On July 15, 2016, a group within the Turkish Armed Forces tried to seize control of the Turkish government. The attempted coup ended a day later when forces loyal to President Recep Tayyip Erdoğan beat back the dissidents.

Predictably, there was an explosion of news articles and commentary about the coup, its implications for Turkey, and how this coup compared to previous coups. However, we noticed something odd in these articles. They were terribly inconsistent. Some news sites proclaimed the coup astonishing while others dismissed it as just another Turkish coup (maybe even a staged one). Some predicted economic doom for Turkey while others said the effect of the coup would be an unnoticeable blip in its economic history. In short, the conversation around Turkey’s coup had confusion in spades.

The quickest way to cut through confusion is to get data. That’s why we decided to analyze data on 475 attempted coups all over the world from 1950 to the present. This post highlights some awesome insights that our analysis threw up, including coup organizers’ uncanny love of certain dates, just how common a failed coup is, and the likelihood of Turkey having another coup in the future (hint: it’s pretty high).

Methodology

We used Jonathan Powell & Clayton Thyne’s “Coup D’etat Dataset” (1950-2016) as the base for our analysis. This data set merges all previous classifications and lists of coups. For more information on Powell and Thyne’s definition of a coup, check out their article.

The Powell/Thyne data set has two important parts — a list of over 2,000 possible coups (1950 to 2010) with supplementary details, and a final list of over 470 global coups (updated in real time). We merged the two data sets to create a master data set with details on every coup from 1950 to 2016. We also pulled historical exchange rate data from the PACIFIC Exchange Rate Service, provided by Professor Werner Antweiler.

Note that the Powell/Thyne data set uses “coups” to refer to both attempted and successful coups. We will do the same thing in this article for consistency.

Coups are fairly common…

Many news outlets and commenters expressed surprise at Turkey’s coup, calling it unexpected or surreal, even considering Turkey’s tumultuous history. Meanwhile, others seemed unimpressed with yet another coup in our increasingly chaotic world.

On the surface, these positions seem mutually exclusive — are coups really that unexpected or not? There is truth in both sides.

Coups are pretty common. Since 1950, there have been 475 coups across 95 countries. That’s 7.2 coups per year on average. Here’s the kicker: At least one coup has occurred every single year since 1950 except 2007. Let that sink in.

The same pattern is true at a country level. For countries with at least 1 coup, the average number of coups per country is 5. Bolivia topped the charts with 23 coups in just 35 years (from 1950 to its last coup in 1984). On average, that means the country couldn’t even get through two years without a coup.

How does Turkey fit into this narrative? TIME claimed that Turkey “has seen more than its fair share of coups”. However, Turkey is not far from average. It has experienced a total of 6 coups (including the coup this month), which is only slightly higher than the average of 5 coups per country.

…but coups are becoming rarer.

With all the terrorist activity and violence in the news, it can seem like volatility and instability are the new normal. However, if coups are any measure of instability, the world has actually become more stable. Over the past few decades, coups have shown a steady decline.

The average number of coups per year dropped from its peak of 12.3 in the 1960s to 2.9 in the years since 2000. What do you think could be the reason for this?

Coups are often unsuccessful.

Some news sites expressed surprise over the failure of Turkey’s attempted coup, insisting that the coup must have been poorly planned (or even a conspiracy). As Aaron Stein summarized, “The narrative following the coup is that this was a small, ill-conceived group of plotters who failed to overthrow the elected government”.

However, it’s fairly normal for a coup to be unsuccessful. Since 1950, 50.3% of all attempted coups have failed. That’s 239 failed coups out of 475 total. Within a given country, only 49.6% of all attempted coups are successful on average.

If anything, the percentage of successful coups has declined in the short term. Fun fact: 2016 is the second year in a row with no successful coups.

Coups happen exactly at the beginning, middle, and end of the month.

Three of the most common days of the month for coups are the 1st, 15th, and 30th. Yes, exactly the beginning, middle, and end of the month. Turkey’s attempted coup fits with this pattern, since it occurred on July 15th. We won’t speculate on why this happens, but perhaps coup instigators secretly want a memorable date?

A coup’s success doesn’t correlate with the number of previous coups.

One popular misconception is that countries with more coups are more unstable, so future coups are more likely to succeed. However, the data shows this link to be false. There appears to be no correlation between the number of coups in a given country and the success rate of those coups.

It shouldn’t be surprising that Turkey had a 6th coup…

Turkey’s 6th coup shocked many, since it came after 36 years of an apparently stable democratic government. However, the data shows that it was reasonable that Turkey, having gone through five coups, would have a sixth coup.

A staggering 86.8% of countries with 5 attempted coups went on to have a 6th coup. Those countries had an average of 4 additional coups (within the timespan of our data set).

What does this chart mean? 48.7% of the countries that had 0 coups went on to have at least 1 coup. 80.0% of the countries that had 1 coup went on to have at least 2 coups. And so on…

The big takeaway from this data — countries that have at least one coup often continue having more coups in the future, at least based on historical coup data. Once a country has that first coup, it’s reasonable to expect that the country will continue having coups. After all, for countries with at least 1 coup, the average number of coups per country is 5.

…and it won’t be surprising if Turkey has a 7th coup.

After July 15, Turkey has now experienced 6 coups. According to our data, 81.8% of countries with 6 coups went on to have a 7th coup. This means that we shouldn’t be surprised if Turkey has a 7th coup in the future.


If anyone knows of a good data set classifying historical coups by their regime type pre- or post-coup, let us know! We’re on the hunt for more data.


We’ve also dug into economic data to predict how Turkey’s failed coup will affect its economy. Check out that blog here.

5 Comments

  1. Gaurav Doshi Reply

    Nice dataset and cool maps and visualisations. Although you can use statistical methods to predict the likelihood of a coup. Your statement, ‘There is no way to predict a coup’s success by the number of previous coups.’ is incorrect as you base this argument on correlation which should NOT be used as causation. Another statement, ‘A staggering 86.8% of countries with 5 attempted coups went on to have a 6th coup.’ is scientifically incorrect method of predicting the likelihood of a coup in a country. The likelihood should be based on economic conditions, foreign relations, political ideologies and the international climate rather than finding % of countries with successive coups.
    ‘We’ve also dug into economic data to predict how Turkey’s failed coup will affect its economy. That analysis will feature in our next blog.’
    One of the very scientific methods developed in economics to study such events is to use the methods of treatment-effects. Developed primarily in drug testing by conducting RCTs, it is being used extensively in arenas of public policy and development economics. You can find huge amount of economic literature on this matter and the ways of applying these methods. I hope you use a better statistical methodology to base your conclusions.

  2. Christine Garcia Reply

    Hey Gaurav, thanks for the detailed comment. Good catch on the “There is no way to predict…” line. That was a couple of poorly phrased sentences on my part. I’ll fix that up shortly.

    As for predicting coups — if you look at those sections of the blog, you’ll notice I didn’t mention the likelihood, prediction, or chances of a coup. That was purposeful because, like you mentioned, this analysis wouldn’t be a good way to analyze the likelihood of a coup. It was just a way of tapping into how coups have occurred in history. The key takeaway from that section was that Turkey having a 6th coup shouldn’t be surprising (as many people said), not about the exact likelihood of this 6th coup.

    • Hi Christine,

      I presume what Gaurav is trying to point out is that statements like “Turkey having a 6th coup shouldn’t be surprising”, do not really have a scientific backing. The statement itself [ and this is perhaps my nitpicking 🙂 ] hints at a probability of another coup.

      Another paradox that I could find in your blog was the significant decrease in coups in recent times. A statement of Turkey (or any country for that matter) having any probability of another coup should possibly take into account the downward trend as well as the economic parameters (that Gaurav mentioned) into consideration.

      Perhaps this is just a nitpick, since I believe if there are any remotely conclusive statements then they should be backed by slightly higher rigour than simple extrapolation.

      The blog was a wonderful read & an eye opener as to what we can do with data. Do keep up the good work. Waiting for the next blog post 🙂

      • Christine Garcia Reply

        Hey Yogesh,

        Good point about accounting for the decrease in coups. That definitely would be interesting to take a look at it.

        I think we’re on the same page that there are better ways to predict the probability of another coup. (And I definitely see your point that some people might read into my statements about the data on subsequent coups.) But that wasn’t quite my goal.

        My goal in writing this blog was to combat some of the confusing or flat-out wrong media rhetoric around Turkey’s coup. I saw so many statements expressing surprise at the coup, acting like it’s unheard of for a country to have 6 coups. That particularly was what I was trying to fight back against.

        I was just trying to show that going from 5 to 6 coups, or from 6 to 7 coups shouldn’t seem that unheard of, and that the media needs to stop spreading that idea. I thought that more people would engage with this idea — and thus the idea could spread farther — if the data analysis behind it was simpler. That’s just my opinion. And believe me, I’m a fan of more complex analysis. I was just afraid that it would make people disengage because this topic is already fairly complex.

        Really appreciate the discussion and feedback! It’s all good to know, and I’ll be careful with the next blog 🙂

  3. Tupeni Baba( tupeni @gmail.com) Reply

    Christine,
    This is an interesting reading especially for those of us who do not have a lot of background on Turkey.
    The one point of interest is Turkeys wish to join the EU and its attempt to improve its Constitution and remove the powers of the military as was provided in the its ‘old’ Constitution which I understand has been revised so as to make Turkey more democratic.This was seen to be to be in line with the requirements of the EU.Specifically ,it got rid of the Immunity Provisions in the old Constitution which gave those that successfully carried out the coups from prosecution.The removal of the immunity provisions meant that the members of the military who carried out earlier coups were required to face persecution and in itself may have been a disincentive for the member’s of the military from moving forward and embracing the new Constitution.
    I wonder if this connection contributed to the motives of the sixth (July 15,2016) Coup?
    All the best
    Tupeni

Write A Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.