A few comments on democracy indices and Turkey

I am grateful to the feedback on my previous post, whether it has been in public or in private.

I think of the V-Dem data as an important contribution to how social scientists can measure institutions, and democracy in particular. I do, however, have some thoughts on the V-Dem data and the main other datasets on the same topic as well.

As for the specific time series for liberal democracy in Turkey, one thing that struck me as particularly interesting (and something many others have commented on as well) is the relatively positive trend during most of the 1990s, especially the later part. Given my priors (and many others’) this at first seemed rather counterintuitive. The 1990s were, after all, a period of significant political instability, plagued by recurring economic crises and violent conflicts. How is it possible that a measure of liberal democracy can be increasing during such a period?

Well, what we know is that in 1989, the EU Commission’s opinion on Turkey stated that:

“although there have been developments in recent years in the human rights situation and in respect for the identity of minorities, these have not yet reached the level required in a democracy.”

In 1995, however, Turkey is admitted to the EU Customs Union, and  in 1999, the EU grants Turkey candidacy status. Thus, somewhere in between, there must have been some sort of improvement. In 1995, The New York Times had the following to say of the customs union agreement:

“For Turkey, the customs union was a test of its acceptance by Europe as a political as well as economic partner, and as a possible step toward full membership in the European Union. For critics of Turkey, it was seen as an endorsement of a human rights record that, while improved, still falls short of European standards”… “in the final debate, a majority of European parliamentarians were swayed by Ankara’s argument that rejection would jeopardize Turkey’s recent democratic reforms.

Last summer, the Turkish Parliament adopted a series of constitutional amendments that broadened political participation, and in the fall, it amended an anti-terrorist law that had been widely used to silence critics of the government’s 11-year-old battle with Kurdish separatists in southeastern Turkey.”

The 1995 amendment, by the way, eliminated several restrictions on political rights and liberties. Amongst others, it included lowering the voting age from 20 to 18 and university staff and students were allowed to engage in political activities etc. In 1993, the state monopoly on radio and television broadcasting was also eliminated (for more, see here).

A possible reason why the index isn’t directly weighed down by the conflict and human rights abuse that went on during the period could be that these reflect the low existing levels of institutions permitting such abuse rather than significant adverse changes in institutions associated with it. For example, in the 1990s, was the institutional facilitation of torture a new phenomenon in Turkey? (The reader should feel free to consult and interpret the data on Turkey’s human rights record from the Political Terror database available here. I should also note in passing that “Midnight Express”, a movie that significantly contributed to Turkey becoming synonymous with torture for many, is from 1978).

Moreover, whereas the 90s is famous for political instability, it’s not clear that these necessarily reflect different institutions than previous years per se. (The fact that Belgium went 589 days without a government didn’t necessarily make it less of a liberal democracy). The broad electoral rules (at least as far as I am aware) have remained pretty much the same ever since the coup. In the 1990s, Turkish citizens voted in a way that resulted in a parliamentary representation fragmented across many different parties. After the 2002 election, this changed. Is this because of institutional rules regulating liberal democracy or because of different voter preferences?

Nonetheless, it’s unfortunate that there are no accompanying explanations or notes for the coding in V-Dem’s data series, which would have been a useful aid in this situation.

A related matter is how the V-Dem Liberal democracy index stacks up against the other main sources of democracy data, specifically the Polity IV database produced by Systemic Peace Institute and the Freedom of the World publication from Freedom House. Fortunately, the V-Dem downloadable content includes these too so it’s easy to compare:


The first graph is the V-Dem data, where I’m now showing both the main Liberal Democracy (LD) Index, as well as its two components, Electoral Democracy (ED) Index (aka polyarchy) and “Liberal component”. The former is the polyarchy series described in the previous post, and the latter is the part that focuses on to what extent is the liberal principle of democracy achieved (see the V-Dem codebook for details). The reason why the LD index is rising during the 1990s is mostly because the ED index is rising (consistent with the reforms mentioned above). The liberal component is however rahter flat during most of the 90s. After the coming to power of the AKP they have roughly the same trend, although the liberal component lags one or two years.

The middle graph shows Polity data, the combined Polity score which runs from -10 (most autocratic) to 10 (most democratic) and is made up of two subscores, one measuring democracy and the other measuring autocracy (see here for details). First of all, Polity considers Turkey a near-perfect democracy with a score of 9 in 2012. Second, it cannot explain how the EU thinks Turkey is democratic enough to become an EU candidate in 1989 but is democratic enough for it in 1999 (the levels of democracy are identical in the two years). Third, the coming to power of AKP had no effect on Turkey’s democratic institutions until a small positive jump occurs in 2011, incidentally the same year OSCE found that Turkey had 57 journalists in jail, “apparently more than any other country.” Frankly, this result from Polity looks rather awkward when one compares with alternative sources.

The lowermost graph shows the two main indicators from Freedom House, the Civil Liberties and Political Rights ratings. These ratings originate from two separate scores ranging from 0-40, and the  ratings are constructed by assigning numbers to intervals (scores 0-5 get a 7, 6-11 get 6 etc). Again, these ratings don’t explain how Turkey improved its democracy enough to become an EU candidate country between 1989 and 1999. Even more concerning, according to Freedom House, the best AKP could do in terms of civil liberties was to return in 2005 to the level Turkey enjoyed in 1989, only to fall below this level again in 2011. The Political Rights has remained the same throughout AKP’s tenure. Similar to the V-Dem, however, the 1990s shows a marked improvement along both measures, especially Political Rights (which is probably more similar to the polyarchy series from V-Dem than any of its other ones). Interestingly, Freedom House considers 1990 and 1991 the heyday of Turkish democracy in terms of Political Rights. That’s also a tad awkward.

An unfortunate methodological aspect of the Freedom House data is that it doesn’t publish the aggregate scores (from which the ratings are constructed) for as many years as the ratings. I have only been able to find this kind of data from 2005 onwards and I’ve written about this here and here. Once you start to look at the more disaggregated Freedom House data, you also start to see how some components of Turkish democracy started to recede, sometimes as early as 2006 (although the absence of pre-2005 makes it impossible to determined if the slide may have started to occur earlier than that). Thus, even when you dig deeper into the Freedom House data, the early deterioration in key components of Turkey’s democracy during AKP’s rule is quite apparent.

In summary, I find the V-Dem data a welcome contribution to an otherwise established group of datasets where there is much room for improvement, in general and for Turkey in particular. The reader should feel free to develop own preferences over which of the different series gives the more illuminating picture of democracy in Turkey over time.