Recent news reports have been dominated by stories about the sale of mislabeled processed meat products offered by well-known brands in supermarkets across the UK.
Assurances from the Food Standards agency suggest that no real risks to health have been identified thus far. Of far more concern is the loss of trust from consumers and businesses alike in the integrity of the food supply chain.
But what does that supply chain look like? And what sorts of data are available that allow us to explore the operation of that supply chain?
One approach to take in a data investigation is to come up with a very specific question, and then look for the data to answer it. A more open-ended approach is to try to map out the territory or system of interest, looking to see what data is available, and then trying to find ways of pulling it together.
Having aggregated the data, you can then start to look for patterns, structures, outliers and discontinuities that emerge and use these to frame one or more specific questions that can be applied more to surgically the data.
I'd learned a little about the value chain surrounding processed foods from the OU/BBC series The Foods That Make Billions (for example, the clip Added value in processed food shows how 'value' is added to cheap grain in order to sell it as expensive breakfast cereal); and whilst I was familiar with the notion of food miles (how far foodstuffs travel between the points of production and consumption), I hadn't really thought too deeply about about food supply chains.
So I thought I'd try to survey the landscape: what follows is something of a postcard describing some of the things I found that were new to me...
The news story broke initially with reports about beef burgers tainted with horse DNA. At a first guess, we might imagine that the food supply chain in reverse goes something like: consumer - supermarket - wholesaler - convenience food processor - importer/dealer* - meat processor - importer/dealer* - abattoir - importer/dealer* - farm. (The * indicates there may be several steps in this part, as dealers sell meat on to each other).
A more refined view is given in the figure below, taken from a 2006 working paper by Samarthia Thankappan and Andrew Flynn entitled "Exploring the UK Red Meat Supply Chain".
Another take on the beef supply chain is provided in the Safefood Review of the Beef Food Chain on IOI (the "island of Ireland"). The diagram below includes additional routes into and out of the supply chain, for example by means of imports or exports.
Diagrams like these represent a particular view of a system, in this case views over the meat supply chain. To a data sleuth, each box, each layer of the diagram and each connecting line represents something that might have one or more data sets associated with it. As such, diagrams such as these can give us clues as to what sorts of data sets might be available, or what sorts of agencies might be responsible for collecting it.
I thought I'd start my exploration by seeing what I could learn about ABP Food Group, one of the food processing companies mentioned in the original news reports. A "half-question" I set myself was to see whether the companies associated ABP Food Group fitted neatly into one box in this diagram, or whether they straddled several.
One of the techniques I've started developing to help consider this question resembles the approach I use for mapping out social networks (for example, Communities and Connections: Social Interest Mapping) with a slight twist: rather than looking for people commonly followed by the followers of a particular user or set of users, I look for companies that have have common directors with a company or companies of interest. Here's what I found looking around the ABP Food Group:
One of the things I noticed looking at the companies that appeared to be associated with directors of a couple of the companies identified on the ABP Food Group website was that there were a lot of companies associated with oil, and the recycling cooking oils for the production of biofuels.
Reading around the subject I also learned that animal fats are also used for the production of biodiesel, which I hadn't realised before. The ABP Food Group corporate structure seems to extend outside the meat supply chain, including the processing of vegetable - and possibly animal - waste.
More specifically, ABP Food Group seems to demonstrate how the food supply chain also leads in to the energy supply chain. Recalling another food related story from earlier thuis year - that "between 30% and 50% of the four billion tonnes of food produced around the world each year went to waste" - it makes me think that maybe this food is "wasted" in the food supply chain, but it actually forms a useful part of the energy supply chain? But that's an investigation for another day!
A key concern around the current scandal relates to the provenance of the meat that found its way into processed food products. This is very definitely a data sort of question, but if the data were to be publicly available, where could it be found?
My first thought was DEFRA, the Department for Environment, Food and Rural Affairs. I quickly learned that the transport of live cattle, pigs and sheep and goats is tracked in the England and Wales, and in turn discovered that cattle movements are tracked through British Cattle Movement Service and pig movements through the Electronic Movement Licensing website.
(Sheep tracking appears to be a little bit more cumbersome, revolving around the filing of spreadsheet, and I couldn't find any information about sheep tracking. Horse owners are required to have a horse passport for each of their horses. The Trading Standards guidance on horse passports describes how their role is "to prevent horses entering the human food chain if they have been treated with medicines that must not be administered to food-producing animals". It also notes that "[a]lthough we do not consume horsemeat in any great quantity in the UK, a large number of horses are slaughtered in Britain each year for export for human consumption." Horse passports are also used to regulate the movement of horses across the EU.)
Unfortunately, I couldn't find detailed and publicly available animal movement data anywhere that would allow us to map out how animals actually move. What I could find, though, was data related to how many animals are slaughtered in the UK each month, and via a tweet to the Food Standards Agency (@foodgov), a list of the company names, and the towns they are located in, for registered abattoirs and meat processing establishments.
This sort of data would allow us to start building up our own version of something like the EBLEX food chain map which "plots data about the red meat food chain in the UK, ... show[ing] a range of information, including the location of auction markets, abattoirs, meat processors and meat traders around the UK".
Unfortunately, there is no easy way of tying the two datasets together, so that was another loose end, at least for now. The DEFRA website also points to the EBLEX website for data relating to market prices for animals - the EBLEX - UK Statistics webpage allows you to download spreadsheets relating to the spread of farm to retail prices for beef, lamb, pork and bacon, as well as data relating to UK slaughterings, the monthly volume (in tonnes) of beef imports and exports from EU and non-EU countries, and the monthly retail expenditure (as well as price per kilo) on beef.
However, the license conditions under which the data is made available are not clear. We can presumably make use of it for research purposes though. So for example, if we look back to the supply chain diagram, we could annotate it with the farm to retail price spread data to see get a crude value chain mapping.
These datasets are still fragmented though, dots without numbers that we can't quite yet connect together. Might we have more look at the European level? Again using DEFRA as a starting point, I learned that live animal movements across the EU seem to be tracked using a system called TRACES, although again, no data was available.
Meat and meat products are regulated by import/export licenses, but aside from the overall volume figures on the EBLEX website, I didn't find a source of data relating to how companies actually traded meat. I did though find a way of categorising meat imports and exports by means of HMRC (Her Majesty's Revenue and Customs - aka the VAT man): Trade Tariff - Live animals; animal products.
Whilst not numerical data, these classifications do provide a clue about what sorts of products we might try to find more specific data about, as well as identification codes that describe the categories if we do ever find some data! And so the search goes on.
Using the supply chain diagrams as a guide, as we find more data sets we can add them to the appropriate part of the diagram - adding in the dots - looking for opportunities as we do so to join the dots together or use one dataset to help reveal stories in the context of another. But those are stories for another day...