Upload
ucf
View
0
Download
0
Embed Size (px)
Citation preview
Electronic copy available at: http://ssrn.com/abstract=1664030
“Does the Type of Federalism Make a Difference for Electoral Behavior?
Comparing Spain and Germany”
Kerstin Hamann
and
Philip H. Pollock
Department of Political Science
University of Central Florida
Orlando, FL 32816-1356
Phone: 407-823-2608
Prepared for delivery at the American Political Science Association Meeting, Washington D.C.,
2010. This is work in progress and comments and suggestions are welcome. We would like to
thank Donald Plungis for valuable research assistance.
Electronic copy available at: http://ssrn.com/abstract=1664030
1
Abstract
In multi-level systems, voters have several opportunities to cast their ballots. This paper focuses
on the relationship between electoral behavior at the regional and state levels. The paper assesses
the ramifications of the way in which state-level and regional-level elections are linked for
electoral behavior analyzing the cases of Germany and Spain. In Germany, regional (Länder)
level elections are directly linked to state-level policy making as the parties composing the
Länder governments constitute the upper house (Bundesrat), which has considerable policy-
making powers. Thus, regional elections are more than just “second-order” elections, but have
direct and explicit ramifications for state-level governance. In contrast, Spain lacks a powerful
upper house representing the regional governments; regional-level elections are thus of less
immediate consequence for state-level governance. This paper explores whether these
institutional differences are reflected in voting behavior. In particular, the paper analyzes
whether differences in electoral turnout in regional and statewide elections exist across the two
cases. We hypothesize that given the higher significance of regional elections for state-level
policymaking in Germany, voter turnout at regional-level elections should be more closely
linked to voter turnout in statewide elections compared to Spain, where regional elections are of
less import for state-level politics than in Germany. Our initial findings confirm this hypothesis.
2
I. Introduction
Multi-level governance in Europe affords voters opportunities for electoral participation at
different levels of government. Voters can cast their ballots in elections at the state level, but also
at the suprastate level (EU) and substate levels (regional and local). The literature has established
that these elections, though taking place in different arenas, are also connected as voters
oftentimes use suprastate or substate elections to send signals to the state-level government and
parties. Thus, the nexus between different levels of government is complex, and consequently,
electoral behavior is similarly complex.
In this paper, we explore one aspect of the links between different levels of electoral
behavior. We provide a preliminary and tentative analysis of the links between voter turnout in
statewide and regional elections. This study is motivated by the fact that the electoral
significance of regional elections varies across countries depending on the relevance regional
elections have for politics at the state level. This institutional analysis probes differences in
federalist structures to gain insights into electoral behavior at the substate level.
We pursue these questions for two cases, Germany and Spain. While both cases have
institutions of multi-level governance, they differ in the types of federalist institutions. We are
interested in finding out whether these institutional differences are reflected in electoral
behavior, in particular the relationship between turnout in elections at the state and regional
level.
The paper proceeds as follows: The next section describes federalist institutions in Spain
and Germany, which forms the basis for our hypothesis. The third section briefly outlines
differences in regional elections for our two cases. We then proceed to present our data and
exploratory analysis. The Conclusion points to directions for further research.
3
II. Federalist Institutions in Spain and Germany
Spain and Germany show marked differences in the way their multi-level systems are set up.
These differences are perhaps most conspicuous in the fact that Germany’s federal system grants
the regions, or Länder, signficiant influence on state-level policymaking through the upper house
of parliament, the Bundesrat. Spain, on the other hand, has what is sometimes referred to as a
quasi-federalist system, in which the regions—the autonomous communities—have no such
representation at the state level. This section briefly describes these differences in the
institutional setup of multi-level governance in Spain and Germany.
Quasi-Federalism in Spain
Spain has a long history of regional differences and calls for autonomy. While a few regions
gained their autonomy statutes during the Second Republic established in 1931, these rights were
soon buried when the Franco dictatorship took over in 1939 and refused to recognize regional
differences. However, while this policy perhaps succeeded in masking existing differences and
tensions, it failed to suppress them. As the Basque terrorist organization ETA illustrated vividly,
regional identities stayed alive and were politicized despite the official politics that ignored their
existence. Not until Spain redemocratized after Franco’s death in December 1975 were conflicts
and cleavages along regional lines officially acknowledged and reflected in the territorial
organization of the state. The Spanish Constitution, ratified in 1979, provides for the existence of
regions, but left the details of the establishment of the regions open to legislation.
In practice, this meant that in the early years of the democracy, 17 regions or
“autonomous communities” were established. Since then, Spain has undergone an extensive
4
process of decentralization together with territorial pluralism, resulting in what Colomer (1999)
calls “non-institutional federalism.” Regions gained autonomy via different routes. The “historic
nationalities” of Catalonia and the Basque Country gained autonomy in 1979; the “historic
region” of Galicia and also Andalucía followed in 1981; three more regions obtained an
intermediate status or gained special rights, and the remaining 10 regions were granted autonomy
via the “normal” or “slow” path, completing the process of setting up autonomous regions in
February of 1983 (see Hamann and Mershon 2008: 117). Each region holds elections to a
unicameral regional parliament, which in turn elects a regional “president” to head the executive.
Regional elections are generally scheduled every four years though not all regions vote at the
same time: the first four regions to gain autonomy vote on separate dates, while the remaining 13
regions hold elections simultaneously at fixed four-year intervals and concurrent with local
elections. All regions use a proportional representation system though some differences exist
concerning thresholds, definition of the unit constituting the electoral district, and mean district
magnitude (see Colomer 1999; Hamann and Mershon 2008: 116).
Perhaps most importantly, Spanish regions lack representation at the state level. Spain
has a bicameral legislature. The lower house, the Congress of Deputies, is elected at regular
intervals that cannot exceed four years. The upper house, the Senate, was established in the 1979
Constitution as a chamber of territorial representation, but in practice only about one-fifth of the
senators directly represent their regions. Furthermore, the Senate is a weak institution and has
few powers to influence legislation (see Field and Hamann 2008: 16-17). Regional governments
mostly exert influence on statewide politics through bilateral negotiations with the statewide
government, especially in cases where a minority government is dependent on legislative support
5
from regional parties (see Heller 2002). Spain is thus a highly decentralized country yet lacks
territorial representation of the regions at the level of the state.
Federalism in Germany
Germany has a federalist system developed after the end of World War II. Some states were
drawn based on historical regions, others were newly created, and the Saarland was the last state
to join in 1957 after having been under French administration. Until unification with East
Germany occurred in 1990, the Federal Republic was comprised of 10 states, or Länder, plus
Berlin; unification added five new states for a total of 16. Much like in Spain, each state has its
own unicameral parliament, headed by a premier (Ministerpräsident). Elections in each state also
take place roughly every four years although they are staggered and do not coincide to the extent
that 13 of the 17 Spanish regions vote together in regional elections. German states also use a
proportional representation system.
Similar to Spain, Germany also features a bicameral legislature where elections to the
lower house, the Bundestag, take place no more than four years apart. Perhaps the most
important difference to the Spanish system is that the German Länder are represented in the
powerful upper house of the German parliament, the Bundesrat, and the “federal states
participate through the Bundesrat in the legislation and administration of the Federation.”1 The
Bundesrat is composed of 69 members, which are allocated to the states roughly proportionate to
their population size. No state has fewer than three seats and the largest states have six seats. The
states fill their seats to represent their governments, meaning that the overall seat distribution
reflects the party composition of the state governments. Members must be government members
11
This quote is on the homepage of the German Bundesrat
(http://www.bundesrat.de/EN/Home/homepage__node.html).
6
at the state level. Each state casts a bloc vote in Bundesrat votes, regardless of the party
affiliation of the individual members (coalition governments are common in German Länder).
Bundesrat members thus represent their state rather than their political party.
The fact that the states have direct representation in the Bundesrat is of significance
especially given the importance of the upper house for state policy making. The Bundesrat has
extensive policy-making powers, possessing veto power for about half of all bills and suspensive
veto power for all other bills (Lohmann et al. 1997: 426; Patzelt 1999: 61).2 This means that the
composition of the Bundesrat is crucial for the policy-making process as the lower house, the
Bundestag, is dependent on approval by the upper house for a large part of its legislation.
Therefore, elections for state governments have obvious and significant implications for the
policy-making process at the state level. To quote the Bundesrat homepage concerning the
significance of elections at the state level: “This means that state elections always have
nationwide political significance too. Even in the Fifties there was a saying ‘Your election in the
state of Hesse counts in the Bundesrat in Bonn.’”3 Voters are thus keenly aware that their vote
can have ramifications beyond the composition of the next government in the Land they reside
in.
This is in stark contrast to Spain, where no direct connections between regional elections
and state-level representation exists. Certainly, regional governments matter as they are
important conduits in facilitating the formation of state-level support coalitions for the
government, especially when the government rules as a minority executive (see Field 2009), but
2 If the Bundesrat casts a suspensive veto, the Bundestag can either amend the bill until the Bundesrat accepts it, or it
can overrule the veto if the entire Bundestag votes for the bill (Silvia1999: 173). 3 Bundesrat homepage (http://www.bundesrat.de/cln_152/nn_11006/EN/organisation-en/plenum-en/plenum-en-
inhalt.html?__nnn=true).
7
the results of regional elections cannot formally and directly help or hinder state-level
policymaking. These differences may affect electoral behavior.
III. Regional Elections in Spain and Germany
Given the institutional differences concerning the structure of the significance of regional
elections for state-level politics between Spain and Germany, we are curious whether, and how,
electoral behavior is affected. In particular, we are interested in turnout patterns for regional and
statewide elections in Spain and Germany. The literature has established that by and large,
turnout levels for state-level elections tend to be higher than for suprastate or substate elections
(see Hough and Jeffery 2006a: 8). This pattern also holds true for Spain and Germany, where,
aside from a few exceptions, more voters tend to go to the polls in elections for the lower house
of the state parliament compared to elections for regional legislatures. However, here we probe
further into the relationship between turnout at the state and regional levels based on the
institutional differences in federalism we have observed in our two cases.
One argument in the literature suggests that turnout in regional elections tends to be
lower because these elections constitute “second-order” elections, of less relevance to state-level
policymaking than state-level, or “first-order,” elections. Pallarés and Keating (2006: 104), for
example, state for the Spanish case that the abstention level for regional elections is on average
“above that for state elections … consistent with the second order notion.” For the German case,
Hough and Jeffery (2006b) discuss evidence for the interpretation of Land elections as second-
order elections particularly for the pre-unification era, though perhaps the logic holds less so
since 1990. Generally, voters cast their vote in second-order elections based on state-level issues
and government performance rather than based on issues that are of particular relevance at the
8
subnational or supranational level. Thus, second-order elections sometimes take on the character
of “barometer” elections for the state-level government (Anderson and Ward 1996).
At the same time, given the difference in the amount that is at stake for voters in different
types of second-order elections, “Some second order elections are more second order than
others” (McLean et al. 1996:4, cited in Hough and Jeffery 2006a: 9). While this differentiation
was aimed at different types of second-order elections—specifically, elections to the European
Parliament compared to local elections—we can draw a similar argument for differences in the
amount that is at stake for voters in regional elections: not all regional elections are the same,
depending on the importance regional elections have for state-level government. In Germany,
regional elections are of considerably higher relevance for state-level politics than in Spain—
regional elections in Germany determine the composition of the Bundesrat, which in turn plays
an important role in state-level policymaking. In Spain, in contrast, the composition of regional
governments is of much less direct impact for state-level policymaking as the regions have no
direct representation in a powerful upper house.
Do these differences affect patterns in voter turnout between regional and state-level
elections? We expect that the relationship between turnout at regional and state elections will be
stronger in Germany than in Spain as regional elections have a more direct significance for state-
level policymaking in Germany. In Spain, where regional elections have no state-level
component, the factors that affect regional turnouts should be different from the factors that
affect state-level turnouts. Thus, we should find that regional and state elections are distinct,
weakly correlated electoral events. In Germany, where regional elections have a significant state-
level component, many of the same factors that affect state-level turnout will also be present in
regional elections. Therefore, we should find that regional and state elections are strongly
9
correlated electoral events. Note that we are not interested in explaining voter turnout or
abstention at the regional level per se. The literature on voter turnout is large and cites
individual-level factors as well as institutional factors to account for differences in electoral
participation within and across countries. However, we are not interested in here in explaining
why some regions have higher or lower turnout level than others, or why turnout differ between
Germany and Spain. Instead, we are interested in analyzing the linkages between turnout in
elections at the regional and state level. We now turn to a preliminary exploration of this
question.
IV. Data and Exploratory Analysis
We compiled, for Spain and Germany, turnout data in state and regional elections, 1982-2008 for
Spain and 1980-2009 for Germany. (See Data Sources.) This spans all regional elections in Spain
since the restoration of democracy in the late 1970s until 2008 and all regional elections in
Germany for the equivalent time frame. We excluded cases where the regional elections and
state-level elections coincided, e.g. Andalucía held simultaneous elections for the regional
parliament and the state-level Congress in 1996, 2000, 2004, and 2008; in Germany,
Mecklenburg-Vorpommern held simultaneous elections to the Land parliament and the
Bundestag in 1994, 1998, and 2002. We exclude these cases because simultaneous elections do
not permit us to distinguish between the effects of state-level and of regional-level forces on
voter turnout. We are not interested in explaining voter turnout at different levels as such, but
rather in the relationship between turnout at the state level and at the regional level. Furthermore,
we exclude Berlin prior to 1990 because deputies from Berlin had no voting rights in the
10
Bundesrat or the Bundestag due to the special status of the city prior to unification. Thus, the
elections for Berlin prior to 1990 are not comparable to the other cases.
These data allowed us to determine, for each state election, turnouts in the preceding—
and the succeeding—regional election for each region. For example, in the Spanish region
Andalucía, turnout was 76.2 percent in the 1993 state election. In the preceding regional election,
turnout was 55.30 percent, and in the succeeding regional election it was 67.3 percent. In
Germany’s Baden-Württemberg, state-level turnout was 79.7 in 1994, preceded and followed by
regional turnouts of 70.1 and 67.6, respectively. If the state-regional electoral link is weak, we
should find that state turnout rates bear little relationship to regional rates in preceding or
succeeding elections. For instance, a higher-than-usual state turnout may be followed by a high
regional turnout or a low regional turnout—depending on what region-level factors are present in
the regional election. If the state-regional link is strong, then higher turnouts in state elections
should also occasion higher succeeding regional turnouts, and lower state turnouts should be
followed by lower regional turnouts. We hypothesize that the same factors that occasion higher
(or lower) state turnouts should also translate into higher (or lower) regional turnouts. We
explore these ideas graphically, first presenting disaggregated data on each region, then showing
patterns in the data aggregated by region.
Figure 1 presents the disaggregated data for Spain. To accommodate data from all 17
regions, Figure 1 has six panels, five showing turnout statistics for three regions and one
displaying data for two regions. Each set of axes has state turnout on the horizontal axis and
succeeding regional turnout on the vertical. Are we able to discern a relationship within regions?
As state turnouts move, do regional turnouts follow? Consider the upper-left panel, which
displays scatterplots for Andalucía, Aragón, and Canarias. State turnouts in Canarias vary
11
markedly, from about 60 percent to nearly 80 percent. Yet succeeding regional turnouts change
very little, staying in the 60-65 percent range. Aragón posts higher means on both axes but shows
a similarly weak state-regional connection. The data for Andalucía are somewhat more
systematic—as state turnouts go up, regional turnouts increase—save for a large inconsistency at
about 70 percent on the horizontal axis. Perusing the panels, we can see the different varieties of
non-relationship. For example, the scatters in the right-hand panel of the middle row show
variable state turnouts and stable regional turnouts (Galicia), stable state-variable regional
(Valenciana), or variable state-variable regional (Extremadura).
[Figure 1 about here]
Table 1 presents a regression analysis that has succeeding regional turnout as the
dependent variable and uses state turnout and a series of regional dummies as independents.
Eight regions are significantly different from the base category (Andalucía)—three have higher
turnouts, on average, and five have lower turnouts. But the effect of state turnout (b = .05, sig. =
.66) is statistically indistinguishable from zero. This is seen, as well, in the aggregated data,
created by collapsing the mean values of state turnout, succeeding regional turnout, and
preceding regional turnout. The top panel of Figure 2 shows the correlational scatter of state
turnout (horizontal axis) and succeeding regional turnout (vertical axis). The bottom panel
displays the scatter for preceding regional turnout (horizontal axis) and state turnout (vertical
axis). To be sure, the data are not without coherence. However, the linkages appear weak,
particularly in the lower ranges of turnout. The weight of the analysis suggests that regional
turnouts in Spain vary for any number of reasons, but state-level turnout does not appear to be
among those reasons.
[Table 1 about here]
12
[Figure 2 about here]
Figure 3 displays the disaggregated data for Germany. Again we can ask whether
variations in state turnout are related to variations in regional turnout. Are movements from left
to right along the horizontal axis associated with lower-to-higher movements along the vertical?
For many regions, the answer is clearly yes. The regions in the upper-right panel (Bremen,
Hamburg, and Hessen) and those in the left-hand panel in the middle row (Niedersachsen,
Nordrhein-Westfalen, and Rheinland-Pfalz) provide the strongest evidence. In all of these
cases, as state turnouts wax and wane, so do succeeding regional turnouts. Note also that these
regions (unlike the Spanish regions) do not group themselves in regionally distinct positions
along either axis. Rather, the regions mix together in a pattern suggesting that supra-regional
factors are more important is structuring turnout. Of course, not all of the regional data are as
well behaved. Two of the regions in the upper-left panel (Baden-Württemberg and Bayern)
show the expected relationship, although not as convincingly as the six regions discussed
above. Furthermore, we visually separated out the five new regions in Germany, joining the 11
“old” ones (including Berlin) after unification in 1990, since electoral behavior might well be
different in the new Länder, where democratic procedures were much less institutionalized and
familiarity with the consequences of regional voting perhaps less well known than in the 11
“old” regions. In fact, the new Länder, which show in the two panels along the bottom row of
Figure 3, appear not to follow the expected pattern. However, these regions’ comparatively low
turnouts—combined with our decision to use the same axis scales for all the scatters—may
obscure a more systematic relationship.
[Figure 3 about here]
13
A clearer picture emerges from the regression analysis, which (as with Spain) models
succeeding regional turnout as a function of state turnout and region-specific mean differences,
measured by dummy variables. Table 2 shows the results. These estimates stand in stark
contrast to the analysis of the Spanish microdata. Aside from two region-specific effects—a
sizable adjustment for Saarland (7.39 points) and a gargantuan adjustment for Mecklenburg-
Vorpommern (13.40 points)—none of the regional coefficients is significant. However, the
coefficient on state turnout is highly significant (b = 1.11, sig. = .00). Thus, state-level factors
are emergent while region-specific factors are muted. The aggregated correlational data (Figure
4) clarify the contrast with Spain. The process of aggregation has removed much interesting
variation, but it has also revealed relationships that were difficult to see in the disaggregated
data, particularly with regard to Germany’s new regions. They in the aggregate resemble the
pattern of the “old” regions in Germany much more closely, although at a considerably lower
turnout level overall, while at the microlevel reveal a much more distinct picture. Mecklenburg-
Vorpommern clearly constitutes an outlier, which will need to be examined more closely.
[Table 2 about here]
[Figure 4 about here]
V. Conclusions
The research question underlying this paper was based on the idea that voting behavior is
influenced by institutional factors. More specifically, we examined whether the institutional
differences of different types of federalisms are reflected in patterns of voter turnout. Our
analysis demonstrates that voter turnout patterns between regional and state elections are not the
same across countries with multi-level governance. The cases of Spain and Germany suggest that
14
where regional elections have a direct and important influence on state-level policymaking, as in
Germany, turnout in regional elections tends to be much closer correlated to turnout in state
election than in cases where regional elections lack such import. Subnational elections may well
all be second-order elections, subordinate to state-level elections in the minds of the voters and
driven by state-level issues, but they are of different import for state-level policymaking and thus
have different consequences for voter turnout.
The analysis and results in this paper are limited, tentative, and preliminary. Future work
will extend the analysis in several directions. For one, we are interested in looking at the early
years of the Federal Republic. Did the patterns since 1980, observed in this paper, develop
slowly over time, or were they present already in the early years of democracy? In other words,
does the stage of democratization make a difference? A comparison of the early years of the
Federal Republic with the early years of democratic Spain might be able to provide some initial
answers. Second, we will explore further the temporal connections between different levels of
elections by comparing regional voter turnout with the statewide election closest to the
respective regional election rather than the next regional election—sometimes the previous
regional election might be closer in time than the subsequent regional election. Related, we will
borrow from the “electoral balancing” literature on Germany (see Kern and Hainmueller 2006)
further analyze whether the relationship in Germany between regional and state-level elections
differs depending on whether the regional election is a “crucial” one, i.e. whether it can change
existing majorities in the Bundesrat. Furthermore, we will refine our analysis by controlling for
factors other to the relevance of regional elections for state-level policymaking. For example, the
timing of elections may be important: Where sub-state elections are held simultaneously, voter
behavior may resemble state-level electoral behavior more closely than in cases where regional
15
elections take place within a regional context. Jeffery and Hough (2006: 249) suggest that this is
the case for issues and electoral results, and we will explore whether this logic also holds for
turnout. We are also interested in looking at other manifestations of voting behavior, such as vote
choice. For example, if state-level governments in Germany are more closely linked to state-
level policymaking, are voters more likely to “punish” parties in regional elections that are
incumbents at the state level?
In sum, the implications of institutional features of multi-level governance are complex;
this paper has begun to assess their effects on electoral behavior. Further research will help refine
and expand our hypotheses, and the inclusion of additional cases will test the extent to which our
findings hold for other countries.
16
Table 1. Spain: Explaining Differences in Regional Turnout
Parameter Regression coefficient Standard error t-ratio Signif.
Constanta 63.94 8.08 7.91 .00
State turnout .05 .11 .44 .66
Aragón .44 2.90 .15 .88
Canarias -4.13 2.98 -1.38 .17
Cantabria 5.40* 2.91 1.85 .07
Castilla-La Mancha 7.39** 2.96 2.50 .01
Castilla y León 3.71 2.91 1.27 .21
Cataluña -7.29** 2.90 -2.51 .01
C. Foral de Navarra -4.69 2.90 -1.61 .11
C. de Madrid -12.52** 2.92 -4.29 .00
C. Valenciana -7.15** 2.93 -2.44 .02
Extramadura 7.05** 2.93 2.41 .02
Galicia -5.48* 2.95 -1.86 .07
Islas Baleares -4.84* 2.95 -1.64 .10
La Rioja 4.64 2.92 1.59 .12
País Vasco .52 2.86 .18 .86
Principado de Asturias -3.16 2.91 -1.09 .28
Región de Murcia 2.55 2.92 .87 .39 aAndalucía is the base category.
*p<=.10 / **p<=.05
Note: Dependent variable is regional turnout in succeeding election. Adjusted R2 = .538.
17
Table 2. Germany: Explaining Differences in Regional Turnout
Parameter Regression coefficient Standard error t-ratio Signif.
Constanta -24.18 16.11 -1.50 .14
State turnout 1.11** .19 5.69 .00
Bayern 1.26 3.73 .34 .74
Berlin 1.71 4.35 .39 .70
Brandenburg -.19 4.52 -.04 .97
Bremen 3.28 3.74 .88 .38
Hamburg 5.32 3.73 1.43 .16
Hessen 4.18 3.65 1.15 .26
Mecklenburg-Vorp. 13.40** 4.61 2.91 .01
Niedersachsen 3.20 3.75 .85 .40
Nordrhein-Westfalen -1.49 3.87 -.38 .70
Rheinland-Pfalz 5.23 4.11 1.27 .21
Saarland 7.39* 3.91 1.89 .06
Sachsen -2.84 4.43 -.64 .52
Sachsen-Anhalt .69 4.67 .15 .88
Schlesweig-Holstein 6.28 3.87 1.62 .11
Thüringen .01 4.42 .00 1.00 aBaden-Württemberg is the base category. Coefficient estimates regional turnout in Baden-
Württemberg when state-level turnout is equal to 0.
*p<=.10 / **p<=.05
Note: Dependent variable is regional turnout in succeeding election. Adjusted R2 = .463.
23
Figure 4. Germany: State Turnout and Regional Turnout (Aggregated data)
r = .76
r = .93 w/o outlier
r = .90
r = .95 w/o outlier
24
References
Anderson, Christopher J. and Dan S. Ward. 1996. “Barometer Elections in Comparative
Perspective.” Electoral Studies 15(4): 447-460.
Colomer, Josep. 1999. “The Spanish ‘State of Autonomies’: Non-Institutional Federalism.” In
Paul Heywood (ed.), Politics and Policy in Democratic Spain: No Longer Different?,
London: Frank Cass, pp. 40-52.
Deutscher Bundesrat. N.d. Homepage, available at http://www.bundesrat.de/.
Field, Bonnie N. 2009. “Minority Government and Legislative Politics in a Multilevel State:
Spain under Zapatero.” South European Society and Politics 14(4): 417-434.
Hamann, Kerstin and Carol Mershon. 2008. “Regional Governments in Spain: Exploring
Theories of Government Formation.” In Bonnie N. Field and Kerstin Hamann (eds),
Democracy and Institutional Development: Spain in Comparative Theoretical
Perspective, Houndmills, Basingstoke: Palgrave Macmillan, pp. 110-134.
Hough, Dan and Charlie Jeffery. 2006a. “An Introduction to Multi-Level Electoral
Competition.” In Dan Hough and Charlie Jeffery (eds), Devolution and Electoral
Politics. Manchester University Press, pp. 2-13.
Hough, Dan and Charlie Jeffery. 2006b. “Germany: An Erosion of Federal-Länder linkages?” In
Dan Hough and Charlie Jeffery (eds), Devolution and Electoral Politics. Manchester
University Press, pp. 119-139.
Jeffery, Charlie and Dan Hough. 2006. “Devolution and Electoral Politics: Where Does the UK
Fit In?” In Dan Hough and Charlie Jeffery (eds), Devolution and Electoral Politics.
Manchester University Press, pp. 248-256.
Kern, Holger Lutz and Jens Hainmueller. 2006. “Electoral Balancing, Divided Government and
‘Midterm’ Loss in German Elections.” The Journal of Legislative Studies 12(2): 127-149.
Lohmann, Susanne, David W. Brady and Douglas Rivers. 1997. “Party Identification,
Retrospective Voting, and Moderating Elections in a Federal System: West Germany,
1961-1989.” Comparative Political Studies 30(4): 420-449.
Pallarés, Francesc and Michael Keating. 2006. “Multi-level Electoral Competition: Sub-state
Elections and Party Systems in Spain.” In Dan Hough and Charlie Jeffery (eds),
Devolution and Electoral Politics. Manchester University Press, pp. 96-118.
Patzelt, Werner J. 1999. “The Very Federal House: The German Bundesrat,” in Samuel C.
Patterson and Anthony Mughan (eds,) Senates: Bicameralism in the Contemporary
World, Columbus, OH: Ohio State University Press, pp. 59-92.
25
Silvia, Stephen J. 1999. “Reform Gridlock and the Role of the Bundesrat in German Politics,” in
Joanne B. Brzinski, Thomas D. Lancaster and Christian Tuschhoff (eds), Compounded
Representation in Western Federations, London: Routledge, pp. 167-181.
26
Data Sources:
All data for electoral turnout in Germany were taken from http://www.election.de/cgi-
bin/tab.pl?datafile=btw05l.txt.
Data for Spain were taken from www.pre.gva.es/argos/archivo/index.html for regional elections
and from the Ministerio del Interior
(http://www.elecciones.mir.es/MIR/jsp/resultados/index.htm) for state-level elections.