announced on 24 August that a new legislative election would be
held on 1 November. As prescribed by the Constitution in case of
the absence of a government in the lead-up to an election, a provisional
election cabinet was formed on 28 August. The Constitution
stipulates that the provisional government must include
members from all parties in the parliament in accordance with
their seat shares. Arguing that Davutoglu was under the tutelage of
Erdogan who violated the constitutional constraints many times
since the elections, the CHP and the MHP refused to join the government.
Hence the cabinet led by Davutoglu consisted of 11
ministers from the AKP, 2 ministers from the HDP, a rebel MHP MP,
and 11 independent ministers. The MHP soon expelled its rebel MP
who accepted Davutoglu's invitation to take part in the cabinet
whereas the two ministers from the HDP resigned on 22 September
on the grounds that the government was pursuing aggressive
policies against Kurdish people.
References
Ahmad, F., 2008. Politics and political parties in Republican Turkey. In: Kasaba, Resat
(Ed.), The Cambridge History of Turkey, Turkey in the Modern World, vol. 4.
Cambridge University Press, New York.
Çarkoglu, A., Kalaycıoglu, E., 2007. Turkish Democracy Today. I. B. Tauris, New York.
Erdogan Urges Change in Charter due to de Facto Change in President's New Role, 14
August 2015. Hürriyet Daily News. http://www.hurriyetdailynews.com/
erdogan-urges-change-in-charter-due-to-de-facto-change-in-presidents-newrole.aspx?PageID¼238&NID¼86992&NewsCatID¼338
(accessed 20.10.15.).
Erdogan Vows to Transform Entire Political System, Again Demands ‘400 MPs’, 30
April 2015. Hürriyet Daily News. http://www.hurriyetdailynews.com/erdogan-
vows-to-transformentire-political-system-again-demands-400-mpse.aspx?
pageID¼238&nID¼81780&NewsCatID¼338 (accessed 17.10.15.).
7 Haziran: Sandık ve seçmen analizi, 18 June 2015. Konda. http://survey.konda.com.
tr/rapor/KONDA_7HaziranSand%C4%B1kveSe%C3%A7menAnaliziRaporu.pdf
(accessed 18.08.15.).
Konda Issues Apology for Poll Favoring Erdogan, 15 August 2014. Today's Zaman.
http://www.todayszaman.com/anasayfa_konda-issues-apology-for-pollfavoringerdogan_355837.html
(accessed 16.08.15.).
Opposition Decries ‛Coup’ After Erdogan's Regime Change Remarks, 16 August 2015.
Today's Zaman. http://www.todayszaman.com/anasayfa_opposition-decriescoup-after-erdogans-regime-change-remarks_396663.html
(accessed
20.10.15.).
€Oktem, K., 2011. Angry Nation: Turkey Since 1989. Zed Books, London and New
York.
€Ozbudun, E., 2015. Turkey's judiciary and the drift toward competitive authoritarianism.
Int. Spect. Ital. J. Int. Aff. 50 (2), 42e55.
Proﬁle: Fethullah Gulen's Hizmet Movement, 18 December 2013. BBC News. http://
www.bbc.com/news/world-13503361 (accessed 20.10.15.).
Selahattin Demirtas¸ : “2023 bas¸ kanlık vizyonuysa, onu da yaptırmayacagız”, 12
August 2015. CNN Türk. http://www.cnnturk.com/haber/turkiye/selahattindemirtas-2023-baskanlik-vizyonuysa-onu-da-yaptirmayacagiz
(accessed
20.10.15.).
6 s¸ irketin Cumhurbas¸ kanı seçimi anket sonucu, 15 July 2014. _Internet Haber. http://
www.internethaber.com/6-sirketin-cumhurbaskani-secimi-anket-sonucu-
697817h.htm (accessed 16.08.15.).
http://dx.doi.org/10.1016/j.electstud.2015.10.009
Only conservatives are voting in the rain: Evidence from German local
and state elections
Felix Arnold a, b, *
, Ronny Freier a, b
a
DIW Berlin, Mohrenstraße 58, 10117 Berlin, Germany
b
FU Berlin, Garystraße 21, 14195 Berlin, Germany
a r t i c l e i n f o
Article history:
Received 5 June 2015
Received in revised form
13 November 2015
Accepted 16 November 2015
Available online 30 November 2015
JEL classiﬁcation:
D72
H70
Keywords:
Turnout
Partisan effects
Rain
Germany
Municipalities
Elections
a b s t r a c t
In this note, we use data from different elections in the German state of North-Rhine Westphalia between
1975 and 2010 to show that the Conservatives proﬁt from lower voter turnout at the expense of
the Social Democrats. We deal with the endogeneity of voter turnout by using election day rain as an
instrumental variable. Our particular contribution is the comparison of municipal and state elections.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
The quality of a democracy is often judged by how many people
go to vote. High turnout is interpreted as a signal for satisfaction
with the democratic system and many organizations state the* Corresponding author. DIW Berlin, Mohrenstraße 58, 10117 Berlin, Germany.
E-mail addresses: farnold@diw.de (F. Arnold), rfreier@diw.de (R. Freier).
F. Arnold et al. / Electoral Studies 41 (2016) 213e224216
promotion of electoral participation as a goal. But what are the
consequences of higher turnout regarding election outcomes? How
is the distribution of vote shares affected? Which parties (if any)
proﬁt from higher turnout?1
We examine the partisan effects of voter turnout in the German
state of North-Rhine Westphalia between 1975 and 2010. Our
contribution is threefold. First, we are ﬁrst to document partisan
effects for Conservatives and Social Democrats in Germany e two
parties for which, given differences in voter composition, we would
a priori expect large effects. Second, we compare the effects at
different tiers of government (municipal and state elections) within
the same institutional framework. Third, we survey the literature
and compare the effects to other countries.
Importantly, we address the endogeneity of voter turnout by
using election day rain as an instrumental variable (IV).2
Our
ﬁndings suggest that a one percentage point increase in voter
turnout signiﬁcantly increases Social Democratic vote shares by
0.76 (0.69) percentage points in municipal elections (state elections)
and decreases Conservative vote shares accordingly. The IV
estimates are much larger than conventional OLS estimates, suggesting
that failing to address the endogeneity of voter turnout can
lead to substantial bias. In the ﬁrst stage, the effect of rain is
signiﬁcantly more pronounced in local elections compared to state
elections, something that is consistent with the calculus of voting.
The second stage effect of turnout on party vote shares, however, is
comparable in size for the two tiers. The effects are larger than
results for the US, but similar in size to estimates for some other
European countries.
Our work relates to the literature on partisan effects of voter
turnout (Martinez and Gill, 2005; Knack, 1994; Hansford and
Gomez, 2010; Nagel and McNulty, 1996) and to the literature on
election day weather and voter turnout (Gomez et al., 2007;
Shachar and Nalebuff, 1999; Fraga and Hersh, 2010).3
These
studies focus on the US, where it seems to be an empirical regularity
that Democrats proﬁt from higher turnout rates while Conservatives
suffer. For Europe, evidence is more scarce, although
partisan effects studies have been conducted for Spain (Artes,
2014), Italy (Lo Prete and Revelli, 2014) as well as Norway (Lind,
2014) and the effect of inclement weather on voter turnout has
been investigated for the Netherlands (Eisinga et al., 2012) as well
as Sweden (Persson et al., 2014). Table 1 in the appendix summarizes
the literature on election day rain and turnout.4
2. Data and empirical strategy
We collected panel data on all municipal and state elections in
North-Rhine Westphalia, Germany's most populous state, between
1975 and 2010. For each of the 396 municipalities, we observe 16
election outcomes: eight municipal council and eight state
parliament elections. On both levels, one electoral term lasts 5
years.
The political landscape in North-Rhine Westphalia is dominated
by two major parties: The Conservatives Christlich-Demokratische
Union (CDU) and the Social Democrats Sozialdemokratische Partei
Deutschlands (SPD). In our sample, these two parties together get
more than 80 percent of the total vote. Smaller parties like the
Table 1
Related literature: rain on election day.
Paper Country Election Time
frame
Rain on
turnout
Main ﬁnding Method
Panel 1: rain as an explanatory variable
Eisinga et al. (2012) Holland Federal 1971
e2010
e*** Rainfall depresses turnout by a rate of one half percent per centimeter RE
Fraga and Hersh
(2010)
United States Presidential 1948
e2000
e*** Rain depresses turnout on average, but not in close elections RE, FE
Gomez et al. (2007) United States Presidential 1948
e2000
e*** Election day rainfall reduces turnout in a county at roughly 0.8% per inch RE
Knack (1994) United States House 1984
e1988
e0
/þ0
No effect of rain on turnout or on partisan outcomes Logit
Arnold (2015) Germany
(Bavaria)
Mayoral 1946
e2009
e*** Rainfall reduces turnout on average, but not if the race is close FE
Persson et al. (2014) Sweden Federal 1976
e2010
e0
/þ0
No signiﬁcant effect of rain on turnout is found in three Swedish datasets OLS
Shachar and
Nalebuff (1999)
United States Presidential 1948
e1988
e** Electoral closeness stimulates party leaders' effort and thereby increases turnout;
rain negatively impacts turnout
Structural
model
Panel 2: rain as an instrumental variable
Artes (2014) Spain Federal 1986
e2011
e*** Conservatives are hurt by higher turnout whereas other smaller parties gain in terms
of vote share
IV
Hansford and Gomez
(2010)
United States Presidential 1948
e2000
e*** Higher turnout helps the Democrats, harms the incumbent, and makes vote shares
less predictable
IV
Lind (2014) Norway Municipal 1972
e2010
þ*** Higher turnout hurts the left wing parties IV
Lo Prete and Revelli
(2014)
Italy Mayoral 2001
e2010
e0
Higher voter turnout has a negative effect on city performance and the valence of
mayors
IV
Notes: À stands for a negative effect, þ for a positive effect. Stars denote the usual level of signiﬁcance. A zero marks an insigniﬁcant effect.
* denotes signiﬁcance at the 10% level. ** denotes signiﬁcance at the 5% level. *** denotes signiﬁcance at the 1% level.
Source: Own Research.
1
The effects of voter turnout are an ongoing subject of debate in the academic
literature. For recent contributions, see Michelsen et al. (2014); Pelkonen (2012);
Jewitt (2014); Bhatti and Hansen (2013); De Paola and Scoppa (2014) as well as
Martins and Veiga (2013).
2
Turnout may be endogenous due simultaneity considerations: If parties carry
out mobilization efforts, causality also runs in the opposite direction.
3
Our note also provides crucial evidence for the literature on close election RDDs
(work for Germany include Ade and Freier (2013); Ade et al. (2014); Freier (2015);
Freier and Odendahl (2015)), in which weather conditions are argued to provide
one source of election outcome randomness.
4
In a companion paper, Arnold (2015) investigates the effects of electoral
closeness on turnout. This paper also uses rainfall to investigate interaction effects
of electoral closeness and other driving forces on turnout.
F. Arnold et al. / Electoral Studies 41 (2016) 213e224 217
Greens (Bündnis 90/Die Grünen), the Liberals (FDP) or the Left Party
(Die Linke) account for the rest of the vote. Importantly, the voters
of Social Democrats and Conservatives are clearly divided with
pensioners and high-income voters favoring the Conservatives and
young families as well as the working-class favoring the Social
Democrats (Thaidigsmann, 2004). It is this relatively clear division
that may motivate strong partisan effects of turnout.
The precipitation data were obtained from the German Weather
Service (Deutscher Wetterdienst, DWD). The state-run DWD maintains
a grid of several thousand weather stations across Germany.
Fig. 1. Weather stations in North-Rhine Westphalia.
Table 2
Summary statistics.
Variable Mean Std. Dev. Min. Max. N
Panel 1: municipal elections
Election data:
SPD vote share 0.34 0.11 0.04 0.67 3162
CDU vote share 0.47 0.11 0.14 0.85 3165
Turnout 0.71 0.12 0.42 0.94 3165
Eligible Voters 32992.9 65953.14 2353 764,876 3165
Share Elderly 0.15 0.03 0.08 0.3 3168
Weather data:
Rain (mm) 1.6 4.23 0 25.8 3087
No rain 0.53 0.5 0 1 3168
Average rain (mm) 2.75 1.27 0.22 7.26 3089
Distance to next weather station (km) 7.83 4.18 0.63 21.17 3168
Panel 2: state elections
Election data:
SPD vote share 0.39 0.11 0.09 0.72 3168
CDU vote share 0.46 0.12 0.2 0.88 3168
Turnout 0.71 0.09 0.48 0.89 3168
Eligible voters 32424.24 64355.06 2463 705,339 3168
Share elderly 0.15 0.03 0.08 0.3 3168
Weather data:
Rain (mm) 1.69 3.75 0 36.3 3113
No rain 0.61 0.49 0 1 3168
Average rain (mm) 1.88 0.57 0.61 4.25 3113
Distance to next weather station (km) 7.83 4.18 0.63 21.17 3168
Notes: The tables highlights the descriptive statistics for the two independent samples of municipal elections (Panel 1) and state elections (Panel 2).
Source: Own calculations.
F. Arnold et al. / Electoral Studies 41 (2016) 213e224218
We employ data from all 121 weather stations in North-Rhine
Westphalia that consistently reported data between 1975 and
2010. Each municipality is assigned to the closest weather station.5
We thus employ variation in rain showers across time and space.
Average municipality-station distance is 7.83 km Fig.1 shows a map
of North-Rhine Westphalia with the location of all weather stations
used in the analysis.
Descriptive statistics of all variables can be found in Table 2,
separate by election type. The CDU is somewhat stronger than the
SPD in terms of vote share, with a smaller difference for state
elections. Turnout is at 71 percent on average. The smallest municipality
has 2353 eligible voters, while Cologne has more than
750,000 eligible voters. The share of elderly people (those above
the legal retirement age of 65) is approximately 15 percent. Average
rain on election day is similar for both election types and at about
1.6e1.7 mm.6
The variation in this measure is quite high, however.
In 53 percent of all municipal elections, it did not rain at all. Conditional
on some rain at all, average rain on election day then increases
to 1:6=0:53 ¼ 3:02 millimeters.
To estimate a causal effect of turnout on party vote shares, we
employ an instrumental variable approach. Turnout is instrumented
by election day rain. For this to be a valid design, two
conditions need to be fulﬁlled: First, rain has to have a signiﬁcant
impact on turnout. This condition of instrument relevance is testable.
Second, rain must not affect party vote shares through any
other channel except turnout. That is, rain is exogenous in the
outcome equation. This condition of instrument exogeneity is not
testable. However, given that rainfall is essentially random, we can
think of no reason why this assumption should not hold.
To be more precise, in the ﬁrst stage we estimate the following
relationship
Titj ¼ a1j þ b1jRitj þ X
0
itjg1j þ qij þ ttj þ uitj; (1)
where T is turnout, R is rain on election day7
(measured in millimeters),
X is a vector of covariates, q and t are municipal and year
ﬁxed effects and u is an error term. The indices i and t stand for
municipalities and (election) years, respectively. We run separate
regressions for each election level, such that j ¼ {Municipal Election,
State Election}.
In the second stage, we estimate
VS
p
itj
¼ a2j þ b2j
bTitj þ X
0
itjg2j þ qij þ ttj þ εitj; (2)
where VSp is the vote share of party p (p ¼ {SPD,CDU}), bT are the
ﬁtted values from the ﬁrst stage and ε is an error term.
The coefﬁcients of interest are b1 (Does rain affect turnout?) and
b2 (What are the partisan effects of voter turnout?). The vector of
controls X includes a third order population polynomial, the share
of elderly people and average rain on election day.
3. Results
Table 3 holds the main results. We estimate separate models for
SPD and CDU vote shares. Furthermore, we compare the baseline
OLS estimate with the IV approach for each case. Panel (1) includes
the results for municipal elections whereas Panel (2) focuses on
state elections.
The following interpretation refers to municipal elections, i.e.
Panel (1). Column (1) and (2) show a positive and signiﬁcant correlation
between SPD vote shares and voter turnout (with and
without rain as a confounding variable). To evaluate whether this
effect is causal, we resort to our IV estimates. Column (3) holds the
Table 3
Main results.
Dependent variable: (OLS) (IV) (OLS) (IV)
SPD SPD Turnout SPD CDU CDU Turnout CDU
Panel (1): municipal elections
Turnout 0.335*** 0.325*** 0.755*** À0.381*** À0.380*** À0.851***
(0.041) (0.042) (0.253) (0.046) (0.049) (0.254)
Rain in mm À0.004 À0.012*** 0.006* À0.012***
(0.003) (0.002) (0.003) (0.002)
[26.25] [26.21]
N 3162 3081 3084 3081 3165 3084 3084 3084
R2
0.57 0.57 0.95 0.54 0.49 0.50 0.95 0.46
Panel (2): state elections
Turnout 0.069* 0.063 0.694*** À0.107*** À0.099** À0.548**
(0.038) (0.038) (0.258) (0.040) (0.041) (0.262)
Rain in cm À0.006*** À0.005*** 0.003* À0.005***
(0.002) (0.001) (0.002) (0.001)
[16.67] [16.67]
N 3168 3113 3113 3113 3168 3113 3113 3113
R2
0.80 0.80 0.96 0.75 0.83 0.83 0.96 0.81
Notes: Standard errors in parentheses, ﬁrst stage F-Test statistics in square brackets. Standard errors are robust to heteroscedasticity. Signiﬁcance Levels: *p < 0.10, **p < 0.05,
***p < 0.01. The data cover all local and state elections in the 396 municipalities of North-Rhine Westphalia between 1975 and 2010. Municipal and year ﬁxed effects, a third
order population polynomial and the share of elderly people are included as controls in each speciﬁcation. Additionally, all models include a control for average rain on election
day. The coefﬁcient of the original variableRain in mm has been multiplied by 10 to ease readability. We therefore denote the variable to relate to rain in cm.
Source: Own calculations.
5
The distance between municipality and closest weather station is calculated
with the STATA package -geonear-. This routine returns the “nearest neighbor” for
each municipality, where potential “neighbors” are all weather stations in NorthRhine
Westphalia. The location of the municipality is deﬁned as the area centroid
and the location of the weather station is a single point (deﬁned by latitude and
longitude).
6
This remarkable similarity is unexpected because of seasonal differences:
Municipal elections take place in late September/early October whereas state
elections happen in May.
7
In addition, we include a dummy that takes the value 1 if it did not rain at all on
election day. We also experimented with a quadratic rain term, but the results
turned out to be insigniﬁcant.
F. Arnold et al. / Electoral Studies 41 (2016) 213e224 219
ﬁrst stage results. One can see that election day rain signiﬁcantly
reduces voter turnout. The point estimate suggests that 10 mm of
rain reduce turnout by 1.2 percentage points.8
The ﬁrst stage Fstatistic
of more than 26 suggests that the instrument is highly
relevant. Column (4) holds the results from the second stage, where
voter turnout is replaced with the ﬁtted values from the ﬁrst stage.
The IV coefﬁcient is twice as large as the OLS coefﬁcient and highly
signiﬁcant.9
It is estimated that a one percentage point increase in
voter turnout increases SPD vote shares by 0.76 percentage points.
Analogous results for CDU vote shares can be found in columns
(5)e(8). While the Social Democrats seem to proﬁt from higher
turnout rates, the Conservatives fare worse when electoral participation
is higher. The IV estimate in column (8) suggests that a one
percentage point increase in voter turnout translates into a 0.85
percentage point decrease in CDU vote shares.
A validation of these ﬁndings is provided in Panel (2), where we
estimate exactly the same relationships for a different set of elections
within the same municipalities. We ﬁnd that also in state
elections, the SPD gains and the CDU loses in terms of vote shares
when turnout is higher. The effects are only insigniﬁcantly smaller
and generally consistent with prior results in terms of sign and
signiﬁcance. The fact that we ﬁnd similar effects for turnout in two
different sets of elections and on two levels of government makes it
more plausible that the results are generalizable. A note is also
warranted on the comparison of the ﬁrst stage effects in the two
different tiers of government. Rain exerts a signiﬁcantly more
pronounced effect in municipal elections compared to state elections,
something that we believe is consistent with the calculus of
voting (Downs,1957; Riker and Ordeshook,1968). We presume that
the probability of being pivotal is quite similar (average election
district size is about the same). As more is at stake in state elections
this should make the beneﬁts of voting larger at the state level.
Now, we view rain as a cost shifter in the calculus of voting. The
induced costs of rain should be similar in municipal and state
elections, however, for municipal elections it is more often that the
shift in costs (rainfall) induces people to forfeit the beneﬁts of
voting. Note that the fact that the ﬁrst stage is smaller in the state
elections does not have consequences on the interpretation of the
second stage effects. The similarities in the second stage effects,
despite the differences in the ﬁrst stage, speak further to the general
nature of the turnout e partisan vote share relationship.
The control variables also merit a short discussion. We include
municipal ﬁxed effects in all models we estimate, using only the
within variation in the data. Electorate-speciﬁc unobserved heterogeneity
(like political culture, historical strength of a certain
party, etc.) is thus implicitly controlled for. Furthermore, we control
for location-speciﬁc precipitation patterns by including average
rain on election day as a control variable. Amounts of rain above (or
below) this average cannot be anticipated by the public and
therefore act as an exogenous cost (beneﬁt) on the act of voting.
Controlling for population size accounts for the fact that turnout is
generally lower in larger electorates (Geys, 2006). The year
dummies capture the downward trend in turnout rates that is
common to Western democracies (Gentzkow, 2006).
We show the results from six additional robustness and supplementary
tests in an online appendix. The ﬁrst part explains in
more detail how we assigned towns to the closest weather station
and what happens when we artiﬁcially decrease the number of
weather stations to obtain a less ﬁne grid. Part two shows that the
main effects in the municipal elections are driven by small towns
(with small electoral districts) and that the effects are more similar
by town size in the state elections (where the district sizes are more
equal). We also show the rain effects on other parties in part three,
notifying the reader that such an analysis is more complicated
because of missing values in many towns (and as such less efﬁcient).
Moreover, we show the full model (including all control
variables) and discuss the impact of early voting (in more recent
years), providing additional material on the ﬁrst stage effects in
different time periods. Finally, we address a potential concern that
higher turnout does not hurt a speciﬁc party, but the incumbent
party, something that is not supported by our data.10
4. Conclusion
In this note, we show that e in a setting of elections on different
levels in Germany e Social Democrats proﬁt from higher turnout
rates whereas Conservative vote shares decline if electoral participation
increases. The endogeneity of voter turnout is addressed by
using election day rain as an instrumental variable. Reassuringly,
the general nature of this effect is conﬁrmed in two independent
sets of elections at different levels of government.
Generally, our ﬁndings are mostly in line with research from the
United States, where Democrats proﬁt from higher turnout rates
(Gomez et al., 2007; Hansford and Gomez, 2010). However, in
comparison to earlier ﬁndings, our ﬁrst stage effects are quite large:
Hansford and Gomez (2010) estimate that an inch of rain (z25
mm) reduces turnout by a bit more than one percentage point e
rendering our effects almost twice as large. While we can only
speculate on the reasons for this, two potential explanations could
be relevant. First, we have a much denser grid of weather stations
that may allow for less measurement error in the rain variable.
Second, in the US, Republican and Democratic voters may not be as
strongly divided by income as poor minority groups favor Democrats
but the poor white working class is also often inclined to vote
for the Republicans.
Our results are informative insofar as they conﬁrm conventional
wisdom and help understanding observed politician behavior in
recent elections. For example, German chancellor Angela Merkel (a
member of the conservative CDU) has become well-known for her
election campaign strategy of “asymmetric demobilization”
(Denkler, 2012). By avoiding statements on controversial issues and
holding still, the CDU prevents Social Democratic voters from getting
engaged. Furthermore, the CDU puts core Social Democratic
issues like minimum wages on its own agenda, weakening potential
SPD voters' reasons to go vote. These actions also imply weak
mobilization of the own clientele, but demobilization of the
opposing political camp e potentially due to known differences in
the clientele's characteristics11
e is larger. It is hence “asymmetric”.
The result is an overall lower turnout rate that helps the Conservatives
win elections.
8
In terms of elasticities, this effect seems rather small: A one standard deviation
increase in rain (þ4.32 mm) reduces turnout by 4:32$0:0012z0:5 percentage
points, which corresponds to 0:005
0:12 z4 percent of a standard deviation in this
variable.
9
When we compare the IV (consistent model) against OLS (efﬁcient model under
H0) in a Hausman endogeneity test, we fail to reject the null hypothesis of no
systematic difference between OLS and IV coefﬁcients. This is partly surprising, as
the coefﬁcients of the two models are quite different. Due to these large differences,
we opt to still favor the IV over the OLS. Note that we also tested additional weak
identiﬁcation and underidentiﬁcation tests (Cragg Donald Wald F and the Kleibergen
Paap LM) and all come out in favor of our IV.
10
See Bechtel and Hainmueller (2011) for an account of how extreme weather
events and subsequent policy action can beneﬁt the incumbent party.
11
This also links our work to the known turnout effects of increased social
inequality, see Sch€afer (2012).
F. Arnold et al. / Electoral Studies 41 (2016) 213e224220
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.electstud.2015.11.005.
References
Ade, F., Freier, R., 2013. Divided government versus incumbency externality effect:
quasi-experimental evidence on multiple voting decisions. Eur. Econ. Rev. 64
(C), 1e20.
Ade, F., Freier, R., Odendahl, C., 2014. Incumbency effects in government and opposition:
evidence from Germany. Eur. J. Political Econ. 36, 117e134.
Arnold, F., 2015. Turnout and closeness: evidence from 60 years of Bavarian mayoral
elections. DIW Discuss. Pap. 1462.
Artes, J., 2014. The rain in Spain: turnout and partisan voting in Spanish elections.
Eur. J. Political Econ. 34, 126e141.
Bechtel, M.M., Hainmueller, J., 2011. How lasting is voter gratitude? an analysis of
the short-and long-term electoral returns to beneﬁcial policy. Am. J. Political
Sci. 55 (4), 852e868.
Bhatti, Y., Hansen, K.M., 2013. Public employees lining up at the pollsethe conditional
effect of living and working in the same municipality. Public Choice 156
(3e4), 611e629.
De Paola, M., Scoppa, V., 2014. The impact of closeness on electoral participation
exploiting the Italian double ballot system. Public Choice 160 (3e4), 467e479.
Denkler, T., 2012. So kann die SPD die Kanzlerin doch noch schlagen. Sueddeutsche
Zeitung. September 20, 2012, retrieved from. http://www.sueddeutsche.de/.
Downs, A., 1957. An Economic Theory of Democracy. Harper and Row, New York.
Eisinga, R., Te Grotenhuis, M., Pelzer, B., 2012. Weather conditions and voter turnout
in Dutch national parliament elections, 1971e2010. Int. J. Biometeorol. 56 (4),
783e786.
Fraga, B., Hersh, E., 2010. Voting costs and voter turnout in competitive elections.
Q. J. Political Sci. 5 (4), 339e356.
Freier, R., December 2015. The major's advantage: causal evidence on incumbency
effects in German mayoral elections. Eur. J. Political Econ. 40 (Part A), 16e30.
Freier, R., Odendahl, C., 2015. Do parties matter? estimating the effect of political
representation in multi-party systems. Eur. Econ. Rev. forthcoming.
Gentzkow, M., 2006. Television and voter turnout. Q. J. Econ. 121 (3), 931e972.
Geys, B., 2006. Explaining voter turnout: a review of aggregate-level research. Elect.
Stud. 25 (4), 637e663.
Gomez, B.T., Hansford, T.G., Krause, G.A., 2007. The Republicans should pray for rain:
weather, turnout, and voting in US presidential elections. J. Politics 69 (3),
649e663.
Hansford, T.G., Gomez, B.T., 2010. Estimating the electoral effects of voter turnout.
Am. Political Sci. Rev. 104 (02), 268e288.
Jewitt, C.E., 2014. Packed primaries and empty caucuses: voter turnout in presidential
nominations. Public Choice 160 (3e4), 295e312.
Knack, S., 1994. Does rain help the Republicans? theory and evidence on turnout
and the vote. Public Choice 79 (1e2), 187e209.
Lind, J.T., 2014. Rainy Day Politics: an Instrumental Variables Approach to the Effect
of Parties on Political Outcomes. University of Oslo, Mimeo.
Lo Prete, A., Revelli, F., 2014. Voter Turnout and City Performance. Siep Working
Paper No.676.
Martinez, M.D., Gill, J., 2005. The effects of turnout on partisan outcomes in US
presidential elections 1960e2000. J. Politics 67 (4), 1248e1274.
Martins, R., Veiga, F.J., 2013. Economic performance and turnout at national and
local elections. Public Choice 157 (3e4), 429e448.
Michelsen, C., Boenisch, P., Geys, B., 2014. (De) centralization and voter turnout:
theory and evidence from German municipalities. Public Choice 159 (3e4),
469e483.
Nagel, J.H., McNulty, J.E., 1996. Partisan effects of voter turnout in senatorial and
gubernatorial elections. Am. Political Sci. Rev. 90 (4), 780e793.
Pelkonen, P., 2012. Length of compulsory education and voter turnouteevidence
from a staged reform. Public Choice 150 (1e2), 51e75.
Persson, M., Sundell, A., €Ohrvall, R., 2014. Does election day weather affect voter
turnout? evidence from Swedish elections. Elect. Stud. 33, 335e342.
Riker, W.H., Ordeshook, P.C., 1968. A theory of the calculus of voting. Am. Political
Sci. Rev. 62 (1), 25e42.
Sch€afer, A., 2012. Consequences of social inequality for democracy in Western
Europe. Z. für Vgl. Politikwiss. 6 (2), 23e45.
Shachar, R., Nalebuff, B., 1999. Follow the leader: theory and evidence on political
participation. Am. Econ. Rev. 89 (3), 525e547.
Thaidigsmann, I., 2004. Sozialstruktur und W€ahlerverhalten. Das Ende einer alten
Beziehung? Arbeitspapier Konrad Adenauer Stift. Nr. 126/2004.
http://dx.doi.org/10.1016/j.electstud.2015.11.005
The parliamentary election in Poland, October 2015
Kamil Marcinkiewicz a, *
, Mary Stegmaier b
a
Universit€at Hamburg, Institut für Politikwissenschaft, Allendeplatz 1, 20146 Hamburg, Germany
b
Truman School of Public Affairs, University of Missouri, USA
a r t i c l e i n f o
Article history:
Received 6 December 2015
Received in revised form
8 January 2016
Accepted 8 January 2016
Available online 11 January 2016
1. Introduction
The October 25, 2015 Polish parliamentary election revealed
new dynamics in the Polish party system. The right-wing Law and
Justice party (PiS) became the ﬁrst party since 1989 to secure an
absolute majority of seats in the lower house of parliament, the
Sejm. The coalition parties that governed over the past eight years,
the centrist Civic Platform (PO) and the agrarian Polish People's
Party (PSL), lost substantial support and now serve in opposition.
The left-of-center parties, the Democratic Left Alliance (SLD) and
Your Movement (TR), which ran together as the United Left (ZL),
lost their parliamentary representation completely, while two new
groups entered the parliament, the liberal Nowoczesna and the
populist Kukiz'15.
2. Background
The PO-PSL coalition that governed from 2007 to 2015 was led
by the PO, with Donald Tusk as Prime Minister (PM) from 2007 to
2014, followed by Ewa Kopacz as PM until 2015. Tusk came to power
promising stability after two years of chaotic PiS-dominated
cabinets led by Jarosław Kaczynski and Kazimierz Marcinkiewicz
(Gwiazda, 2008; Millard, 2007). Tusk formed a stable coalition with
the PSL and during his ﬁrst term earned high approval ratings by
avoiding controversial decisions. As a response to the political inﬂuence
exerted on the judiciary under the PiS-led government, the
* Corresponding author.
E-mail addresses: kamil.marcinkiewicz@wiso.uni-hamburg.de
(K. Marcinkiewicz), stegmaierm@missouri.edu (M. Stegmaier).
K. Marcinkiewicz et al. / Electoral Studies 41 (2016) 213e224 221