This article employs the Cox proportional hazards model to discover the factors of survival of 162 party systems from 1792 to 2009. In order to avoid the endogeneity problem, the analysis employs the level of democracy as a control variable. The impact of the overall level of party system fragmentation is found negligible, even though excessive fragmentation is conducive to a higher hazard of party system termination. More importantly, systems with close competition among leading parties – including two-party systems and systems with close competition among up to four parties – are more likely to survive across time. The article introduces a methodological innovation by disaggregating the effective number of parties into two components, the leading parties’ balance and the residual fragmentation.
DumontPatrickCaulierJean-FrançoisThe Effective Number of Relevant Parties: How Voting Power Improves Laakso-Taagepera’s Index2003BrusselsCERECCenter for Research in Economics cerec working paper 2003/7
Dunleavy and Boucek (2003) Golosov (2010). The index of Dumont and Caulier (2003) while being also based on HH to an extent departs from it by replacing the proportions of seats in the calculus for NLT with the Banzhaf voting power measures but it is not a truly continuous measure because it does not allow for making distinctions among party constellations with s1 > 0.5. See however Kline (2009).