A new/better way of measuring the fragility of states?

Tolstoy opened Anna Karenina with the much-quoted line ‘Happy families are all alike; every unhappy family is unhappy in its fragile stampown way.’ Does the same apply to states?

The OECD’s new report on Fragile States goes some way down that route. Instead of its past (much criticised) single dimension of ‘fragile/non fragile’, it assesses fragility across five dimensions: Violence, Justice, Institutions, Economic Foundations and Resilience. This is definitely progress – with the new framework, it is clear that Benin — vulnerable in resilience and economics — has different development needs than Cambodia — vulnerable in justice and institutions.

There is of course, a price to pay. Instead of a simple list of fragile states, we have a funky, but more complicated, five pointed star.

SoF Venn Diagram June 03 2015

Only 9 benighted countries tick all five dimensions and end up in the middle of the star:

  • Central African Republic
  • Chad
  • Democratic Republic of Congo
  • Cote d’Ivoire
  • Guinea
  • Haiti
  • Sudan
  • Swaziland
  • Yemen


The others are ranged across different combinations of dimensions.

On the Washington Post Monkey Cage blog, Thomas Scherer promptly jumped on the report, claiming that the OECD had got its numbers wrong and that ‘Only 30 of the 70 listed states are classified correctly.’ The OECD replied, defending the data (they said Scherer had used different datasets and years to those used by the OECD, but acknowledging some mistakes with the graphic – I’ve used their cleaned up one).

As always there’s a tension – the more precise you are, the closer you get to saying ‘every state is fragile in its own way’ a la Tolstoy. That might be a better way to acknowledge that these issues apply to all countries, not just a fairly arbitrary subset, but then it becomes subsumed within general discussions of development, whereas the Fragile States lobbyists want to argue that a subset of countries are so fragile that they need special treatment.

Certainly, simply having a fragile/non-fragile binary is too crude, but predictably the choice of the five dimensions and the indicators provoked challenges. Frauke de Weijer of ECDPM felt that the most important gap was external stresses – arms trade, illicit financial flows, organized crime etc.

According to report author Jolanda Profos ‘We have had a lot of feedback on extra dimensions that we should add in 2016, but less feedback on which of the current dimensions we might drop – the problem with a multi-dimensional model of fragility is where do you stop in terms of numbers of issues? It does seem likely that we will try to add a `social dimension’ as the lack of one has been the most consistent area of feedback.   But there has also been a suggestion that we should look at the peacebuilding goals.’

At least having a discussion about the different dimensions of fragility allows harder thinking about the whole issue of the role of the state, how effective institutions emerge and the role of aid/other external forces. And if some of the projections are true that fragile states are going to be home to the vast majority of the world’s poor in a few years’ time, thinking about states and institutions is likely to form an ever larger part of the development discussion.

Your views?

Subscribe to our Newsletter

You can unsubscribe at any time by clicking the link in the footer of our emails. For information about our privacy practices, please see our .

We use MailChimp as our marketing platform. By subscribing, you acknowledge that your information will be transferred to MailChimp for processing. Learn more about MailChimp's privacy practices here.


10 Responses to “A new/better way of measuring the fragility of states?”
  1. Tom

    Given the projections on poverty, this is an important discussion. A few points:

    1. A movement away from a binary of fragile / non-fragile is welcome. Aside from being politically contentious beyond OECD States (including with emerging donors), it risks encouraging generic responses where context specificity is so crucial and overlooking the risk of conflict in countries not on The List (e.g. Look who’s missing as late as the 2013 fragile States list).

    2. Frauke’s quite right to question the absence of external stresses. But domestic factors obviously matter: we see different impacts of the same transnational factors in different countries. S we still need to better understand how transnational factors interact with domestic factors (e.g. How have illicit flows contributed to conflict in Yemen or South Sudan? What’s different?).

    3. Notions of fragility have tended to be too focused on the weakness of state institutions. Of course they’re important, but it is their type as well as their interaction with society that matters. We’ve tended to equate state building with peace building and, worse for conflict, approached state building in an apolitical and technical way (and, further, worried too much about getting aid money out the door). So a focus on social fragility would be welcome.

    4. The data underpinning the above list (and the choice of indicators, Frauke’s other point) makes a huge difference. We still lack adequate information and have to fall back on blunt proxies. Hopefully the data revoltion and SDG process can help addres this, but we need to remember that the numbers will never tell the full story.

    5. Really welcome in the OECD report is the acceptance of how geopolitical or national security agendas have affected Western engagement in “fragile” states, I.e. which ones get more aid. Moreover, short term Western national security objectives do not always dovetail with people’s security in other countries. Indeed they risk distorting the longer-term processes needed to build sustainable peace. How we can get long-term peace and conflict prevention – which is in everyone’s interest – to trump short-term national security objectives is a huge challenge…


  2. Catherine Dom

    Interesting, but there surely are some strange groupings (e.g. Ethiopia and Somalia in the same group), suggesting that there may also be issues of “level” of fragility in each dimension – which in turn may matter re role of state, aid and other forces? Also – again this is a within-dimension issue, I think forms of and possible/actual responses to institutional, violence and economic foundations fragility would differ a lot between natural resource-rich and other countries (e.g. another strange grouping: South Sudan and Nepal). I guess my overall reaction is that these dimensions (and potentially others) are very useful from an analytical point of view, but when it comes to “what to do”, I’m afraid I’m of the opinion that then each context is unique etc.

  3. Jo

    This is definitely a huge step forward AND every case is still unique! It’s much improved from the first version – but still a bit of a mystery why South Sudan would fall outside the resilience dimension with current levels of food insecurity! But the point in my view is not to put any given country into a specific category – after all, circumstances are constantly changing for the better or worse – but for better analysis and theories of change to underpin any agency response, and coordinated responses, and for better understanding that no one aspect will provide ‘solutions’ within this kind of complexity. I share a concern about the absence of the social dimension. I wouldn’t necessarily see a need to add a peace-building dimension to this – but the social dimension would help get a clearer analysis of ‘root causes’, which I would hope would inform contributions to change processes. Ultimately, though, there will need to be political solutions on many of these dimensions, and how to get that to happen is a whole other debate.

    • Duncan Green

      Thanks Jo, OECD keen to see what kind of responses the new framework gets, and to respond, so hopefully these comments will feed into an even better version for next year

  4. David Carment

    none of this is new

    we introduced the multidimensional measurement of fragility over seven years ago (and used the Tolstoy quote to introduce the concept in our book Security. Development and the Fragile State: Bridging the Gap Between Theory and Policy in 2009) the names and terms were changed a bit by the OECD but the concept and the ideas behind it were introduced by us

    You can access all relevant assessments, rankings and documents at



    rankings 2014


    rankings 2012

    a standard brief combing events data, structural and narrative


    democracy and governance audit

  5. David Carment

    that last link should be

    democracy and governance audit


    here you can play around with the categories – we use 6 clusters instead of OECD’s 5 plus gender as a cross cutting theme plus the overarching categories of A, L and C and the combined fragility ranking

    it doesn’t have spatial layout as the graphic above but the concept is the same


  6. Dominik Schweizer

    Thinking of Swaziland as being the worst of the fragile countries also doesn’t feel right. I spent a couple of days there with a teenage kid – something I couldn’t imagine doing in Somalia, Congo or Yemen. The national cohesion supported by single language and single ethnicity was quite impressive so I suspect that the dimension “Violence” should not apply here.

  7. James Cox

    I am both excited and a little bit worried by the OECD’s proposal. My excitement comes from the idea itself. The idea of multidimensional fragility, though not new (evidenced by both David Carment’s comment above and the New Deal’s Peacebuilding and Statebuilding Goals and fragility spectrum), is expressed here in an engaging and compelling way. The use of clusters of fragility drivers to define the experience of fragility in any country is potentially a liberating one. Rather than be marked with an unwanted ‘fragile state’ label, it’s possible that governments could be more receptive to the more nuanced analysis that the OECD proposal can offer.

    My worry has two aspects. The first, as others have noted, is that the five dimensions chosen by the OECD represent only one possible set. A social dimension has been suggested, and my own take is that I would like to see poverty addressed in some way – possibly underlying the framework as a filter within any of the fragility clusters.

    My second worry is that for all its potential nuance, the framework as presented is another externally generated, top-down view of fragility. The data errors identified as a result of the Washington Post’s review highlight the risks inherent with this. Some countries may get unfairly tarred, while the shortcomings of others go unnoticed. For example, I am always intrigued that these indices commonly fail to highlight the violence against women that is endemic to society in Papua New Guinea and the Solomon Islands. If ‘violence’ is to be more than a proxy for ‘conflict’ then this needs to be dealt with. That’s just one example though. The more important question is ‘how can a model like this be designed to ensure that weight is given to the perspectives of the state and its most vulnerable citizens, and not only to the view from outside?’

  8. Thomas Scherer

    Congratulations Duncan for hosting such a thoughtful discussion about the use of fragility indicators. I am honored that my Washington Post piece was included.

    As an update, I have redone the replication incorporating the OECD’s response and find that many discrepancies remain. I have posted a brief write-up with the updated data work at the following link: http://bit.ly/1BGY0Ts

  9. MJ

    Is this a purely academic exercise or is it intended to drive policy responses (especially, but hopefully not limited to ODA)? Because if, as I assume, it is the latter then it seems to me important to start by asking what elements of fragility affect ODA efficacy and other policy interventions, and then work back from there. Further, as Catherine Dom suggests, this analysis appears to omit any measurement of severity. If you look hard enough just about every country in the world could be identified as fragile in some sense, e.g. ref current centrifugal political forces in UK. So more rigour appears to be needed in determining when fragility becomes overwhelming and thus a central issue to be addressed.

    In short multi-dimensional thinking is good, but woolly-thinking conflation of countries facing challenges of vastly different orders of magnitude is not so helpful.

Leave a Reply to Dominik Schweizer Cancel reply

Your email address will not be published. Required fields are marked *