The 2018 Italian election had a notable geographic split in voting behaviour, with Lega having more support in the north of the country and the Five Star Movement proving more successful in the south. Monica Langella digs deeper into the regional variations underpinning the result by carrying out an analysis of the link between local socio-economic factors and support for the country’s four main parties.
Orta San Giulio in Piedmont, Italy, Credit: Leonardo (CC BY-NC-ND 2.0)
Although Italy is a country with a long tradition of geographical segmentation, the pattern of support in the Italian election on 4 March was more sharply divided than usual, with a concentration of votes for Lega in the North and the Five Star Movement in the Centre-South. Wage disparities have been proposed as an explanation for this pattern, paired with the main economic proposals of both Lega and the Five Star Movement. Their electoral campaigns clearly tried to tap into popular discontent, with some hot topics including immigration control, taxation, and unemployment.
To assess how local socio-economic conditions affected the distribution of votes, I have analysed the vote shares of each of the four main parties, Forza Italia, Lega, the Five Star Movement, and the Democratic Party. Using the number of votes each party received at the constituency level for the lower chamber of the Italian Parliament (the Camera dei Deputati), standardised by the size of the electorate, I ran regressions based on the socio-economic characteristics of each area, assessing the percentage point changes in voting support for each party with respect to the previous Italian election in 2013.
The dimensions I used in the analysis were:
- Economic conditions and levels of deprivation: To capture this dimension, I used the provincial rate of unemployment, the regional share of people at risk of poverty, regional life expectancy at birth, and the crime rate at the provincial level.
- Migration: To capture this I used the share of extra EU regular migrants and asylum seekers assigned to each province.
- Levels of trust: This incorporated the share of people at the regional level who report that they generally trust others.
- Demographics: For this I used the share of people with university degrees and with high school degrees, and the share of the working age population – all at the constituency level – with binary controls for constituencies with large cities (above 300,000 people) and for macro areas (south-centre-north).
It should be noted that this analysis is largely descriptive in nature and, although informative, it does not determine any causal effect of socio-economic conditions on voting. For the sake of brevity, tables with detailed results are not reproduced here, but these are available on request.
The results of the analysis indicated that support for Forza Italia, led by Silvio Berlusconi, was not particularly sensitive to any of the controls. With respect to 2013, Forza Italia has seen its vote share decrease by around 5.7 percentage points on average. However, Forza Italia vote shares decreased less where unemployment rates and trust levels had grown. On the other hand, vote shares for the party decreased more rapidly in areas where crime rates had increased.
The campaign run by Lega’s leader, Matteo Salvini, was heavily focused on migration control, security, and on the need for a simpler taxation system that would lower average tax rates. Nevertheless, support for Lega does not appear to have had any link to higher rates of regular migration or to a higher presence of asylum seekers in a given area. The relationship between crime rates and support for Lega was also negative, suggesting that despite the apparently effective campaigning by the party on these topics, the reason for their success likely lies somewhere else.
The vote share for Lega was higher in slightly less deprived and non-urban areas, but support was lower in areas with more highly educated people and in areas with a lower level of trust. Controls for macro regions were strongly significant. Lega increased its vote share by 9.6 percent on average from 2013, however this change was lower in places where crime and migration rates had increased, while it was higher where unemployment had increased.
Five Star Movement
Luigi Di Maio and the Five Star Movement ran a campaign that emphasised the need for some form of general unemployment benefit. In line with the party’s history, the campaign also had notable anti-establishment elements. The analysis showed that the party had higher levels of support in more deprived areas, particularly those where life expectancy is lower and crime rates are higher. The relationship with unemployment was also positive, although not statistically significant. Support for the Five Star Movement was also higher in areas with low levels of trust, and where the share of regular migrants is lower.
Overall, the party increased its share of the vote by 4.7 percent on average. This increase was higher in areas where the numbers of people at risk of poverty have increased, and where life expectancy has fallen. The vote share also increased more in areas where the crime rate had risen. A change in levels of trust also has a positive relationship with a change in the Five Star Movement’s vote share.
The poor performance of Matteo Renzi’s Democratic Party was one of the most striking features of the 2018 election. The result has been interpreted as a personal failure on the part of Renzi, but also as a symptom of a more general crisis affecting social democratic parties across continental Europe.
In my analysis there were some signals that the Democratic Party’s poor performance can be partly read in tandem with the rise of anti-establishment sentiment. The vote share for the party was lower in areas where the risk of poverty is higher. The relationship with the unemployment rate was also negative, although insignificant. But the rise of anti-establishment sentiments may not be the full story behind the Democratic Party’s decline. The party also had a lower share in areas where life expectancy rates are higher and crime rates are lower, as well as in areas where the share of people with university degrees is lower.
Overall, the Democratic Party lost an average of 5.1 percentage points of support from 2013. The change was quite geographically dispersed and it does not appear to be strongly related to any of the socio-economic conditions studied, besides migration (places where regular extra-EU migration has increased registered a lower decline in support for the party) and the risk of poverty (regions where the risk of poverty had increased registered a higher vote loss for the party).
In general, the geographical division of the country that is evident from the Lega vs Five Star Movement North-South divide is apparent in my analysis, and it is strong even when controlling for a large number of socio-economic measures (alternative measures have been experimented with, but none of them appeared to have a strong impact on votes).
This analysis suggests that, on the one hand voting for the Five Star Movement appears to have been driven by adverse economic conditions and a lack of generalised trust, and therefore seems to fit with the definition of an anti-establishment and ‘protest’ vote, which was perhaps in direct competition with the centre-left. On the other hand, the vote for Lega does not fit this pattern so readily, and, although Salvini’s campaign heavily emphasised certain topics like migration, this does not seem to have been the driver of support at the local level, while a stronger relationship is observed for measures of trust.
Please read our comments policy before commenting.
Note: This article gives the views of the author, not the position of EUROPP – European Politics and Policy or the London School of Economics.
Monica Langella – LSE
Monica Langella is a Research Officer in the Centre for Economic Performance at the London School of Economics.
Very interesting Thanks for sharing this analysis, after every Italian read the simpler version (“Northern people voted for Lega, who promised to lower taxes, while Southern people voted for M5S, who promised a universal income support”). Beyond being more figures-based and articulated, it gives a more detailed overview.
I was just wondering what results there would be in case, instead of the actual number of migrants, the perceived number of migrants was used in the regression.
Is your model available on github or elsewhere? Did you use ecological regression? I am planning a similar research in the recent Hungarian elections & looking out for sound modelling between votes & aggregated social data.