Southeast European and Black Sea Studies

ISSN: 1468-3857 (Print) 1743-9639 (Online) Journal homepage: https://www.tandfonline.com/loi/fbss20

Economic voting in Turkey: single- vs. multi-party

governments

Ali T. Akarca

To cite this article: Ali T. Akarca (2019): Economic voting in Turkey: single- vs. multi-party
governments, Southeast European and Black Sea Studies, DOI: 10.1080/14683857.2019.1702621

To link to this article: https://doi.org/10.1080/14683857.2019.1702621

Published online: 23 Dec 2019.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=fbss20
SOUTHEAST EUROPEAN AND BLACK SEA STUDIES
https://doi.org/10.1080/14683857.2019.1702621

ARTICLE

Economic voting in Turkey: single- vs. multi-party

governments
Ali T. Akarca
Department of Economics, University of Illinois at Chicago, Chicago, IL, USA

ABSTRACT KEYWORDS
This study investigates the strength of economic voting in Turkey Elections; Voter behaviour;
under various types of governments. A vote equation developed for Economic voting; Coalition
this purpose, which takes into account all of the key factors men- governments; Governance;
Turkey
tioned in the economic voting literature, shows that voters hold
coalition governments less responsible for their economic perfor-
mance than single-party governments and minor partners of
a coalition government less responsible than its major member.
These gaps widen as political fragmentation of the government
increases. In governments involving many parties and parties with
significantly different ideologies, some of the junior coalition mem-
bers benefit rather than suffer from a bad economy.

1. Introduction
Well-informed voters that assess economic performance of governments and reward or
punish them through their ballots are essential for a well-functioning democracy and
economic system. Economic voting literature indicates that voters indeed do that.1
However, how voters make their evaluations is of utmost importance as well. For
example, economic voting literature shows that voters base their evaluations only on
the recent past, and give far more weight to growth than inflation. This provides the
politicians with an incentive to create political business cycles.2
Another issue of importance is whether voters evaluate economic performances of
single-party and multi-party governments similarly and whether they treat all of the
parties in a coalition government the same way.3 Ruling parties that are rewarded less for
a good economy will have less incentive to perform well. They will be more likely to
sacrifice economic goals for other considerations and try to get votes through populist
policies. When the parties in a coalition government are not held equally accountable,
this would create friction between them, delay critical decisions and reduce the expected
life of the government. Ruling parties with nothing or very little to lose can drag their feet
even on reforms they approve of just to deny their main coalition partner a vote gain. All
of these are things that would reduce economic performance of a government.4
Thus, it is not surprising that a number of studies investigated whether the structure of
the government effects the strength of economic voting. Powell and Whitten (1993),
Whitten and Palmer (1999), Anderson (2000), Nadeau et al. (2002), Hellwig and Samuels

CONTACT Ali T. Akarca akarca@uic.edu Department of Economics, University of Illinois at Chicago, Chicago, IL, USA
© 2019 Informa UK Limited, trading as Taylor & Francis Group
2 A. T. AKARCA

(2008) and Hobolt et al. (2013) argued that the strength of economic voting depends on
the ‘clarity of responsibility’ for economic outcomes. Under coalition governments,
especially under those too fragmented size-wise and ideologically, it becomes more
difficult for voters to assign responsibility and sanction incumbent parties for their
performance. Then, voters reduce the weight they give to the economy. Fisher and
Hobolt (2010), Debus et al. (2014), Duch et al. (2015), Stegmaier and Williams (2016),
Angelova et al. (2016), and Williams et al. (2017) found further that economic voting is
not only weaker under multi-party governments, but it is also not the same for all of its
partners. It appears that voters hold the junior members of a coalition less responsible or
not responsible at all for economic conditions than the primary incumbent party. Despite
these findings, most of the economic studies continue to treat economic voting under
single-party and coalition governments as if they are identical.
Turkey has rich electoral data. Its democracy spans more than two-thirds of a century,
during which single-party governments, as well as coalitions, partnered by two or more
parties and by ideologically compatible and incompatible parties, ruled the country. More
than two dozen studies investigated economic voting in Turkey, using these data.5 However,
none of them examined empirically whether the strength of economic voting is the same for
different types of governments and for all partners in a coalition government. Two studies
considered it but only as a side issue. Akarca and Tansel (2007) found that the economy
affects minor members of a ruling coalition less than its major member, but that study was
based on the 1995 election only. Akarca and Tansel (2006) found the impact of economic
growth on the vote shares of minor and major incumbent parties to differ, but they also found
its effect on the latter not to vary depending on whether the party rules in a coalition or alone.
However, that study used a shorter time-series and less precise government fragmentation
variables. The aim of the current paper is to fill this gap in the literature by investigating more
thoroughly, whether, and how, economic voting differs between single-party and coalition
governments in Turkey. This will be done within the context of a vote equation, which also
takes into account other key factors that affect the vote shares of parties, such as incumbency
advantage, political inertia, depreciation in the political capitals of ruling parties, strategic
voting by the electorate to balance the power of the government, and political realignments.
Next section presents this model. Section 3 provides the empirical results obtained from
fitting it to data. Section 4 summarizes the key findings and discusses their implications.

2. Modelling voter behaviour

A model that takes into account all of the key factors mentioned in the literature that
cause incumbent party vote to swing from one election to another is the following:
Vt ¼ a þ b1 Vtk þ b2 ΔLt  Vtk þ b3 ΔBt  Vtk þ b4 rt  Vtk þ b5 Z73t þ b6 D15t
þ b7 D19t þ b8 D02t þ b9 D04-11t  Qtk þ b10 NEWt þ b11 gt þ b12 pt þ b13 gt
 ðENOP-1Þt þ b14 pt  ðENOP-1Þt þ b15 gt  MIXEDt þ b16 pt  MIXEDt þ et (1)

where Δ is the differencing operator (Δ Xt = Xt - Xt-k), and the variables are defined as
follows:

Vt : vote share of the major incumbent party (or the aggregate vote share of all
parties in the government) in election held at time t,
 SOUTHEAST EUROPEAN AND BLACK SEA STUDIES 3

Vt-k : vote share of the major incumbent party (or the aggregate vote share of
all parties in the government) in the previous election held k years
earlier,
Lt : a dummy variable, which takes on the value of one if the election involved is
for local administrations or for parliament under presidential system, and
zero otherwise,
Bt : a dummy variable, which takes on the value of one if the election involved is a
National Assembly by-election only (that is, not held simultaneously with a
Senate election), and zero otherwise,
rt : number of years the major incumbent party or government was in power
since the previous election,
Z73t : a dummy variable, which takes on the value of one in 1973 election, and
minus one in 1975 election, and zero in all other elections,
D15t : a dummy variable, which takes on the value of one in November 2015
election, and zero in all other elections,
D19t : a dummy variable, which takes on the value of one in 2019 election, and zero
in all other elections,
D02t : a dummy variable, which takes on the value of one in 2002 election, and zero
in all other elections,
D04-11t: a dummy variable, which takes on the value of one in elections held between
2004 and 2011, and zero in all other elections,
Qt-k : the aggregate vote share of the independent candidates and right-wing
parties other than the incumbent Justice and Development Party (AKP) in
the previous election, or 100 minus aggregate vote share of the AKP, the
Republican People’s Party (CHP), Democratic Left Party (DSP) and the
ethnic Kurdish party in the previous election,
NEWt : a dummy variable which takes on the value of one in the 1965 and 1991
elections, and zero in all other elections,
gt : growth rate of the per capita real GDP during the four quarters preceding the
election held at time t (henceforth referred to as the growth rate),
pt : inflation rate in GDP implicit price deflator during the four quarters
preceding the election held at time t (henceforth referred to as the inflation
rate),
ENOPt : effective number of parties in government (as defined by Laakso and
Taagepera, 1979) at time t,
MIXEDt: a dummy variable, which takes on the value of one if the incumbent govern-
ment is a minority government or a coalition largely made up of ideologically
incompatible parties (a government in which at least one half of the junior
partners are from the other side of the political spectrum than the primary
incumbent party), and zero otherwise,
et : error term.
Most voters align themselves with a party that they identify as representing their
interests and ideology. The latter depends on their ethnicity, religious sect, birthplace,
and other demographic and socio-economic characteristics. Since these usually change
very gradually, most voters continue to stick with the party they voted for in the previous
4 A. T. AKARCA

election. Consequently, there is a great amount of inertia in the political system. The
lagged vote share variable is in the equation to account for that. Its coefficient b1 should
be very close to unity but less than that due to strategic voting which is discussed below.
However, although it occurs rarely, occasionally voters can change their political alle-
giances. Such a realignment took place in Turkey between 2002 and 2011, during which
time religious and economic right voters were gradually consolidated under the Justice
and Development Party (AKP), after all parties in the government and in opposition were
ousted from the parliament in 2002.6 The terms D02t and D04-11t * Qt-k are included in
the equation to approximate this process. We would expect the parameter of the first (b8)
to be negative and the second one (b9) to be positive. The value of the latter should take
on a value in between zero and one, probably much closer to the former than to the latter.
In every election, a portion of the electorate votes for a party other than their first
choice for strategic reasons. They behave this way mainly for two reasons: to check the
power of the incumbent party or parties, and to avoid wasting their vote by voting for
a party which is not likely to gain representation in the elected body. In some elections,
such as local administrations and parliamentary by-elections in Turkey, supporters of
incumbent parties get a chance to check the power of the central government, without
toppling it. Then, even more of them vote with the intention of diluting the power of the
government. Consequently, incumbent parties tend to do poorly in these types of
elections. Existence of threshold regulations in parliamentary general elections, such as
the minimum 10 percent nationwide vote share requirement to gain representation in the
Turkish Grand National Assembly, contributes to this effect as well. Some of the small
party supporters, who had voted strategically for one of the major parties in the previous
parliamentary election in order not to waste their vote, return to their first choices in
elections where no such handicaps apply, such as local administrations elections in
Turkey. In a parliamentary election, with the control of government at stake, the
incumbent party experiences fewer deserters. Furthermore, the party attracts some
supporters from its smaller ideological cousins, who fear wasting their vote if they vote
for their first choice. Therefore, holding other factors constant, we should expect the vote
losses of an incumbent party, due to strategic voting, to be higher in a local administra-
tion election that follow a parliamentary one, and lower in a parliamentary election that
follow a local administration election, and to be in between these when the two elections
involved are of the same type. Incumbent party vote losses due to strategic voting in
parliamentary by-elections should be even higher than in local elections since the control
of neither central nor local administrations are at stake then.
What is said above applies to elections held before 2017. At that date, Turkey switched
from a parliamentary system to a presidential one. As a result, now it is possible for
a voter to pick a party to form the government while simultaneously balancing it by
voting for another party for the parliament. In addition, in 2018, the election law was
changed to allow alliances between parties in parliamentary and presidential elections.
This enabled small parties to escape the 10 percent threshold by simply partnering with
one of the larger parties. In other words, voting in the parliamentary election in 2018 and
after should be treated the same way as voting in local elections.
The coefficients of the Vt-k, ΔLt * Vt-k and ΔBt * Vt-k terms help determine the degree of
strategic voting in various types of elections. (b1–1) represents the change in the vote share
of the major ruling party (or all government parties) between two pre-2018 parliamentary
 SOUTHEAST EUROPEAN AND BLACK SEA STUDIES 5

elections due to strategic voting. The corresponding change between a parliamentary and
a local administrations election or between the November 2015 and 2018 parliamentary
elections is given by (b1–1+ b2), between a local administration election and
a parliamentary election by (b1–1–b2), between a parliamentary general and by-election by
(b1–1+ b3), and between a parliamentary by and a general election by (b1–1–b3). The
expected signs of b1, b2 and b3 are positive, negative and negative, respectively.
In some elections, factors specific to those contests can make strategic voting larger or
smaller than usual. For example, in the June 2015 election, no party, including the
incumbent AKP, was able to gain a majority of the seats in the parliament. When
a coalition government could not be formed, a government crisis ensued, at a time
outlawed PKK organization resumed its terror activities. This resulted in a snap election
being called in November 2015. Many of the voters blamed the MHP for the crisis, which
refused to enter any coalition, even the care-taker election government, and the ethnic
Kurdish People’s Democracy Party (HDP), for not condemning the PKK. A lot of people
who voted for these parties in June strategically switched to the AKP in November. The
coefficient of D15t (b6) aims to capture this. The AKP and its ideological cousin MHP
entered the 2019 election in joint candidate lists. Most of these lists were under the
banner of the AKP. Consequently, the AKP supporters who would have voted strategi-
cally for the MHP could not, and many MHP supporters who would have voted for their
party were forced to vote for the AKP. The coefficient of D19 (b7) measures how much
the vote share of the AKP rose due to its alliance with the MHP.7
Ruling involves making compromises and unpopular or bad decisions, and shelving
some promises. These erode political capital of the incumbent parties. The ‘cost of
ruling’, as some refer to it in the literature, rises with the time spent in power, as
disappointments with the government accumulate. This depreciation in the political
capital of the major incumbent party (or all incumbent parties) per year spent in
power is given by b4, which should be negative. However, sometimes a government
action can incense the voters. Then, to exhibit their outrage, they may react in a bigger
way than usual. Such an incident occurred in 1973. When faced with a coup threat, the
ruling Justice Party (AP) leadership decided not to pursue in earnest a proposal granting
amnesty to the leaders of its predecessor, the Democrat Party (DP), who were banned
from politics by the 1960 junta. This infuriated many of its supporters and their
representatives, causing a faction of the party to split and form the Democratic Party
(DP2). The new party siphoned off considerable amount of votes from the AP in the 1973
election. However, after the amnesty law passed, fences were mended and these voters
returned to the AP in the following election held in 1975. The DP2 virtually disappeared
from the political scene after that. The coefficient b5 aims to capture the extraordinary
vote movements between the 1973 and 1975 elections. Its sign is expected to be negative.
Incumbency has advantages, which can offset part of the losses due to strategic-voting
and cost of ruling. Besides things like access to the media and name recognition, the
incumbency advantage involves ability of the ruling party (or parties) to indulge in transfer
activities such as providing services, subsidies and patronage, and picking locations of
government investment and public work projects to attract supporters of other parties and
mobilize its own base. The constant term in the equation captures this factor.
A change in the leadership of an incumbent party may bring an additional advantage
to the party, and offset some of the cost of ruling by wiping the slate clean. On the other
6 A. T. AKARCA

hand, the loss of the experience and talents of the departing leader may prove to be
disadvantageous. NEWt is included in the equation to capture the impacts of changes in
the leaderships of the major incumbent parties in the 1965 and 1991 elections.8 The sign
of this variable’s coefficient (b10) cannot be determined a priori.
The effects of economic growth and inflation on election outcomes are measured by
b11 and b12, respectively. We would expect the sign of the first to be positive and
the second negative. The parameters b13, b14, b15 and b16 determine whether the impact
of the economy varies with fragmentation in the government. The first two parameters
check the importance of numerical fragmentation, and the latter two, ideological frag-
mentation. If the theoretical arguments and empirical findings of studies on other
countries are any guide, the signs of these four parameters should be negative, positive,
negative and positive, respectively.

3. Empirical results
Parameter estimates of Equation (1) and its subsets are presented in Table 1 both for the
major incumbent party and for the government. These are obtained using the Ordinary
Least Squares method.9 Included in the table are also the t-statistics, R-square, adjusted
R-square, and F values for judging the fits of the equations, and Durbin’s (1970) h and
White’s (1980) chi-square statistics and their probability values for checking autocorrela-
tion and heteroskedasticity in the residuals and any misspecification in the model. The
data pools 33 national assembly general and by-elections, Senate elections, and Provincial
Council elections held between 1950 and 2019. These elections are not equidistant from
each other and thus, the data used is not a typical time-series. Table A1 of the Appendix
presents the data and its sources and explains in detail how the variables are measured.10
A comparison of regressions numbered 1 and 2 reveals that as the effective number of
parties in the government increases, the effect of the economy on the vote shares of the
government and its major partner decreases. A comparison of regressions numbered 2
and 3 shows that, while the ideological fragmentation in the government reduces the
impact of the economy on its vote share, it does not for its major partner. These findings
and the summary statistics presented indicate that regression 2 is the most appropriate
one for the major incumbent party and regression 3 for the government as a whole.
According to the second column of Table 1, a percentage point increase in the growth
rate during the four quarters preceding an election raises the vote share of the major
incumbent party by slightly more than one-percentage point if it is in power by itself.
However, this figure drops by 0.46 points for each additional unit of effective number of
parties (ENOP) in the government. ENOP varies between 1.18 and 2.88 in our sample in
case of multi-party governments. Thus, we can say that during the period examined, the
impact of a percentage growth on the leading coalition member’s vote share varied between
0.19 and 0.96 points, depending on the number of partners it had. The effect of a percentage
increase in the inflation rate on vote share of the major incumbent party is 0.13 points of
drop and is not altered significantly by a change in the number of the parties in the
government.
The regression in question shows also that the major incumbent party enjoys about
a 6-point incumbency advantage. However, the party’s vote share depreciates at the rate
of 4 percent per year while in office. In addition, it loses 13 percent of its support in the
 SOUTHEAST EUROPEAN AND BLACK SEA STUDIES 7

Table 1. Regression results.

Major incumbent party All incumbent parties
Variables 1 2 3 1 2 3
Constant 4.14 (1.30) 5.62 (1.78) 5.90 (1.68) 12.14 (1.51) 4.73 (0.71) −0.03 (0.01)
Vt-k 0.90 (13.91) 0.87 (13.26) 0.86 (12.03) 0.78 (4.96) 0.93 (7.28) 0.97 (12.93)
ΔLt * Vt-k −0.04 (3.18) −0.03 (2.60) −0.03 (2.23) −0.06 (1.88) −0.03 (1.66) −0.02 (1.92)
ΔBt * Vt-k −0.13 (5.63) −0.13 (6.16) −0.13 (5.79) −0.12 (2.16) −0.13 (3.42) −0.13 (6.24)
rt * Vt-k −0.05 (4.52) −0.04 (4.71) −0.04 (4.48) −0.05 (2.05) −0.05 (3.27) −0.05 (5.34)
Z73t −8.31 (5.10) −8.82 (6.04) −8.82 (4.66) −6.84 (1.85) −9.05 (3.52) −3.02 (1.71)
D15t 8.13 (3.63) 7.89 (3.93) 7.88 (3.68) 5.89 (1.08) 6.68 (1.82) 8.90 (4.18)
D19t 8.40 (3.62) 8.92 (4.28) 9.11 (4.04) 4.38 (0.78) 7.44 (1.95) 10.67 (4.76)
D02t −13.69 (5.73) −17.51 (6.48) −17.04 (5.56) −23.89 (4.21) −37.32 (7.57) −40.59 (12.93)
D04-11t * Qt-k 0.20 (5.47) 0.19 (5.41) 0.19 (5.03) 0.16 (1.65) 0.19 (2.81) 0.23 (5.95)
NEWt 5.80 (3.62) 5.98 (4.08) 6.15 (3.89) 6.78 (1.74) 8.86 (3.36) 10.24 (6.65)
gt 0.95 (7.10) 1.05 (8.26) 1.08 (7.35) 0.48 (1.55) 0.91 (3.78) 1.22 (8.25)
pt −0.11 (4.35) −0.13 (5.42) −0.14 (4.59) −0.14 (2.75) −0.16 (3.29) −0.12 (3.72)
gt * (ENOP-1)t −0.46 (2.54) −0.38 (1.34) −1.64 (4.70) −0.62 (2.20)
pt * (ENOP-1)t 0.04 (1.60) 0.03 (0.69) 0.15 (3.37) 0.22 (5.88)
gt * MIXEDt −0.13 (0.36) −2.05 (5.69)
pt * MIXEDt 0.02 (0.43) −0.05 (1.43)
F 104.61 113.23 89.04 11.51 23.89 67.03
Prob > F (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Durbin-h −1.28 −1.08 −1.69 1.55 0.73 0.10
Prob > h (0.10) (0.14) (0.05) (0.06) (0.23) (0.46)
White Chi-square 27.64 27.40 31.14 27.42 31.84 28.40
Prob > Chi-sq. (0.91) (0.92) (0.74) (0.82) (0.62) (0.81)
R-square 0.99 0.99 0.99 0.88 0.95 0.99
Adj. R-sq 0.98 0.98 0.98 0.80 0.91 0.97
In the first three regressions, the dependent variable (Vt) is the vote share of the sole incumbent party in case of single-
party governments and of the major incumbent party in case of multi-party governments. In the last three regressions,
the dependent variable (Vt) is the vote share of the sole incumbent party in case of single-party governments and the
aggregate vote share of all incumbent parties in case of multi-party governments. For the definitions of independent
variables, see Section 3, and for their measurement, the notes to the Appendix Table. The data covers 33 local and
parliamentary elections held between 1950 and 2019. Estimates are obtained using the Ordinary Least Squares method.
The numbers in parentheses, next to the parameter estimates, are the t-statistics in absolute value. The dark-grey
shaded cells indicate significance of the parameter estimates at one percent level, and the light-grey shaded cells, at
five percent level, in one-tailed tests.
Source: Author’s computations using the data given in the Appendix.

previous election due to strategic voting, if the two elections in question are of the same
type. This figure rises to 16 percent in local elections and parliamentary elections held
under a presidential system that follow parliamentary elections held under
a parliamentary system. It rises to 26 percent in by-elections that follow a regular
parliamentary election, and go down to 10 percent in parliamentary elections that follow
a local election and to zero in regular parliamentary elections that follow a by-election.
Furthermore, it is estimated that the extraordinary events prior to the 1973 election,
discussed in the previous section, cost the AP, the major incumbent party then, about
9 percent of its supporters which it gained back in the next election. Extraordinary events
preceding the November 2015 election brought to the ruling AKP about eight points of
extra votes. It appears that about nine of the 44 percent vote the AKP received in 2019
was its ally MHP’s contribution. In 2002, the primary incumbent party lost 17.5 percent
more votes than it should have, given the incumbency and economic circumstances
prevailing at that time. In each election between 2004 and 2011, the old right-wing parties
collectively lost about 19 percent of their remaining supporters to the newly formed AKP.
The changes in the leaderships of the major incumbent parties in 1965 and 1991 seem to
have brought them almost six-points of extra votes.
8 A. T. AKARCA

The last column of Table 1 implies that in case of single-party governments, each
additional percentage of economic growth brings to the government 1.2 points of extra
vote share, and each additional percentage of inflation takes away 0.12. Each additional party
in the government reduces the impact of a percentage growth by 0.6 points and the impact of
a percentage inflation by 0.22 points, as long as its ideology is similar to that of the lead party.
When the effective number of parties in the government exceeds two, the effect of growth on
the government’s vote share turns negative and the effect of inflation positive, even when the
ruling parties are from the same political wing. When the incumbent parties exhibit sub-
stantial amount of ideological differences, the government’s vote share becomes inversely
related to economic growth and positively related to the inflation rate, regardless of the
number of parties in the government. It looks like when they are at the opposite side of the
political spectrum than the major incumbent party, minor ruling parties are considered by
the electorate as if they are not part of the government. Probably this is due to such parties
often acting as opposition within the government or treated as such by the dominant
government party.Comparing the second and the last columns of Table 1, we can say that
for the junior members of a coalition government, strategic voting is weaker and cost of
ruling slightly stronger than for its leading member, and the incumbency advantage turns
into incumbency disadvantage.
The claims made above concerning economic voting, incumbency advantage, cost of
ruling and strategic voting are strongly supported by the t-tests given in Table 1.
However, to check their validity further and to establish their robustness, we also
compared these to the results obtained from fitting the regressions to data covering
1950 to 1999. This took the 17-year AKP rule entirely out of the sample and eliminated
from the equation four dummy variables relating to special events, which occurred
during the post-2000 period. The coefficients obtained from the subsample, which are
not listed here for brevity, were almost identical to those obtained from the full sample.
For example, the coefficients of g and g*(ENOP-1), which were 1.05 and −0.46 in
the second column of Table 1 became 0.97 and −0.48 in the regression fitted to the
subsample. Coefficients of g, g*(ENOP-1) and g*MIXED which were 1.22, −0.62 and
−2.05, respectively, in the last column of Table 1 changed to 1.21, −0.64 and −2.05 when
the regression was estimated using the subsample. Thus, we can clearly state that the key
conclusions reached in this paper do not depend on unusually long AKP rule or the
extraordinary events experienced since 2002.

4. Conclusion
There are many studies on Turkey, which establish the existence of economic voting. The
present study confirms this but goes a step further by showing that the strength of
economic voting decreases as the number of parties in the government and the ideolo-
gical distance between them increase. It also finds that voters hold minor partners of
a coalition government, especially those differing in ideology, less responsible than its
major member. The paper’s findings may shed light also on the literature studying
political determinants of economic performance. That literature finds that
a government’s performance decays as the number of parties in it and ideological
distance between them increase.11 Various reasons are given for this.12 Perhaps, the
 SOUTHEAST EUROPEAN AND BLACK SEA STUDIES 9

incentives provided by the electorate to coalition governments being weaker and con-
flicted are culprits in this as well and should be explored.

Notes
1. According to Lewis-Beck and Stegmaier (2019), this literature involves more than 600 books
and articles. Surveys of these are provided by Lewis-Beck and Paldam (2000), Akarca and
Tansel (2006, 2007), Lewis-Beck and Stegmaier (2000, 2008, 2015, 2019), Lewis-Beck and
Lobo (2017), Stegmaier and Lewis-Beck (2013) and Stegmaier et al. (2017).
2. Akarca and Tansel (2006, 2007) show that Turkish electorate’s behaviour in that regard is
not that different from their counterparts in other countries.
3. The term single-party government should not be confused with one-party state. The term
refers to a government formed solely by one of the multiple parties in the country.
4. However, the topic of economic performance under different types of governments is
beyond the scope of the present study.
5. Akarca (2009, 2010, 2011, 2015a, 2015b), Akarca and Başlevent (2009), Akarca and
Tansel (2006, 2007, 2016), Başlevent (2012), Başlevent and Kirmanoğlu (2016),
Başlevent et al. (2009), Çarkoğlu (2008, 2012), Çınar (2016), Hazama (2007, 2009, 2012,
2018), Işık and Pınarcıoğlu (2010), Kalaycıoğlu (2014, 2018), Köksal et al. (2010), Toros
(2011, 2012), and Yüksel and Civan (2013) are the noteworthy studies on economic
voting in Turkey.
6. Causes of that realignment are beyond the scope of the present paper but are discussed in
many studies such as Akarca and Başlevent (2011) and Akarca (2011).
7. As mentioned above, alliances existed in the 2018 election as well. However, unlike the
2019 election, in 2018 every party entered the election under its own banner and
candidates. Alliances mattered only in the allocation of the parliamentary seats among
the parties in the alliance. Nevertheless, the existence of alliances may have affected the
outcome of the 2018 election. To check for this possibility dummy variable for that
election was considered as well but came out highly insignificant in all of the regressions
presented in the next section.
8. Leader of the major incumbent party changed in 1993 as well. However, in preliminary
investigation of major party leadership changes individually, it was found that while the
1965 and 1991 changes made large and statistically significant effects on the vote shares of
the government and the major incumbent party, the one in 1993 did not, perhaps because it
coincided with the change in the leadership of its junior coalition partner. Similar examina-
tion of the 1995 election showed the change in the leadership of the minor incumbent party
created no noticeable impacts. The changes in the chair of the incumbent party in 2014 and
2016 were not considered leadership changes because the previous chair who became
president of the country remained as de facto leader of his party and resumed his position
officially in 2017.
9. A unidirectional causality running from economic activity to electoral outcomes is assumed.
This is justifiable in view of Luca (2016) who investigated this issue in the case of Turkey
using panel data, and found that causality in the reverse direction to be negligible.
10. The 1961 and 1983 elections are not included in the sample because then the military was
the incumbent for which there is no lagged vote share value.
11. Akarca (2019) verifies this for the Turkish case.
12. According to this literature, more inefficient projects are undertaken under coalition
governments, in particular, those involving partners with conflicting ideologies, than
under single-party governments, because under the former less of the costs are internalized
(Bawn and Rosenbluth 2006). This causes, in addition, government spending and deficit,
and thus inflation and interest rates to be higher and investment to be lower. Also, fiscal
discipline is weaker under coalitions because such governments have shorter lives and thus
shorter time horizons and lower discount factors (Bejar et al. 2011). With each member of
10 A. T. AKARCA

coalition acting as a veto player, it becomes more difficult to respond to changing condi-
tions, to correct policy mistakes, and to resist rent-seeking (Tsebelis 1995, 2002, Hall and
Nishikawa 2014). However, one should not conclude that economic growth is always better
under single-party governments. The literature studying political determinants of economic
performance shows that economic growth typically exhibits an inverted-U type of pattern
over the life of a government. All long-lasting governments in Turkey were single-party
governments. As Akarca (2019) points out, the growth rate of per capita real GDP under
such governments has been three to five times higher during their first terms than in their
later terms. This has to do more with lack of institutions providing checks and balances, in
particular lack of intra-party democracy, than voter behaviour.

Acknowledgments
This work was sponsored by the Economic Research Forum (ERF) and has benefited from both its
financial and intellectual support. However, the contents and recommendations do not necessarily
reflect ERF’s views. An earlier version of this paper, then titled “Economic Voting under Single-
party and Coalition Governments: Evidence from the Turkish Case” was presented at the 23rd
Annual Conference of the ERF held in Amman, Jordan, 18–20 March 2017. I have benefited
greatly from the comments I have received from the audience and the discussant at that meeting.

Disclosure statement
No potential conflict of interest was reported by the author.

Notes on contributor
Ali T. Akarca is a Professor (Emeritus) of Economics at the University of Illinois at Chicago with
which he has been affiliated since 1975. He has also worked as a Research Associate at the
Department of Public Policy Studies of the University of Chicago during 1978-1980. His articles
on the behavior of oil prices and their impact on the economy; the effect of tax rate changes on tax
revenues; economic, social and political determinants of election outcomes and of government
performance; causes and consequences of coups; and economic and political consequences of
internal migration are published in various academic journals and books. His analyses and
accurate predictions about Turkish elections regularly appear in the media.

References
Akarca, A., and C. Başlevent. 2009. Inter-party vote movements in Turkey: The sources of AKP
Votes in 2007. Iktisat, Işletme Ve Finans 24, no. 285: 32–47. doi:10.3848/iif.2009.285.2511.
Akarca, A.T. 2009. A prediction for AKP’s nationwide vote share in the 29 March 2009 Turkish
local elections. Iktisat Işletme Ve Finans 24, no. 276: 7–22. doi:10.3848/iif.2009.276.4172.
Akarca, A.T. 2010. Analysis of the 2009 Turkish election results from an economic voting
perspective. European Research Studies Journal 13: 3–38.
Akarca, A.T. 2011. A prediction for AKP’s nationwide vote share in the 12 June 2011 Turkish
parliamentary election. Iktisat Işletme Ve Finans 26, no. 302: 53–74. doi:10.3848/
iif.2011.302.5374.
Akarca, A.T. 2015a. Putting Turkey’s June and November 2015 election outcomes in perspective.
Insight Turkey 17: 81–104.
Akarca, A.T. 2015b. Modeling political performance of Islamist and Islamist-rooted parties in
Turkey. Middle East Development Journal 7: 49–69. doi:10.1080/17938120.2015.1019295.
 SOUTHEAST EUROPEAN AND BLACK SEA STUDIES 11

Akarca, A.T. 2019. Political determinants of Turkish government structure and economic perfor-
mance since 1950. In Economic miracles in european economies, ed. M. Osinska, 35–60. Cham,
Switzerland: Springer Nature.
Akarca, A.T., and A. Tansel. 2006. Economic performance and political outcomes: An analysis of
the Turkish parliamentary and local election results between 1950 and 2004. Public Choice 129:
77–105. doi:10.1007/s11127-005-9013-9.
Akarca, A.T., and A. Tansel. 2007. Social and economic determinants of Turkish voter choice in
the 1995 parliamentary election. Electoral Studies 26: 633–47. doi:10.1016/j.
electstud.2007.06.001.
Akarca, A.T., and A. Tansel. 2016. Turkish voter response to government incompetence and
corruption related to the 1999 earthquakes. Journal of Economic Studies 43: 309–35.
doi:10.1108/JES-07-2014-0115.
Akarca, A.T., and C. Başlevent. 2011. Persistence in regional voting patterns in Turkey during
a period of major political realignment. European Urban and Regional Studies 18: 184–202.
doi:10.1177/0969776411399342.
Anderson, C.J. 2000. Economic voting and political context: A comparative perspective. Electoral
Studies 19: 151–70. doi:10.1016/S0261-3794(99)00045-1.
Angelova, M., T. Konig, and S. Proksch. 2016. Responsibility attribution in coalition governments:
Evidence from Germany. Electoral Studies 43: 133–49. doi:10.1016/j.electstud.2016.06.004.
Başlevent, C. 2012. The region-of-origin effect revisited: More on the behavior of Turkey’s internal
migrants. Economics Bulletin 32: 112–21.
Başlevent, C., and H. Kirmanoğlu. 2016. Economic voting in Turkey: Perceptions, expectations,
and the party choice. Research and Policy on Turkey 1: 88–101. doi:10.1080/
23760818.2015.1099784.
Başlevent, C., H. Kirmanoğlu, and B. Şenatalar. 2009. Party preferences and economic voting in
Turkey (Now That the Crisis Is Over). Party Politics 15: 377–91. doi:10.1177/
1354068808097896.
Bawn, K., and F. Rosenbluth. 2006. Short versus long coalitions: Electoral Accountability and the
size of the public sector. American Journal of Political Science 50: 251–65. doi:10.1111/
ajps.2006.50.issue-2.
Bejar, S., B. Mukherjee, and W.H. Moore. 2011. Time horizons matter: The hazard rate of coalition
governments and the size of government. Economics of Governance 12: 201–35. doi:10.1007/
s10101-011-0096-0.
Çarkoğlu, A. 2008. Ideology or economic pragmatism: Profiling Turkish voters in 2007. Turkish
Studies 9: 317–44. doi:10.1080/14683840802012074.
Çarkoğlu, A. 2012. Economic evaluations vs. ideology: Diagnosing the sources of electoral change
in Turkey, 2002–2011. Electoral Studies 31: 513–21. doi:10.1016/j.electstud.2012.02.005.
Çınar, K. 2016. Local determinants of an emerging electoral hegemony: The case of justice and
development party (AKP) in Turkey. Democratization 23: 1216–35. doi:10.1080/
13510347.2015.1077228.
Debus, M., M. Stegmaier, and J. Tosun. 2014. Economic voting under coalition governments:
Evidence from Germany. Political Science Research and Methods 2: 49–67. doi:10.1017/
psrm.2013.16.
Duch, R., W. Przepiorka, and R.T. Stevenson. 2015. Responsibility attribution for collective
decision makers. American Journal of Political Science 59: 372–89. doi:10.1111/ajps.12140.
Durbin, J. 1970. Testing for serial correlation in least squares regression when some of the
regressors are lagged dependent variables. Econometrica 38: 410–21. doi:10.2307/1909547.
Fisher, S.D., and S.B. Hobolt. 2010. Coalition government and electoral accountability. Electoral
Studies 29: 358–69. doi:10.1016/j.electstud.2010.03.003.
Hall, S., and M. Nishikawa. 2014. The dynamics of within-regime stability: Party tenure and
economic performance. International Political Science Review 35: 430–47. doi:10.1177/
0192512113492261.
Hazama, Y. 2007. Electoral volatility in Turkey: Cleavages vs. the economy. Institute of Developing
Economies: Chiba, Japan.
12 A. T. AKARCA

Hazama, Y., 2009. Economic voting and electoral volatility in Turkish provinces. Institute of
Developing Economies Discussion Paper No. 202, Chiba, Japan.
Hazama, Y., 2012. Non-economic voting and incumbent strength in Turkey. Institute of
developing economies. Discussion Paper No. 340, Chiba, Japan. doi:10.1094/PDIS-11-11-
0999-PDN
Hazama, Y. 2018. Economic and corruption voting in a predominant party system: The case of
Turkey. Acta Politica 53: 121–48. doi:10.1057/s41269-017-0041-5.
Hellwig, T., and D. Samuels. 2008. Electoral accountability and the variety of democratic regimes.
British Journal of Political Science 38: 65–90. doi:10.1017/S0007123408000045.
Hobolt, S., J. Tilley, and S. Banducci. 2013. Clarity of responsibility: How government cohesion
conditions performance voting. European Journal of Political Research 52: 164–87. doi:10.1111/
ejpr.2013.52.issue-2.
Işık, O., and M.M. Pınarcıoğlu. 2010. Back to the year when it all started: Local determinants of
party preferences in 2002 Turkish elections. METU Journal of the Faculty of Architecture 27:
161–83. doi:10.4305/METU.JFA.2010.1.9.
Kalaycıoğlu, E. 2014. Local elections and the Turkish voter: Looking for the determinants of party
choice. South European Society and Politics 19: 583–600. doi:10.1080/13608746.2014.993511.
Kalaycıoğlu, E. 2018. Two elections and a political regime in crisis: Turkish politics at the
crossroads. Southeast European and Black Sea Studies 18: 21–51. doi:10.1080/
14683857.2017.1379148.
Köksal, B., A. Civan, and S. Genç. 2010. An analysis of the shift in the voter preferences in 2009
Turkish local elections. Iktisat Işletme Ve Finans 25: 9–28. doi:10.3848/iif.2010.289.2683.
Laakso, M., and R. Taagepera. 1979. Effective number of parties: A measure with application to
west europe. Comparative Political Studies 12: 3–27.
Lewis-Beck, M.S., and M. Stegmaier. 2015. Economic evaluations and electoral consequences. In
International encyclopedia of the social and behavioral sciences. 2nd. ed. J.D. Wright. Vol. 7,
26–32. Oxford: Elsevier.
Lewis-Beck, M.S., and M.C. Lobo. 2017. The economic vote: Ordinary vs. extraordinary times. In
The Sage Handbook of electoral behaviour, ed. K. Arzheimer, J. Evans, and M.S. Lewis-Beck,
606–29. London: Sage Publications.
Lewis-Beck, M.S., and M. Stegmaier. 2019. Economic voting. In The Oxford Handbook of public
choice, ed. R.D. Congleton, B. Grofman, and S. Voight. Vol. 1, 247–65. New York: Oxford
University Press.
Lewis-Beck, M.S., and M. Paldam. 2000. Economic voting: An introduction. Electoral Studies 19:
113–21. doi:10.1016/S0261-3794(99)00042-6.
Lewis-Beck, M.S., and M. Stegmaier. 2000. Economic determinants of electoral outcomes. Annual
Review of Political Science 3: 183–219. doi:10.1146/annurev.polisci.3.1.183.
Lewis-Beck, M.S., and M. Stegmaier. 2008. The economic vote in transitional democracies. Journal
of Elections, Public Opinion and Parties 18: 303–23. doi:10.1080/17457280802227710.
Luca, D. 2016. Votes and regional economic growth: Evidence from Turkey. World Development
78: 477–95. doi:10.1016/j.worlddev.2015.10.033.
Nadeau, R., R.G. Niemi, and A. Yoshinaka. 2002. A cross-national analysis of economic voting:
Taking account of the political context across time and nations. Electoral Studies 21: 403–23.
doi:10.1016/S0261-3794(01)00002-6.
Powell, G.B., and G.D. Whitten. 1993. A cross-national analysis of economic voting: Taking
account of the political context. American Journal of Political Science 37: 391–414.
doi:10.2307/2111378.
Stegmaier, M., and M.S. Lewis-Beck. 2013. Economic voting. In Oxford bibliographies in political
science, ed. R. Valelly, 1–25. New York: Oxford University Press.
Stegmaier, M., M.S. Lewis-Beck, and B. Park. 2017. The VP-function: A review. In The Sage
Handbook of electoral behaviour, ed. K. Arzheimer, J. Evans, and M.S. Lewis-Beck, 584–605.
London: Sage Publications.
Stegmaier, M., and L.K. Williams. 2016. Forecasting the 2015 British election through party
popularity functions. Electoral Studies 41: 260–63. doi:10.1016/j.electstud.2015.11.011.
 SOUTHEAST EUROPEAN AND BLACK SEA STUDIES 13

Toros, E. 2011. Forecasting elections in Turkey. International Journal of Forecasting 27: 1248–58.
doi:10.1016/j.ijforecast.2011.01.002.
Toros, E. 2012. Forecasting local elections in Turkey. International Journal of Forecasting 28:
813–21. doi:10.1016/j.ijforecast.2012.05.002.
Tsebelis, G.,. 1995. Decision making in political systems: Veto players in presidentialism, parlia-
mentarism, multicameralism and multipartyism. British Journal of Political Science 25: 289–325.
doi:10.1017/S0007123400007225.
Tsebelis, G. 2002. Veto players: How political institutions work. Princeton: Princeton University
Press/Russell Sage.
Tuncer, E. 2002. Osmanlıdan Günümüze Seçimler: 1877–1999. Ankara, Turkey: TESAV.
Tuncer, E. 2007. 22 Temmuz 2007 Milletvekili Genel Seçimleri: Sayısal ve Siyasal Değerlendirme.
Ankara: TESAV.
Tuncer, E. 2009. 29 Mart 2009 İl Genel Meclisi ve Belediye Seçimleri: Sayısal ve Siyasal
Değerlendirme. Ankara: TESAV.
Tuncer, E. 2010. 1950 Seçimleri. Ankara: TESAV.
Tuncer, E. 2011a. 1954 Seçimleri. Ankara: TESAV.
Tuncer, E. 2011b. 12 Haziran 2011 Milletvekili Genel Seçimleri: Sayısal ve Siyasal Değerlendirme.
Ankara: TESAV.
Tuncer, E. 2012a. 1957 Seçimleri. Ankara: TESAV.
Tuncer, E. 2012b. 1961 Seçimleri. Ankara: TESAV.
Tuncer, E., and B. Tuncer. 2016. 1 Kasım 2015 Milletvekili Genel Seçimleri: Sayısal ve Siyasal
Değerlendirme. Ankara: TESAV.
Tuncer, E.,, and C. Kasapbaş. 2004. 28 Mart 2004 İl Genel Meclisi ve Belediye Seçimleri: Sayısal ve
Siyasal Değerlendirme. Ankara: TESAV.
Tuncer, E. C. Kasapbaş, and B. Tuncer. 2003. 3 Kasım 2002 Milletvekili Genel Seçimleri: Sayısal ve
Siyasal Değerlendirme. Ankara: TESAV.
Tuncer, E., and H. Yurtseven. 2018. 24 Haziran 2018 Cumhurbaşkanlığı ve Milletvekili Genel
Seçimleri: Sayısal ve Siyasal Değerlendirme. Ankara: TESAV.
Tuncer, E., H. Yurtseven, and B. Tuncer. 2014. 30 Mart 2014 İl Genel Meclisi ve Belediye Seçimleri:
Sayısal ve Siyasal Değerlendirme. Ankara: TESAV.
Tuncer, E., H. Yurtseven, and B. Tuncer. 2015. 7 Haziran 2015 Milletvekili Genel Seçimleri: Sayısal
ve Siyasal Değerlendirme. Ankara: TESAV.
White, H. 1980. A heteroskedasticity-consistent covariance matrix estimator and a direct test for
heteroskedasticity. Econometrica 48: 817–38. doi:10.2307/1912934.
Whitten, G.D.,, and H.D. Palmer. 1999. Cross-national analyses of economic voting. Electoral
Studies 18: 49–67. doi:10.1016/S0261-3794(98)00043-2.
Williams, L.K., M. Stegmaier, and M. Debus. 2017. Relaxing the constant economic vote restric-
tion: Economic evaluations and party support in Germany. Party Politics 23: 286–96.
doi:10.1177/1354068815593458.
Yüksel, H., and A. Civan. 2013. The impact of economic factors on the 2011 Turkish general
election. Boğaziçi Journal: Review of Social, Economic and Administrative Studies 27: 53–67.
 14

Appendix

Table A1. Political and economic conditions: 1950-2019.

Time in Power since last
A. T. AKARCA

Vote Share (%) Previous Vote Share (%) election (years)

Provinces cov- Major All Major All Effective num- Growth Inf.
ered by the Incumbent Incumbent Incumbent Incumbent Major All ber. of parties in Rateh Ratei
Election Date Elec. Typea election Incumbent Partiesb Party Partiesc Party Partiesd Incumbente Incumbentsf gov.g (%) (%)
May.14, 1950 A 63 of 63 CHP 39.59 39.59 3.75 3.75 1.00 -3.5 -0.2
Sep. 16, 1951 B 17 of 63 DP 52.73 52.73 55.22 55.22 1.25 1.25 1.00 9.2 4.3
May 2, 1954 A 64 of 64 DP 58.42 58.42 52.73 52.73 2.50 2.50 1.00 4.8 4.9
Oct. 27, 1957 A 67 of 67 DP 48.62 48.62 58.42 58.42 3.50 3.50 1.00 3.6 20.3
Oct. 15, 1961 A 67 of 67 Military rule
Nov. 17, 1963 L 67 of 67 CHP/YTP/CKMP 36.20 45.80 36.74 64.43 2.25 1.50 2.40 6.8 5.7
June 7, 1964 S 26 of 67 CHP 40.85 40.85 36.20 36.20 0.50 0.50 1.00 4.2 4.1
Oct. 10, 1965 A 67 of 67 AP/CKMP/YTP/MP 52.87 65.10 50.28 56.81 0.50 0.50 2.45 0.1 4.2
June 7, 1966 S+B 24 of 67 AP 56.49 56.49 52.87 52.87 0.75 0.75 1.00 4.6 5.2
June 2, 1968 L 67 of 67 AP 49.06 49.06 56.49 56.49 2.00 2.00 1.00 3.7 5.3
Oct. 12, 1969 A 67 of 67 AP 46.53 46.53 49.06 49.06 1.25 1.25 1.00 2.5 6.5
Oct. 14, 1973 A 67 of 67 AP/CGP 29.82 35.08 46.53 53.11 4.00 2.50 1.28 1.7 19.2
Oct. 12, 1975 S+B 27 of 67 AP/MSP/CGP/MHP 41.34 52.98 29.82 50.26 0.75 0.50 2.37 4.4 21.0
June 5, 1977 A 67 of 67 AP/MSP/CGP/MHP 36.88 53.73 41.34 52.98 1.75 1.75 2.37 4.7 19.4
Dec. 11, 1977 L 67 of 67 AP/MSP/MHP 37.08 50.59 36.88 51.86 0.50 0.50 1.83 1.3 23.7
Oct. 14, 1979 S+B 29 of 67 CHP/CGP/DP2 29.22 31.59 41.81 43.42 1.75 1.75 1.18 -2.2 68.4
Nov. 6, 1983 A 67 of 67 Military rule
Mar. 25, 1984 L 67 of 67 ANAP 41.48 41.48 45.14 45.14 0.25 0.25 1.00 4.1 48.2
Sep. 28, 1986 B 10 of 67 ANAP 32.12 32.12 41.48 41.48 2.50 2.50 1.00 3.9 40.3
Nov. 29, 1987 A 67 of 67 ANAP 36.31 36.31 32.12 32.12 1.25 1.25 1.00 7.1 33.6
Mar. 26, 1989 L 71 of 71 ANAP 21.80 21.80 36.31 36.31 1.25 1.25 1.00 -2.5 69.2
Oct. 20, 1991 A 74 of 74 ANAP 24.01 24.01 21.80 21.80 2.50 2.50 1.00 1.5 55.2
Mar. 27, 1994 L 76 of 76 DYP/SHP 21.44 35.01 27.03 47.78 2.25 2.25 1.97 6.4 66.1
Dec. 24, 1995 A 79 of 79 DYP/CHP 19.18 29.89 21.44 39.64 1.75 1.75 1.97 5.6 87.2
Apr. 18, 1999 A 80 of 80 ANAP/DSP/DTP 13.22 35.99 19.65 34.29 2.00 1.75 1.96 -1.8 68.4
Nov. 3, 2002 A 81 of 81 DSP/MHP/ANAP 1.22 14.71 22.19 53.39 3.50 3.25 2.88 -0.6 46.4
Mar. 28, 2004 L 81 of 81 AKP 41.67 41.67 34.28 34.28 1.25 1.25 1.00 4.5 18.8
(Continued)
Table A1. (Continued).
Time in Power since last
Vote Share (%) Previous Vote Share (%) election (years)
Provinces cov- Major All Major All Effective num- Growth Inf.
ered by the Incumbent Incumbent Incumbent Incumbent Major All ber. of parties in Rateh Ratei
Election Date Elec. Typea election Incumbent Partiesb Party Partiesc Party Partiesd Incumbente Incumbentsf gov.g (%) (%)
July 22, 2007 A 81 of 81 AKP 46.58 46.58 41.67 41.67 3.25 3.25 1.00 4.6 9.0
Mar. 29, 2009 L 81 of 81 AKP 38.39 38.39 46.58 46.58 1.75 1.75 1.00 -5.6 13.3
June 12, 2011 A 81 of 81 AKP 49.83 49.83 38.39 38.39 2.25 2.25 1.00 7.3 7.6
Mar. 30, 2014 L 81 of 81 AKP 43.40 43.40 49.83 49.83 2.75 2.75 1.00 3.5 7.1
June 7, 2015 A 81 of 81 AKP 40.87 40.87 43.40 43.40 1.25 1.25 1.00 1.4 7.4
Nov. 1, 2015 A 81 of 81 AKP 49.50 49.50 40.87 40.87 0.25 0.25 1.00 1.9 7.2
June 24, 2018 A 81 of 81 AKP 42.56 42.56 49.50 49.50 2.75 2.75 1.00 4.2 12.6
Mar. 31, 2019 L 81 of 81 AKP 44.33 44.33 42.56 42.56 0.75 0.75 1.00 -3.2 17.5
a/A: National Assembly general election.
B: National Assembly by-election.
S: Senate election
L: Local administrations election (election for Provincial Councils until 2014, and for metropolitan mayors in provinces, which have metropolitan status, and for Provincial councils in other
provinces, since 2014).
S+B: Senate election plus National Assembly by-election (only in provinces whereno Senate election was held simultaneously).
In instances when different types of elections are held simultaneously or almost simultaneously, the priority for inclusion in the sample was given first to the National Assembly general
elections, next to local elections, then to the Senate elections, and last to the by-elections. The Senate and by-elections were given lower priorities because, unlike the National Assembly
general elections and local elections, they did not cover the whole country. The Senate elections involved only a third of the provinces and only a third of the seats in the Senate that were
subject to election. The coverage of by-elections were even less, about 15-27 percent of the provinces when they did not coincide with a Senate election. When the Senate and by elections
were held simultaneously, their results were aggregated to increase the coverage of the country. In such aggregation, for provinces where the two elections overlapped, the outcome of the
Senate election is considered.
b/The party listed first in the Table is the major incumbent party. The Turkish acronyms used in the table and the parties they represent are as follows:
CHP: Republican People’s Party
DP: Democrat Party
YTP: New Turkey Party
CKMP: Republican Peasant’s Nation Party
AP: Justice Party
MP: Nation Party
CGP: Republican Reliance Party
MSP: National Salvation Party
MHP: Nationalist Action Party
SOUTHEAST EUROPEAN AND BLACK SEA STUDIES

DP2: Democratic Party

ANAP: Motherland Party
DYP: True Path Party
15

SHP: Social Democratic People’s Party

 16

DSP: Democratic Left Party

DTP: Democrat Turkey Party
AKP: Justice and Development Party
A minority government formed by DSP was in power during the four months preceding the 1999 election but it was just a caretaker government. For that reason, the coalition government in
power prior to that for over eighteen months is taken as the incumbent for that election.
Of the parties listed, CHP, CGP, SHP and DSP are considered to fall in the left side, and the rest in the right side of the political spectrum. Thus, governments at the time of 1963, 1973, 1979,
1994, 1995, 1999, and 2002 elections are treated as ideologically mixed.
A. T. AKARCA

c/The vote share given for the 1975 election is for the AP, MSP and MHP only. The CGP did not enter the 1975 election.
The vote share given for the 1979 election is for the CHP and CGP only. DP2 did not enter the 1979 election.
d/The lagged vote share given for 1965 is the aggregate vote share of AP, CKMP and YTP in 1964. The MP did not enter the 1964 election.
The lagged vote share given for the June 1977 election is the aggregate vote share of the AP, MSP and MHP in 1975. The CGP did not enter the 1975 election.
The lagged vote share given for 1995 is the aggregate vote share of DYP, CHP and SHP in 1994. As the SHP and CHP merged before the 1995 election, the SHP and CHP are treated as if they
were one party in 1994.
The lagged vote share given for 1999 is the aggregate vote share of the ANAP and DSP. DTP was formed in 1997 and thus did not enter the 1995 election.
e/0.25 times the number of quarters since last election during which the major incumbent party was in power majority of time, either alone or with other parties.
f/0.25 times the number of quarters since last election during which all incumbent parties were in power simultaneously majority of time, with or without other parties.
As the CGP was formed by the merger of National Reliance Party (MGP) with the Republican Party (CP). In computing CGP’s time in power, CGP and MGP are treated as if they were the same
party.
g/Effective number of parties in government (ENOP) is computed according to the definition suggested by Laakso and Taagepera (1979):
where sk stands for the proportion of votes the kth party in government received in the previous national assembly general election, relative to the aggregate vote share of all incumbent
parties, and l is the total number of parties in the government.
During the 1964 election, a CHP minority government was in power. Since this government got the extra support it needed from the independent deputies rather than opposition parties, it is
treated as a single-party government.
For the 1965 election, CKMP and MP were treated as if they are one party because MP split from the CKMP after the previous general assembly election in 1961.
For the 1995 election, ENOP is computed treating the SHP vote share in 1991 as if it was for the CHP’s. CHP and SHP merged before the 1995 election.
For the 1999 election, DTP is assumed to be not in the government as that party did not exist in 1995 and was very small anyway.
h/The growth rate, gt, is taken as the growth rate of per capita real GDP during the four-quarter period preceding the election. The latter is obtained by adjusting the growth rate of real GDP
during the four-quarter period before the election with the annual growth rate of the population during the year of the election if the election was held in the second half of the year and
during the year before if the election was held in the first half of the year. The quarter of the election is included in the four-quarter period if the election was held in the second half of the
quarter and not, if otherwise.
For elections prior to 1989 when quarterly data was not available, gt is computed as follows:
where Gt and Gt-1 are the annual growth rates for the year in which the election was held, and the one prior to that.
m = 0.00 if the election is held between January 1 and February 14,
m = 0.25 if the election is held between February 15 and May 15,
m = 0.50 if the election is held between May 16 and August 15,
m = 0.75 if the election is held between August 16 and November 15,
m = 1.00 if the election is held between November 16 and December 31,
except for elections in 1965, 1975 and 1984, when m is taken as unity because the governments then were either not in power during the year preceding the election or were in power for less
than half a quarter.
For the year 1968, growth rate of per capita real GNP is substituted for the missing growth rate for per capita real GDP.
i/The inflation rate, pt, is taken as the growth rate of the GDP implicit price deflator during the four-quarter period preceding the election. The quarter of the election is included in the four-
quarter period if the election was held in the second half of the quarter and not if otherwise. For the elections prior to 1989, when quarterly data was not available, pt is computed as weighted
average of the annual inflation rates during the election year and the one before it, in a similar way the gt was computed as explained above
For the year 1968, rate of change in GNP deflator is substituted for the missing rate of change in GDP deflator.
Sources of data:
The dates and the coverage of elections, make-up of governments and their time in power are determined using the information given in Tuncer (2002, 2007, 2009, 2010, 2011a, 2011b, 2012a,
2012b), Tuncer and Kasapbaş (2004), Tuncer et al. (2003), Tuncer and Tuncer (2016), Tuncer et al. (2014, 2015) and Tuncer and Yurtseven (2018).
Vote shares in parliamentary elections are computed by the author, using the data provided by Tuncer (2010) for the 1950 election, by Tuncer (2011a) for the 1954 election, by Tuncer (2012a)
for the 1957 election, by Tuncer (2012b) for the 1961 election, by Tuncer (2002) for elections between 1965 and 1999 (including by-elections), by Tuncer et al. (2003) for the 2002 election, by
Tuncer (2007) for the 2007 election, by Tuncer (2011b) for the 2011 election, by Tuncer et al. (2015) for the June 2015 election, by Tuncer and Tuncer (2016) for the November 2015 election,
and by Tuncer and Yurtseven (2018) for the 2018 election. In aggregating the Grand National Assembly By and Senate elections held in 1975 and 1979, the province level vote data provided by
Turkish Institute of Statistics (TurkStat) were also utilized. It should be noted that for 1950, 1954, 1957 and 1961 elections, data in Tuncer (2002) differ slightly from those given in Tuncer (2010,
2011a, 2012a, 2012b). Here the latter are used as they are based on more detailed and more recent research. When Tuncer publishes his planned individual volumes on elections between 1965
and 1987, vote share figures on some of those may be revised as well.
Vote shares in local administrations elections are obtained from Tuncer and Kasapbaş (2004) for the 2004 election, from Tuncer (2009) for the 2009 election, and Tuncer et al. (2014) for the
2014 election. Source of data for all other local administrations elections is TurkStat. The figures given for all elections, except the one for 2014, are for Provincial General Councils. For the 2014
election, the sum of the votes cast for District Municipal Councils in 30 provinces which are officially classified as Metropolises and for Provincial General Councils in the remaining 51 provinces
is used.
The growth rates are computed by the author, as explained in note (e), using the data provided by the TurkStat for all years except 1948 and 1968. For latter two years, per capita real GNP
growth rate is substituted for the missing growth rate in per capita real GDP. In computing the former, the population growth rate, provided by the TurkStat, and the real GNP growth rate,
provided by the State Planning Organization (SPO) of the Republic of Turkey are utilized.
The inflation rates are computed by the author, as explained in note (f) above, using the data provided by the TurkStat for all years except 1948 and 1968, for which the rate of change in GNP
price deflator was used instead. The rate of change in GNP deflator is provided by the SPO.
The GDP series, from which growth and inflation rates are obtained, is 1987-based for the years prior to 1998, 1998-based for years between 1999 and 2015 and 2009-based for the years after
2015. The new 2009-based GDP series goes back only as far as 1998, and for the period after 2010, it differs from the old series substantially, not only in level, but in growth as well. For the
years 2011, 2012, 2013, 2014 and 2015, the GDP growth rates given by the new series exceed those given by the old series by 2.3, 2.7, 4.3, 2.2 and 2.1 points, respectively. For that reason, the
new series is utilized in the computation of real GDP growth rates after 2015 only, for which period the old series is not available. However, the rates obtained were reduced by 2.3 points, the
average difference between the two series during 2011–2015, excluding 2013 which is treated as an outlier. No such adjustment was made for the inflation series, as the inflation rates
obtained from the two series were very close.
SOUTHEAST EUROPEAN AND BLACK SEA STUDIES
17