Last-minute electoral announcements promising tax breaks can successfully alter the voting behaviour of those who would benefit from the policy change. That is the central finding of research by Matteo Alpino, to be presented at the Royal Economic Society''s annual conference in Brighton in March 2016.

His study considers the 2006 Italian elections, in which six days before the vote, Silvio Berlusconi announced that he would abolish property tax on main residencies if he won. Having trailed far behind in opinion polls, his coalition caught up to the point where they almost closed the gap and lost by only 24,000 votes. This study looks at an election survey by the Italian National Election Studies to see whether these two events were linked.

People were asked their opinions twice: first, a few weeks before the announcement, and then after the election itself. Importantly, the survey also asked people whether they owned or rented their main home, an indication of whether they would benefit from Berlusconi''s declaration.

The study finds that renters generally did not change their minds in between the surveys, but the number of house owners who supported Berlusconi''s coalition rose by five percentage points. In addition, the number of votes for Berlusconi''s coalition grew by more in regions with relatively higher levels of homeownership.


Very often politicians make sensational fiscal promises during the electoral campaign, especially in the last few days before the polls open. Despite the fact that these announcements are clearly motivated by the desire to boost their odds of victory, it is not clear if they are successful in doing so. This research suggests that they are, at least in the episode under investigation.

On 3 April 2006, six days before the Italian national elections, Silvio Berlusconi, back then the incumbent Prime Minister, caught everybody by surprise when he announced on TV that he was going to abolish the property tax on main residences in case of victory. Despite lagging behind according to opinion polls, Berlusconi''s coalition eventually almost closed the gap vis-à-vis the centre-left block, and lost by only about 24,000 votes in the lower chamber.

By analysing an election survey conducted by the Italian National Election Studies (ITANES), this study show that the announcement had indeed a sizable effect on the final electoral outcome.

Participants in the survey were interviewed twice: first, a few weeks before the property tax announcement, they were asked about their voting intention; the second time, after the election, they were asked about their actual voting choice.

In other words, Mr. Berlusconi made the announcement in between the two interviews. Crucially the survey also records whether respondents are house owners, or whether they rent their main residence. As the abolition of the property tax would have led a tax break only for the house owners, but no direct private economic advantage for the renters, the announcement had the potential to appeal only to the former group.

Instead of merely looking at the temporal evolution of the support for Berlusconi''s coalition between the first and second interviews, which could be caused by several different factors beyond the property tax proposal, the research calculates the temporal evolution of support for Berlusconi''s coalition only among the house owners, that is, those affected by the policy announcement, and compares it with the renters, used as control group, given that they would not pay the property tax irrespective on the electoral outcome. Any difference in the temporal voting pattern between the two groups can be attributed to the property tax announcement.

The research shows that the number of renters willing to vote for Berlusconi''s coalition at the time of the first interview – that is, before the announcement – is not significantly different from the number of renters who actually voted for Berlusconi''s coalition in the actual election, after having learned about the announcement.

In contrast, five percentage points more house owners declared to have voted for Berlusconi''s coalition in the second interview, compared with those who intended to vote for it at the time of the first interview – that is, before the announcement.

The result from the electoral survey is corroborated by additional analysis of municipality-level data from the Ministry of the Interior and the Italian Bureau of Statistics: in 2006 the votes for Berlusconi''s coalition grew differentially more (compared with previous elections) in those municipalities with a higher house ownership rate, than in municipalities where more inhabitants are renting.

We should be cautious in generalising this result to different times or countries, but the research shows that last-minute electoral announcements promising tax breaks can indeed alter the voting behaviour of those who would benefit from the policy change.

Matteo Alpino

PhD Student & Research Fellow Department of Economics & ESOP, University of Oslo, Norway