Tuesday, May 27, 2014

Three Points (and a bunch of links) about the European Parliament Elections

1. The big winner was abstention.

The good news for European leaders was that the non-stop turnout decline since the first EP elections in 1979 was halted.

The bad news? At 43%.

There are several reasons why nearly 3 in 5 Europeans didn't vote. But such low turnout surely reflects broad indifference to the outcome. Many Europeans don't turn out to vote simply because they don't see how it matters.

Consider the following example. Ask 1000 British voters before their next general election what difference it will make it Cameron or Miliband (or Farage!) becomes the next Prime Minister, and you'll get some pretty good answers: taxes, social spending, transport policy, etc. Or, phrase it differently, and ask them if it matters whether Labour or the Tories form the next government. Of course, there will be a substantial group who think it doesn't matter (because all politicians are the same), but most will have an opinion. But ask 1000 Europeans if it matters whether Jean-Claude Juncker or Martin Schulz becomes the next Commission President, and you're likely to get blank stares. And not without reason.

In truth, the decision of who becomes the next Commission President will be made in deals among the leading European party groups (and possibly national leaders), far removed from any direct link to voter accountability. And the actual business of governance at the European level will work in a pragmatic and consensual manner, blurring the differences between any of the major party groups and reducing accountability. The end result, as Sara Hobolt and James Tilley describe, is that there is 'responsibility without accountability' at the European level.


In short, the deeper problem facing the EU is not an 'earthquake' of Euroskepticism, but a 'meh' of widespread indifference. Take a look at this chart, which shows the breakdown of the party vote with abstention included. You can see what little support any party grouping actually received, which is what you get when only 43% of eligible voters bother to turn out.



2. Euroskeptical parties had their best results ever, but they're a pretty diverse crowd.

No, the radical right isn't taking over, and, no, this isn't like the 1930s all over again. 

I'll outsource the work of explaining this to several excellent commentaries:

Here's Cas Mudde providing a nice overview of the results for Euroskeptical parties.

Duncan McDonnell similarly points out that any talk of an 'earthquake' election relies on an exaggerated (and geographically-limited) reading of the results. And, he points out that apathy was the big 'winner.'

Matthew Yglesias throws water on the 'earthquake' fire in an explainer about the election results.

Need help making sense of the different Euroskeptical, nationalist, populist, radical parties? Here is a nice guide rating each party on several dimensions.

3. Rise of the 'left-authoritarians'?

Matthew Goodwin & Rob Ford have been arguing for a while that most British commentators misunderstand the rise of the UK Independence Party. UKIP voters are not Euroskeptical Tories. Instead, they are older, blue-collar voters who feel abandoned by the mainstream parties--all of which offer essentially 'middle-class' programs that are friendly towards bankers, the EU, free trade, and immigration. These voters used to support Labour, until it moved decisively towards the center in the 1990s. As a result, UKIP had its best results in areas that traditionally supported Labour.

Here is a better summary of their findings.

A similarly worrying result from France (if you are a Socialist). Those who identify as workers voted 43% for the Front National, 8% for the Socialist Party, and 8% for the far left. (Link here in French).

In an academic study (ungated here), Markus Wagner, Zoe Lefkofridi, and Johanna Willmann argue that 'left-authoritarians'--those who hold left-of-center economic views but conservative/nationalist socio-cultural views--are broadly unrepresented in most European party systems. Thus, left-authoritarians face a choice of voting for a party that either represents their economic (but not cultural) views, or vice versa. Their analysis of 2009 data show that most choose to follow their economic views, but that may have well changed in the past 5 years.

And I've argued (here behind a paywall and here in blog form) that authoritarian dispositions predict anti-EU beliefs.

Conversely, Theresa Kuhn and Florian Stockel argue that the growth of European economic governance since the crisis has resulted in only those who truly identify with Europe supporting further integration (something like the converse of my argument).

In short, the rise of support for parties like FN and UKIP seems to be driven by predominantly working-class and socially conservative voters who feel abandoned by the mainstream parties economically and threatened by the social changes they see around them. The challenge for European leaders ought to be in considering how to connect with these voters in order to understand how national and European policies could address their concerns. It seems unlikely, however, that they will. Instead, the more commonplace view seems to be offering lip-service to addressing their concerns while dismissing them as being uneducated.

Thursday, May 15, 2014

Here is Your Chance, CHP

Turkey is still reeling from the Soma mine disaster, which has resulted in the deaths of 282 (and probably more) miners. There has been an outpouring of rage towards the government of Recep Tayyip Erdogan for failing to take necessary steps to ensure proper regulation of the mines. Reports are disseminating about the lack of regulatory oversight and the lack of proper safety measures in Soma and other Turkish mines.

This anger towards the government has been exacerbated by the Erdogan government's response. Bizarrely, Erdogan gave a speech at Soma yesterday in which he compared the death total of this disaster to those in Europe and North America from decades, or even over a century, ago. The anger in Soma towards Erdogan was documented by journalists, and the actions of an Erdogan adviser, who was seen kicking a protestor (who was already been subdued by several policemen) can't help.

It can feel inappropriate to "play politics" in the aftermath of a human tragedy, but it is important that the proper questions be asked. Namely, is the government taking adequate steps to ensure miner safety? Was the privatization of the Soma mine in 2005 handled in a proper fashion? Why did AKP MPs reject an opposition motion to start a parliamentary inquiry into mine safety conditions?

Given the anger in Soma, the renewal of anti-government protests after the disaster, and the general strike held by several unions today, it is worth asking whether this disaster will have any lasting political consequences. The writer(s) at Ataturk's Republic argue, and rightly so, that this disaster will not by itself bring down the Erdogan government. However, it may create a genuine political opportunity for the opposition Republican People's Party (CHP), if it can manage the opportunity.

Observers have spent the past year expecting each new challenge--the Gezi Park protests, the corruption investigations, the leaked conversations, etc--to bring down Erdogan, and they haven't. So why would this time be any different? The reason is in who is affected and how.

The Gezi Park protests consisted mostly of opposition supporters and others with deep dissatisfaction with the government. As with similar protests in Europe and North American in the late 1960s, it was relatively easy for Erdogan to label the protestors as troublemakers while claiming to have the backing of a large and silent national majority.

The corruption scandals, while being great fodder for opposition politicians and journalists, also probably didn't hit average Turks in a way that would resonate. After all, politicians taking bribes from businessmen: this is remote, and it's hard to see how it affects ordinary people.

To put it all a bit differently, these protests and scandals largely reinforced existing party divides in Turkey. They didn't necessarily give many AKP supporters a strong reason to reconsider their party loyalty.

The Soma disaster presents a different political opportunity. I can't claim any specific evidence about how miners vote in Turkey, but existing patterns of party support suggest that they--and many voters like them--lean heavily AKP. This is probably in large part due to deeper worldviews that emphasize authority, tradition, and social cohesion (as I have argued before), but it is also the result of a virtually reversed class divide in Turkish electoral politics.

The CHP calls itself a social democratic party, and the AKP calls itself a conservative party. Yet who actually supports each party? Using data from the Sixth Wave of the World Values Survey (conducted in 2011 in Turkey--near the peak of AKP support), we can examine this question. The survey doesn't contain too many detailed questions about class or occupation, but two are illustrative. The first is education. The chart shows support for AKP (blue column) and CHP (red column) by level of education (the parentheses show the number of survey respondents in each category).



The results are pretty stark. Those with less than a secondary education (the plurality category) overwhelmingly supported the AKP, while a plurality of university-educated voters lean CHP. This is not news to anybody who follows Turkish politics closely, but it illustrates just how severe the disconnect between CHP's social democratic label and its actual social support is.

We can get a second look by examining the type of work performed by the respondent. Respondents were asked to describe their work on a scale ranging from manual to intellectual tasks.



The same pattern appears. AKP support is much higher among those who describe their work as involving manual, rather than intellectual, tasks. In other words, miners and many others Turkish workers like them have supported the AKP heavily.

The opportunity for CHP, BDP/HDP, or a new opposition party is clear. Changing a predominant pattern of electoral politics requires some sort of realignment of party support. Some significant bloc of present AKP supporters need to be persuaded to defect to another party. Therein lies the crucial opportunity in the aftermath of this disaster and the government's response. While the Erdogan government relies heavily on the support of voters like the miners, this disaster has given the opportunity for the opposition to demonstrate that it can better represent the interests of working-class Turkish voters.

To do so successfully, the CHP needs to connect its (claimed) social democratic identity to actual policies. It needs to explain, in terms that voters can understand, how a CHP-led government would improve the standard of living for working-class Turks. This requires more than just criticism of the Erdogan government. It requires a concrete, positive vision. And, given the nature of the Turkish media, it requires taking that message into the doorways and living rooms of voters, so that it will actually be heard.

It's worth adding, of course, that this need not be a complete transformation. But simply peeling away 10-15 percentage points of the AKP's massive lead among working-class voters would transform electoral politics into a genuinely competitive era.

I don't want to make light of a horrible human tragedy, but a combination of tragic circumstances and government (in)action has handed the CHP the best opportunity that it has had in nearly 12 years of Erdogan rule. It had best take it.

Thursday, May 1, 2014

Evaluating the Proposed Electoral Reform in Turkey


In the topsy-turvy world of Turkish politics, one item that has been on the agenda occasionally has been electoral reform. Many observers of Turkish politics agree that the current electoral system is problematic, and any scholar of electoral politics would surely agree! As I argued in a previous post, the current Turkish electoral system manages to capture the “worst of both worlds”: a highly disproportional arrangement that hurts small parties but does not encourage the consolidation of a stable national party system.

So, electoral reform would be a good thing. But it is, of course, possible to take something that is bad and make it even worse. So when Turkish Prime Minister Tayyip Erdogan proposed a couple of possible electoral reforms last autumn and again recently, this was the first question that came to my mind. In addition, one has to imagine that any reform that Erdogan proposes will be to his party’s benefit.

The Current System

As noted above, the current system is a rather poor design. There are three problematic features (in order of severity):

1.     The presence of the 10% election threshold, which allows only those parties that obtain over 10% of the vote nationally to gain any seats in parliament. (Note that BDP bypasses this by running all of its candidates as independents, which makes a mockery of threshold—never a good thing in institutional design).
2.     The division of Turkey into electoral districts based on provinces, resulting in a large number of low-population provinces that return fewer than 5 MPs. When the number of MPs is so low, the potential for a proportional representation of the population is reduced. (Meanwhile, Turkey’s three largest cities are subdivided into smaller constituencies so that no district returns more than 30 MPs).
3.     The use of the d’Hondt electoral formula, which slightly favors larger parties over smaller ones. This is by far the least consequential of the three, and is not necessarily a problem unless you support a small party.

Evaluating The Proposed System

Erdogan’s proposal appears to be this: divide Turkey into 110 electoral districts that return 5 members of parliament each. Concurrently, lower the electoral threshold, which currently is 10%, to a lower number of 7% or perhaps 5%. Other features of the system would presumably remain the same.

What effect would this reform have? Answering this question definitively is complex. For one thing, strategic parties change their behavior in response to new institutions. So parties might begin to campaign or invest their resources differently under new electoral rules. Second, the creation of 110 new 5-seat districts opens up many possibilities for the drawing of those boundaries. Not knowing what those would look like, I make some simplifying assumptions. Below, I use the d’Hondt formula to estimate the distribution of seats from the 2011 parliamentary elections under the current rules for several different electoral districts (which are selected to represent different areas and patterns of voting) and what those would look like under the new rules. The key simplifying assumption that I make is that the distribution of votes in an electoral district (as constituted now) is uniform throughout. That is, of course, not true. But making this simplifying assumption allows us to isolate the effect of the electoral system.

I choose several current electoral districts to examine the effects of this proposed reform. While certainly not exhaustive or representative of every part of Turkey, these allow for an examination of what this electoral reform might do and why.

A quick primer on the d’Hondt electoral method. All PR methods use a sequential method of allocating seats using a formula to determine which party gains each seat in turn. The d’Hondt formula is V/(S+1) where V=total # of votes cast for that party and S=the number of seats that the party has already been awarded in that district. In other words, the formula starts with the number of votes cast for that party (because S+1=1 when S=0) and then increases the divisor by 1 every time that party wins a seat. Seats are awarded, applying that formula at each stage to all parties, until all of the available seats have been assigned. As an example, consider Balikesir, which awards 8 seats.

Party
Votes
(S=0)
S=1
S=2
S=3
S=4
AKP
356649
178325
118883
89162
71330
CHP
259104
129552
86368
64776
51821
MHP
106663
53332
35554
26666
21332

The seats won by each party are in bold. The first seat is awarded to AKP, after which it is pushed right on the chart to S=1. Then, the largest total belongs to CHP (which is still at S=0), so it gains the 2nd seat. The third seat then goes to AKP, pushing it the S=2 column, and so on. This produces a reasonably proportional distribution of the seats.

What would happen if Balikesir were divided into one 5-seat district and its three remaining seats were combined with another distict with (hypothetically) equal voting patterns? To see the result, just consider the awarding of the first five seats, which are both bolded and italicized. Under those rules, AKP would win 3 seats, CHP 2 seats, and MHP 0. Two of the remaining three seats of this new hypothetical district would go to AKP and 1 to CHP. The loser of this new arrangement would be the third-largest party, MHP, which would no longer gain any seats despite winning close to 15% of the vote.

Again, it is important to note that this result is not what would happen in reality, because patterns of voting are not uniform across neighborhoods and districts, but it gives us a sense of what effect the electoral reform would have.

Let’s consider some other examples. Take the Istanbul 1 district, which covers the Anatolian side of the city. At present, it sends 30 MPs to parliament, from which AKP had 16, CHP 11, MHP 2, and 1 was an independent. I won’t walk through the math of awarding 30 seats via d’Hondt, but the results under the current electoral system and what they would look like in this counterfactual using the proposed system with 5 MPs per district are shown. (Note that I make one more assumption due to a lack of better information: I treat all votes for independent candidates as though they were for a single party. It makes no difference in this case, though it is not exactly correct).

The results are pretty clear. The current system does a reasonable job of producing a proportional representation of the vote. The main reason why it is so proportional is the high number of seats being distributed! By contrast, the proposed system with only 5 seats per district would produce a much less equitable outcome. In this hypothetical example, all seats would go to the two largest parties, while voters of the MHP and Independent candidates (totally nearly 15% of the vote) would get no representation. (Again, this is not exactly what would happen because voters are concentrated unevenly in certain neighborhoods).

Istanbul 1st District
Party
Vote
Current Seats
Seats under new system
Difference
AKP
1,391,558
16
18
2
CHP
969,038
11
12
1
MHP
260,825
2
0
-2
Independent
133,617
1
0
-1

It is a similar story if one looks at Ankara’s (15 seat) 1st electoral district instead. The winner of this reform would be the largest party—the AKP—while the loser would be the third party (MHP).

Ankara 1st District
Party
Vote
Current Seats
Seats under new system
Difference
AKP
693,413
7
9
2
CHP
550,710
6
6
0
MHP
228,825
2
0
-2

The same pattern—gains for the largest party, losses for the third and any smaller party—generally applies to other parts of the country as well. Consider Izmir, a CHP stronghold. The reform here might well bring gains to the CHP at the other two parties. Of course, it’s important to point out that there are relatively few CHP strongholds—mostly in Thrace and the west coast—so there are fewer places where the CHP could gain seats.

Izmir 1st District
Party
Vote
Current Seats
Seats under new system
Difference
AKP
454,390
6
5
-1
CHP
528,001
6
8
2
MHP
134,473
1
0
-1

However, consider what happens when one looks at Diyarbakir—a stronghold of the pro-Kurdish BDP. Under the current system, BDP has to run its candidates as independents in order to circumvent the 10% electoral threshold. This is a difficult process that requires an impressive amount of coordination among candidates and voters. The result is a high number of wasted votes, which does a disservice to BDP supporters. In the last election, AKP won 6 of the 11 seats in Diyarbakir despite gaining only about one-third of the vote. Under this proposed reform, BDP would benefit in two ways. First, as the largest party, it would benefit from the disproportional nature of this electoral system. Second, it would be able to compete as a “normal” party rather than running its candidates as independents if the threshold were reduced to 5%, meaning its votes would be allocated more efficiently to representatives.

Diyarbakir
Party
Votes
Current Seats
Seats under new system
Difference
AKP
218,552
6
4
-2
CHP
15,882
0
0
0
MHP
5,733
0
0
0
Independent (BDP)
419,095
5
7
2

Finally, there is the issue of the many provinces in Turkey that currently elect fewer than 5 MPs. Under the proposed reform, these populations will be combined to form larger districts, which may actually increase the proportionality of some. Consider the neighboring provinces of Aksaray and Nevsehir, each of which currently have 3 MPs. If we imagine that some proportion of their respective populations were combined to create a district with 5 MPs (again, assuming a uniform population distribution), what effect would that have? Presently, both districts awarded all three of their respective seats to the AKP and none to any of the other parties. However, AKP received about 69% in Aksaray and 63.5% of the vote in Nevsehir. Thus, the 30-35% of the electorates in each province who voted for an opposition party were denied representation.

What would a “combined” Aksaray/Nevsehir district look like? The answer is that it would be marginally more proportional. The MHP, second-largest party in these districts, would gain a seat under this new arrangement. Nonetheless, it must still be noted that this system would translate AKP’s vote share of about 65% into 83% of the seats, leaving the other 35% with just 17% of the seats. Not very proportional.

Aksaray/Nevsehir
Party
Votes
Current Seats
Seats under new system
Difference
AKP
234,839
6
5
-1
CHP
51,277
0
0
0
MHP
67,061
0
1
1

What can we conclude about this proposed electoral reform? I think three conclusions are in order:
1.     This system would benefit the largest party in each district. For the majority of the country, that is the AKP. In a smaller number of western districts, the CHP would benefit, and the BDP would gain in the heavily Kurdish southeast. The loser under this new arrangement almost everywhere would be the MHP, which does not really have a true geographic stronghold anywhere in Turkey.
2.     Reducing the 10% threshold to 7% or (better) 5% would be an improvement by any measure. Allowing BDP (or any other such small party) to compete on equal terms would be an improvement for Turkish electoral democracy.
3.     The proportionality of the system would be marginally improved in those districts that currently have fewer than 5 MPs.

The problem is that the downside of #1 certainly outweighs the benefits of #3, while #2 can be accomplished easily and directly without any other reforms being necessary. On the whole, this proposed change would replace one bad system with another. Moreover, it would create a new system that further strengthens the electoral power of the governing AKP, which is surely one of the motivations for this proposal. I’ve seen estimates claiming the AKP would gain anywhere from 10-25 seats in the 550-seat parliament. Obviously, those claims depend on assumptions about just how the boundaries of these new districts would be drawn, as well as about the share of the vote accruing to each party in future elections.

A Better Electoral System for Turkey?

So what would a better alternative look like for Turkey? The two main problems that can be addressed via electoral reform are:

1.     The high election threshold, which creates a series of perverse incentives for parties and inhibits effective representation
2.     The presence of large numbers of districts with small populations.

Turkey would therefore be much better off with a simpler reform bill that addressed those two concerns. And this could be done in a straightforward legislative fashion. First, lower the threshold to 5%. Even better, created a two-tiered threshold of the sort that Sweden has. There, a party can win seats if either it earns 4% of the vote nationally or 12% in one district. This ensures that parties with a strong geographic base of support (such as the BDP in Turkey’s case) can still gain representation. Second, reduce the total number of electoral districts by about half. Consolidate the small districts into larger ones so that each district has a minimum of 8 seats or so. In combination, these rules would ensure effective proportional representation while still alleviating concerns that too many small and/or extremist parties could enter the parliament.

It is unlikely that the governing AKP would willingly endorse such a reform at the present time. However, the opposition parties would be wise to embrace and publicize this as part of a moderate and democratizing series of proposals that would improve representation for all Turks.