More parties, but less power to extreme wings: it's the dual ballot, my dear!
Understood by the number of lists that participate in elections, fragmentation of the political landscape increases with runoffs in the electoral system, but the influence of extreme parties on policies that are actually implemented by winning parties decreases. This is supported and empirically verified by Massimo Bordignon (Università Cattolica) and Guido Tabellini (Università Bocconi) in Moderating Political Extremism: Single Round vs Runoff Elections under Plurality Rule, an Igier Bocconi working paper.
"The reason," write the two economists, "is that the dual ballot reduces the bargaining power of the extremist parties, that typically appeal to a smaller electorate." In single round elections lesser and extremist parties running autonomously can win the most ideologically-similar votes. This takes votes which are essential for the final success away from the moderate large parties. In runoff elections this threat is empty, provided that when the second vote is cast some extremist voters are willing to vote for the closest moderate (rather than abstain). Observations confirm that almost half of electors of the parties excluded from the second ballot do come back to vote at the runoff.
When the bargaining power of smaller parties is limited, larger parties are not interested in joining forces with them before elections, which increases the number of formations running for election. But they are not interested in inserting smaller parties' extremist programs on their electoral platforms either, which brings about more moderate policies after elections.
After building a model to predict the described outcome, Bordignon and Tabellini test it with data regarding municipal elections in Italy. This data lends perfectly to the verification because a 1993 reform introduced different systems for municipalities smaller than 15,000 inhabitants (single round) and for cities with more than 15,000 inhabitants (runoff). Using data prior to and after 1993 for Lombardy municipalities with similar sizes (a little under or over the 15,000 inhabitant threshold), the two authors observe that the introduction of the reform coincided with a reduction in the number of lists in smaller municipalities that adopted the single round. In larger cities, the number of lists increases, but not the number of candidates for mayor, each of which can be supported by more parties.
The influence of extreme parties on policies adopted by winning parties is tested using another data set, which involves the municipal real estate-tax (ICI) rate on commercial real estate. Fixing this rate is an important choice for municipal finances and, unlike the rate for residential real estate, it is strongly influenced by political ideology, with the right preferring to keep the rate low and the left to keep it high. Where extreme parties have a higher influence, each change in city government should correspond with a clear change in the rate, while changes should be marginal when extreme parties do not influence city politics. Bordignon and Tabellini would therefore expect higher rate volatility in cities with single round electoral systems (where extreme parties have more negotiating power and a higher possibility of influencing adopted policies) and less volatility where a runoff system is in force. The data analyzed confirms this hypothesis.