Turkey has been rocked by an election storm. In less than two years since March 2014, the country has had four elections, the most recent taking place on November 1st, 2015. The ruling Justice and Development Party (Adalet ve Kalkınma Partisi or AK Party) won three consecutive elections in 2002, 2007 and 2011 by continuously increasing its vote share from about 34 percent to nearly 50 percent; in doing so it became a rare example of a predominant party in a competitive democracy.1 Following the local and presidential elections in March and August 2014, the AK Party incurred its first significant electoral loss in June 2015, leading to a parliamentary outlook that did not allow for the formation of any government. President Recep Tayyip Erdoğan eventually took the decision to hold early or “repeat” elections which eventually led the AK Party to recover its losses in November 2015.
This Turkish election storm is significant in many respects. We will touch upon a selection of issues that the Turkish experience could shed light upon. First, we show how a predominant party builds its electoral base, loses, and then recovers votes to consolidate its support base. We show geographical patterns of voting across the country to illustrate how the electoral scene shifted in less than four months. Second, we illustrate the power and limitations of performance politics as a force that shapes electoral outcomes in contexts where security concerns override concerns about economic and social policy performance. We argue that lacking or diminished influence of performance politics is inherently harmful for Turkish democracy and given the divided nature of the electorate a consensus building approach to policy reform and constitution writing is more likely to succeed.
The opposition coalition stuck with their erroneous calculations and was defeated handsomely by Erdoğan, who received 21 million votes or 51.8 percent of the valid votes, securing the presidency in the first round of elections
The Results of the June and November 2015 General Elections
Figure 1 summarizes the developments in the electoral scene for local, parliamentary and presidential elections since 2002. This familiar scene shows how the AK Party increased its vote share in consecutive elections in a zigzag pattern. In the initial phase a respectable 34 percent support was expanded to reach nearly 42 percent in the local elections of 2004. In the aftermath of the political turmoil prior to the 2007 elections, AK Party support was already about 47 percent. When the AK Party support was on the rise, mass public evaluations concerning the economic policy performance of the incumbent party was steadily high and even improving.
The first electoral challenge to the AK Party’s tenure came with the global economic crisis that arose just prior to the 2009 local elections, leading to about eight percentage points of electoral support decline. The AK Party recovered from this decline by continuously improving its economic condition evaluations within the mass electorate. By the 2011 parliamentary elections the recovery was more than complete and the AK Party scored a record-breaking level of support, reaching almost 50 percent of the valid votes with about 21.4 million total votes.
By the time of the 2014 local elections, AK Party performance had led to a decline in electoral support, pushing the level of support down to about 43 percent. It was clear however, that even in times of crisis with local electoral forces in play, AK Party support was still higher than its original 34 percent by about 9 percentage points. Even in one of the worst electoral contexts, the number of AK Party votes in provincial council elections was about 17.8 million, up by more than 4.4 million votes compared to the 2004 local elections.
Then came the presidential elections of August 2014. This was the first ever popular election for the Turkish presidency. So, the whole party system was unprepared for a suis generis electoral campaign necessitated by the presidential election. The coalition determined the direction of the campaign in total rejection of the fact that this was a watershed election for the future of the country. The assumption that constitutional restrictions upon the presidency could be maintained, even for a popularly elected president, was unrealistic and untenable. The two large opposition parties, the Republican People’s Party (Cumhuriyet Halk Partisi or CHP) and the Nationalist Action Party (Milliyetçi Hareket Partisi or MHP), wanted to imitate the profile of the AK Party candidate Recep Tayyip Erdoğan by jointly picking a conservative intellectual of high caliber, Ekmeleddin İhsanoğlu, thinking that he could not only keep the opposition party constituencies together but also convert some of the AK Party voters away from their candidate. The opposition coalition stuck with their erroneous calculations and was defeated handsomely by Erdoğan, who received 21 million votes or 51.8 percent of the valid votes, securing the presidency in the first round of elections.
The presidential election was a watershed election that changed the very dynamics of Turkish electoral campaigns. Now that the biggest vote winner in the Turkish electoral system became a single individual, the constraints that had regulated campaign dynamics were de facto modified. As Turkey moved into the parliamentary elections of June 2015, president-elect Erdoğan remained an active campaigner often perceived as supporting the ruling AK Party government. However, the influence of his active presence in the campaign was later questioned as being counterproductive for the AK Party’s electoral support.2
In the November 1st general elections, 85.18 percent of the Turks went to the ballots to exert their right of suffrage. | AA PHOTO / AHMET İZGİ
The details of the pre-election campaign dynamics in the form of media content analyses for the June 2015 election are yet not available for researchers to study. However, the impression a casual observer gets from the 2015 election campaign is that President-elect Recep Tayyip Erdoğan actively campaigned for the AK Party and openly criticized the three major opposing parties and their leaders.3 Among these, the Peoples’ Democratic Party (Halkların Demokratik Partisi, HDP) was perhaps the most significant. The HDP is a democratic socialist and anti-capitalist pro-Kurdish party and follows the heritage of some earlier Kurdish nationalist political parties that typically obtained about 5-7 percent of the votes in earlier elections. The co-president of the HDP and its presidential candidate Selahattin Demirtaş was quite successful in building a support base of 9.8 percent of the valid votes in the August 2014 presidential election. This strong showing signaled the voters and the leadership of all parties in the Turkish party system that the military imposed 10 percent threshold needed to secure representation in parliament might be surpassed by the HDP in June 2015. In fact, the HDP did obtain about 13 percent of the vote in June.
A significant shift in the electoral dynamics from June to November concerns the number of actual voters or turnout. We see that from June to November about one million more voters cast their votes, reaching a total of about 48.5 million actual voters. The participation rate was the highest since 1995 with 85.2 percent of the eligible voters participating, up from 83.9 percent in June.
The parliamentary election on June 7, 2015 resulted in an approximately nine percentage point loss for the ruling AK Party compared to the 2011 general election, necessitating a minority or coalition government for the first time since 1999. The AK Party nevertheless remained the largest group with nearly 41 percent, followed by the main opposition CHP with about 25 percent of the valid votes (down by about 1 percentage point from 2011). The Turkish and Kurdish ethnic nationalist parties at the opposing ends of the ideological spectrum, and the MHP and HDP jointly scored 3.3 and 6.6 percent point increases, respectively.
As a result primarily of the terror attacks and military action against the PKK, the political agenda in the country dramatically shifted between June and November and this benefitted the AK Party considerably
Since the two parties that scored significant gains in June 2015 elections, the MHP and the HDP, are at the polar opposites of the ideological spectrum, coalition negotiations proved to be difficult to conduct. The inherent ideological incongruence between the AK Party and the CHP, as well as the fact that both had incurred electoral losses, did not create a permissive ground for a grand coalition. Both the MHP and the HDP party leadership had already undertaken adversarial positions, and remained distant to the idea of a coalition with the AK Party.
The Election Context from June to November 2015
Coalition negotiations ultimately failed and the first popularly elected president of the country called for a “repeat” election. In the meantime, the stagnating peace process for the resolution of the Kurdish conflict took a sharp turn toward increased militarized action. On July 20, only six weeks after the election, a terror attack in the border district of Şanlıurfa, Suruç killed 34 activists who had been carrying humanitarian aid to Kobani. Most of these activists came from socialist platforms and youth groups. Investigations revealed that ISIL (the Islamic State of Iraq and the Levant) had conducted this attack. On the same day, the PKK also killed two soldiers in a terror attack and two days later, the PKK claimed responsibility for the killing of two police officers at their homes in Şanlıurfa. After July 20, the conflict between the Kurdish PKK and the Turkish armed forces re-intensified and the media covered this conflict extensively, aggravating the public’s security concerns.
The conflict went on from July 20 until October 10, when the largest terror attack in the history of modern Turkey occurred in the capital city of Ankara. The attack happened during a rally which was organized by leftist unions and confederations to protest the intensifying conflict between the PKK and Turkey’s armed forces. 102 people were killed and more than 400 were wounded. Among those killed were some HDP politicians and youth branch members of the CHP. The CHP and HDP announced that they would not hold rallies or large campaign meetings because of the threats. An intriguing development following the Ankara bombing was that the responsibility for the event was not clearly attributed neither by the interim government nor the AK Party to which the majority of the ministers belonged. Prime Minister and leader of the AK Party, Ahmet Davutoğlu, successfully shifted the blame to the electoral context that did not allow a single party government. After the attacks on October 10, there were no serious clashes between the armed forces and the PKK or any other terror plot against civilians that captured mass attention prior to the elections on November 1.
From June to November, the conservative constituencies of the MHP as well as the HDP appear to have shifted to the AK Party
The campaign period prior to the June elections was similarly violent. Two days before the elections, two bombs exploded in the HDP’s rally at its Diyarbakır meeting, killing four civilians. This attack was attributed to ISIL in Syria as well. But before this last attack, there had also been numerous small-scale attacks against parties. According to the Human Rights Association of Turkey (İHD) there were 187 attacks against various parties. 168 of them were against the HDP, 12 against the AK Party, 5 against the CHP and 2 against the MHP. According to the same statement, 97 party activists were wounded, 10 attackers were detained and only one attacker was imprisoned because of these 187 attacks.4
Besides these violent attacks in both June and November, there was also polarization concerning fraud in June elections. There was no substantial evidence to make a claim about fraud before the elections, but the opposition party supporters overwhelmingly expected some fraud and as an overall judgment, they claimed that election would not be fair. On the other hand, supporters of the AK Party remained confident about the free and fair nature of the elections.5 A civil initiative called Oy ve Ötesi (Vote and Beyond) monitored various provinces during both elections and according to their reports, there were no serious electoral frauds or irregularities in the ballot boxes.6
An issue potentially linked to allegations of fraud in the elections concerns the unusually large number of invalid votes in the June elections. 2.8 percent of all the votes were invalid in June. This number adds up to around 1.3 million votes. However, in November this figure decreased to 1.4 percent or 680 thousand votes. Most of this difference stems from İstanbul’s three electoral districts, where the ballot design was confusing for most voters in June. The ballot was divided into two parts; in the upper part parties were written and in the lower part candidates were written. Reportedly, this confused many voters and the percentage of invalid votes increased in İstanbul as a consequence.7 This ballot design was changed in November elections and in fact, this is what we see in our election results dataset as well. Although the percentage of invalid votes decreased in all of the country’s 970 districts, the highest decrease took place in the districts of Istanbul. The percentage of invalid votes decreased 5 points in İstanbul and only 2 points in the rest of the country.
As a result primarily of the terror attacks and military action against the PKK, the political agenda in the country dramatically shifted between June and November and this benefitted the AK Party considerably. More than half of the voting age respondents saw economic problems as the most important concern during the months going into the June election.8 Terror and national security figured most prominently for slightly less than one in ten respondents, and those who voiced the Kurdish question or peace process as the primary issue on Turkey’s agenda amounted to about one in twenty people. By early Fall 2015 however, the composition of the political agenda had been completely reshaped. Terror more than tripled its presence on the agenda. Together with the Kurdish question or peace process, terror came to occupy the forefront of the agenda for nearly one in every three respondents. While ISIL, foreign policy issues or the Syrian conflict did not gain ground, we observe that political instability increased to the primary concern for nearly one in ten of those polled.
While economic issues lost the most ground on the country’s agenda, we also see a decline in the emphasis on social problems such as education and health issues. Nevertheless, as of early September 2015, economic issues still constituted slightly more than one third of the answers to the question concerning problems on Turkey’s agenda and, as such, still occupied as large a place in the mind of the people as did terror and national security. However, as the country moved closer towards the November 1st election day and terror attacks influencing the feeling of insecurity among the mass voters increased, the prominence of terror and national security and political instability is expected to have continuously risen at the expense of economic and social issues.
From a more subtle perspective, one observes that the past economic performance of the AK Party provided a solid foundation of credibility from an economic perspective. As the salience of security concerns rose, uncertainty surrounding economic policy also expectedly increased. Given the reliable past economic performance of the AK Party, the public’s rising insecurity is likely to have boosted its credibility compared to the opposition parties that do not have a tangible performance record for the last decade.
Such security focus in the minds of the voters inevitably led to an enlarged emphasis on stability. The nature of the ideological misalignment between the political parties and their respective commitments concerning coalition options appear to have led to the conclusion that unless a significant shift were to occur, a stable coalition was not likely to result in a parliamentary setting similar to the aftermath of the June elections. As a result of this projection, a significant shift did take place, primarily amongst right-wing voters. From June to November, the conservative constituencies of the MHP as well as the HDP appear to have shifted to the AK Party. Below we will delve into the details of this shift primarily from a geographical perspective since, as of now, micro-individual level survey evidence of the nature of these shifts is not available.
Socio-economic development patterns in Turkey, which typically finds the higher levels of development in the coastal provinces, suggest that the AK Party received most of its shifting support from the most developed regions of the country
Table 1 below summarizes the raw vote shifts from June to November across geographic regions. AK Party is the only party that scored gains across all regions. All the other parties incurred some losses in at least one or more regions. We see that from June to November, the AK Party raised its votes by about 4.6 million, which constitutes nearly 9.9 percent of the valid votes cast in November, 2015. About 21 percent of these new votes came from İstanbul and nearly 65 percent of all the gains came from coastal regions from the eastern Black Sea to Thrace, the Aegean and Mediterranean provinces. AK Party gains from these coastal regions total about 3 million votes. However, the total vote loss of the MHP, HDP, SP and BBP is about 1.8 million votes. Considering the slight gain of about 221 thousand votes for the CHP in the coastal provinces, we see that these additional votes predominantly came from new voters in these provinces who did not vote or whose votes were invalid in the June elections.
In comparison, the CHP had about half a million more votes in November and all the rest of the parties effectively lost votes. What is striking in this picture is that those parties which did lose votes from June to November did so mostly in the coastal regions. Even the HDP, which typically gets most of its support from the Eastern and Southeastern Anatolia regions instead of the coastal regions, lost more votes in the coastal regions than in its core support regions in the east and southeast. Similarly, the MHP losses mostly came from the Aegean and eastern Marmara regions.
Another important observation concerns the losses that came from the conservative SP and BBP electoral base. A total of about 366 thousand voters who voted for either the SP or the BBP did not cast their vote for these two parties in November. About 75 percent of these vote switches took place again in the coastal regions. Given the size and more volatile nature of the population of the coastal regions, the predominance of vote changers being found in these provinces is hardly surprising. However, socio-economic development patterns in Turkey, which typically finds the higher levels of development in the coastal provinces, suggest that the AK Party received most of its shifting support from the most developed regions of the country. However, we should note here that by looking solely at the geographic election returns it is simply not possible to infer which party constituency shifted to which parties. This is a well-known fallacy in data analysis. Simply put, looking into the aggregate vote switches we cannot infer anything about the individual level. Voters who chose the MHP in June are reduced in November. That we know. However, did they predominantly switch to the AK Party? How about those SP and BBP voters? It is conceivable that some SP/BBP voters in June did cast their vote in November for the MHP. However, some MHP voters might also have voted for the CHP as well as switching to the AK Party. We only observe the final election result across geographical locations and cannot match any of these with individual choices in an earlier election. Only with individual level survey data such inferences are appropriate. Even with those data one would need to have the same individuals observed at different points in time in a panel design and compare their party choice declarations from one point to the other to make inferences about individual level shifts. With only one cross-sectional observation one could obviously obtain measurements about vote choice at two points in time. However, at the time of writing this article such data were not available to us.
Source: Calculated by the authors based on official ballot box level data from the Higher Election Council's Webpage (www.ysk.gov.tr).
Provincial level data is summarized in graphical form for the AK Party vote switches in Figure 2. The vote share of the AK Party in the June election is shown on the horizontal axis and the November results are shown on the vertical axis. As such, any point lying above the main diagonal will show an increase in AK Party support from June to November. Indeed, only in Tunceli do we see no change in AK Party support from June to November. In all other provinces we do observe some positive change. The east and southeastern Anatolia provinces are shown as a group on the graph. The provinces within this group exhibit positive shifts in the AK Party vote that range between 28 to 62 percent from June to November. One exception is the case of Iğdır where the AK Party vote is more than 2.6 times higher in November (with about 31 percent) than in June (with about 12 percent). Within this group, Gaziantep, Şanlıurfa, Kilis and Bingöl had 47 to 49 percent support for AK Party in the June elections, increasing to 61-65 percent in the November election. As such these four provinces form a sub-group of provinces in southeastern Anatolia where AK Party support is particularly high.
In the four provinces where the AK Party obtained more than 60 percent of the vote in November (Rize, Konya, Kahramanmaraş and Bayburt), one would expect that a ceiling effect would be in place to limit the vote increases in the November elections. However, gains in the AK Party support in those four provinces ranged between 14-21 percent reaching 72-76 percent.
The second group of provinces shown in Figure 2 brings together most of the Western Anatolian provinces, where the range of vote share for the AK Party in the June elections was between 23 to 48 percent. These provinces increased their range to 27-57 percent in November. Table 2 provides a summary of this linkage between the range of support in June and vote increase in November. The overall trend is, as expected, a negative one. As the range of support for the AK Party in June increases, the average percentage of increase in its vote share in November declines. For those provinces where the AK Party obtained 10 to 20 percent of support in June, the average gain is about 51 percent. However, among the four provinces where the vote share is above 60 percent, the average gain in November is only 17 percent over the June vote share.
Table 2: Range of Change in the AK Party Vote Share between June and November Elections
Looking into the same picture of vote shifts from June to November we can also use sub-provincial district level data from both elections. Below we use Google Maps and Google fusion tables to plot vote differences from June to November at the district level. Green districts indicate a positive vote change while red districts are where parties lost votes moving from June to November. Darker colors indicate higher percentages of vote gain or loss. For ENP (Effective Number of Party) and volatility maps, green districts indicate higher ENP and volatility while red indicates lower statistics.
Figure 3: AK Party Vote Shifts from June to November at District Level
Compared with other parties, the AK Party was able to gain new votes in almost every district (see Figure 3). The overall pattern of vote share changes in November compared to the June election results show that the AK Party scored the highest vote share increases in the Eastern and Southeastern Anatolia regions. However, pockets of high gain districts are also found in and around Bayburt, Gümüşhane, Trabzon, Sivas, Kayseri, Karabük, Kütahya and Afyon. There were only eight districts in which the party lost votes. These were Diyarbakır Dicle, five districts of Tunceli such as the central district, Nazımiye, Ovacık, Pülümür, Mazgirt as well as Datça from Muğla and Beytüşşebap in Şırnak. In every other district in the country, the party increased its vote percentage. On average, the party was performing below its country average in these eight districts where the AKP got 9.8 percent in June and 8.5 percent in November. On the other side of the spectrum are districts where the party increased its votes. There are 8 districts where the party increased its vote more than 20 percentage points. These are Hamur and Eleşkirt from Ağrı, Köprüköy from Erzurum, and the central district of Iğdır. All of these districts are located at the northeastern part of Turkey. In addition to these, there were four districts of Şanlıurfa, namely Hilvan, Siverek, Harran and Akçakale where the AK Party increased its votes more than 20 points. It is especially curious to see the increase in Harran and Akçakale where the ongoing conflict with ISIL has spilled over from the Syrian border. On average, the AKP got around 34 percent of the votes in June while it increased its support to 60 percent in November in these eight districts.
Figure 4: CHP Vote Shifts from June to November at District Level
An interesting geographical feature of the AK Party’s performance is the fact that most of the districts where the party under- as well as over- performed in November are from the east or southeastern part of the country. Five of the nine districts where the AK Party lost votes are from Tunceli where Alevi voters form the majority. On the other hand, all of the over-performing districts are conservative districts with an AK Party constituency. Some of these districts were predominantly pro-Kurdish HDP supporters, such as districts from Ağrı, Erzurum and Iğdır –but the two border districts with Syria did not support the HDP. On the contrary, the MHP lost more than 20 percent points in both districts.
The HDP’s gains were not considerable but it lost substantial support in the northeastern and southeastern regions
The overall pattern of vote changes for the CHP indicates that, although very modestly, the party gained a few votes in most of the east and southeastern provinces. However, except for several districts in Tunceli, Ardahan and Edirne, the CHP did not gain considerable votes in any of the districts. In several districts of the Kahramanmaraş and Kars provinces, the party was able to increase its share; but given its poor performance, this increase did not translate into any new seats in the parliament. The CHP lost votes in 490 districts out of 970 and on average its loss was 1.6 points in these districts. There are 10 districts where the party lost more than 6 percentage points. Half of these districts are from the Black Sea region. The largest underperformer was the Karabük Yenice district where the party lost more than 10 points. Other similar districts from the Black Sea are Kastamonu Doğanyurt (-9.5 percent), Samsun Tekeköy (-7.2 percent), Ordu Altınordu (-7.1 percent) and central Rize (-6.1 percent). Other districts outside of the Black Sea region were Mersin Gülnar, Adıyaman Besni, Isparta Aksu and Eskişehir Beylikova.
On the other side of the spectrum, the CHP increased its votes more than 6 points in 19 districts. Five of these districts are from Tunceli, the same districts where the AK Party was the underperformer. Aside from these districts in Tunceli, the CHP increased its vote 13 points in Hatay Samandağ, 12.4 points in Ardahan Hanak and 11.8 points in Kırklareli Pehlivanköy. All of these districts are close to different borders of the country but they do not share any other feature besides this. Nonetheless, the CHP increased its votes from 31 percent to 40 percent in districts where it performed better in November with a 6 point or more increase in its vote share. Except for the predominantly Alevi districts of Tunceli, these districts do not share any socioeconomic or demographic features.
The MHP lost more than 4 percentage points in November across Turkey in total and lost half of its seats in the parliament. In the November election, the MHP won new votes only in 19 districts and on average it won 1.2 percentage points in these districts over its vote share in June. It increased its vote share from 11 percent to 12.1 percent on average in these districts. Eight of these districts are in the western and inner Anatolian regions. Four of them are in Ardahan province and the rest are in eastern provinces such as Kars, Siirt, Bingöl, Van, Malatya and Şanlıurfa. However, on average, these were outlier districts as the party lost considerable support in the rest of the country. There were 11 districts where the party lost 15 percent points or more in November compared to June. In these districts, the MHP got 33 percent of the votes in June while it could only get 13.7 percent in November, losing around 20 percentage points, or more than 58 percent of its vote in five months. In these same 11 districts where the MHP losses were highest, the AK Party increased its share from 50 to 64.6 percent, making up for most of what the MHP lost. Geographically, most of these districts were underdeveloped, conservative districts in the Anatolian heartland, such as 4 districts of Yozgat, one district from Konya, Karaman and Kilis provinces and two border districts of Şanlıurfa, Harran and Akçakale, which were also the two largest over-performers for the AK Party. There were also 66 additional districts in which the MHP lost between 15 to 10 percent of the votes in November compared to June. In addition to the 4 districts in Yozgat where the MHP lost more than 15 percent points in June, 9 of these districts were also from Yozgat. As such, in 13 Yozgat districts the MHP losses were above 10 points in November, implying that voters in Yozgat switched from the party at large. The other provinces with various districts that stopped supporting the MHP were conservative, Anatolian districts from Aksaray, Erzurum, Bilecik, Afyon and Kayseri.
Figure 5: MHP Vote Shifts from June to November at District Level
The HDP also lost support similarly to the MHP (see Figure 6). The party gained new support only in 22 districts and as a curious case, it received exactly the same vote share in both the June and November elections in Kastamonu Ağlı. In districts where it gained new votes, it gained 0.54 percent new support on average except from Hatay Samandağ where it increased its vote from 12.46 percent to 19.34 percent. Samandağ is also the district where the CHP increased its votes by 13 points. In fact, all of the largest four parties increased their votes in Samandağ. This is because Refik Eryılmaz, who ran as an independent candidate in June was very much supported by the Arab Alevi population of the district. Eryılmaz was an active politician opposing the government’s Syrian policy and he even met with President Bashar al- Assad before the elections.9 In Samandağ he earned around 19.4 percent of the votes back in June. This totaled up to 2.2 percent of all the votes in Hatay. His candidacy prevented the CHP from winning an additional seat for Hatay province in June. His decision to run as a CHP candidate in November changed the electoral decisions of voters in Samandağ, and while most of the Arab Alevis supported the CHP against the ruling AK Party’s Syrian policy, some of the votes appear to have gone to the HDP as well. Aside from this exceptional district of Hatay, the HDP lost around 3 percentage points in November. There were 8 districts in northeastern Anatolia from provinces such as Elazığ, Ardahan, Bingöl, Erzurum and Ağrı where the party lost considerable support. Its support decreased from 54 percent to 35 percent in these districts. There were also 43 districts where the HDP lost between 15 to 10 percentage points. These 43 districts are from 14 provinces and, in Şanlıurfa alone, there are 6 districts in this category. In Bingöl, Diyarbakır and Van each, there were 5 such districts. In summary, the HDP’s gains were not considerable but it lost substantial support in the northeastern and southeastern regions. However, its loss cannot be explained by the Kurdish cleavage only, due to the fact that in the heartland of the Kurdish population in provinces such as Hakkari and Şırnak, the party remained strong. The HDP lost support predominately in districts where a conservative tendency remained strong and thus voters could choose the AK Party over the HDP.
Figure 6: HDP Vote Shifts from June to November at District Level
Table 3 below summarizes parties’ performances in 970 districts, comparing November to June elections.
We also plotted the effective number of party (ENP) measurements for both elections, as well as volatility from June to November.10 ENP measures the number of different parties, weighted by their relative electoral performance. It is calculated by summing the squared vote shares and then taking its reciprocal.
It gives a comparable figure for the level of competitiveness in a given electoral district. Higher figures imply greater competition while lower figures mean that fewer parties get a larger share of the vote. The ENP is only equal to the total number of parties if all parties were to receive the same share of vote. For instance, if five parties got 20 percent in a district, ENP would be 5 (ENP=(1/(0,04+0,04+0,04+0,04+
Table 4: Changes in the Effective Number of Parties from the June to November 2015 Elections
Table 4 indicates that there are no districts which had an ENP of 4 or above in November whereas in June there were 12 such districts. In fact, 11 of the 12 districts which had such a high competition in June decreased to a 3-4 ENP level and the other one (Malatya Akçadağ) decreased to 2.69 (2-3 category) in November. This extreme decrease indicates a peculiar situation and can be explained only by referring to the specifics of the district. Just like the peculiar HDP performance in Hatay Samandağ, the decrease in competitiveness is due to an independent candidate. Münir Doğan Ölmeztoprak ran as an independent in June and was able to get 13.21 percent of all the votes in Akçadağ. Although this was not enough for him to be elected from the larger Malatya electoral district, his decision not to participate in the November elections excited local AK Party and MHP officials who competed for votes coming from Ölmeztoprak. Thus, a large chunk of votes in Akçadağ was transferred to other parties which were already competing in the district in June, which caused the high decrease of competition in the district.
The November “repeat” election was a true victory for the AK Party, which was able to raise its total number of votes to an all-time high of almost 23.7 million
In 101 districts, the ENP increased from June to November but the average increase in these districts was a minute 0.16. The number of districts which had a higher competition, measured by an ENP difference of 0.3 or higher, was 15. These were from provinces of Van, Kars, Iğdır, Erzurum, Diyarbakır, Ağrı, Bingöl and Siirt. In other words, the ENP increased in eastern districts dominated by the HDP in June. This increase was caused by a higher competition between the HDP and the AK Party. In 869 out of 970 districts, the ENP decreased, implying that there was less electoral competition due to a strengthening of electoral support behind fewer parties in the system. The biggest decrease occurred in Akçadağ as explained above; there were five districts in total from Erzurum, Ardahan and Şanlıurfa where the ENP decreased for a full point or more, which is analogous to wiping away a party from the competition. Again, these districts were from eastern Turkey. In both elections, the region had less competition compared to the rest of the country and both the biggest increase and decrease also happened there as well. Even if there were increases in the ENP because of the AK Party’s better performance, southeastern Turkey in general had lower level of party competition compared to the rest of the country.
More salient concerns became the restoration of stability and the punishment of the opposition parties who were seen as responsible for the failure of coalition negotiations (in the case of the MHP) and rising conflict on the Kurdish front (in the case of the HDP)
Volatility shows us where vote shifts occur. Electoral volatility is also called the Pedersen Index12 and it measures the extent of vote shifts between parties. It ranges from 0 (meaning that all parties received the same vote share) to 100 (all parties in the system ceased to exist and totally new parties were formed). Since volatility measures changes in votes for all of the parties from one election to the next, measuring it necessitates continuity in the parties participating in consecutive elections. However, five minor parties did not participate in November, namely Anadolu Partisi, HAP, MEP, TURKPAR and YURT.13 This lack of participation automatically inflates volatility figures somewhat for districts, but it should be noted that these small parties had negligible support in June (0.3 percent in total).
The average volatility at the district level was 10 percent of votes, but it varied from 0.7 percent to 35 percent. Districts where the MHP and HDP lost considerable support are those where volatility is high. In 20 districts, volatility was 20 percent or higher. Three of these districts were central districts of Kilis, Yozgat and Iğdır provinces. Also, there were four districts from Şanlıurfa, three from Yozgat, two from Ağrı and Ardahan and one district each from Bingöl, Bitlis, Diyarbakır, Erzurum, Hatay and Karaman. Comparing the volatility map with the AK Party’s vote shifts indicate that the party benefitted from volatility in most of the Inner Anatolian districts where MHP support diminished, and the eastern districts where the HDP lost support.
The November “repeat” election was a true victory for the AK Party, which was able to raise its total number of votes to an all-time high of almost 23.7 million. With rising turnout and declining invalid votes, the AK Party’s supporters peaked. A significant share of these votes appears to have come from the conservative constituencies of the MHP and HDP voters. However, a non-trivial sum of votes is likely to have come from the SP and BBP, as well as those who did not vote in June or whose votes (especially in Istanbul) were counted as invalid in June.
The geographic pattern that emerges from examining the June to November shifts in election outcomes show that while the AK Party scored gains in almost all districts, in several pockets its performance was much more significant. This is likely to be attributable to effective candidate selection and voter mobilization by the AK Party. However, the positive upward swing across all geographic regions is likely to be attributable to the shifting political agenda of the country.
The AK Party built its predominant party position primarily by delivering a successful performance on the economic front. As perceptions of its economic policy performance lagged behind earlier periods in the AK Party’s tenure since 2002, the electoral performance of the party suffered repeatedly in the 2009 local as well as the June 2015 elections. It appears that the opposition parties have caught up with the AK Party on this front and were well prepared to argue on the economic policy front and make policy proposals that appear to have resonated well with the masses in the June elections. Yet, the agenda of the country changed dramatically between June and November due to rising tensions on the Kurdish and Syrian front, with terror attacks on civilian as well as military targets. These developments, together with intensifying threats and attacks on political parties, impeded an effective campaign by the opposition which also got stuck with economic policy discourse and an uncooperative image that portrayed them as responsible for the rising political instability in the country.
The inability of effective campaigning and the rising feeling of insecurity among the voters inevitably kept performance considerations and responsibility attribution aside in the minds of the electorate. More salient concerns became the restoration of stability and the punishment of the opposition parties who were seen as responsible for the failure of coalition negotiations (in the case of the MHP) and rising conflict on the Kurdish front (in the case of the HDP). Both of these parties lost considerable electoral support as a consequence.
Nevertheless, both the MHP as well as the HDP managed to survive this backlash and remained above the critical 10 percent threshold, enabling them to remain in parliament. Passing the threshold was especially important in curbing the AK Party’s legislative power. The parliamentary seat distribution after the November elections leaves the AK Party only 13 seats short of a constitutional referendum. It remains to be seen if the AK Party will form an alliance with one of the opposition parties to write a new constitution or if it will try to persuade several individual legislators to support the party’s revived constitutional ambitions.
The strategic choice currently lying before the AK Party means a lot for the future direction of Turkish democracy. Given the parliamentary seat distribution in the aftermath of the November elections, an inclusive grand coalition as the foundation of support for the new constitution may be difficult to create. However, a majoritarian approach that would unilaterally push a new constitution through, or the imposition of critical amendments that would have to be taken to a referendum would likely further polarize the country. The experience of the 2015 election campaigns provide clues as to the nature and likely outcomes of a polarizing strategy.
Various AK Party elites have been supportive of a presidential system since Erdoğan won the presidential election last year. However, both the CHP and the MHP have fervently opposed such a fundamental change to the Turkish political system. Only the HDP has shown mixed signs of support and even in their case, their provisional approval rests on the condition of local government reform and more local freedoms for Kurds in the country. Given the constrained relationship between the AK Party and the HDP, the AK Party may not get the HDP’s support for a presidential constitution in the foreseeable future unless the government relinquishes some of its centralized political power to local governments.
While polarization has been on the rise for well more than a decade, these were eased by the AK Party’s tenure in power and its bold emphasis on performance politics
The significant electoral shift from June to November was unexpected for pundits and scholars. As noted above, Turkey has gone through an election storm and this storm has culminated in a dramatic finale. More research is necessary to understand why an unexpectedly high portion of Turkish voters changed their minds in just five months. The usual suspect is security concerns in an increasingly conflict-hit Middle East. As we explained above, Turkey experienced the worst terrorist attack in its history three weeks before the November elections. It is clear that the resulting security concerns were not beneficial to the opposition parties, especially when these parties gave the impression of not being competent enough to form a government after the June elections. Moreover, these security concerns were coupled with the waning importance of performance politics. Rather than focusing on punishment for the incumbent party for worsening security and continuing economic uncertainty, voters appeared to be concerned about a more basic ontological issue: their personal safety and the public order. The AK Party appears to have been perceived as competent in these policy areas and it benefitted from the voters’ confidence in the party to solve Turkey’s security issues.
What does the rise of security and other ontological concerns mean for the future of Turkish democracy? Polarization along secular vs. conservative lines, or Turkish vs. other minority identities such the Kurdish identity or simply the so-called center vs. the periphery in the cultural context, have all been observed over the past few decades.14 Many of the anxieties at the heart of polarized debate concern life-style trepidations or perceived threats. As such, ontological concerns are not new to Turkish politics.15However, while polarization has been on the rise for well more than a decade, these were eased by the AK Party’s tenure in power and its bold emphasis on performance politics. Without acceptable performance providing for the demands and expectations of the masses a government could not stay in power. The last experience in the November elections however, could be taken as evidence of electoral success without a noticeable performance advantage. The AK Party’s electoral success in November appears not to be driven by performance evaluations but rather by its successful management of the changing agenda, its credibility in the face of an uncertain future, and a de facto constrained campaign effort on the part of the opposition.
The media is a crucial tool for fair elections in a democratic setting. However, polarized and slanted media has the inverse effect of creating echo chambers and further distances different voter groups from each other. From June to November the agenda of the election campaign shifted from performance evaluations concerning the economy to an increasingly polarized debate around security concerns. In a sound democratic context such difficulties are expected to be handled with a balanced public debate in the media. However, the Turkish media sector is already divided and polarized.16 The access of the opposition parties to media outlets remained limited and the AK Party effectively controlled the media coverage of the campaign. Although the specific extent of the AK Party’s control is yet to be obtained from the content analysis of the media outlets, our earlier analysis of the 2011 campaign clearly shows that rising polarization in a media setting favors the incumbent party.17 In the June as well as the November elections, the opposition parties’ access to media coverage was severely limited even in the public outlets. A favorable incumbent-party bias was also evident among the private channels.18 Hence, although the claims of fraudulent practices in the elections remained unfounded, given the condition of the media sector and pressures and attacks on all political parties, the fairness of the elections is likely to remain questionable. Such constraints upon the political parties and the media at large can only mean deteriorating democratic standards in the country.
At a higher level of abstraction, the rising violence and pressures upon democratic competition in the most recent Turkish elections could only be taken as a warning for the well-being of Turkish democracy. Interpersonal trust in the country is amongst the lowest in the world.19 Tolerance concerning minorities, unconstrained freedoms of expression and social progress is already very low. In a polarized political setting such a background creates a deep mistrust between groups and especially between the government and its opposition. On many surprising occasions, such as the polarized and eventually unfounded expectations of election fraud noted above, a wide cleavage of trust or mutual lack of trust is repeatedly observed. Given the public’s ontological security concerns, together with rising economic uncertainty under increasingly more self-imposing conservatism, a free and competitive democracy may not be viable. Mass support for minority rights, freedom of speech and political expression, or simply put, a modern empathetic tolerance as the foundation of Turkish democracy may not be present. The developments observed over the course of the 2015 election campaign could be a warning in this respect. For an optimist, the new AK Party government can be seen as having a fresh new mandate for economic and democratic reforms. However, whether the international and domestic security environment will allow the AK Party leadership to pursue an inclusive, cooperative consensus-building reform process remains to be seen.
Appendix on Data Sources, Visualization and Calculations
of ENP and Volatility
We used official ballot box level election results for both the June and November 2015 elections. Data were taken from the Higher Election Council’s online results page. These results are published based on the tallying of votes on Election day. The number of ballot boxes in a district depends on the registered population. On average, there were 310 voters per ballot box in both elections and the median was 355 voters. The number of ballot boxes per district ranged from 6 in Konya Yalihüyük to 1,868 ballot boxes in Ankara Çankaya. In order to conduct our analyses on the district level, ballot box data were aggregated to the district level. This aggregate data does not include votes distributed to the electoral districts from votes cast outside the country.
Figures 3 to 6 were created using Google Maps and Google fusion tables. After 12 provinces became metropolitan municipalities in 2012, the number of districts increased from 957 to 970. However, the necessary geocoding and KML files for these new districts are still not publicly available. Therefore, we used 957 districts in our maps. For the newly created 13 districts, we reshaped the data and aggregated newly created districts with other new ones to account for the difference.
Effective Number of Parties (ENP) was calculated using Laakso and Taagepera’s (1979) measurement. It is an inverse Herfindahl index, also called an inverse Simspon index which measures level of party competition and diversity. The formula for calculating ENP is:
where n is the number of parties competing in the district and p is the vote share of each party.
In addition to ENP, we also measured the Volatility Index, also called the Pedersen Index (1979), which measures the total number of vote shifts in two consecutive elections. The formula for Volatility index is:
where n is the number of parties competing in the district, p is the vote share for each party and t is the latest election. Since one party’s gain is another party’s loss, sum of the absolute vote differences in consecutive elections is divided by two. The index ranges from 0 (no shifts in the district) to 100 (all the votes shifted in the district).
- On the predominant party system in Turkey, see Ali Çarkoğlu, “Turkey’s 2011 General Elections: Towards a Dominant Party System?” Insight Turkey, Vol. 13, No. 3 (2011), pp. 43-62; Şebnem Gümüşcü, “The Emerging Predominant Party System in Turkey,” Government and Opposition, Vol. 48, No. 2 (April 2013), pp. 223-244.
- See, Özge Kemahlıoğlu, “Winds of Change? The June 2015 Parliamentary Election in Turkey,” South European Society and Politics, Vol. 20, No. 4 (2015), p. 453; Muqtedar Khan, “Reflections on Turkey’s Parliamentary Elections of 2015,” Turkey Agenda(2015), retrieved from http://works.bepress.com/
- Such an intervention that created a double election campaign by the AK Party might have backfired, pushing some of the potential AK Party supporters towards other parties. This is a difficult claim to test empirically since it involves a comparison with a hypothetical circumstance where the president remained neutral during the election campaign. Otherwise all other evaluations are likely to be driven by partisan bias and do not correspond to a true test of the hypothesis that if the President remained neutral, the AK Party would have gotten a higher percentage of support. See http://america.
8/akp-loss-leaves-erdogan-and- his-former-party-floundering. html
for such claims. For the specifics of Erdoğan’s approach to the opposition parties before the June elections, see: http://www.voanews.com/
content/in-turkey-erdogan- election-rhetoric-draws-ire/ 2775833.html.
- “Emek Ve Demokrasi Platformu Üyeleri, Diyarbakır’da Yaşanan Patlamayı Kınadı,” Milliyet, retrieved from http://www.milliyet.com.tr/
emek-ve-demokrasi-platformu- uyeleri-mersin-yerelhaber- 826431/.
- See Daniel Pipes, “On Turkey’s Fraudulent Elections Tomorrow,” Middle East Forum, (June 6, 2015), http://www.meforum.org/blog/
- For Oy ve Ötesi’s coverage of the June election, see: http://oyveotesi.org/gecmis-
haziran-2015-genel-secimleri- degerlendirme-raporu/. For November, see: http://oyveotesi.org/1-kasim- 2015-genel-secimleri/1-kasim- 2015-secim-sonuc- degerlendirmeleri/.
- “Tricky Ballot Tickets Main Cause for Half a Million Invalid Votes in Istanbul,” Daily Sabah, (June
08, 2015), http://www.dailysabah.com/
elections/2015/06/08/tricky- ballot-tickets-main-cause-for- half-
- See the presentation of pre-election survey results by Erdem Aytaç and Ali Çarkoğlu of Koç University at https://csr.ku.edu.tr/public-
- For details of the CHP’s envoy to Syria including Eryılmaz and his impact in Samandağ, see:
- See Appendix for definitions of the ENP and electoral volatility.
- Markku Laakso and Rein Taagepera, “Effective Number of Parties: A Measure with Application to West Europe,” Comparative Political Studies, Vol. 12, No. 1 (1979), pp. 3-27.
- Mogens N. Pedersen, excerpted from “The Dynamics of European Party Systems: Changing Patterns of Electoral Volatility,” European Journal of Political Research, Vol. 7, No. 1 (1979), pp. 1-26.
- The SP and BBP were taken together for volatility measurement in November for comparability, because of their electoral alliance in June.
- On the secular vs. conservative polarization see Ali Çarkoğlu and Binnaz Toprak, Değişen Türkiye’de Din, Toplum ve Siyaset (Religion, Society and Politics in Changing Turkey), Türkiye Ekonomik ve Sosyal Etüdler Vakfı (TESEV), (2006), retrieved from http://www.tesev.org.tr/en/
- By ontological concerns, we refer to deeply embedded anxieties that are fundamental to the political and social attitudes of citizens. Based on a critical textual analysis, Grayson defines a similar concept in the following way: “Threats provide citizens with ontological security, that is, security about their own identity, their rightful position in the world and who (or what) poses a danger to them.” See Kyle Grayson, “How to Read Paddington Bear: Liberalism and the Foreign Subject in A Bear Called Paddington,” The British Journal of Politics and International Relations, Vol. 15 (2013), p. 390.
- See Salih Bayram, “Political Parallelism in the Turkish Press, a Historical Interpretation,” Turkish Studies, Vol. 11, No. 4 (2010), pp. 579-611.
- See A. Çarkoğlu, L. Baruh and K. Yıldırım, “Press-Party Parallelism and Polarization of News Media during an Election Campaign: The Case of the 2011 Turkish Elections,” International Journal of Press / Politics, Vol. 19 (2014), pp. 295-317.
- See http://www.mynet.com/haber/
guncel/trtnin-25-gunluk-yayin- tablosu-akp-30-chp-5-mhp-1- saat-hdp-18-dakika-2156446-1 on biased coverage of the media favoring the incumbent AK Party.
- See World Values Survey data for international comparisons: http://www.worldvaluessurvey.