Over at the Guardian, Damian Carrington has a blog post arguing that “Food is the ultimate security need.”  He bases this argument on a map produced by risk analysts Maplecroft, which sounds quite rigorous:

The Maplecroft index [represented on the map], reviewed last year by the World Food Programme, uses 12 types of data to derive a measure of food risk that is based on the UN FAO’s concept. That covers the availability, access and stability of food supplies, as well as the nutritional and health status of populations.

I’m going to leave aside the question of whether we can or should be linking food security to conflict – Marc Bellemare is covering this issue in his research and has anice short post up that you should be reading.  He also has a link to a longer technical paper where he interrogates this relationship…I am still wading through it, as it involves a somewhat frightening amount of math, but if you are statistically inclined, check it out.

Instead, I would like to quickly raise some questions about this index and the map that results. First, the construction of the index itself is opaque (I assume because it is seen as a proprietary product), so I have no idea what is actually in there.  Given the character of the map, though, it looks like it was constructed from national-level data.  If it was, it is not particularly useful – food insecurity is not only about the amount of food, but access to that food and entitlement to get access to the food, and these are things that tend to be determined locally.  You cannot aggregate entitlement at the national level and get a meaningful understanding of food insecurity – and certainly not actionable information.

Further, you can’t aggregate food markets or prices at the national level and get anything meaningful with regard to food security – let’s compare Maplecroft’s map with FEWS-NETs maps for the immediate future (August-September 2011):

First Maplecroft:

Now FEWS-NET:

While FEWS-NET does not have global coverage, compare their maps to those of Maplecroft and you see two things: One, FEWS is clearly working at a much finer geographic scale, because they have on-the-ground information about actual markets and access, as well as a deep understanding of climate and livelihoods through which to contextualize their grounded data.  This is what it takes to represent variable vulnerability within a country.  The variability you see on their map illustrates my point about the problems of national-level statistics – clearly food insecurity is a regional-to-local problem in every country, even Somalia.  Two, FEWS is not projecting major risk in the same places as Maplecroft, whose map has painted most of equatorial and dryland Africa as problematic at best.  Now, FEWS-NET’s medium-term projections (October-December 2011):

Again, no real resemblance to the Maplecroft map.

Now, you can argue that the Maplecroft map is aimed at a different goal than the FEWS-NET maps, as Maplecroft is trying to create a risk-assessment picture of food security in the region.  However, Maplecroft’s timescale is unclear (does it cover the next 6 months? 1 year? 5 years?), and its data is so over-aggregated as to be non-actionable.  You can’t build policy or programs from this, and I would argue that you can’t really assess the risk of food insecurity from the map or the underlying index either.  FEWS-NET’s maps are what actionable information looks like . . .

I appreciate the point Carrington is trying to make on his blog – food security is a really important issue.  But if we are to address the challenges of hunger and conflict, we need to build our understanding of the connection between them from meaningful data . . . and probably work from the outstanding material already available via FEWS-NET and others.