Feeding Britain after Brexit


  • Author: Chris Smaje
  • Date: 27 Feb 2017
  • Copyright: Image appears courtesy of Getty Images

Various origins have been proposed for the word ‘statistics’ but they all invoke the concept of the political state. Statistics, originally, was what states did in order to inform their rule – perhaps the most obvious pioneering example in England being the Domesday Book compiled by the country’s rulers in 1086 in order to get a handle on their new dominions.

Statistics may have come a long way since then, but recent events in global politics, an associated questioning of ‘expert’ knowledge in statistics and economics, and the rise of digital ‘big data’ has brought the relationship between statistical knowledge and the political state back to centre stage, as has been argued in a thought-provoking article by William Davies (1). In the UK, the decision to leave the European Union has raised a plethora of questions about how the country will make its way in the world economically and in every other sphere. The long-dominant economic idea that everybody’s interests are best served by lowering the barriers to global trade was always somewhat vitiated by the machinations of supra-national organisations like the EU or the G8, but has now been thrown into disarray by politicians adopting a much older ‘statistical’ calculus of costs and benefits. This resembles nothing so much as the competitive mercantilism of the early modern European state system – really, war by other means – in which states tried to maximise their exports and minimise their imports in pursuit of greater power.

thumbnail image: Feeding Britain after Brexit

I don’t propose to take a position here on the rights and wrongs of our current politics, something I’ve done elsewhere (2). Instead I’m going to take the neo-mercantilist political turn as a given and look at its implications in my own main area of interest, agriculture. What I propose to do, using our current versions of the Domesday Book in the form of national statistics is to ask whether Britain could, if its government so chose, feed its population entirely from its own indigenous resources – something that it certainly doesn’t do at the moment (3). If nothing else, in demanding a series of assumptions about the future shape of the country and its agriculture, the exercise may at least help to focus attention on some more explicit areas for debate. And if the answer to the question of whether Britain can or wishes to feed itself is no, it prompts sober reflection on the country’s prospects in the emerging neo-mercantilist world.

Modelling future scenarios can involve complex statistical techniques and data manipulation, though ultimately these still involve little more than guesswork concerning the future trajectory of given parameters and their overall relationships in the model. One way of building a modicum of robustness into such a model is by trying to deliberately bias the parameter estimates uniformly in a given direction as appropriate, so that the model then stands as a ‘worst case’ or ‘best case’ scenario establishing a limit for the actual expected range. In the present case, I’d suggest that an appropriate bias would be towards underestimating agricultural productivity in order to establish a lower limit of possibility – after all, farmers are notoriously cautious and when it comes to the feeding of the population it’s probably better to be a bear than a bull.

I’m now going to quickly run through various key parameters in the model. Each one could easily command a whole treatise to themselves – I hope my brief remarks will give the reader enough information to gauge the broad plausibility of the analysis. Full productivity assumptions involve a level of detail that can’t appropriately discussed here.

Population: the Office for National Statistics publishes projections for the UK population as far forward as 2039 – at which point it expects there to be an additional 10.5 million residents to the population measured in the 2011 census, yielding a total population of 74.3 million. About half the difference is projected to arise from natural increase and half from net in-migration (the latter, of course, being a contentious issue in contemporary politics). I propose to accept the ONS 2039 figure as a realistic basis for projecting future food demands.

Crop yield: crop yields have increased considerably in modern times, but these increases have come to some extent through breeding for yield in response to inputs, and continued yield increase may not be achievable in the future. Here I’m assuming static yields, generally based on average crop yields over the past five years.

Crop mix: in keeping with free trade economic ideology, the present focus of British agriculture is largely on maximising income rather than national food self-sufficiency. In view of the relative prices of energy and labour, this results in a concentration on cereal crops for cultivated land and meat from ruminants (cattle and sheep) on grassland. More labour intensive sectors such as fruit and vegetables are mostly import-reliant. Cereal crops produce the highest protein and energy returns per unit area of any crops, but are otherwise nutritionally inferior to fruit and vegetables – they also act as a safety margin for increasing food productivity in accordance with population increase. The relationship between greenhouse gas emissions and land use is complex, but I’d venture to say that ploughing up extra grassland for crops should ideally be avoided. Before the advent of oilseed rape a few decades ago, the only sources of dietary oil and fat in the British climate came from meat and dairy products. Dairying is the most productive form of livestock agriculture in terms of food value per unit area, but it requires good quality grass. Modern dairying also depends on concentrate feeds grown on cropland that could otherwise be used to feed people more efficiently – and the same is true of other meat products such as pork and chicken. But it’s possible to raise lower input/lower output dairy cows on grass alone, and to raise pigs and chickens on food waste. Ruminant meat can be raised on poor quality (mostly upland) grass with few other inputs, but at low levels of productivity and with some environmental disadvantages. Summarising all of that, I propose in the model to retain the existing balance of cropland and grassland, to grow as much fruit and vegetables on the cropland as possible at the expense of cereals, to raise mostly grass-fed dairy cows on the better quality grass and ruminant meat on poor grass, and to raise pigs and chickens on a mix of food waste and fodder crops.

Nutrition: the output variables for a model of national food self-sufficiency need to be nutritional ones. People tend to think in terms of energy or calories as the key measure, but a nutritionally adequate diet involves a wide range of other demands – fat, protein, vitamins and minerals – which aren’t reducible to a summary indicator. Including all possible variables would make for an unwieldy analysis so I’ve chosen four indicator variables – energy, protein, Vitamin A and magnesium – to capture at least part of the required diversity in the diet. Official daily recommendations for an ‘average’ person have been aggregated into an annual demand for the projected population.

Production method: organic agriculture usually involves less energy input than conventional methods, arguably with lower downstream environmental costs, but it also produces lower crop yields. I propose to use data for organic inputs and yields in the analysis. This isn’t necessarily intended to suggest that UK agriculture should be entirely organic in the future, but it does build in a margin for underestimating potential productivity mentioned above. A debate has unfolded about the ‘conventionalization’ of organic agriculture as it has sought to scale up to meet national and international markets, in which higher yields and lower energy use may be sacrificed in organic agriculture to economic ends. Thus, various claims are made for much higher organic yields on ‘best practice’ farms. For comparative purposes, however, these could only be used if set alongside ‘best practice’ conventional farms. Using aggregate data better fits the underestimation bias approach adopted here.

Energy: it seems likely that conventional sources of energy will be less available or higher priced in the future. They’re also associated with high greenhouse gas emissions. Since agriculture uses a relatively small proportion of total UK energy use (less than 1%) and since its business is to capture energy, it makes sense to build into the model a requirement for agriculture to furnish its own energy requirements. This is probably most easily done by making methane from the anaerobic digestion of grass silage. Of course, the grass that goes to make methane can’t be used to feed livestock. So this assumption builds another possible margin of underestimation into the model.

Table 1 shows supply/demand ratios for the four output measures mentioned above – so a figure of less than one suggests that supply would be unable to meet demand (or, at least, need) and a figure above one suggests a surfeit of supply over demand. The figures in the first row show the results for an entirely organic agriculture and horticulture, and suggest that Britain could meet its nutritional needs from organic production across three of the four indicators, the exception being the rather important one of energy. Note that this is referring to a population around 10 million in excess of present levels, however – at present levels, organic production could just about meet existing need.

The second row in Table 1 shows ratios based on conventional/non-organic arable production, with livestock and horticulture remaining organic. In this case, nutritional needs are fully met. It’s debatable how to adjust the various parameters between the organic and non-organic cases – here I’ve decreased the proportion of arable relative to horticulture due to the higher productivity of the conventional arable, and reduced the amount of grass available for livestock to allow for more direct grass-methane production in order to fund the energetic cost of fertiliser production. The rotation of crops projected in this conventional arable system is of a fairly extensive kind, more typical of organic systems.

Table 1: Production/need ratios for UK agriculture

Energy Protein Vitamin A Magnesium
Fully organic 0.84 1.66 2.02 1.41
With conventional arable 1.09 1.99 3.08 1.97

Both scenarios involve a diet that’s very low in fat (about 6% of dietary energy) – essentially because livestock numbers are low relative to present meat consumption as a result of prioritising producing human food for national food self-reliance. The only ways around this are either to increase the size of the area devoted to arable cropping and grow animal fodder on it, or to grow oil crops for direct human consumption, the only feasible one probably being oilseed rape.

Before running the analysis, my expectation was that a conventional-arable rather than an organic-arable system might be necessary in order to meet energetic needs, but that the other nutritional indicators might be worse because the resulting diet would be heavier in less nutritious staples. This hasn’t been borne out by the exercise as it stands – partly because the greater productivity of the conventional-arable system allows relatively more horticultural crops to be grown so that the ratio of horticultural to arable crops consumed in the two cases is much the same, and partly because the staple crops themselves do contain some of the nutrients measured here. A fuller nutritional analysis may reveal a different picture, but – other things being equal – the extra productivity of the conventional-arable crops provides a cushion for other forms of nutrition-boosting husbandry.

It’s a truism of broad-brush data analyses like this that ‘other things’ aren’t, in fact, equal – hence the recurrent cry of the analyst that ‘more research is needed’. There are many upstream and downstream effects of the agricultural choices discussed here which haven’t been accounted for in this simple modelling exercise. One of them that’s pertinent to current times is the relative price of energy and labour in relation to the balance of arable and horticultural production. The high cost of labour in the British economy has prompted a concentration on energy- (and carbon-) intensive arable production over labour-intensive horticulture – the demand for horticultural produce being satisfied through importing from abroad either the produce or low-cost temporary labour. It’s not hard to imagine various scenarios in which the availability or relative price of energy and labour might change in the future – including the possibility that the price of both increases. The downstream effects of fertiliser runoff, soil loss and pesticide residues also need consideration. Exercises like the one I’ve undertaken here can at least be a good starting point for identifying a basic analytical framework for considering agricultural futures – perhaps it’s best if I leave it to the reader to judge its implications if any for the present jockeying between states in the global political system. But there’s a case for more sophisticated modelling incorporating more variables and building in some projections for their future covariance in the light of this contemporary ‘statistics’ of the state. Or, to put it another way, more research is needed.


(1). Davies, W. (2017) ‘How statistics lost their power – and why we should fear what comes next’ https://www.theguardian.com/politics/2017/jan/19/crisis-of-statistics-big-data-democracy

(2). See http://smallfarmfuture.org.uk passim   

(3). http://foodresearch.org.uk/wp-content/uploads/2016/03/Food-and-Brexit-briefing-paper-2.pdf  

Related Topics

Related Publications

Related Content

Site Footer


This website is provided by John Wiley & Sons Limited, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ (Company No: 00641132, VAT No: 376766987)

Published features on StatisticsViews.com are checked for statistical accuracy by a panel from the European Network for Business and Industrial Statistics (ENBIS)   to whom Wiley and StatisticsViews.com express their gratitude. This panel are: Ron Kenett, David Steinberg, Shirley Coleman, Irena Ograjenšek, Fabrizio Ruggeri, Rainer Göb, Philippe Castagliola, Xavier Tort-Martorell, Bart De Ketelaere, Antonio Pievatolo, Martina Vandebroek, Lance Mitchell, Gilbert Saporta, Helmut Waldl and Stelios Psarakis.