Friday, June 19, 2015

Turkey's Stable Electorate and AKP Dominance

A lot of the analysis of the Turkish parliamentary elections has focused on the votes that have changed hands--mostly from AKP to HDP (see here for a good example).

But, lost in all of that, is an equally important story. That story is just how stubbornly polarized the Turkish electorate has become. Since the electoral turmoil of the 1991-2002 period, Turkish politics has evolved an increasingly stable four-party system. Moreover, each party has an increasingly clear voter profile. With the defection of many conservative Kurds from AKP to MHP, those profiles have become even clearer.

To illustrate the point, consider this chart. (Note: all charts here come from the excellent election survey analysis by polling firm KONDA).

 This chart shows the shifts in (self-reported) vote from the 2011 parliamentary elections to the 2015 elections. Keep in mind that 2011 was the high point of AKP support during a fast-growing economy and with an opposition that was unready to govern.

In spite of everything that happened since 2011--the slowing economy, the Gezi protests, the corruption scandals, Erdogan's presidential ambitions--less than 20% of AKP's 2011 voters abandoned the party.

The numbers were more remarkable for the other parties. CHP and MHP both retained about 90% of their past voters. And, incredibly, almost all who reported voted for an independent BDP candidate in 2011 voted HDP in 2015.

(It should be noted that self-reported past voting does lead to some amount of selective misreporting to achieve internal consistency with current behavior).

Why So Stable?

The simple reason for this stability is that each party has by now developed a clear electoral profile. In part, recent shifts have just been consolidation of those profiles: more Kurdish voters have moved to HDP, more 'liberal' voters to CHP, and so on. There seem to have been two main illustrations of tactical voting: some defections from AKP to MHP (the party most similar to it ideologically), and the small numerically (though much discussed) defection of some younger Turkish leftists to HDP.

The profile of each party is pretty straightforward. The following KONDA charts illustrate some key points.


First, CHP is something like one-half of a 'social democratic' party. Its core electorate is affluent, educated, and urban. While it has weak support among the working poor and the less-educated, it is the strongest party among those in the highest income groups and among the university educated. In effect, the party has consolidated the 'urban intellectual' part of the traditional social democratic party base; it just lacks the mass working-class support. 

That HDP support is highest among those in the lowest income groups speaks to the condition of large number of Kurds in Turkey.

It is also worth noting that MHP is the one party for which there is no clear pattern by class or education; its support is roughly constant across all groups.

For the Opposition to Win

A key reason for the pattern in the above two charts is the close relationship between religiosity and support of AKP or CHP. Typically, CHP support is strongest among the least religious voters, while AKP is dominant among those who self-describe as strongly religious.

The reason for this strong pattern, as I've described before, is that Turkish electoral politics is structure more about cultural values and 'identity' than around interests. AKP is the party of religious conservatives, CHP the party of urban secularists, HDP the party of Kurds, and MHP the party of nationalists. AKP became the dominant party by tying their religious program to a strategy of development that focused on providing infrastructure and services to poor but religious voters.

Meaningful electoral change will not occur until one of the opposition parties successfully carries out a strategy of electoral realignment. HDP made an important step in this regard. By capturing the votes of many conservative Kurds, it broke the effective AKP monopoly on poor, religious voters. (See the chart using data from earlier years here to see what I mean).

[As an aside, it sometimes seems that one of the best ways to signal your standing as a Turkey expert is to refer to CHP as 'feckless' or 'incompetent'. In fact, I sometimes wonder if the party should just change its name to "the feckless and incompetent Republican People's Party (known by its Turkish initials CHP)". Alternatively, one can simply criticize party leader Kemal Kilicdaroglu for being uncharismatic, or for failing to resign after electoral defeat (presumably so that Deniz Baykal could return to the post?). This is not to minimize the problems that the party has, only to point out that such criticisms are typically superficial (what does 'incompetent' even mean in this context?) and that they miss the point.]

If CHP are to become a governing party, they will have to drive a wedge into AKP's core support--an electoral realignment. I've argued before, and will suggest again, that the most plausible way of doing so is pretty clear: to emphasize (and re-emphasize, and really mean it) the party's social democratic commitment. There are a lot of (relatively) poor voters in Turkey, and a sustained program of policies that will improve job security, wages, and upward mobility would appeal to many of them. This strategy effectively shifts the axis of conflict in electoral politics from questions of identity ('white Turks' vs. 'black Turks') to questions of economic interests. It would also makes cooperation with HDP (a party whose voters are also predominantly poorer) much easier.

Interestingly, CHP under Kilicdaroglu have taken steps in that direction recently. Those steps had little effect in this election, but that does not mean that they were unwise. Instead, I think the problem is that an entire generation of Turkish voters has grown up with the experience that CHP does not speak for the interests of working-poor voters. Voting is habitual, and so many older voters in this demographic are unlikely to give CHP a chance. The good news is that Turkey has a young electorate, so the opportunity to persuade voters who loyalties are not as entrenched remains. Obviously, there is no guarantee that CHP can successfully manage this realignment.

In the short term, AKP will remain the dominant party in Turkish politics until the opposition successfully manages such a realignment. If snap elections are held in the fall, there will not be significant change in the election results. AKP may regain some votes that they lost to MHP and (less likely) HDP--quite possibly enough to secure a parliamentary majority. Of course, turnout would be a factor; whether the opposition parties could mobilize effectively a second time would be a big question. Nonetheless, the basic contours of party support wouldn't change dramatically in the short term.

Wednesday, June 10, 2015

False Dawn for Turkey's Left?

The post-election euphoria among Turkish leftists over the election results has given way to more sober analysis. Despite the historic performance by the HDP (breaking 10% for the first time ever) and the cumulative 38% for CHP/HDP (the best performance by the left since the pre-coup 1977 election), there are various causes for concern (not least the specter of a AKP/MHP coalition).

One specific reason is articulated by economist Erik Meyersson in a recent blog post. While much of the talk was of cross-over voters going from CHP to HDP, he rightly points out that the real story behind the HDP surge was Kurdish voters defecting from AKP. Many of these Kurdish voters are quite religious and conservative. I quote at length here:
[I]t’s unlikely that yesterday’s HDP’s electoral success represents some kind of revival of Turkey’s left, or its liberals. Much of the boost to HDP’s political clout instead seem to be coming from former AKP voters, both in the East and inside Turkey’s largest cities. Especially among the former, who were not already voting for the pro-Kurdish parties’ independent candidates, these are likely the ones AKP were able to win over catering to their religiously conservative values....
[D]espite the liberal and leftist appeal of HDP frontpersons, they’re likely up for a significant challenge in steering a party that is secular and progressive at the top whilst pious and socially conservative at the bottom. And so, observers hoping that the Kurds will provide the liberal alternative that the existing Turkish parties have so far failed to provide, are likely to be disappointed.
The Leftist Dilemma?
I think Meyersson's characterization of the composition of the HDP electorate is correct. Many of its voters are probably quite poor, low-educated, and deeply conservative on cultural and religious questions. But this may not be such a big problem.

Why? In short, this is the problem that left-wing/social democratic parties everywhere have faced. Most are a coalition of (1) highly left/liberal urban middle/upper class voters, (2) working-class and--where such a dynamic exists--(3) minority voters. Of these three groups, only the first typically endorse the mixture of leftist economic views and liberal social/cultural views that one associates with the left. The latter two groups tend to support the left for strictly interest-based reasons (either economic redistribution or greater civil rights/autonomy). In fact, the latter two groups commonly endorse rather conservative/authoritarian social values.

As a brief illustration of the point, I had a quick look at data from the 2000 American National Election Study, a survey of voters in the US. It includes a 4-item battery measuring authoritarianism, which is a good predictor of whether an individual is intolerant, xenophobic, and religiously dogmatic. The four-item scale ranges from -1 to 1, with the positive value indicate higher authoritarian values. The overall sample of American voters scored at 0.17, slightly to the authoritarian end of the midpoint. African-American respondents? They scored 0.50, by far the highest of any major demographic group.

Observers of US politics will know that African-Americans vote overwhelmingly for the Democratic Party (the more left/liberal of America's two parties). Typically, African-Americans vote 85% or more for Democratic candidates. It is then perhaps a bit surprising that African-Americans are substantially more authoritarian than people who voted for Republican George W. Bush in that election, who averaged only 0.23. (Whites who voted for Al Gore had an average score of -0.30). So the most authoritarian demographic of voters is also the most loyal to the left/liberal of the two main parties, and it was paired with the least authoritarian voters in supporting that party.

It's unlikely that this pattern exists only in the US. In 1959, political sociologist Seymour Martin Lipset wrote of "working-class authoritarianism," noting that low-education, low-income voters were likely to display those traits of intolerance, dogmatism, and submission to authority. Instead, it is probably a dilemma of left-wing parties that their supporters include the most and least liberal (in socio-cultural terms) of the electorate.

In this respect, what HDP is moving towards is perhaps not that different than other left-wing parties (particularly if the survey evidence that 22% of those aged 18-24 voted HDP is indicative of its future growth). In the US and other democracies, left-liberal leaders adopt a broad language of inclusion and social justice and legislate it when in power. Even though the leaders and a relatively small segment of affluent voters actually internalize those values, it still works because it is in the interests of the party's supporters. In the case of HDP, it would be pretty straightforward: Kurds are (on the whole) relatively poorer and politically marginalized in Turkish society. Whether or not many truly believe in the high-minded liberal/leftist values their party endorses, it makes sense for them to continue to vote HDP as long as it presents itself as the best vehicle for improving their political and economic condition. And that message in time may attract more left/liberal and low-income Turkish voters into a workable electoral coalition.

Of course, this is not to ignore the challenges that Meyersson raises. To refer back to the American example, white Southerners (who at the time were poorer than average) were a core voting bloc of the Democratic Party from the 1930s-50s. During that time, they effectively blocked progress on civil rights legislation. Such scenarios could happen within HDP too. Nonetheless, that era of Democratic Party dominance from 1932-68 was one of major economic and political advancement for poor whites and (in the 1960s) African-Americans, despite the strong cultural conservatism of many parts of the party's electoral base.

Monday, June 8, 2015

What Have We Learned from Turkey's Elections?

Some thoughts on what the results of Turkey's elections tell us about Turkish voters...and elections and voting in general:

(Note, this is to say nothing of what will happen after these elections. Good analyses on those questions would include this and this by Aaron Stein and Michael Koplow, respectively, among many others. I offered a brief take here for Vocal International). 

1. Turnout goes up when the stakes are high, the choices clear, and the outcome uncertain!

Yesterday's election checked all three of those boxes, and Turkey's voters responded by turning out at a rate of 86.5%. This contrasts with declining turnout in many Western democracies, where there is a growing sense that elections no longer offer meaningful choices. (In countries like the US and Switzerland, elections offer clear party choices, but they don't always matter very much because of how governing actually works; hence turnout is very low there).

To get a sense of these different dimension, consider that the two lowest-turnout elections in recent Turkish history were the 2002 parliamentary election and the 2014 presidential election. The former was conducted in a very uncertain political environment; many of the old political parties were in shambles (ANAP, DYP, DSP) while several new parties had entered the contest (AKP and the 'Genc Parti'). In that election, the choices and potential outcomes were deeply confusing. In the latter, the outcome (Erdogan's victory) was a virtual certainty, so the motivation to vote was lower.

This brings to mind an old debate about whether high turnout is a healthy sign or not (this is the type of question we ask students in an introductory class on political science). Many observers of Turkey will say so; this result shows how much Turks care about preserving democracy. Fair enough. But the high turnout is also the result of the deeper problems in Turkey: the "all or nothing" 10% election threshold, President Erdogan's grand designs of a 'presidential' system, and so on. Turkey would benefit from having slightly lower-stakes elections in the future!

2. What moved voters?

A big pitfall after any election is the temptation to over-interpret the findings and speak of the "will of the voters" or a "mandate" or whatever.

In this case, the temptation is to argue that Turkish voters rejected Erdogan's presidential dream. And there is probably a lot of truth to that. But one may also be reading too much into that. Polls showed that Turks were against Erdogan's presidential system by about a 2-to-1 margin. Without thinking too hard, we can guess that most of those in favor were probably AKP voters, and most of those against were opposition party supporters. So, in truth, the real 'action' was in that 10-15% of the electorate who had voted AKP in prior elections but were against Erdogan's presidential ambitions.

My attempt at a predictive model based on the national economy suggested that AKP would get about 42.5% to 43.2% of the vote. The virtue of such a model is that it isolates one factor's effect on the vote--the economy--while treating all other factors (mobilization efforts, other political issues, leadership population, etc) as constants. In fact, AKP got about 40.9%. While there are plenty of reasons to treat my predictive model with caution, one way of looking at the results is that the non-economic factors cost AKP an additional 1.6 to 2.3 percentage points of the vote. Those are more likely where the concerns over Erdogan's authoritarianism and Kurdish anger over Kobane and the peace process actually mattered.

Beyond that, voters also switched between opposition parties. HDP gained a lot of votes in major cities like Istanbul and Izmir; many of these were probably former CHP voters. CHP probably also lost a number of nationalist voters to MHP in some of the interior districts.

3. Go Left!

Last spring and summer, I argued in several blogs (here, here, and here) that the opposition parties needed to adopt a 'social democratic' strategy to break AKP's electoral dominance. Why? The reason could be summarized in the following chart, which shows that AKP maintained (as of 2010) a virtual monopoly on the votes of those in lower-income, less-educated demographics. This includes many conservative rural voters, of course, but it also includes many urban 'working-class' voters.


Turkey's electorate is more like that of a Western country in the 1960s than today. It is a much more working-class, lower-income, lower-education electorate. It is impossible to win an election in Turkey--and will be for several decades--without competing effectively in this demographic.

Many of these voters resemble Seymour Martin Lipset's old "working-class authoritarian" description: left-wing on economic preferences, deeply conservative on social/cultural matters. CHP and HDP cannot hope to capture these voters by competing heavily on socio-cultural issues. Instead, they must adopt social democratic messages on economic questions. This is one reason why I was encouraged by CHP's moves to the left during this campaign. While some of the promises about minimum wages and other economic promises were clumsy and unconvincing, it was (in general) the right idea.

In other words, Turkey has a particular electoral alignment, in which AKP dominates among rural and conservative voters and among these "working-class authoritarians." As all the research on "issue evolution" and "heresthetics" would tell us, the only way for Turkey's left-wing opposition parties to start winning is to fracture AKP's winning electoral coalition in order to build one of their own.

4. Warning: Voting may be habit-forming!

Turkey has a very young population, and a lot of new voters have entered the electorate since 2011. It will be very interesting to see how these new voters voted.

This is an important question, because research by political scientist Mark Franklin (among others) show that there is a strong early-life socialization/habit-forming aspect to voting. This is true for turnout as well as for party support. Those young adults who voted yesterday are more likely to continue voting in future elections throughout their lives. And they are more likely to continue voting for the party (or its ideological successor) that they voted for yesterday.

If we assume that first-time non-Kurdish voters supported HDP in larger numbers than older cohorts, then this matters for future elections as they will likely continue to do so in future elections, and their fingerprint will grow larger as they replace older generations in the electorate.

Thursday, June 4, 2015

The Two Opposing Trends in Turkey, and What It Means for 8 June (and Beyond)

Building on two interesting and good articles about what will happen after Turkey's 7 June parliamentary elections (here and here), I think it's worthwhile to emphasize the two contradictory trends occurring in contemporary Turkey (points I've made before here).

The Authoritarian Trend
Turkey is governed by a president who sees no problem with breaking the law, suppressing criticism, and using the organs of the state to increase his own political power.

While Turkey's democratic credentials have always been questionable (at best), the trend since 2007 has clearly been towards a centralization of power in President Recep Tayyip Erdogan's hands. This trend is well-documented. I've written about it here. Economist Erik Meyersson has documented it nicely in comparative institutional terms here. The problem is three-fold: first, Erdogan and the ruling AKP have weakened formal institutional checks on their governing power (e.g., the judiciary, civil service, media), so that there are fewer actors who can legitimately oppose them. Second, every branch of the government is more or less "stuffed" with AKP loyalists after nearly 13 years of AKP single-party rule, so their ideal points may often be identical to those of Erdogan anyway. And, third, Erdogan, in true competitive authoritarian style, routinely abuses his formal institutional powers anyway. No better evidence is needed of this than his open campaigning in favor of AKP (and against the opposition parties), despite the constitutional requirement that the president be neutral.

Whatever happens on 7 June, these facts will not change. Erdogan will still be president until 2019. If he doesn't have the 330 votes in parliament to push through a constitutional referendum to create a presidential system, he will still continue to rule by fiat when he can. And his/AKP's control of many major institutions will often enable him to do this (witness the unwillingness of Turkey's Supreme Electoral Council to uphold opposition complaints about Erdogan's partisan campaigning).

The Democratizing Trend
At the same time, Turkey is a rapidly modernizing country, and this is reflected in its society. Turkish society has become more pluralistic, more liberal, and there is now a much more robust set of civil society organizations dedicated to protecting individual rights (see, for example, Oy ve Otesi, an organization dedicated to monitoring the upcoming elections for fraud).

AKP has both contributed to and benefited from this liberalization of Turkish society. In its early years, it gained both domestic support and international approval by using the language of rights expansion to justify its policies. And there were many things about pre-2002 Turkey--the denial of rights to Kurds, vague 'anti-terrorism' laws, the mistreatment of women wearing headcovers, etc--that were deeply undemocratic. The partial expansion of rights in the early years was important. Turkey is a significantly more liberal society than it was in 2002.

The 2013 Gezi Park protests similarly both illustrated how much more liberal important segments of Turkish society had become and helped to spur on that process of liberalization. One common refrain from Turks has been that Gezi helped them to realize just how many like-minded people were out there. After the overwhelming AKP victory in the 2011 elections and in the environment of increased media self-censorship, being a liberal opponent of the Erdogan government was a lonely calling. Gezi destroyed that image.

A youthful population that has increasingly grown up in a more affluent and urban Turkey is producing a more liberal population. In many ways, this is a similar development to that of 1968 in the United States and many other Western countries. This is all very consistent with the predictions of 'modernization theory' in political science.

Importantly, the opposition political parties have begun to respond to these developments. This is most obvious with the Kurdish-associated HDP (People's Democracy Party), the big story of these upcoming elections (in more ways than one). But the main opposition CHP (Republican People's Party) has also taken a distinctly more progressive turn in these elections. During the early AKP years, CHP were on the wrong side of many important debates over political rights in Turkey--allowing women with headcovers into universities, increased cultural and political rights for Kurds, etc. That is far less the case today.

Whatever happens on 7 June, these developments will also not change. The demands from Turkish society for liberalization will continue.

The Collision Course?
The trouble for Turkey is that these two trends push towards a showdown: the rulers attempting to centralize power, and growing segments of society demanding that it be liberalized. 

During the Gezi Park protests, political scientist Burak Kadercan wrote an insightful blog warning of an impending 'Turkish winter' to come. This may well be the scenario, as Michael Koplow now also warns, after the elections.

The 'best-case' scenario is probably that HDP are sufficiently above the 10% threshold that targeted electoral fraud cannot keep them out of parliament, and AKP/Erdogan are delivered an electoral setback. Whether AKP still control a narrow parliamentary majority or not is probably less important. In this scenario (and assuming that HDP remain true to pre-election promises), Erdogan's dream of a constitutional referendum to create his 'presidential' system is effectively dead. But this is a very limited best-case scenario: still in power until 2019, Erdogan will still seek every opportunity to exercise his own power (legally or not). And the temptation to foment crisis for his political benefit will be significant. Nonetheless, Turkey's opposition would--for the first time since 2002--actually have a sense of positive momentum and feel further emboldened to oppose Erdogan's agenda. Internal dissent within AKP would probably also increase, and the economic storm clouds on the horizon could mean further trouble for AKP's popularity.

If HDP do not enter parliament, and especially if Erdogan gets his desired 330-seat supermajority to push forward with the constitutional referendum, the scenario will indeed be dire. Accusations of AKP electoral fraud will be widespread. A cycle of mass protests and police violence would seem quite likely, and it would all likely get far worse than in 2013. The Gezi Park protests were tempered in part by the expectation of upcoming elections in 2014/15. Now, with no potential elections until 2019 (by which point Erdogan's centralization of power would be too far advanced), the stakes are much higher for opposition groups. And there is always the PKK factor.

In the long run...
I believe that there are good reasons to be optimistic about Turkey's long-run democratic prospects. Political scientists Steven Levitsky & Lucan Way, who developed the concept of "competitive authoritarianism" (which I believe describes contemporary Turkey), note that a crucial factor in whether such states ultimately democratize or descend into full authoritarianism is the closeness of their ties to the West. Institutionally and socially, Turkey's ties to the West are comparatively strong. And Turkey does not have the mineral resources to allow its regime to 'buy off' domestic opposition without continued economic development. These facts suggest that Turkey is more likely to continue heading down a path (however bumpy) towards increased social liberalization and development--even as political centralization may also continue.

However, this long-run optimism says nothing about the short-term prospects, which remain far less promising. Even if one could imagine a post-Erdogan Turkey in the near future, it would be one in need of deep institutional and constitutional reform to move towards a genuine liberal democratic future.