probability for those who lost their jobs of finding new work was extremely low. This is shown in the following Figure which shows the short- and long-term unemployment rate. Long-term unemployment (the blue line) starts to rise exactly one year after short-term unemployment. The large marjority of those who lost their jobs in the period 2007-2009 did not recover them so that they are now long-term unemployed.

Long term unemployment trap in Spain

The explosion of unemployment in the crisis, its shocking persistence and its concentration in construction allows us to understand the effect of unemployment on the various outcomes we are interested in. Our empirical strategy was to look at individuals or regions who were most exposed to the decline to employment in construction.

Effects on Health

The main part of the project was dedicated to look at changes in health due to crisis. In order to study health we collected data on deaths by cause in the 52 Spanish provinces, population data and all available waves from the National Health Survey.

Feeling Useless

In Farré et al (2015) we look at the National Health Surveys, in particular we look at questions on mental health in the two last waves of the survey, 2006 (just before the crisis) and 2011. This survey reports self-assessed general and mental health status and diagnosis on illnesses, such as chronic back pain, chronic headache, heart attack and stroke. We looked at most of these outcomes but found only correlations between unemployment, a general health measure and a whole myriad of questions that relate to mental health. People who were unemployed were much more likely to report feeling worthless, useless to society and unable to make decisions. All indicators for mental disorders. It was therefore no surprise that these individuals were much more likely to report being diagnosed with mental disorders as well.

In order to understand whether it was unemployment causing mental disorders we used the exogenous shock in the construction sector as an instrument for unemployment in the estimation of the effect of unemployment on mental health. This method hinges on the size of the construction sector in a particular cell (for example: male/under 40 years old/in Barcelona) in 2006 being a strong predictor of the subsequent unemployment rate increase when the financial crisis hit. What we are identifying this way is the effect of being hit by an exogenous shock that made employment opportunities disappear rapidly, i.e. we estimate the effect of being pushed into long-term unemployment for those who were previously employed.

Our results reveal that a 10 percentage point increase in unemployment driven by this exogenous shock increased mental disorders by around 3 percentage points. This is a very large effect which is also much larger than the basic correlation we observe. However, this finding was corroborated by trhe detailed mental health questionnaires which show a general worsening on several dimensions. Individuals reported feeling under strain and that they did not feel they were playing a useful part in things. They were more likely to report that they were unable to concentrate or to overcome their difficulties.

In summary: Our study provides evidence for a decline in mental health in the aftermath of the economic recession in Spain. The decline is identified from a group which is caught in long-term unemployment since the collapse of the Spanish construction sector. This implies that individuals in this group were caught in unemployment for reasons beyond their own control. Mental health in this group seems to have deteriorated dramatically.

The Puzzling Changes in Suicide Rates

In the years following the financial crisis of 2007, media reports and academic articles alike have warned of the disastrous social repercussions of the Spanish economic crisis. One issue that has received particular attention are suicides. Using the data we collected on suicides we tried to confirm previous papers. Our idea was particularly to see whether our instrument would have power in this context. To our surprise we found that previous research on the paper had reported only sub-sets of the data.

The full picture is discussed in this note. On the one hand, our results suggest that the peak in suicide rates around 2008 could indeed be linked to the economic shock. On the other hand, this does not help us understand the higher rates in the period 2002-2004 or the decline after 2009. It also does not necessarily explain an explosion in the suicide rate after 2011. What is it that drives, mostly men, into suicide 5 years after the start of the economic crisis? It is possible that the large effects on mental health found in Farré et al (2015) provide an explanation. It is not unemployment per se but desperation linked to the persistence of unemployment that triggers anxiety disorders or depression. Unemployment duration or debt levels of a small group might therefore be more relevant in explaining suicides than general unemployment levels. In any case the links are not straightforward.


The Spanish Crisis in the News

An important part of the project was to collect news data to understand whether certain news topics systematically received more attention as a result of the rise in unemployment in the Spanish provinces. For this purpose, we downloaded more than 850,000 newspaper articles from Lexis Nexis. The archive allowed us to search for name of the 52 Spanish provinces in the title of articles in the newspapers “El Mundo” and “El País” for the period 2002-2014.

The resulting text was cleaned and put into a form such that each word, for example “woman”, could be counted in each article. We then looked at counts of words relative to all words written in province in a year to see whether a topic was expanding with unemployment (keywords we looked at were related to: crime, conflict, emigration, mental health, violence, divorce and unemployment). To our surprise we found very few consistent relationships with unemployment. What we did find, if anything, were negative relationships with violence and violence against women.

This was not, however, because the reporting data in general was useless. We found an extremely strong relationship between reporting on unemployment and the unemployment rate in Spanish provinces. Interestingly, reporting on unemployment anticipated the actual increase of unemployment by at least a year. In other words, reports on unemployment increased considerably in 2006 already. This is a finding which we aim to explore in further research as it is consistent with findings by Macroeconomists like Valerie Ramey (2011, QJE), for example. However, typically news data is used in situations in which state actions are anticipated by the market. In our case, news reports were anticipating a change which was not widely anticipated by banks, the government or economists.