What went wrong with rental voting in the 2013 German election?

By Lukas Stötzer, Steffen Zittlau & Thomas Gschwend, University of Mannheim

What is the story?

Rental voting is a specific form of strategic voting that occurs in proportional representation systems with stable coalitions.1 It supposes that supporters of a senior coalition party switch their vote in favour of the junior coalition partner if this junior partner is scoring so low in polls that is it likely to be ejected from parliament, and if this implies that the overall coalition does not get a majority of parliamentary seats.

In Germany, the pro-business liberal FDP, the traditional junior partner of the conservative CDU, has traditionally been the main beneficiary of rental votes. CDU supporters have oftentimes helped the FDP to overcome the 5% threshold. One of the most impressive examples of vote-coordination in the history of German politics occurred during the January 2013 state election in Lower Saxony. The FDP was polling at around 4-6 % during the run-up to the election but ultimately received 10% of the votes. The gap between the projected and actual results comes from the decision of many of the CDU supporters to cast a rental vote (see our previous blog entry). However, the CDU and FDP narrowly failed to gain enough seats to form the Lower Saxony government. This caused quite some stir inside the CDU. Party leaders in Lower Saxony, as well as in Berlin, felt that their party had been weakened by this strategic desertion to the FDP.

For the first time in post-war Germany, the September 2013 federal election resulted in the ejection from the lower house of the national parliament of the FDP. With its 4.8% of the votes (and only 4.2% in Lower Saxony), the party fell below the 5% threshold. This came as a big surprise to many political analysts (like us…) that anticipated rental voting. The seats brought by its junior partner would have been crucial the CDU. Without their favorite coalition partner in parliament and all the wasted votes that were casted for the FDP, the CDU now faces a costly government-formation process as they will have to enter a Grand Coalition with the social democrat SPD.

How could we explain the huge difference between the Lower Saxony and the federal election results? In this entry, we offer two competing explanations of rental voting and take a look at the empirical evidence brought by MEDW survey data.

Two competing explanations of rental voting

There are two competing explanations of rental voting. On the one hand, supporters of a senior coalition partner might choose to switch their vote in favor of the junior coalition partner by themselves. If they perceive the junior partner to be in danger and if they support the coalition as a whole, they should operate this strategic decision. The difference observed between Lower Saxony and German results should thus comes from variations in the level of support to the CDU-FDP coalition, or variations in the perceived likelihood of the FDP to fall below the 5% threshold.

On the other hand, rental voting might be driven by party campaign strategies. The difference observed between Lower Saxony and German results should thus be explained by variations in the signal given by the CDU to their supporters. In Lower Saxony in January, some CDU candidates encouraged their supporters to cast rental votes in favour of the FDP. The subsequent defeat in this state election however pushed the party leaders to send a clear signal to their supporters not to operate such type of strategic voting during the September federal elections.

Evidence from both elections

To test these two competing explanations, we use pre- and post-election panel survey data collected within the framework of the MEDW project during the Lower Saxony state election in January 2013 and during the German federal election in September 2013. Both of these surveys were conducted with only Lower Saxony citizens.

From Figure 1, we see that the FDP and the CDU-FDP coalition were just as popular in the Lower Saxony during the state (above) and the federal election (below). Only the perceived likelihood that the FDP would enter parliament is slightly lower for the German federal election (by one point on the ten-point scale, the red bar is the mean). It thus seems likely that variations in the number of rental votes effectively casted in the two elections cannot be explained by variations in these factors.










Figure 1: FDP support, CDU-FDP coalition support, and perceived probability of the FDP to enter parliament (Lower Saxony election, above; federal election, below)

To test this with more certainty, we estimate, for both elections, a hybrid multinomial-conditional logit model predicting voting behavior as a function of the factors at play in rental voting (various controls are also used). Figures 2 and 3 present the simulated probability of voting for the FDP minus the probability to vote for the CDU for CDU supporters who poorly rate the FDP. For each election, we present two scenarios: one in which the voter perceives the probability that the FDP enters the parliament as low, and one where the perceived probability is high. The results for the Lower Saxony state election (Figure 1) clearly reveals the expected pattern: The probability to cast a vote for the FDP as opposed to the CDU increases substantively for voters that support a CDU-FDP coalition, but only if they perceive the FDP to be at risk of falling below the electoral threshold (left-hand side). These voters are almost twice as likely to cast a vote for the FDP than to cast a vote for the CDU.








Figure 2: Probability to vote for the FDP (compared to CDU) as a function of levels of support to the CDU-FDP coalition (Lower Saxony election)

From Figure 3, we see that the coalition support even has a marginally negative effect on the probability to vote for the FDP (compared to the CDU) at the federal election. More importantly, we see that this effect is not influenced by how the voter evaluates the chances of the FDP to receive more than 5% of the votes.








Figure 3: Probability to vote for the FDP (compared to CDU) as a function of levels of support to the CDU-FDP coalition (federal election)


For the first time in history of post-war Germany, the FDP was ejected from the lower house of the German parliament. It fell below the 5% threshold and, contrary to what happened during Lower Saxony elections held last January, CDU supporters did not save the junior coalition partner. From our analysis, it appears that the reason for this difference does not come from the distribution of factors usually associated with rental voting. The more likely explanation is thus that the CDU campaign strategy played a crucial role in the FDP defeat. While the FDP has campaigned heavily for rental votes from CDU supporters during the last week before the election, the CDU has massively counteracted these attempts. Throughout the campaign, the CDU sent the message that their supporters should cast all their votes for the CDU if they want to keep Angela Merkel as their chancellor. Ironically, if the FDP had received 0.2 percent additional votes, this gamble would have paid off perfectly for the CDU, since it would have been able to form their preferred coalition government, in which they would have played the all-commanding role (see our previous blog entry). As it now stands, the CDU’s success in preventing rental votes resembles more the won battles of a certain Greek King, Pyrrhus of Epirus.

Follow us on Facebook and Twitter.

  1. The term ‘rental vote’ was coined by Michael Meffert in Meffert, Michael F., and Thomas Gschwend. 2010. Strategic Coalition Voting: Evidence from Austria. Electoral Studies 29: 339-349. 

This entry was posted in Blog. Bookmark the permalink. Both comments and trackbacks are currently closed.