Dominant Party Regimes in Canada and Australia: Going the Way of the Dodo Bird?
Paper prepared for delivery at the Annual Meeting of the American Political Science Association in Seattle, WA, September 1-4, 2011. Copyright by the American Political Science Association
Steffen G. Schneider
Amir Abedi
TranState Research Center University of Bremen P.B. 33 04 40 28334 Bremen Germany
Department of Political Science Western Washington University 516 High Street, MS-9082 Bellingham, WA 98225-9082 USA
Phone: Fax:
+1-360-650 4143 +1-360-650-2800
+49-421-218 87 15 +49-421-218 87 21
[email protected]
[email protected]
Draft version – Please do not quote – Comments welcome
Electronic copy available at: http://ssrn.com/abstract=1910276
Introduction The existence of dominant party regimes – in which one and the same party controls government over an extended period of time, whether alone or as the major player in a coalition – is usually depicted as normatively objectionable and empirically puzzling, as something that is “not supposed to happen” (Pempel 1990: 5-6), in the context of liberal democracies. The focus of the literature on single-party dominance is very much on a limited number of cases such as Japan, Italy, and Sweden; the phenomenon is supposedly more prevalent in continental Europe, including Scandinavia, than in Anglo-Saxon and Westminster-type countries. Yet dominant party regimes are perhaps not as anomalous in Western democracies as it may appear to casual observers, especially if the subnational jurisdictions of federations are considered alongside national party systems. If we consider subnational jurisdictions together with the national level, single-party dominance in federal regimes turns out to be a relatively frequent and substantively important phenomenon. After all, the partisan complexion of subnational governments may have a strong impact on federal institutions and decision-making processes, and hence on the political dynamics and democratic quality of entire federal regimes (Weaver 2004; Abedi/Schneider 2010). In this paper, we explore the incidence of dominant party regimes in two countries – Australia and Canada – that combine federalism with Westminster-type institutions. The claim that single-party dominance is less frequent in the Anglo-Saxon world is partially belied by these cases, and especially by Canada and its provinces. Furthermore, our analysis of provincial and state party systems and governments over more than a century reveals a trend, namely that single-party dominance may well become a thing of the past. We discuss possible explanations for this trend and for the greater prevalence of dominant party regimes in Canada than in Australia. An original dataset comprising all subnational election results and government formation outcomes in the provinces and states of the two federations since January 1, 1901 (the date of the formation of the Commonwealth of Australia) and a dataset of government episodes of varying length – defined by the identity of the sole or largest government party – are the empirical basis of our study. The next section introduces and operationalizes our definition of single-party dominance, which departs from the extant literature by drawing on the game theoretical concept of dominant players, power indices, and survival estimates. This operationalization is then used for a number of inferences on the prevalence, scope, and nature of regional dominant party regimes
1
Electronic copy available at: http://ssrn.com/abstract=1910276
in Australia and Canada. Finally, we review extant hypotheses on the rise and fall of dominant party regimes and discuss their plausibility against the backdrop of our cases. While we suggest that there are a few universally applicable factors that make it more difficult to establish dominant party regimes in recent times, our data suggest that there are few individually necessary or sufficient causes of single-party dominance, and that nationally or perhaps even regionally specific configurations of variables seem to play a particularly important role in the rise and fall of dominant party regimes – a finding that in our view underlines the necessity of further multimethod research on the topic.
Single-Party Dominance: Definition, Measurement, and Descriptive Inferences Applying the standard definition of single-party dominance in parliamentary systems (Boucek 1998: 103; Cox 1998: 238; Weaver/Rockman 1993: 20) to the subnational level, we qualify a party as dominant if it controls the government of a state (Australia) or province (Canada) over an extended period of time, whether alone or as the most powerful member of a coalition.1 Hence we are primarily interested in executive rather than electoral or parliamentary dominance (Boucek 1998: 105-8), treating opposition status as a sufficient indicator of non-dominance. Yet even the largest government party may or may not be dominant – depending, first, on its relative bargaining power vis-à-vis coalition partners and other parliamentary competitors, and secondly, on the length of its government experience.
Operationalization and measurement of the power dimension: The power dimension is usually measured by way of vote or seat shares of the relevant party, or indicators derived from them; we provide an extensive critique of these approaches – which, in a nutshell, tend to be rather ad hoc and do not explicitly measure relative power – elsewhere (Abedi/Schneider 2010). In our own research, we draw on the dominant player concept and indices of voting power to operationalize the first dimension of our definition.2 The game-theoretical literature has modeled processes of
1
Our definition does of course not imply that a party dominating one or more subnational jurisdictions will also be dominant at the national level; subnational dominant party regimes are “embedded” and may not always have a strong impact on national politics. Yet national and subnational cases of single-party dominance can be understood and examined as instances of essentially the same phenomenon, drawing on the same definition, even if the range and weight of explanatory factors is likely to vary between the national and subnational levels. 2 We are grateful to Patrick Dunleavy and Jean-François Caulier, participants in the 2005 ECPR Workshop on Dominant Parties and Democracy, for bringing the dominant player concept and power indices to our attention.
2
decision-making and government formation in parliamentary systems as weighted majority games with at most one so-called dominant player. The power structures of dominated and nondominated games differ qualitatively. Since dominant player status is merely an a priori indicator of power that neglects the effects of ideological polarization and policy-based coalition formation, a party enjoying it may nevertheless be excluded from government. Yet a dominant player has strictly more options to form a winning coalition than any of her competitors. This comparative advantage in government formation processes should turn into a particularly credible exit threat vis-à-vis less powerful coalition partners once a dominant player joins government, or into a similarly credible “disruption” potential where it is excluded.3 Some minimal vote share (electoral dominance) is neither a necessary nor a sufficient precondition of dominant player status.4 A plurality of seats and a weight (seat share) of at least half the (50 percent + x) majority quota of parliamentary games are necessary but not sufficient preconditions. Among the members of a government coalition only the formation with the largest weight may thus be a dominant player. Even a government party with a plurality of seats is no more powerful than some or all of its coalition partners and competitors, though, if it lacks this status. By contrast, a dominant player with a majority of seats is a “dictator” and its competitors are entirely powerless “dummies.” In short, a party’s relative power may be much lower or higher than suggested by its weight.5 The dominant player concept is easily linked with power indices based on the related idea that power in parliamentary games “rests on how often [a party] can add [its] votes to a losing More detailed and technical presentations of the game-theoretical background can, for instance, be found in Deemen (1989: 316-25) and Roozendaal (1992: 6-11). 3 If the exclusion of dominant players from government were anything but rare and short-lived in the real world, one would of course have to reconsider the usefulness of this a priori concept for empirical research. Yet data on national elections and party systems (Deemen 1989: 314), as well as our own findings, suggest that this is not the case. The rational and unitary actor assumptions implicit in the concept also seem unproblematic in the context of parliamentary systems. 4 Moreover, the presumptive consequences of single-party dominance do certainly not depend on the way it was achieved. Hence there is no reason to exclude cases that owe their status to a manufactured or even artificial plurality or majority of seats. 5 Thus the power differentials between dominated and non-dominated games, or between a dominant player and its competitors, are not captured by indicators of a party’s relative size or opposition fractionalization either (as suggested, for instance, by Blondel 1968: 196; Sartori 1976: 197-9; Ware 1996: 159-60, 165-6), and they may even be obscured by it. A party may clearly “outdistance[…] all the others” (Sartori 1976: 193) in terms of its seat share without being a dominant player or a dictator, and vice versa. By the same token, and despite their intuitive plausibility, measures of opposition fractionalization do not appropriately convey the power structures of parliamentary games. A dictator, for instance, is not made any less powerful by a divided opposition than she is in a two-party scenario. Moreover, some of the literature inappropriately uses opposition fractionalization both as a definitional criterion and as an explanatory variable, which obviously yields tautological findings.
3
coalition so that it wins” (Leech 2002: 5). The normalized Banzhaf index employed here (in line with Caulier/Dumont 2010) takes values between 0 for dummies and 1 for dictators. The values of dominant and non-dominant players in games without a dictator vary with their number and relative weight. A government party has a “power surplus” if its own value is larger than any other player’s – this is the case if and only if the party is a dominant player, and the power surplus of a dictator is, again, 1. If at least one other player has the same value, the power surplus is 0, indicating a non-dominated game. Finally, the government party with the largest weight – and a fortiori, any smaller coalition partner – has a “power deficit” if the dominant player is in the opposition.6 In the subsequent analysis, we examine a population consisting of the sole or major government party in each parliamentary game (i.e., each instance of government formation) and divide it into four groups of positive and negative cases:
The party is a dictatorial dominant player, with a majority of seats and a power surplus of 1 (D1).
The party is a non-dictatorial dominant player, with a plurality of seats and a power surplus of 0 < x < 1 (D2).
The party – which may or may not have a plurality of seats – has a power surplus of 0 (D3), and no dominant player exists.
The party – which holds less than a plurality of seats – has a power deficit of -1 < x < 0 (D4), and the dominant player is in the opposition.7
Our operationalization of single-party dominance, then, combines the requirement of government membership with the necessary preconditions of dominant player status. The proposed ordinal ranking captures non-arbitrary, genuinely qualitative power differentials, and the D1 and D2 scenarios represent types and gradations of single-party dominance for our purposes. A player in the D1 category can govern alone while players of the D2 type may be found in coalition or minority governments (Bogaards 2004: 188). Finally, players of the D3 and D4 type are 6
To be precise, the index value of the largest government party and the highest value achieved by any of its coalition partners or competitors are used for the calculation of the power surplus or deficit. 7 There is a minor complication: A party may have a power surplus while having a weight (seat share) below the q/2 (half the majority) criterion for dominant player status. Thus having a power surplus is a necessary but insufficient precondition of being a dominant player, and conversely, a power surplus value