Return to: Search Page or to: Table of Contents Vol. 19, Issue 3

David Lowery, Arjen van Witteloostuijn, Gábor Péli, Holly Brasher, Simon Otjes, and Sergiu Gherghina, "Policy agendas and births and deaths of political parties," Party Politics, 19 (May 2013), 381-407. [Available at http://ppq.sagepub.com/content/vol19/issue3/ ]

First paragraph:
The now standard model of the density or number of parties found in political party systems combines the insights of political sociologists such as Grumm (1958) and Lipset and Rokkan (1967), who emphasized the role of social cleavages, and those following Duverger (1954; Lijphart, 1990; Riker, 1982) highlighting the role of electoral institutions, especially district magnitude (M) or the number of legislators elected from a district (Cox, 1997; Neto and Cox, 1997; Ordeshook and Shvetsova, 1994; Powell, 1982; Taagepera and Grofman, 1985; Taagepera and Shugart, 1993). Despite this very real success, there are several reasons to be less than fully satisfied with the standard model. It does not, for example, account for the numbers of political parties found in political systemswith high district magnitudes or values of M. The standardmodel hides away these troublesome cases under the blanket of an arbitrary logged specification. When the logged measures of district magnitude are unpacked using a polynomial specification, number of parties does not respond to changes in the value of district magnitude after M values of 15 or so (Lowery et al., 2010), a value considerably lower than the extreme values ofMfound in Israel (120) and theNetherlands (150) and, indeed, to about a fifth of the cases used in the usual cross-sectional tests of the standard model (Cox, 1997; Neto and Cox, 1997; Ordeshook and Shvetsova, 1994). This suggests that the model is at least incomplete in terms of the party systems it can account for and that something other than electoral institutions and/or cleavages may matter.

Figures and Tables:
Figure 1. Number of elections in which 161 parties have competed, 1946-2006.
Figure 2. Predicted and actual number of parties competing by election, 1952-2006.
Figure 3. Herfinhdahl index of policy concentration in annual Queen speeches, 1952-2006.
Table 1. OLS analysis of number of parties competing in elections, 1952-2006.
Figure 4. Estimated issue party slack, 1952-2004.
Table 2. Hazard of party births and deaths predicted by lagged residual values.

Last Paragraph:
As a final thought experiment, then, we can reconsider the opposite extreme of party density -- the one party systems of the Solid South studied by V. O. Key in Southern Politics (1949). In accounting for variations in party factionalism found in the southern states, Key highlighted their differences in economic and social heterogeneity. Arkansas produced no discernible factions and a quiet, limited policy agenda because of few policy disagreements among its homogeneous white voters. The more heterogeneous Florida, in contrast, produced both a more active policy agenda and a complex form of friends-and-neighbours factionalism. In Key's view, social heterogeneity raises new issues, which in turn provides a rich foundation for competing factions. While Canon (1978) has attempted to reinterpret Key's analysis using the kinds of election rule variables cited in the standard model, we think that Key's original analysis based on social and economic heterogeneity and issue agenda complexity is far more plausible. It is certainly consistent with the results presented here. It could well be that changes in levels of policy-issue slack in party systems with low and moderate levels of M are expressed not via influencing party births and deaths, but by influencing party faction births and deaths.

Last updated November 2013