UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Connected by purse strings : fiscal decentralization and the nationalization of provincial party systems… Bettencourt, Daniel 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2017_november_bettencourt_daniel.pdf [ 1.24MB ]
JSON: 24-1.0355535.json
JSON-LD: 24-1.0355535-ld.json
RDF/XML (Pretty): 24-1.0355535-rdf.xml
RDF/JSON: 24-1.0355535-rdf.json
Turtle: 24-1.0355535-turtle.txt
N-Triples: 24-1.0355535-rdf-ntriples.txt
Original Record: 24-1.0355535-source.json
Full Text

Full Text

CONNECTED BY PURSE STRINGS: FISCAL DECENTRALIZATION AND THE NATIONALIZATION OF PROVINCIAL PARTY SYSTEMS IN CANADA by  Daniel Bettencourt  B.A., Carleton University, 2015  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF ARTS in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Political Science)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  September 2017  © Daniel Bettencourt, 2017  ii Abstract While some provincial party systems in Canada diverge significantly from the national level, others are relatively congruent with the federal party system. Congruence is one aspect of provincial party system nationalization. Why does provincial party competition mirror federal party competition in certain provinces but not in others? Some scholars theorize that strong national governments encourage the same parties to be competitive on both levels, while the decentralization of power to regional governments encourages electoral patterns to diverge between arenas. However, this relationship has been underexplored and warrants further examination. This thesis tests whether two aspects of fiscal decentralization, provincial spending power and fiscal dependence on the federal government, encourage provincial party system nationalization and congruence. No concrete evidence of this relationship is found, suggesting that fiscal decentralization does not encourage divergence between federal and provincial party systems or increase support for non-statewide parties on the provincial level.  iii Lay Summary In some Canadian provinces, the parties that are competitive on the federal level are also competitive on the provincial level. In other provinces, different parties are competitive in provincial and federal elections. Why does provincial party competition mirror federal party competition in some provinces but not in others? Some scholars theorize that strong national governments encourage the same parties to be competitive on both levels, while the decentralization of power to regional governments encourages electoral patterns to diverge between arenas. However, this relationship has been underexplored and warrants further examination. This thesis tests whether two aspects of fiscal decentralization, provincial spending power and fiscal dependence on the federal government, encourage the same parties that are competitive federally to also be competitive provincially. No concrete evidence of this relationship is found.   iv Preface This thesis is original, unpublished, independent work by the author, Daniel Bettencourt.    v Table of Contents Abstract ..................................................................................................................... ii Lay Summary ............................................................................................................ iii Preface ....................................................................................................................... iv Table of Contents ....................................................................................................... v List of Tables ........................................................................................................... vii List of Figures ......................................................................................................... viii Acknowledgements ................................................................................................... ix Chapter 1: Introduction .............................................................................................. 1 1.1 A Note on Terminology ............................................................................................. 5 Chapter 2: Theory and Literature............................................................................... 6 2.1 Regional Authority and Second-Order Elections .....................................................6 2.2 Regional Authority and Nationalization .................................................................. 8 2.3 Other Explanations ................................................................................................. 12 2.4 The Nationalization of Regions ............................................................................... 13 2.5 Fiscal Dependence ....................................................................................................15 2.6 Nationalization and Fiscal Dependence in India................................................... 17 2.7 Theoretical Concerns .............................................................................................. 19 Chapter 3: Measurement, Hypotheses, and Models ................................................ 23 3.1 Dimensions of Fiscal Decentralization .................................................................. 23 3.2 Hypotheses .............................................................................................................. 24  vi 3.3 Dataset ..................................................................................................................... 25 3.4 Dependent Variables ............................................................................................... 26 3.5 Independent Variables ............................................................................................ 27 3.6 Control Variables ..................................................................................................... 30 3.7 Estimation Models ................................................................................................... 32 Chapter 4: Results and Discussion........................................................................... 34 4.1 Results ...................................................................................................................... 34 4.2 Discussion ................................................................................................................ 39 4.2.1 Theoretical Explanations ............................................................................................... 39 4.2.2 Methodological Explanations ........................................................................................ 42 Chapter 5: Conclusion .............................................................................................. 44 References ................................................................................................................ 45 Appendix A ............................................................................................................................ 49 A.1 Data Sources......................................................................................................... 49 A.2 Initial Calculation of Independent Variables .................................................... 49 A.3 Further Adjustments to Independent Variables ................................................ 50   vii List of Tables Table 1: Correlation Between Provincial Expenditure and Transfer Revenue ..................... 33 Table 2: Impact of Fiscal Decentralization on Provincial Party System Nationalization .... 34    viii List of Figures Figure 1: Provincial Share of Federal and Provincial Expenditure, 1960-2015 ..................... 29 Figure 2: Transfers as Share of Provincial Revenue, 1960-2015   .......................................... 29 Figure 3: Provincial Expenditure by Dissimilarity (Distance from Provincial Average) ..... 36 Figure 4: Transfer Dependence by Dissimilarity (Distance from Provincial Average) ....... 36 Figure 5: Average Provincial Expenditure by Average Provincial Dissimilarity .................. 38 Figure 6: Average Provincial Transfer Dependence by Average Provincial Dissimilarity ... 38   ix Acknowledgements I thank my supervisor, Christopher Kam, for his attentiveness and guidance throughout this process. I also thank Richard Johnston for providing me with data and direction in the early phases of this project. Finally, I thank my family, who had the unfortunate privilege of picking up the phone every time my anxiety got the better of me this year. It means the world.  1 Chapter 1: Introduction Referred to by Blake (1985) as “two political worlds”, the division between Canadian federal and provincial politics is uniquely distinct. This is demonstrated through Canada’s provincial party systems, which are relatively disconnected from the federal party system. For instance, it is not uncommon for a major party in the federal arena to be a comparatively small player in provincial elections. An extreme example of this occurred in British Columbia, with the Conservative Party securing 47 percent of the province’s vote in the 1984 federal election but only 1 percent of the vote in both the 1983 and 1986 provincial elections (Jeffery and Hough 2003). Parties that dominate the provincial landscape may not exist on the federal level. A current example is the Saskatchewan Party, which has held power in the province for nearly 10 years despite never competing for federal office. On the other side of the spectrum, certain provincial party systems are highly symmetrical with the federal system. This is notable in Atlantic Canada, with the Liberals and Conservatives continuing to dominate the region with relatively few exceptions (Adamson and Stewart 2001). The distinctiveness of Canadian provincial politics becomes clear when placed within the literature on multilevel elections. The dominant view of regional elections is that they are second-order, meaning that they are largely determined by national criteria (Reif and Schmitt 1980; Reif 1984). While often subject to qualification, the fundamentals of second-order election theory have been supported in the literature (Van der Eijk et al. 1996; Marsh 1998; Heath et al. 1999; Pallarés and Keating 2003). However, research has demonstrated that not all regional elections conform to second-order expectations;  2 Canadian provincial elections are among the exceptions. This leads to questions about why regional elections in some countries are more independent while others are a function of national politics. To explain these differences, one school of thought focuses on a country’s institutional structure. For example, it has been argued a strict division of constitutional powers encourages fewer vertical linkages between parties in federations (Chandler and Chandler 1987; Thorlakson 2009) and that countries without regular alterations in government tend to exhibit fewer second-order effects (Marsh 1998; Selb 2006). Most notable, however, is the argument that elections become less second-order when the power of regional governments increases. Higher stakes in regional elections encourage voters and candidates to base voting choices and electoral strategies on arena-specific rather than national considerations (Heath et al. 1999; Thorlakson 2007; Cutler 2008). This subject overlaps heavily with the literature on nationalization, which at its base “refers to the territorial homogenisation of voting behaviour” (Schakel 2013, 212). Similar to the literature on second-order elections, it has been argued that the institutional distribution of power influences the nationalization of electorates. While the nationalization literature often neglects regional party systems, the concept can also be applied vertically to refer to the nationalization of regions within a federation (Schakel 2013). In Canada, a province is highly nationalized in part if parties receive similar vote shares in both provincial and federal elections. Adjacent to this concept is party system congruence, defined as “the similarity of party systems across jurisdictions” (Thorlakson  3 2007, 69). Although the literature on second-order elections and nationalization does not always overlap, the two concepts are strongly related. If regional elections are truly second-order, parties have little incentive to differentiate themselves from their national counterparts. While second-order outcomes do not directly mirror first-order outcomes, regional party systems ought to remain relatively congruent. In contrast, when regional elections are more important, parties are encouraged to develop territorially differentiated electoral strategies. As electoral patterns grow increasingly divergent across arenas, regional party systems should grow less nationalized (Jeffery and Hough 2009). Accordingly, as regional authority increases, provincial elections should become less second-order and provincial party systems should become less nationalized and congruent. Less work has focused on variation between regional party systems in a single country. This is particularly relevant for Canada, given that certain provincial party systems are far more congruent with the federal party system than others. One theory used to explain the continued dominance of the Liberals and Conservatives in Atlantic Canada points to the fiscal dependence of these provinces on the federal government. The logic of this argument is that fiscal dependence increases the relative importance of the federal government, encouraging voters to support provincial parties with federal representation and influence. However, this theory is underexplored and has rarely been examined in detail.  4 This thesis investigates the relationship between regional authority and nationalization in Canada’s provinces on two main fronts. First, it assesses whether increases in provincial spending power actually result in less nationalized and congruent provincial party systems. Second, it examines whether fiscal dependence on the federal government has the opposite effect, resulting in more nationalized and congruent provincial party systems. Ultimately, this analysis finds no concrete evidence that either variable impacts nationalization on the provincial level, although further research with a more robust dataset is necessary for greater conclusively. This work contributes to the field by exploring issues neglected within the nationalization literature. Regional elections are often overlooked as a level of analysis and attention is rarely paid to party system variation between regions. By including fiscal dependence as an explanatory variable, this thesis tests a theory that has been repeatedly stated but poorly supported in the literature. Accounting for these oversights is a step towards a fuller understanding of nationalization and its causes. This is important in a broader sense given the democratic implications of both nationalization and decentralization. Nationalization “tends to channel the choices of voters and politicians into a smaller number of coalitions and to force governments to confront national-level problems” (Chhibber and Kollman 2004, 10). Because decentralization may also be desirable, this relationship suggests a normative trade-off exists between them (Harbers 2010). For these reasons, exploring underdeveloped aspects of nationalization is a worthy endeavor.  5 This paper is structured as follows. It begins with a review of the broader literature on second-order elections and nationalization, followed by the work suggesting fiscal dependence as an additional explanatory variable. A number of theoretical concerns are addressed to conclude this section of the paper. The subsequent chapter deconstructs the concept of fiscal decentralization, with this launching into the two hypotheses, the measurement of variables, and the regression models. The results of these models are then displayed and discussed, with a focus on potential theoretical or methodological explanations for the null findings of this study. 1.1 A Note on Terminology Due to inconsistency within the literature, some clarification of terminology is necessary. For ease of reading, the following terms are used throughout this paper even in reference to work that uses a different nomenclature. In this thesis, national and regional refer to the level of government in generic terms. Federal and provincial may be used when discussing Canada specifically. The terms statewide and non-statewide refer to the nature of the parties themselves without reference to the level of government they operate on. A statewide party attempts to win votes across the country on the federal level, whereas a non-statewide party attempts to win votes only within a specific region(s). Although the Ontario Liberal Party by definition only competes within Ontario, it qualifies as a statewide party because its federal affiliate fields candidates across the country. This thesis considers the Liberals, Conservatives, and NDP to be statewide parties while all other parties are considered non-statewide. While this is obviously a simplification, it should not meaningfully impact the analysis.  6 Chapter 2: Theory and Literature 2.1 Regional Authority and Second-Order Elections Building on Dinkel's (1977) analysis of German electoral patterns, Reif and Schmidt (1980) distinguish between national (or "first-order") elections and regional (or "second-order") elections. Although second-order elections partially reflect arena-specific factors, the authors argue that they are primarily determined by national criteria. A number of dynamics differentiate first and second-order outcomes. Because the electorate perceives second-order elections as less consequential, they are usually characterized by lower voter turnout. Based on a similar logic, voters are less concerned about “wasting” their ballots and are more willing to vote for smaller parties. Finally, national governing parties tend to perform less well relative to national opposition parties. This effect is largely contingent on the national electoral cycle, with backlash against the national government tending to be strongest near the middle of its mandate. Second-order election theory has formed the baseline against which other theories of multilevel elections are measured against. Further research shows that regional elections do not always conform to second-order expectations, particularly when regional governments have a significant impact on policy outcomes. Compared to less consequential European Parliament elections, Heath et al. (1999) find that voters in British local elections were more likely to turnout, more likely to vote differently than in the previous general election, and more likely to cite arena-specific issues as the primary reason for their decision. Cutler (2008) reinforces this idea through Canada, one of the most decentralized federations in the world. He demonstrates that vote choice in the 2003 Ontario provincial election was determined by  7 provincial criteria rather than national criteria, with opinions on the federal government having little to no impact on voting behaviour. Cutler argues that this independence of electoral arenas stems from the significant power of Canadian provinces. When regional governments matter, elections are less likely to function as second-order. This raises interesting questions about the ability of voters to evaluate parties separately in each arena. Blake (1982) argues that dual partisanship in Canada can be a rational response to divergent electoral contexts, with split-identification merely suggesting that voters consider provincial and federal arenas to be distinct spheres of competition. However, Stewart and Clarke (1998) argue against the notion that party identification can be effectively partitioned into “two political worlds”. Rather, they contend that voter evaluation of party performance on one level inevitably contributes towards their evaluation of the party overall. Leon (2014) addresses this question in relation to regional government power, although her focus is the autonomous communities of Spain rather than Canada. While she confirms a correlation between the vote share of national parties and their regional counterparts, Leon finds this relationship weakens as political and fiscal decentralization increases. She argues that, by incentivizing parties to differentiate their policy platforms in each of the autonomous communities, decentralization erodes the perceived connection between multilevel parties and encourages voters to evaluate their performance in each arena independently. In this way, decentralization encourages electoral patterns in national and regional elections to diverge.  8 2.2 Regional Authority and Nationalization Morgenstern et al. (2009) outline two traditional dimensions of nationalization referenced in the literature. The first dimension, dynamic nationalization, refers to whether changes in party vote shares follow similar patterns across the country. This dimension can be understood as the nationalization of electoral swings (Jones and Mainwaring 2003). The second dimension, distributional nationalization, is the degree that parties receive similar vote shares across the country. This dimension is usually understood in terms of parties and party systems. Schakel (2013) highlights a third dimension, party-linkage nationalization, which is the extent that voters and candidates coordinate across the country through national party labels. Chhibber and Kollman (2004) refer to this phenomenon as party aggregation and define “a national party system as one in which the same parties compete at different levels of vote aggregation” (4). Chhibber and Kollman’s (2004) rational institutionalist theory of party system nationalization parallels the literature on second-order elections. They demonstrate through a historical analysis of Canada, Great Britain, India, and the United States that national voting patterns tend to be more uniform during periods of greater central authority. The authors theorize that, as the power of the national government increases, voters have an incentive to coordinate/aggregate across districts by supporting statewide parties over non-statewide parties. Voters reason that statewide parties are more likely to form a government and therefore wield influence over policy outcomes. The opposite is also possible, with decentralization resulting in party system fragmentation as voters prioritize local preferences over the formation of a national government. Put simply,  9 Chhibber and Kollman argue that nationalization increases when the federal government has power over issues that matter to citizens. One of the most significant measures used by Chhibber and Kollman (2004) to test their theory is the difference between the effective number of parties overall and the average effective number of parties in each district. A large difference between these two values suggests that different parties are competitive in different districts, indicating a low level of party aggregation. Conversely, little to no difference between these two values suggests that the same parties are competitive across the country, indicating a high level of party aggregation. Chhibber and Kollman first separate the political histories of Canada, Great Britain, India, and the United States into centralizing, provincializing, and ambiguous periods. They then demonstrate that the difference between the effective number of parties overall and the average effective number of parties in each district tends to fall in centralizing periods and rise in provincializing periods, as their theory suggests. Chhibber and Kollman place Canada in a provincializing phase from 1960 to 2000, arguing this contributed to the fragmentation of the federal party system in the 1990s. While not their main focus, Chhibber and Kollman (2004) also extend their theory to the regional level. They claim that voters prefer regional parties with linkages to the national level when power is centralized under the logic that these parties have greater influence over national policy. Chhibber and Kollman argue that, “[i]n federal systems, voters may reason that common party labels not only help politicians from different  10 regions within the national government work together but also help politicians across levels of government work better together” (161-162). If power is relatively decentralized, however, there is less motivation for regional parties to associate themselves with national party labels. As a result, Chhibber and Kollman theorize that non-statewide parties on the regional level should be more likely to arise in decentralized systems, whereas the same parties that exist nationally should dominate on the regional level when power is centralized. As with party aggregation across districts, voters and candidates are more willing to support the same parties on both levels when doing so offers more influence over policy that matters. Unfortunately, this aspect of their argument is neglected relative to their national emphasis. Thorlakson (2007) brings together the literature on second-order elections and nationalization by placing Chhibber and Kollman (2004) more squarely in a multilevel context. She uses patterns of resource allocation to explain differences in party system congruence between six federations: Austria, Australia, Germany, Switzerland, the United States, and Canada. Thorlakson argues that decentralization helps to differentiate the policy space that regional governments operate within. Decentralization provides “parties and voters both incentive and opportunity to mobilize and respond to locally defined issues, leading to the development of ‘unique’ party systems at the state level” (71). When the prize of regional office increases, regional branches of statewide parties have an incentive to distance themselves from their federal counterparts and develop distinct identities. Citizens recognize that regional elections in decentralized federations have significant consequences, encouraging voters to make decisions based on arena-specific  11 rather than national criteria. Thorlakson confirms that regional party systems in centralized federations tend to be more congruent with the national party system than in decentralized federations. She does encounter some outliers, demonstrating this relationship is imperfect. Nevertheless, Thorlakson’s research lends support to Chhibber and Kollman’s argument that multilevel party systems are more congruent when power is centralized.  Other scholars have also found evidence of a relationship between decentralization and nationalization. Harbers (2010) finds that both political and fiscal decentralization relate to regionalized voting in Latin America. She suggests this effect may be greater due to a lack of strong functional cleavages in the region. Harbers focuses on distributional nationalization on the national level and does not address regional party systems in Latin American federations. Brancati’s (2008) conclusion is somewhat different. Whereas both political and fiscal decentralization are often used as explanatory variables, she claims that only the former strengthens non-statewide parties. Brancati argues that the presence of a regional level of government with independent decision-making authority encourages non-statewide parties because it increases their chances of forming government. Once a non-statewide party is established on the regional level, the costs of transitioning into national politics are much lower than if the party began on the national level. Because of this, she argues that political decentralization leads to denationalization in both arenas. Brancati uses an instrumental variable regression to account for endogeneity, concluding that  12 political decentralization does cause non-statewide parties rather than the relationship running in the opposite direction. 2.3 Other Explanations The theory that centralization and decentralization impacts nationalization is not universally supported. Through an analysis of 17 Western European countries, Lago-Peñas and Lago-Peñas (2011) find limited evidence that regional authority shapes the nature of party systems. Bochsler (2010) arrives at a similar conclusion in his study of post-communist democracies in Central and Eastern Europe, although his data on budget decentralization has no time component. Through an analysis of Belgium and Spain, Deschouwer (2009) concludes that decentralization has not resulted in the denationalization of national electorates. Finally, although Brancati (2008) confirms a relationship between political decentralization and the strength of non-statewide parties, she finds that fiscal decentralization has the opposite effect. These conflicting findings cast doubt upon the proposed relationship between institutional power and nationalization, suggesting that further research is necessary. An alternate branch of the literature on both party system nationalization and second-order elections focuses on sociological rather than institutional forces. Caramani (2004) theorizes that nationalization in Western Europe has been facilitated by a shift in electoral salience away from territorially based cleavages and towards homogenizing functional cleavages. He argues that “the survival of territoriality in politics today can therefore be explained principally through cultural cleavages that resisted the  13 homogenizing impact of class politics – the most important cause of nationalization processes – in the early phases of political-electoral mobilization” (292). Other literature has supported the idea that social divisions explain nationalization (Bochsler 2010). Similarly, territorial differentiation has been used to explain deviations from second-order election theory. It is argued that distinctive regional interests help to differentiate the policy space that each level operates within, reducing second-order effects (Jones and Scully 2006; Lineira 2011; Hough and Jeffery 2006). In order to identify whether regional authority has an impact on provincial party systems, this alternative hypothesis must be accounted for. 2.4 The Nationalization of Regions Research on nationalization is often restricted to the national level, even when the devolution of authority to regional governments is a key explanatory variable. Schakel (2013) refers to this as methodological nationalism bias. He contends that researchers should consider nationalization in terms of individual parties and regions in addition to party systems as a whole. A party is completely nationalized if it receives exactly the same share of the vote across regions, while a party is completely denationalized if it only receives votes in a single part of the country. Regional nationalization is based on the extent that the regional electorate votes similarly to the national electorate overall in national elections as well as the extent that the regional electorate votes similarly across regional and national elections (Schakel 2013, 2015). Because this thesis focuses on nationalization in a vertical context, the congruence between national and regional voting behaviour is most important here. Based on this second aspect of regional  14 nationalization, a region is completely nationalized if its regional party system is identical to the national party system and completely denationalized if its regional party system is comprised entirely of different parties than the national party system. By parsing out these different ideas, Schakel (2013) finds that regional authority has an important effect on the extent that regions and regional elections are nationalized. In other words, the difference between regional and national party systems tends to be greater when regional governments are relatively powerful. This is consistent with the theories most pertinent to this thesis, such as the arguments that decentralization decouples electoral dynamics between arenas and bolsters non-statewide parties on the regional level. On the other hand, Schakel does not find evidence that regional authority impacts the territorial distribution of statewide party support in national elections. He argues that, because parties may respond to decentralization in a multitude of different ways, there is no reason that increased regional authority should necessarily fragment statewide party support. Further, Schakel finds that decentralization has a strong impact on regional vote shares but only a marginal impact on national vote shares. These findings demonstrate the importance of extending analysis on nationalization to regional party systems.  By focusing on the nationalization of individual regions, Schakel’s (2013) analysis suggests that the impact of decentralization may not be symmetrical across units of a federation. This is important because it creates the possibility that the degree of nationalization may vary within a single country based on contextual factors (Jeffery and  15 Hough 2009; Diwakar 2010). Some research has explored the potential for regional differences within a single country. This work has primarily approached the issue from a sociological perspective, arguing that elections in less integrated regions or in those with more distinct territorial identities exhibit fewer second-order effects (Pallarés and Keating 2003; Jeffery and Hough 2003; Selb 2006). It is perhaps unsurprising that fewer scholars discuss this potential from an institutional perspective, given that authority is often relatively symmetric across regions. Further, the asymmetric devolution of power may be driven by pressure from distinctive regions and cannot be clearly attributed to decentralization itself. Despite this, it is possible that the level of decentralization may vary between regions for reasons other than territorial distinctiveness.  2.5 Fiscal Dependence A potential cause of such differentiation is the degree to which regions are dependent on the federal government. As mentioned, a number of scholars have flirted with this notion but few have explored it in greater detail. Brancati (2008) suggests that the size of federal transfers may be a misleading indicator of decentralization. She argues that, rather than signifying greater decentralization, large transfers may tie regions to the center and encourage regional party systems to mirror those that exist federally so they can better secure funding. A similar idea has been used to explain the continued dominance of the Liberals and Conservatives in the less wealthy provinces of Atlantic Canada. Chhibber and Kollman (2004) claim that the fiscal dependence of these provinces increases the relative importance of the federal government in that region, thus encouraging greater party aggregation.  16 Unfortunately, Chhibber and Kollman (2004) do not provide strong evidence of this theorized relationship between fiscal dependence and party aggregation. They state that, in Atlantic Canada, federal and provincial party organizations are more integrated, federal politicians are more likely to have provincial experience, and federal political authorities tend to have higher public support. While these facts are consistent with their theory, they do not demonstrate a causal relationship. Chhibber and Kollman’s most direct evidence comes from Stewart (1986), who argues that the reliance of PEI on transfer payments creates a preference for provincial governments that align with the federal government. This is driven by the perception that provincial governments of the same party as the federal government are best equipped to secure resources from the center. To prove his theory, Stewart highlights provincial election campaigns that stress the benefits of partisan alignment, survey evidence that demonstrates this view is publically held, and provincial elections that frequently align with federal outcomes. However, Chhibber and Kollman (2004) mischaracterize Stewart’s work in certain respects, somewhat undermining their argument. They cite Stewart in their claim that, in provincial elections between 1905 and 1984, a higher proportion of the vote was cast for the major federal parties in Atlantic Canada than in other provinces. This is a miscitation of Stewart, who does not provide the average share of votes cast for the major federal parties in general but rather for the current governing party more specifically. When citing Stewart’s main argument on PEI’s preference for partisan alignment, Chhibber and Kollman do not acknowledge that Stewart characterizes this phenomenon as unique to PEI rather than Atlantic Canada as a whole. Finally, they overlook Stewart’s (1994) later  17 revision that this preference for partisan alignment has faded since 1980, despite the economic status of the Island remaining largely unchanged. These issues further demonstrate that the evidence provided by Chhibber and Kollman is insufficient to prove a causal relationship between fiscal dependence and support for statewide parties. Other scholars have made reference to fiscal dependence as a reason for the continued dominance of the Liberals and Conservatives in Atlantic Canada. Bickerton (1994), also quoted by Chhibber and Kollman (2004), states that “the historic affinity of the four Atlantic Provinces for traditional two-party politics is arguably not unrelated to this regional political economy” (Bickerton 1994, 434). Wolinetz and Carty (2006) contrast the alienation experienced by eastern and western Canada, pointing out that the fiscally dependent Atlantic provinces have continued to support the traditional parties while the Western provinces have often turned to protest parties. Stevenson (2004) similarly connects fiscal dependence to the party system, stating that “the traditional two-party system has survived at the provincial level only in the Atlantic provinces, which are so dependent on federal largesse that affiliation with the federal government is still an electoral asset” (218). However, this theorized relationship is only briefly alluded to and is not the focus of each respective work. As a result, despite its recurrence within the Canadian literature, this theory is highly underexplored. 2.6 Nationalization and Fiscal Dependence in India While not in the Canadian context, Diwakar’s (2010) application of Chhibber and Kollman’s (2004) theory to India at the state level is notable both for its focus on regional  18 party systems and its exploration of fiscal dependence as a mediating variable. She argues that, if voters prefer national parties when power is relatively centralized, this preference should also be reflected within state party systems. Using the centralizing and provincializing periods outlined by Chhibber and Kollman, Diwakar examines whether the effective number of parties at the state level relative to the effective number of parties in the districts shifts alongside changes in federal authority. The theory suggests that the effective number of parties on the state level should decrease when power is centralized and increase when power is decentralized. With inconsistent findings between states, Diwakar is unable to confirm a causal relationship between regional authority and the degree of party aggregation.  Most interestingly, however, Diwakar (2010) does find that states with a high dependence on the federal government follow this expected pattern more clearly than other states, suggesting that dependence may increase the desire of voters to influence national policy. This is consistent with Chhibber and Kollman (2004), who also apply the fiscal dependence argument to India. Although the relationship between decentralization and party aggregation is less clear than the relationship between centralization and party aggregation, Diwakar nevertheless concludes that “states that are highly dependent on the national government are expected to have fewer parties at state level and a higher degree of party aggregation” (491). Diwakar’s mixed findings reinforce the need for further study on the relationship between regional authority, fiscal dependence, and regional party systems.  19 2.7 Theoretical Concerns There are a number of complicating factors that may impact the theorized relationship between regional authority and nationalization. While this thesis cannot provide definitive answers to many of these broader concerns, they should be noted. The first concern relates to endogeneity, which Lago-Peñas and Lago-Peñas (2011) describe as “inherent to the [regional authority] argument” (247). The criticism is that greater decentralization may occur due to pressure from non-statewide parties, reversing the causal direction of the theory (Bochsler 2006). Chhibber and Kollman (2004) acknowledge that the relationship between centralization and party system nationalization likely suffers from a degree of endogeneity. However, they maintain that shifts in power distribution tend to predate party system transformation, with this being the case in the United States, Canada, India, and the United Kingdom. Harbers (2010) also discusses these concerns in her work on Latin America but ultimately concurs with Chhibber and Kollman. She highlights the larger trend towards decentralization, arguing that “country-specific approaches are unable to sufficiently account for the global wave of decentralization” (621). While it is important to mention this theoretical disagreement, resolving the issue is beyond the scope of this thesis. The second concern relates to party autonomy and its influence on the relationship between regional authority and nationalization. Hopkin (2009) is critical of both institutional and sociological explanations of party system nationalization, arguing that neither Chhibber and Kollman (2004) nor Caramani (2004) treat political parties as organizations with agency. Hopkin argues that “parties cannot only mitigate the effects of  20 institutional change, [but] are also the architects of these changes and shape institutions to suit their strategic ends” (179). He also highlights the potential for institutional inertia, which he argues may render parties less responsive to decentralizing reforms than theoretically expected. While these criticisms are valid, they are difficult to formally incorporate into this analysis. Lago-Peñas and Lago-Peñas (2011) do not consider the party-centered approach to be a testable theory because it does not provide a clear alternative or null hypothesis. Accordingly, the potential mediating role of party organizations should be kept in mind but will not be directly tested. Finally, the treatment of vertical linkages between federal parties and their provincial counterparts must also be addressed. According Chhibber and Kollman (2004), when authority is relatively centralized, voters prefer regional parties with linkages to the national level due to a desire to influence national policy outcomes. However, this presupposes the existence of strong vertical linkages within parties, an assumption that cannot be easily made in Canada. According to Dyck (1996), Canadian parties can be categorized as integrated, confederal, or truncated. Integrated parties are those with formal linkages between levels and best fit the assumptions made by Chhibber and Kollman. Confederal parties operate both federally and provincially but are organizationally distinct in each arena. Truncated parties only compete on a single level. At the time of his writing, Dyck classifies the NDP as integrated, the Progressive Conservatives as (largely) confederal, and the Liberals as somewhere in between these two. However, given that the Conservative Party of Canada did not establish a formal relationship with the provincial Progressive Conservatives after the merger, they may now  21 be best described as truncated (Pruysers 2015). Given these weak organizational ties, one might question whether Chhibber and Kollman’s causal mechanism can be logically applied to Canada. Unlike the other concerns, this issue is more easily resolved. While it may appear theoretically problematic, the strength of vertical linkages are likely endogenous to the level of decentralization. Thorlakson (2009) argues that decentralization incentivizes regional branches of statewide parties to develop regionally focused electoral strategies and distance themselves from their national counterparts. This appears to correspond well with Canada; for example, certain provincial branches of the Liberal Party have cut ties with the federal Liberals in part due to a desire to establish a unique policy agenda and political identity (Dyck 1996). Therefore, although vertical linkages are in a sense necessary for Chhibber and Kollman’s (2004) causal process to function, their strength is at least partially endogenous to the model. This is further suggested by the fact that Liberal vertical linkages are strongest in the Atlantic provinces. Unlike elsewhere it Canada, it “is the only region in the country where affiliation with the federal government remained advantageous” (Dyck 1991). Because of this, the current lack of strong vertical linkages between Canadian parties is not detrimental to the logic of the theory. However, there remains the complication of the modern Conservative party. It is not immediately clear whether the federal and provincial Conservatives should be considered the “same party” for the purposes of this thesis. While both options are problematic, this thesis considers the parties to be equivalent. If this were not done, there would be a major decrease in party system congruence in some provinces while not in  22 others. While a spike in dissimilarity is not within itself an issue, this shock would be driven by federal rather than provincial developments and would therefore be misleading. The massive increase in dissimilarity that would occur in Atlantic Canada, for example, would not be due to decisions made within the region to distance from the federal level. Further, informal linkages based on electoral cooperation and shared campaign staff and activists should not be discounted. Of the Conservative Party of Canada and the Progressive Conservative Party of Ontario, Esselment (2010) states that, “[i]n spite of the informality of co-operation, each party is finding the other extremely helpful at election time and this goodwill and collaboration may yet become increasingly formal” (883). Although the parties do not coordinate on policy, this aspect is no different than the Ontario Liberals (Esselment 2010). While not a perfect solution, considering the Conservative Party of Canada and the various provincial Progressive Conservatives to be equivalent is the best option for this study. This decision will be revisited later to ensure it does not compromise the regression results.  23 Chapter 3: Measurement, Hypotheses, and Models 3.1 Dimensions of Fiscal Decentralization Fiscal decentralization is a complicated concept that must be deconstructed into its component parts before being measured. This section demonstrates that fiscal dependence can be more accurately understood as an aspect of fiscal decentralization rather than as a completely separate variable. This approach is supported by a group of scholars aiming to create a series of standard measures of decentralization (Schneider 2003; Blume and Voigt 2011; Martinez-Vazquez and Timofeev 2010). While the specific dimensions outlined in this literature vary somewhat, there is significant consistency between the variables identified as most important. Work by Schneider (2003), Blume and Voigt (2011), and Martinez-Vazquez and Timofeev (2010) suggest that the full extent of fiscal decentralization can be captured by three variables: the spending capacity of regional governments, the degree to which regional governments rely on transfers relative to own source revenues, and the extent to which these transfers are unconditional. This schema integrates well into the theoretical framework of this study. Spending power, fiscal dependence, and transfer conditionality can be conceived as three different forms of fiscal decentralization that may impact nationalization and congruence in different ways. This is further reinforced by the lack of significant correlation between suggested measures in all three articles. Schneider (2003) argues that the surprisingly weak relationship between the different dimensions of decentralization supports his “assumption that the dimensions [can] be analyzed as though they were orthogonal” (46).  24 A common message within this literature is that both fiscal and political elements must be considered to fully capture the nature of decentralization. Recommended indicators for the level of political decentralization include the presence of local or provincial elections, the existence of a strong, regionally selected upper chamber, and the division of powers between levels of government (Schneider 2003; Blume and Voigt 2011; Martinez-Vazquez and Timofeev, 2010). However, these political indicators are intended for cross-national usage and often do not significantly vary over time or between provinces within Canada. Even in the nationalization literature, political decentralization is often measured through a simple dummy variable (Harbers 2010; Brancati 2008). As a result, this thesis focuses exclusively on the relationship between fiscal decentralization and nationalization. 3.2 Hypotheses When considered an alternate form of decentralization, the fiscal dependence argument provides an interesting modification to institutional theories of nationalization. If these different aspects of fiscal decentralization are largely independent as the literature suggests, a high degree of decentralization along one axis may be accompanied by a comparatively low degree of decentralization along another. This thesis tests two hypotheses that arise out of this theory. The first hypothesis concerns the relationship between provincial spending power and provincial party systems. When provincial spending power increases, voters should be less concerned with influencing federal politics and more likely to support provincial parties that are not represented on the federal level. In addition, as the spoils of provincial office increase, the  25 provincial wings of statewide parties should have a greater incentive to tailor their electoral strategies to each provincial arena, even if this requires distancing themselves from their federal counterparts. These two effects together should result in less nationalized and less congruent provincial party systems. Accordingly, the first hypothesis is as follows: H1: As provincial spending power increases, provincial party systems grow less similar to the federal party system and support for statewide parties decreases. The second hypothesis focuses on the relationship between fiscal dependence and provincial party systems. Fiscal dependence theory suggests that a high reliance on federal transfers should encourage voters to support statewide provincial parties and reduce the incentive for provincial parties to distance themselves from their national counterparts. The second hypothesis is: H2: As fiscal dependence on the federal government increases, provincial party systems grow more similar to the federal party system and support for statewide parties increases. 3.3 Dataset The dataset used for this analysis consists of vote shares received by party for the 147 provincial elections held from 1962 to 2015. Each of these provincial elections is linked to the vote share received by party in that province in the preceding federal election. Given the causal logic of the theory, it is important that provincial elections are paired  26 with the previous federal election rather than simply the closest federal election. While it would be desirable to extend the period under study, the scope of this analysis is limited by the availability of provincial revenue and expenditure data from Statistics Canada. 3.4 Dependent Variables Three dependent variables are needed to capture the theorized relationship between fiscal decentralization and provincial party system nationalization. The first is the dissimilarity index, which was originally applied to multilevel elections in Canada by Johnston (1980). Schakel (2013) recommends the dissimilarity index as the appropriate measure for the nationalization of regions in a multilevel system. The index can be interpreted as “the minimum percentage of the province’s electorate that would have to be reallocated between parties to transform the provincial outcome into the federal one” (Johnston 1980). A dissimilarity score of 0 percent indicates that provincial election results are completely congruent with the previous federal election and, by extension, that the province is entirely nationalized. Correspondingly, a dissimilarity score of 100 percent suggests that the province is completely denationalized. The index is calculated by taking the difference between party vote shares in each respective election, summing the absolute values of these differences, and dividing by two to avoid double counting. Mathematically, the dissimilarity index can be expressed as:                                         where     is the share of the vote received by party   in the provincial election and     is the share of the vote received by party   in the previous federal election (Schakel 2015).  27 The dissimilarity index is ideal for measuring whether decentralization encourages greater electoral divergence between federal and provincial elections. However, to test the specific argument made by Chhibber and Kollman (2004), a narrower dependent variable is required. Rather than claiming that increases in provincial authority result in divergent electoral patterns, Chhibber and Kollman argue that decentralization makes voters more likely to support non-statewide parties. To test this formation of the theory, one must evaluate support for statewide parties relative to non-statewide parties. This is operationalized as the combined vote share received by the Liberals, Conservatives, and NDP in a given provincial election. However, if a preference for statewide parties when power is centralized is borne from a desire to influence federal politics, voters may prefer only parties that have formed national government. Accordingly, a third dependent variable is operationalized as the combined vote share received by the Liberals and Conservatives in a given provincial election. 3.5 Independent Variables The groundwork for the measurement of the independent variables has already been laid. As discussed, the three aspects of fiscal decentralization are provincial spending power, provincial reliance on transfer payments, and the conditionality of transfer payments. Due to the challenges of measuring transfer conditionality with precision, only the first two are used as independent variables. The spending power of provincial governments is measured in terms of provincial expenditure as a share of combined federal and provincial expenditure. This variable provides an indication of the overall fiscal impact of provincial governments relative to the federal government. The  28 fiscal dependence of provincial governments on the federal government is measured in terms of federal transfers as a share of provincial revenue. This variable, which may also be referred to as the vertical imbalance between the provinces and the federal government, is the inverse of a province’s revenue autonomy (Martinez-Vazquez and Timofeev 2010). Figure 1 and Figure 2 display the variation in the independent variables over time within each province. Provincial expenditure relative to federal expenditure fluctuates but follows a general upward trend throughout the period under study. This indicates that Canada has grown increasingly decentralized over the past fifty years. Variation in provincial reliance on transfer payments is greater in certain provinces than others. The most significant long-term trend emerges in Atlantic Canada, with dependence on federal transfers clearly decreasing over time. While this larger trend is less defined in other provinces, the variable does fluctuates from year to year. If the hypotheses are correct, these fluctuations and trends should impact the extent of provincial nationalization and congruence.         29 Figure 1: Provincial Share of Federal and Provincial Expenditure, 1960-2015  Figure 2: Transfers as Share of Provincial Revenue, 1960-2015   30 3.6 Control Variables The third aspect of fiscal decentralization, the conditionality of transfer payments, is one of the most important control variables in the model. It is theoretically necessary to control for because fiscal dependence may encourage greater nationalization when federal transfers have more conditions. Neither Schneider (2003), Blume and Voigt (2011), nor Martinez-Vazquez and Timofeev (2010) propose useful measures for capturing this dynamic within a single country, nor is detailed data on transfer conditionality readily available. Despite these challenges, it is possible to control for major shifts in the conditionality of federal transfers through a categorical variable that separates the period under study into three distinct eras. In the 1960s and 1970s, federal transfers included a larger number of conditions and were often administered through shared cost and federal matching programs. This changed with the creation of the Established Programs Financing in 1977, which replaced cost sharing programs for health and post-secondary education with a mixture of unconditional cash and tax point transfers. The next major reform to federal transfers came in 1995 when the Canada Assistance Plan, the last major federal-provincial cost sharing program, was rolled together with the Established Programs Financing to create the Canada Health and Social Transfer, later split into the Canada Health Transfer and the Canada Social Transfer (Department of Finance 2014). While other changes to programs have occurred since 1960, the creation of the Established Programs Financing and later the Canada Health and Social Transfer are arguably the most significant. At the risk of oversimplification, the years under study are separated into three periods: 1962 to  31 1976, 1977 to 1994, and 1995 to present. While Canada’s equalization program has changed over time, its payments have always been unconditional and need not be accounted for in these periods. While imperfect, this three-point categorical variable at least roughly controls for the degree of transfer conditionality. It is also theoretically important to control for the electoral cycle. Second-order election theory suggests that dissimilarity should increase when regional elections are held near the middle of the national government’s mandate. In contrast, simultaneous national and regional election results should be congruent, with no second-order effects being observed (Romanova, 2014). While Canadian provincial elections largely do not operate as second-order, it is possible that the distance between federal and provincial elections may influence their dissimilarity. Taking direction from Leon (2014), the electoral cycle is controlled for based on the number of months between paired federal and provincial elections. Given the literature on sociological factors and their influence on nationalization and electoral patterns, it is important to account for the strength of provincial identity. However, similar to political decentralization, the measures used to capture territorial heterogeneity in the literature are less useful when studying a single country. Rather than controlling for distinct regional identities or languages through categorical variables (Schakel 2013), these differences are accounted for through the use of provincial fixed effects. This should be sufficient to control for any unobserved heterogeneity arising from sociological differences between provinces.  32 Finally, the decade of the provincial election is controlled for. This is used to account for broader fiscal trends and other changes in government transfers, such as the universal (but short-lived) reduction in federal transfer payments during the 1990s. 3.7 Estimation Models Three models are used to test the hypotheses, with each using a different dependent variable. As mentioned, these dependent variables are election dissimilarity, support for statewide parties, and support for the Liberals and Conservatives. The model in its generic form is detailed below:                                                                                                                        ,        where provincial expenditure, transfer revenue, conditionality, election cycle, and decade are as described in the preceding section.            represents the inclusion of provincial fixed effects. The first hypothesis is concerned with the impact of provincial spending power. It is theorized that, as provinces gain greater control over policy areas that matter to citizens, provincial party systems should increasingly diverge from the federal party system. Therefore, one should expect a positive relationship between provincial expenditure and election dissimilarity. Chhibber and Kollman (2004) more specifically theorize that fiscal decentralization encourages support for non-statewide parties. Accordingly, one should expect a negative relationship between provincial expenditure and support for statewide parties. More precisely, if fiscal decentralization encourages support for parties that have formed a national government, one should expect a negative  33 relationship between provincial expenditure and support for the Liberals and Conservatives specifically. The second hypothesis is concerned with fiscal dependence. It is theorized that a high dependence on federal transfers increases the relative importance of the federal government. This encourages voters to prefer parties with linkages to the federal level and decreases the incentive for provincial parties to distance themselves from their federal counterparts. Therefore, one should expect a negative relationship between fiscal dependence and election dissimilarity, and a positive relationship between fiscal dependence and support for statewide parties.  34 Chapter 4: Results and Discussion 4.1 Results First, it should be noted that the two independent variables are correlated to a higher degree in certain provinces than expected. The correlation between the provincial share of combined federal and provincial expenditure and federal transfers as a share of provincial revenue is detailed in Table 1: Table 1. Correlation Between Provincial Expenditure and Transfer Revenue. Province Correlation British Columbia -0.12 Alberta -0.39 Saskatchewan -0.41 Manitoba -0.13 Ontario -0.04 Quebec -0.71 New Brunswick -0.66 Nova Scotia -0.62 Prince Edward Island -0.70 Newfoundland and Labrador -0.63  The literature on the measurement of fiscal decentralization suggests that these variables should be only weakly correlated (Schneider 2003; Blume and Voigt 2011; Martinez-Vazquez and Timofeev, 2010). While true for certain provinces, the correlation for others is far greater. This is particularity notable in the Atlantic provinces, in which the correlation between the two variables exceeds -0.6 without exception. While this correlation is not strong enough to create multicollinearity issues, it is worth highlighting given the expectations established by the literature. Estimations for all three models are detailed in Table 2:  35 Table 2. Impact of Fiscal Decentralization on Provincial Party System Nationalization. Variable (1) Election Dissimilarity (2) Support for Statewide Parties (3) Support for Liberals and Conservatives Provincial Expenditure 0.487 -0.345   -0.313  (0.337) (0.226) (0.288) Transfer Revenue 0.013  -0.066 -0.149  (0.356) (0.251) (0.315) Conditionality  5.967*  1.512 1.245  (2.533) (1.806) (2.210)  Election Cycle -0.002  0.027 .001  (0.054) (0.034) (0.044) Decade  -0.182 0.088  -0.007  (0.166) (0.130) (0.154) Constant  345.019*   -72.651 97.046   (156.257)  (58.206) (88.913) Number of observations 137 137 137 R-squared 0.18 0.11 0.01 Notes: Models are estimated in Stata 14 using the xtregar command with fixed provincial effects. This procedure uses a Cochrane–Orcutt transformation to correct for first-order autocorrelation. Formal tests show that heteroskedasticity is not an issue. *p<0.05, **p<0.01, ***p<0.001.  The regression results are uniformly negative. The two null hypotheses cannot be rejected in any of the three models, suggesting that neither provincial spending power nor fiscal dependence on the federal government impact provincial party system nationalization. The low R-squared values for all three models indicate that the independent variables and control variables explain little of the variation in election dissimilarity or support for statewide parties within provinces. The inability of the independent variables to explain changes in party system congruence over time is reinforced through Figure 3 and Figure 4. These graphs plot the distance between provincial expenditure, transfer dependence, and election dissimilarity in a given provincial election from their average provincial values.  36 Figure 3: Provincial Expenditure by Dissimilarity Score (Deviation from Provincial Average)  Figure 4: Transfer Dependence by Dissimilarity Score (Deviation from Provincial Average)   37 If the first hypothesis is correct, increases in the provincial share of federal and provincial expenditure should correspond with increases in the level of election dissimilarity. Figure 3 illustrates that, while this relationship is positive in the majority of provinces, its strength varies considerably. In many provinces, no clear relationship is apparent. If the second hypothesis is correct, a negative relationship should exist between provincial reliance on federal transfers and election dissimilarity. However, as displayed in Figure 4, increases in election dissimilarity are often more significant than changes in transfer dependence would suggest.  One might argue that this approach fails to capitalize on the diversity that exists between provincial party systems. Indeed, the cross-sectional variation between provinces appears consistent with both hypotheses at first glance. This is displayed in Figures 5 and 6, which plot the average election dissimilarity against the average value of the independent variables for each province. These graphs demonstrate that provinces with higher average expenditures relative to the federal government tend to have less congruent provincial party systems, while provinces with a higher average dependence on transfer revenues tend to have more congruent provincial party systems. However, when examined more closely, even cross-sectional variation does not strongly support the hypotheses. When the models are estimated using the between estimator, only the relationship between transfer dependence and support for the Liberals and Conservatives achieves statistical significance at the 0.05 level.1                                                  1 Because the between estimator considers only cross-sectional variation, the conditionality of transfer payments and the decade of the election are not controlled for.  38 Figure 5: Average Provincial Expenditure by Average Provincial Dissimilarity  Figure 6: Average Provincial Transfer Dependence by Average Provincial Dissimilarity   39 Further, even if these results did strongly support the hypotheses, it would be difficult to assess causality because territorial differentiation is a major competing explanation for nationalization in the literature. Cross-sectional variation may be driven by unobserved heterogeneity between provinces, with results merely reflecting these differences rather than indicating a causal relationship. For this reason, the results of the between estimator should be interpreted with caution. It could also be argued that equating the Conservative Party of Canada and the provincial Progressive Conservatives despite a lack of formal affiliation may contribute to these insignificant findings. To account for this, each model has been rerun using the 107 provincial elections from 1962 to 2003. Even with this adjustment, the independent variables remain statistically insignificant in all three models. 4.2 Discussion This thesis finds no concrete evidence that provincial spending power or fiscal dependence on the federal government impact provincial party system nationalization. A variety of theoretical and methodological explanations for this null finding are explored in this section. 4.2.1 Theoretical Explanations The most obvious theoretical explanation is that no connection exists between regional authority and provincial party system nationalization. This is certainly possible given that evidence of this relationship is mixed and generally focused on the national level. Alternatively, Brancati (2008) may be correct that only political decentralization rather than fiscal decentralization impacts the strength of non-statewide parties. This  40 cannot be tested with Canadian data alone due to a lack of variation but may nevertheless be true. While the existence of an elected regional government with decision-making authority may create space for non-statewide parties to compete, further increases in provincial authority (as measured through provincial expenditure) may not have an additional effect. There are a number of reasons why the causal mechanisms outlined in the literature may not operate as theorized. Because Chhibber and Kollman’s (2004) theory is contingent on the vertical linkages between parties, it may only apply up to a certain level of decentralization. As discussed, this thesis considers the strength of vertical linkages to be largely endogenous to the level of decentralization. As provinces gain authority, provincial parties organizationally distance themselves from their federal counterparts to allow for greater policy space. However, once vertical integration falls below a certain threshold, additional decentralizing reforms may not decrease nationalization further. If voters do not consider political parties to be vertically connected in a meaningful way, there is little reason to believe that electing the Ontario Liberal Party would result in greater influence over the Liberal Party of Canada. Therefore, while decentralization may continue even after vertical linkages are significantly weakened, the mechanism connecting regional authority with provincial party system nationalization is essentially broken. Given that Canada is among the most decentralized federations in the world, this causal mechanism may not function for much or all of the period under study. Another explanation for this null finding is the mediating role of political parties. As previously highlighted, critics of nationalization theories have argued that parties are  41 organizations with agency that respond to institutional changes in diverse ways. Both Hopkin (2009) and Schakel (2013) state that one should not expect decentralization to necessarily fragment the party system or impact all parties equally. While this renders theories of nationalization far less parsimonious, it may explain the mixed findings in the literature. For example, provincial parties may choose not to distance themselves from their federal counterparts even as decentralization increases. Conversely, provincial statewide parties may tailor their platforms so well to the provincial electorate that formal affiliation with a federal party does not create space for non-statewide parties to rise. Although the range of strategic decisions that parties make is difficult to account for, it may explain why theoretical expectations are not always borne out in reality. Some possible theoretical issues are unique to the fiscal dependence argument. Namely, fiscal dependence may only result in greater provincial party system nationalization when transfers are highly conditional. Although Chhibber and Kollman (2004) consider Canada’s shift towards unconditional transfers to be a decentralizing reform, they do not discuss transfer conditionality in relation to fiscal dependence. Because federal transfers in Canada have been regular and largely unconditional for a significant period, the level of fiscal dependence on the federal government may not influence voter behaviour. If the vast majority of transfers are guaranteed rather than negotiated on a case-by-case basis, voters in dependent provinces may have no greater incentive to support provincial parties with federal ties. Further, it is not obvious that statewide parties are in the best position to win concessions from the federal government or influence federal policy. In contrast, citizens may prefer provincial parties that do not  42 form the national government, as these parties may be considered more likely to fight in the provincial interest (Mueller 1968). The lack of research on fiscal dependence theory in the literature only increases the possibility that it is not theoretically sound. 4.2.2 Methodological Explanations It is possible that a relationship between these variables does exist but could not be identified due to the limitations of this thesis. That no relationship is found between fiscal dependence and provincial nationalization is not entirely surprising given its lack of empirical support in the literature. However, while previous research on regional authority and nationalization is decidedly mixed, the fact that neither hypothesis can be confirmed does raise questions about this research design. Perhaps the most likely methodological issue is that this analysis does not encompass enough variation to capture the theorized relationships. While provincial spending power changes significantly within provinces, there is far less variation in fiscal dependence over time. Thorlakson (2007) states that “[t]he general degree of decentralization, rather than fluctuations of limited magnitude and duration, likely exert the strongest effect [on party system congruence]” (89). It is possible that a relationship between fiscal decentralization and nationalization can only be observed when the variation within provinces is more extreme. Another possibility is that there are issues with the variables included in the model. It may be that the explanatory variables are not well operationalized. Although this thesis takes its independent variables directly from the literature on fiscal decentralization, these indicators may not capture the concepts they are intended to  43 measure. However, while this must be considered, these variables are both logical in their conception and grounded in the literature. Another possible problem is omitted variable bias, in which an important variable has not been controlled for. For example, non-statewide parties may have more traction in provinces that feel alienated from the federal government. In this way, fiscal decentralization may create the opportunity but not the incentive for voters to support non-statewide parties. Unfortunately, alienation is a challenging variable to quantify, making it difficult to incorporate into the model. While it is hard to know whether issues of operationalization or omitted variable bias influence the null findings of this thesis, these possibilities cannot be entirely discounted.  44 Chapter 5: Conclusion In one sense, the results of this thesis are unambiguous. None of the three models find evidence of a relationship between fiscal decentralization, measured in terms of provincial spending power and fiscal dependence on the federal government, and provincial party system nationalization. However, despite these insignificant results, it is difficult to make a definitive statement on the issue. Due to data limitations, this study incorporates just over fifty years of Canadian provincial and federal elections. Future research should extend the period under study, allowing for greater variation in the data. This would require either finding alternate data sources or measuring the independent variables in new ways. Further research should also consider expanding the study to include regions from other federations. This would also make it easier to test the impact of political decentralization in additional to fiscal decentralization. Nevertheless, this thesis does contribute to the conversation on regional authority and nationalization. By focusing on regional party systems, it explores a frequently neglected aspect of the nationalization literature. While additional research is necessary, the null findings of this thesis should further increase skepticism about the theorized relationship between fiscal decentralization and nationalization.    45 References Adamson, A., & Stewart, I. (2001). Changing Party Politics in Atlantic Canada. In H. G.  Thorburn & A. Whitehorn (Eds.), Party Politics in Canada (8th ed.) (302-320). Toronto: Pearson Education Canada.  Bickerton, J. (1994). Atlantic Canada: Regime Change in a Dependent Region. In J. P.  Bickerton & A. G. Gagnon (Eds.), Canadian Politics (2nd ed.) (426-449). Peterborough: Broadview Press.  Blake, D. E. (1982). The Consistency of Inconsistency: Party Identification in Federal and  Provincial politics. Canadian Journal of Political Science, 15(4), 691-710.  Blake, D. E. (1985). Two Political Worlds: Parties and Voting in British Columbia.  Vancouver: UBC Press.  Bochsler, D. (2006). The Nationalization of Political Parties: A Triangle Model, Applied to  Central and Eastern European Countries. CEU Political Science Journal, (4), 6-37.  Bochsler, D. (2010). The Nationalisation of Post-Communist Party Systems. Europe-Asia  Studies, 62(5), 807-827.  Brancati, D. (2008). The Origins and Strengths of Regional Parties. British Journal of  Political Science, 38(1). 135-159.  Blume, L., & Voigt, S. (2011). Federalism and Decentralization—A Critical Survey of  Frequently Used Indicators. Constitutional Political Economy, 22(3), 238-264.  Caramani, D. (2004). The Nationalization of Politics: The Formation of National  Electorates and Party Systems in Western Europe. Cambridge: Cambridge University Press.  Chandler, W. M., & Chandler, M. A. (1987). Federalism and Political Parties. European  Journal of Political Economy, 3(1-2), 87-109.  Chhibber, P. K., & Kollman, K. (2004). The Formation of National Party Systems:  Federalism and Party Competition in Canada, Great Britain, India, and the United States. Princeton: Princeton University Press.  Cutler, F. (2008). One Voter, Two First-Order Elections? Electoral Studies, 27(3), 492- 504.    46 Department of Finance. (2014). History of Health and Social Transfers. Retrieved from  https://www.fin.gc.ca/fedprov/his-eng.asp.   Deschouwer, K. (2009). Towards a Regionalization of Statewide Electoral Trends in  Decentralized States? The Cases of Belgium and Spain. In W. Swenden & B. Maddens (Eds.), Territorial Party Politics in Western Europe (31-46). New York: Palgrave Macmillan.   Dinkel, R. (1977). Der Zusammenhang zwischen Bundes-und Landtagswahlergebnissen.  Politische Vierteljahresschrift, 18(2/3), 348-359.  Diwakar, R. (2010). Party Aggregation in India: A State Level Analysis. Party Politics,  16(4), 477-496.  Dyck, R. (1991). Links Between Federal and Provincial Parties and Party Systems. In  Herman Bakvis (Ed.), Representation, Integration and Political Parties in Canada (129-177). Toronto: Dundurn Press Limited.  Dyck, R. (1996). Relations Between Federal and Provincial Parties. In A. B. Tanguay & A- G. Gagnon (Eds.), Canadian Parties in Transition (160-189). Scarborough: Nelson Canada.  Esselment, A. L. (2010). Fighting Elections: Cross-level Political Party Integration in  Ontario. Canadian Journal of Political Science, 43(4), 871-892.  Harbers, I. (2010). Decentralization and the Development of Nationalized Party Systems  in New Democracies: Evidence from Latin America. Comparative Political Studies, 43(5), 606-627.  Heath et al. (1999). Between First and Second Order: A Comparison of Voting Behaviour  in European and Local Elections in Britain. European Journal of Political Research, 35(3), 389-414.  Hopkin, J. (2009). Party Matters: Devolution and Party Politics in Britain and Spain. Party  Politics, 15(2), 179-198.  Hough, D., & Jeffery, C. (2006). Germany: An Erosion of Federal-Lander Linkages? In D.  Hough & C. Jeffery (Eds.), Devolution and Electoral Politics (119-139). New York: Manchester University Press.  Jeffery, C., & Hough, D. (2003). Regional Elections in Multi-Level Systems. European  Urban and Regional Studies, 10(3), 199-212.    47 Jeffery, C., & Hough, D. (2009). Understanding Post-Devolution Elections in Scotland  and Wales in Comparative Perspective. Party Politics, 15(2), 219-240.  Jones, M. P., & Mainwaring, S. (2003). The Nationalization of Parties and Party Systems:  An Empirical Measure and an Application to the Americas. Party Politics, 9(2), 139-166.  Jones, R. W., & Scully, R. (2006). Devolution and Electoral Politics in Scotland and Wales.  Publius, 36(1), 115-134.  Johnston, R. (1980). Federal and Provincial Voting: Contemporary Patterns and Historical  Evolution. In D. J. Elkins & R. Simeon (Eds.), Small Worlds: Provinces and Parties in Canadian Political Life (131-178). Toronto: Methuen Publications.  León, S. (2014). How Does Decentralization Affect Electoral Competition of State-wide  Parties? Evidence from Spain. Party Politics, 20(3), 391-402.  Liñeira, R. (2011). ‘Less at Stake’ or a Different Game? Regional Elections in Catalonia and  Scotland. Regional & Federal Studies, 21(3), 283-303.  Lago-Peñas, I., & Lago-Peñas, S. (2011). Decentralization and the Nationalization of Party  Systems. Environment and Planning C: Government and Policy, 29(2), 244-263.  Marsh, M. (1998). Testing the Second-Order Election Model after Four European  Elections. British Journal of Political Science, 28(4), 591-607.  Martinez-Vazquez, J., & Timofeev, A. (2010). Decentralization Measures Revisited. Public  Finance and Management, 10(1), 13-47.  Morgenstern et al. (2009). Party Nationalization and Institutions. The Journal of Politics,  71(4), 1322-1341.  Mueller, S. (1968). Federalism and the Party System in Canada. In J. Peter Meekison (Ed.),  Canadian Federalism: Myth or Reality (119-132). Toronto: Methuen Publications.  Pallarés, F., & Keating, M. (2003). Multi-Level Electoral Competition: Regional Elections  and Party Systems in Spain. European Urban and Regional Studies, 10(3), 239-255.  Pruysers, S. (2015). Two Political Worlds? Multi-level Campaign Integration in Canadian  Constituencies. Regional & Federal Studies, 25(2), 165-182.  Reif, K. (1984). National Electoral Cycles and European Elections 1979 and 1984. Electoral  Studies, 3(3), 244-255.   48 Reif, K., & Schmitt, H. (1980). Nine Second-Order National Elections: A Conceptual  Framework For The Analysis Of European Election Results. European Journal of Political Research, 8(1), 3-44.  Romanova, V. (2014). The Principle of Cyclicality of the Second‐Order Election Theory for  Simultaneous Multi‐level Elections. Politics, 34(2), 161-179.  Schneider, A. (2003). Decentralization: Conceptualization and Measurement. Studies in  Comparative International Development, 38(3), 32-56.  Selb, P. (2006). Multi-Level Elections in Switzerland. Swiss Political Science Review, 12(4),  49-75.  Schakel, A. H. (2013). Nationalisation of Multilevel Party Systems: A Conceptual and  Empirical Analysis. European Journal of Political Research, 52(2), 212-236.  Schakel, A. H. (2015). Nationalisation of Regional Elections in Central and Eastern  Europe. East European Politics, 31(2), 229-247.  Stevenson, G. (2004). Unfulfilled Union: Canadian Federalism and National Unity.  Montréal: McGill-Queen's University Press.  Stewart, I. (1986). Friends at Court: Federalism and Provincial Elections on Prince Edward  Island. Canadian Journal of Political Science, 19(1), 127-150.  Stewart, I. (1994). Roasting Chestnuts: The Mythology of Maritime Political Culture.  Vancouver: UBC Press.  Stewart, M. C., & Clarke, H. D. (1998). The Dynamics of Party Identification in Federal  Systems: The Canadian case. American Journal of Political Science, 42(1), 97-116.  Thorlakson, L. (2007). An Institutional Explanation of Party System Congruence:  Evidence From Six Federations. European Journal of Political Research, 46(1), 69-95.   Thorlakson, L. (2009). Patterns of Party Integration, Influence and Autonomy in Seven  Federations. Party Politics, 15(2), 157-177.  Van der Eijk et al. (1996). What Voters Teach Us About Europe-Wide Elections: What  Europe-Wide Elections Teach Us About Voters. Electoral Studies, 15(2), 149-166.  Wolinetz, S. B., & Carty, R. K. (2006). Disconnected Competition in Canada. In D. Hough  & C. Jeffery (Eds.), Devolution and Electoral Politics (54-75). New York: Manchester University Press.  49 Appendix A   This appendix provides additional information on the data sources and methods used to calculate the two independent variables of this thesis. A.1 Data Sources Four CANSIM tables from Statistics Canada were used to calculate (1) the provincial share of federal and provincial expenditure and (2) federal transfers as a share of provincial revenue:  Table 384-0022: Federal government and government sector revenue and expenditure, annual (1961-1980)  Table 384-0023: Provincial government revenue and expenditure, annual (1961-1980)  Table 384-0004: Government sector revenue and expenditure, provincial economic accounts, annual (1981-2009)  Table 384-0047: Revenue, expenditure and budgetary balance - General governments, provincial and territorial economic accounts, annual (2007-2015) Table 384-0022 and Table 384-0023 respectively provide federal and provincial revenue and expenditure data from 1961-1980. For subsequent periods, Statistics Canada provides federal and provincial revenue data together rather than in separate tables. A.2 Initial Calculation of Independent Variables The provincial share of combined federal and provincial expenditure is calculated by dividing provincial expenditure by the sum of provincial and federal expenditure in a  50 given province. Before this, however, current transfers to provincial governments are subtracted from federal expenditure. This is necessary to avoid double counting, as federal transfers to provincial governments are ultimately reflected in provincial expenditure as well. Capital transfers are listed separately from current transfers for the 2007-2015 period. For these years, both federal current and capital transfers to provincial governments are subtracted from federal government expenditure. It should be noted that provincial transfers to the federal government are not subtracted from provincial expenditure. These transfers are minimal and have no significant impact on the calculations. Transfers to local governments are also not adjusted, meaning that federal transfers to local governments are counted as federal expenditure while provincial transfers to local governments are counted as provincial expenditure. No adjustments are made in the calculation of federal transfers as a share of provincial revenue. Transfers from the federal government are simply divided by total provincial revenue to calculate this statistic. For 2007-2015, both current and capital transfers are considered together as federal transfers to the provinces. A.3 Further Adjustments to Independent Variables In recent years, Statistics Canada migrated from the Financial Management System (FMS) to Government Financial Statistics (GFS) in an effort to align its data with international standards. Because of this change, the data from the first two periods no longer reconciles with the data from the third period. This is evidenced by the three overlapping years of data between the second and third series, which do not correspond  51 directly. To reconcile the data periods, the data from 2007-2015 was regressed on the data from 1961-2009. With an R-squared value of over 0.98, the variance in one period overwhelmingly predicts the variance in the other period. After confirming this, the linear predictions of this fitted model were calculated. This provided adjusted values for the years 1961-2006 that better align with the data from 2007-2015. These adjusted data were ultimately used for the regression analysis. 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items